Breeding Strategies for Sustainable Forage and Turf

Grass Improvement
Susanne Barth • Dan Milbourne
Editors

Breeding Strategies
for Sustainable Forage
and Turf Grass Improvement

2123
Editors
Dr. Susanne Barth Dr. Dan Milbourne
Teagasc Teagasc
Crops Environment and Land Use Crops Environment and Land Use
Programme Programme
Oak Park Research Oak Park Research
Carlow Carlow
Ireland Ireland

ISBN 978-94-007-4554-4 ISBN 978-94-007-4555-1 (eBook)

DOI 10.1007/978-94-007-4555-1
Springer Dordrecht Heidelberg London New York

Library of Congress Control Number: 2012942481

© Springer Science+Business Media Dordrecht 2013

No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by
any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written
permission from the Publisher, with the exception of any material supplied specifically for the purpose of
being entered and executed on a computer system, for exclusive use by the purchaser of the work.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

Preface

From the 4–8th of September 2011, the Eucarpia Fodder Crops and Amenity Grasses
Section held its 29th Meeting in the impressive surroundings of Dublin Castle in
Ireland. Over one hundred and twenty scientists from 21 countries, all working in the
area of the genetics and breeding of forage species, attended the meeting, which was
themed ‘Breeding strategies for sustainable forage and turf grass improvement’.
Why did we choose this theme?
Grasslands cover a significant proportion of the land mass of the world, and play
a pivotal role in global food production. At the same time we are faced with several
challenges that affect the way in which we think about this valuable set of resources.
The population of the world is expected to exceed 9 billion by 2050, and increase of
about one third relative to today’s levels. This population increase will be focused in
urban areas, and in what are currently viewed as “developing” countries, meaning that
the buying power of this increased population will be greater—shifting the balance
of demand from staple crops to high value items such as meat and dairy products.
Overall this means that the world will have to approximately double agricultural
output across all categories of food to meet the demands of this larger, urbanised
population. This is occurring against a backdrop of equally large challenges in terms
of global climate change. Agriculture is already a significant contributor to things
such as greenhouse gas emissions, deforestation and soil erosion. The situation is
made more complex by an increased emphasis on biofuels as a solution for our
imminent oil shortage, resulting in increased competition between land utilised for
food and fuel. In short, agriculture must continue to feed the world, whilst not
contributing to damaging it further. It must be sustainable. Plant breeding plays a
significant but frequently understated role in meeting the challenges presented by
this complex and changing scenario. However, plant breeding and improvement is
itself undergoing radical change, driven by technologies that, quite frankly, seem to
have sprung from the pages of science fiction novels written decades ago.
Thus, it seemed to us, when given the opportunity to organise this meeting, that
it was timely to explore how forage and turf breeding is changing and adapting to
meet these challenges using the technological advances being experienced in plant
breeding as a whole. Consequently, the meeting focused heavily on how next gener-
ation sequencing technologies are interacting with advanced phenotyping strategies
for a variety of increasingly well defined traits. This type of analysis is powerful,
v
vi Preface

potentially telling us a lot about the genetic control of these traits, but also has the
potential to revolutionise plant breeding via approaches such as genomic selection
(GS).
A wonderful characteristic of the membership profile of Eucarpia is that the mem-
bership is composed of a mixture of plant scientists from multiple disciplines and
practical breeders. While some of us wax lyrical about the potential of approaches
such as GS, it’s always useful to have breeders present who can ask pointed ques-
tions about how much this is going to cost them, and how it’s better (i.e. more cost
effective per unit of genetic gain) than what they currently do. This can sometimes
be an uncomfortable experience, but it is through such a frank exchange of ideas that
real progress is made.
As well as the focus on advanced technology, the meeting featured the usual
interesting array of topics that attract the broad audience that attends the section
meetings. Several contribution focused on the use of germplasm of grasses and
legumes to improve the vegetation in different environmental conditions, particularly
under conditions to be expected by climate change—these addressed the theme in a
way in which we hadn’t considered when we discussed it originally (again showing
the advantage in a broad section membership). There were also regular topics such as
the results of the EUCARPIA multi-site rust evaluation, showing that over a period of
11 years there is no evidence that crown rust resistance in individual Lolium cultivars
was overcome by the pathogen), and the Festulolium satellite workshop.
This book contains papers based on many of the oral and poster presentations
presented at the Dublin meeting. With some minor changes to represent the diversity
of material presented, the papers are organised in sections fairly similar to the session
topics, and for the purpose of this volume, are grouped into the following sections:
European grasslands in the future; Breeding strategies; Novel emerging tools for the
breeding of forage and turf crops; Breeding towards breeding objectives; Genetic
variation and adaptation; and Agronomy and performance of forage and turf crops.
We hope they present a good snapshot of a very stimulating meeting, and will be a
useful resource for participants and those who couldn’t attend.
We would like to acknowledge the enormous efforts of the local organising com-
mittee members (Connie Conway, Dermot Forristal, Dermot Grogan, Eleanor Butler,
Patrick Conaghan), with a special mention for Connie Conway and Eleanor Butler,
without whom the meeting would not have run so smoothly and efficiently. Finally,
the work of the scientific committee and referee board for this book (Beat Boller,
Bohumir Cagas, Christian Huyghe, Daniele Rosellini, Danny Thorogood, Dejan
Sokolovic, Dermot Grogan, Dirk Reheul, Jan Nedelnik, Joost Baert, Michael Ab-
berton, Michael Camlin, Niels Roulund, Paolo Annichiarico, Petter Marum, Roland
Kölliker, Trevor Gilliland, Trevor Hodkinson, Ulf Feuerstein and Ulrich Posselt)
must also be acknowledged, especially in providing their time so graciously and
uncomplainingly to review the papers for this volume, and ensuring a high quality
of presentation in these proceedings.

Carlow, Ireland Susanne Barth

Dan Milbourne
Contents

Part I Introduction: European Grasslands in the Future

1 What Global and/or European Agriculture Will Need from Grasslands

and Grassland Breeding over the Next 10–15 Years for a Sustainable
Agriculture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
D. Reheul, B. de Cauwer, M. Cougnon and J. Aper

Part II Breeding Strategies

2 Marker Assisted Selection Made Cheap and Easy . . . . . . . . . . . . . . . . . 21

H. Riday

3 Genome-wide SNP Marker Development and QTL Identification

for Genomic Selection in Red Clover . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
S. Isobe, B. Boller, I. Klimenko, S. Kölliker, J. C. Rana, T. R. Sharma,
K. Shirasawa, H. Hirakawa, S. Sato and S. Tabata

4 Breeding for Resistance to Bacterial Wilt in Ryegrass: Insights into the

Genetic Control of Plant Resistance and Pathogen Virulence . . . . . . . 37
F. Wichmann, F. Widmer, F.-J. Vorhölter, B. Boller and R. Kölliker

5 Mechanisms Utilised Within the IBERS Diploid Lolium perenne

L. Forage Grass Breeding Programmes to Improve Rumen Nitrogen
Use Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
R. C. Hayes, J. A. Lovatt and M. T. Abberton

6 Population Genetics of the Grass Self-incompatibility System—Practical

Implications for Grass Breeding Programmes . . . . . . . . . . . . . . . . . . . . 55
C. Manzanares, B. Studer, R. C. Hayes, S. Barth and D. Thorogood

7 Use of Molecular Marker Information in the Construction of Polycrosses

to Enhance Yield in a Lolium perenne Breeding Programme . . . . . . . . 63
A. Ghesquiere, H. Muylle and J. Baert

vii
viii Contents

8 An Analysis of Chromosome Pairing Behaviour in Newly Synthesized

Alfalfa Tetraploids by Means of SSR Markers . . . . . . . . . . . . . . . . . . . . 69
D. Rosellini, N. Ferradini, S. Allegrucci, A. Nicolia and F. Veronesi

9 Genome Constitution in Selected and Unselected Plants of F2 –F4

Generations Derived from an Allotetraploid Festuca
pratensis × Lolium perenne Hybrid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Z. Zwierzykowski, T. Ksia˛żczyk, M. Taciak, E. Zwierzykowska,
N. Jones and A. Kosmala

10 Estimation of Temporal Allele Frequency Changes in Ryegrass

Populations Selected for Axillary Tiller Development . . . . . . . . . . . . . . 81
G. Brazauskas, I. Pašakinskienė and T. Lübberstedt

11 Understanding the Genetic Basis for Slow Plant-Mediated Proteolysis

in Festulolium Hybrids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
S. A. O’Donovan, A. H. Kingston-Smith and M. W. Humphreys

12 Chromosomal Rearrangements in BC1 Progeny Obtained from Crosses

of Tetraploid Festuca pratensis × Lolium perenne Hybrids
with Tetraploid L. perenne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
T. Ksia˛żczyk, Z. Zwierzykowski and E. Zwierzykowska

Part III Novel Emerging Tools

13 Establishing Chromosome Genomics in Forage and Turf Grasses . . . 105

D. Kopecký, J. Číhalíková, J. Kopecká, J. Vrána, M. Havránková, Š. Stočes,
J. Bartoš, H. Šimková, J. Šafář, M. Kubaláková, P. Navrátil and J. Doležel

14 DArTFest DNA Array—Applications and Perspectives for Grass

Genetics, Genomics and Breeding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
D. Kopecký, J. Bartoš, A. J. Lukaszewski, J. H. Baird, S. R. Sandve,
O. A. Rognli, R. Kölliker, S. L. Byrne, C. Tomaszewski, S. Barth,
A. Kilian, V. Černoch, M. Klíma, P. Azhaguvel, M. Saha and J. Doležel

15 Using DArT Markers in Festuca × Lolium Breeding . . . . . . . . . . . . . . . 121

M. Ghesquière, J.-L. Durand, T. Bourgoin, E. Huttner and A. Kilian

16 Development of an SNP Identification Pipeline for Highly Heterozygous

Crops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
T. Ruttink, L. Sterck, E. Vermeulen, A. Rohde and I. Roldán-Ruiz

17 First Insights into the Mitochondrial Genome of Perennial Ryegrass

(Lolium perenne) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
K. Diekmann, T. R. Hodkinson, K. H. Wolfe and S. Barth
Contents ix

18 Quantifying Early Vigour and Ground Cover using Digital Image

Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
M. Cougnon, J. Verhelst, K. De Dauw and D. Reheul

19 Expression of the Lolium perenne Terminal Flower 1 Gene in Alfalfa

and Tobacco . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
N. Ferradini, A. Nicolia, F. Veronesi and D. Rosellini

20 Morphological and Molecular Characterization of Branching in Red

Clover (Trifolium pratense) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
A. Van Minnebruggen, I. Roldan-Ruiz, J. Van Dingenen, E. Van Bockstaele,
A. Rohde and G. Cnops

Part IV Breeding Towards Breeding Objectives

21 Designing Grass Cultivars for Droughts and Floods . . . . . . . . . . . . . . . 171

M. W. Humphreys, C. J. A. Macleod, W. R. Whalley, L. B. Turner,
M. S. Farrell, M. Ghesquière and P. M. Haygarth

22 Variation and Heritability of α-linolenic Acid Content and Rumen

Escape Protein Fraction in Fodder Grass and Clover . . . . . . . . . . . . . . 181
J. Baert, M. Vandewalle, J. De Riek, J. De Boever, V. Fievez and
C. Van Waes

23 Similarities and Differences in Leaf Proteome Response to Cold

Acclimation Between Festuca pratensis and Lolium perenne . . . . . . . . 189
A. Kosmala, A. Bocian, M. Rapacz, B. Jurczyk, Ł. Marczak and
Z. Zwierzykowski

24 Multi-population QTL Detection for Flowering Time, Stem Elongation

and Quality Traits in Medicago truncatula . . . . . . . . . . . . . . . . . . . . . . . . 197
L. del Carmen Lagunes Espinoza, T. Huguet and B. Julier

25 Role of the RCT1 Gene in Anthracnose Resistance in Alfalfa . . . . . . . . 203

B. Julier, I. Meusnier, L. Alaux, S. Flajoulot, P. Barre and J. Gouzy

26 The EUCARPIA Multi-site Rust Evaluation—Results 2010 . . . . . . . . 209

F. X. Schubiger, J. Baert, T. Ball, B. Cagas, E. Czembor, U. Feuerstein,
A. Gay, S. Hartmann, H. Jakesova, M. Klima, B. Krautzer, H. Leenheer,
C. Persson, W. Pietraszek, L. Poinsard, U. K. Posselt, Y. Quitté, M. Romani,
L. Russi, S. Schulze, M. C. Tardin, M. Van Nes, E. Willner, L. Wolters and
B. Boller

27 The Main Topics of Resistance Breeding in Grasses in the Czech

Republic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
B. Cagaš and M. Svobodová
x Contents

Part V Genetic Variation and Adaptation

28 Origins of Diploid Dactylis from the Canary Islands as Determined

by DNA Sequencing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
A. V. Stewart, N. W. Ellison and V. E. Martín Osorio

29 Introduction and Adaptation of Cynodon L. C. Rich Species

in Australia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
M. C. Jewell, W. F. Anderson, D. S. Loch, I. D. Godwin and C. J.
Lambrides

30 Variation in Traits Associated with Carbon Sequestration for a Range

of Common Amenity Grass Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
S. J. Duller, J. MacDuff, D. Allen and R. Mathews

31 Suitability of Grass Species for Phytoremediation of Soils Polluted

with Heavy-metals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
G. Żurek, M. Pogrzeba, K. Rybka and K. Prokopiuk

32 Targeting Lucerne Cultivars to Saline-soil Environments . . . . . . . . . . . 249

L. Pecetti, P. Annicchiarico, L. De Rosa and S. Proietti

33 Comparison of Seed Mixtures for Technical Revegetation at High

Altitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
P. Spoleto, A. Tosca, G. Della Marianna, F. Gusmeroli, L. Pecetti
and M. Romani

34 Genetic Diversity for Cell Wall Digestibility in a Diverse Lolium perenne

Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
H. Muylle, C. Van Waes, F. Van Parijs, G. Obianugba, J. Baert
and I. Roldán-Ruiz

35 Variability Among Accessions of Forage Vetch for Basic Agronomic and

Morphological Traits under Agro-Ecological Conditions of Serbia. . . 269
Z. Lugić, J. Radović, G. Sokolović, G. Jevtić, T. Vasić and S. Andjelković

36 Genetic Variation of Root Characteristics and Deep Root Production in

Perennial Ryegrass Cultivars Contrasting in Field Persistency . . . . . . 275
D. Sokolovic, S. Babic, J. Radovic, J. Milenkovic, Z. Lugic, S. Andjelkovic
and T. Vasic

37 The Study of Similarities Among Medicago sativa L. Accessions . . . . . 283

D. Knotová, J. Pelikán, T. Vymyslický and S. Raab

38 Genetic Structure and Agronomic Value of Italian Lucerne Landraces:

A Synopsis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
P. Annicchiarico
Contents xi

39 The Use of Genebank Accessions in the Breeding Programme

of Lolium perenne . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
A. Ghesquiere and J. Baert

40 Characterization and Evaluation of Genebank Accessions as a

Pre-selection Instrument for Plant Breeding Objectives and Strategies 301
S. Nehrlich, E. Willner and K. J. Dehmer

41 Exploitation of ‘Site-Specific’Alpine Grass Germplasm for Revegetation

at High Altitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
L. Pecetti, M. Romani, P. Spoleto, A. Tosca, G. Della Marianna
and F. Gusmeroli

Part VI Agronomy/Performance and Compositional Analysis

42 The Impact of Perennial Ryegrass Variety Throughout the Growing

Season on in Vitro Rumen Methane Output . . . . . . . . . . . . . . . . . . . . . . . 315
P. J. Purcell, M. O’ Brien, T. M. Boland, M. McEvoy and P. O’Kiely

43 Origin and Yield of European Perennial Ryegrass (Lolium perenne L.)

Varieties in Ireland . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
D. Grogan

44 Yield Dynamics and Quality in White Clover and Perennial

Ryegrass in the First cut of the Establishment Year . . . . . . . . . . . . . . . . 327
B. Ćupina, A. Mikić, Ð. Krstić, S. Antanasović, P. D’Ottavio and P. Erić

45 Influence of Plant Growth Promoting Rhizobacteria on Alfalfa,

Medicago sativa L.Yield by Inoculation of a Preceding Italian
Ryegrass, Lolium multiflorum Lam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
D. Delić, O. Stajković-Srbinović, D. Kuzmanović, N. Rasulić,
S. Maksimović, J. Radović and A. Simić

46 Optimal Plant Type of Pea for Mixed Cropping with Cereals . . . . . . . 341
P. Annicchiarico, P. Ruda, C. Sulas, M. Pitzalis, M. Salis, M. Romani
and A. M. Carroni

47 Dry Matter Recovery and Aerobic Stability of Maize Whole-Crop, Cob

and Stover Silages—Harvest Date and Cultivar Effects . . . . . . . . . . . . 347
J. P. Lynch, P. O’Kiely and E. M. Doyle

48 Performance of Forage Soya Bean (Glycine max) Cultivars in the

Northern Balkans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
V. Mihailović, A. Mikić, V. Ðordević, B. Ćupina, V. Perić, Ð. Krstić,
M. Srebrić, S. Antanasović and T. E. Devine
xii Contents

49 Effects of Trinexapac-Ethyl (Moddus) on Seed Yields and Its Quality

of Eleven Temperate Grass Species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 359
R. Macháč

50 The Chemical Composition of a Range of Forage Grasses Grown Under

Two Nitrogen Fertiliser Inputs and Harvested at Different Stages of
Maturity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
C. King, J. McEniry, M. Richardson and P. O’Kiely

51 NIRS Calibration Strategies for the Botanical Composition

of Grass-Clover Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
M. Cougnon, C. Van Waes, J. Baert and D. Reheul

52 Comparison of LOCAL and GLOBAL Calibration Models to Predict

Ryegrass Quality Using Near Infrared Reflectance Spectroscopy . . . . 377
G. A. Burns, T. J. Gilliland, D. Grogan and P. O’Kiely

53 Grass for Biogas Production—Anaerobic Methane Production from

Five Common Grassland Species at Sequential Stages of Maturity . . 383
K. O’Riordan, J. McEniry, T. Woodcock, C. King and P. O’Kiely

Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389

 Part I
Introduction: European Grasslands
in the Future
Chapter 1
What Global and/or European Agriculture Will
Need from Grasslands and Grassland Breeding
over the Next 10–15 Years for a Sustainable
Agriculture

D. Reheul, B. de Cauwer, M. Cougnon and J. Aper

Abstract The paper analyses actual trends in (European) ruminant agriculture and
grassland based production systems. Consequences of reduced and/or zero grazing
for grass breeding and grassland management are discussed. The impacts on eco-
efficiency, recycling of minerals and ecosystem services are highlighted as well as
the role of ley-arable farming. Special emphasis is on the potential use of tall fescue
as a component of mixtures or as an interspecific cross. In grazed grassland, the role
of white clover, the disease resistance and the nitrogen use efficiency of the grasses
and the significance of biodiversity are considered. Based on an article published by
Parsons et al. (2011) some reflections on the way ahead in grass and forage breeding
are presented.

1.1 Introduction

At the start of the second decade of the twenty-first century, agriculture is chang-
ing faster than ever in most (European) countries. Attempts to realize some radical
changes in the way we live, confront us with the tremendous complexity of soci-
eties. This results in important gaps between what should happen and what really is
occurring. In theory, sustainable development aims at compromises between socio-
economic and ecological imperatives. The transition from today’s reality to this
new world is a most difficult process passing along several stepping stones (Meer-
burg et al. 2009). It is occurring mostly within existing paradigms improving the
eco-efficiency or eco-productivity (“producing more with less”) of processes and
making them cleaner and more rewarding. Next to this major development new
paths are explored.
Agriculture is changing in line with the major drivers in society. Mainly driven
by European policy, farming has become a very regulated business. To cope with

D. Reheul () · B. de Cauwer · M. Cougnon · J. Aper

Department of Plant Production, Faculty of Bioscience Engineering,
University of Gent, Gent, Belgium
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 3

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_1, © Springer Science+Business Media Dordrecht 2013
4 D. Reheul et al.

this, the most striking development during the past years is the very fast expan-
sion of agricultural enterprises both in terms of means of production, technology
application, management and alliances. While this evolution is going on, scientists
and policy makers already think way ahead of actual evolutions. A striking example
is the newest report of the Standing Committee of Agricultural Research (SCAR)
(Freibauer et al. 2011). The report clearly proposes to move away from the exist-
ing paradigm of productivity and replace it by the paradigm of sufficiency where
consumer-driven, technology-driven and organizational innovation-driven pathways
are building blocks of the transition. The report states “Scientific advance has the
potential to bring forward agro-ecosystems that are both productive, respectful for
ecosystems and resource saving. Demand increases need to be mitigated through be-
havioural changes, the internalization of environmental externalities and appropriate
governance structures”. If one has to reflect on what European agriculture will need
from grassland, one cannot deny this report.
Actual trends form the main frame of this presentation. Grafted upon this main
frame are scientific developments, potential implications of climate change and
personal reflections.
Apart from the mentioned SCAR report, a number of recent high quality pub-
lications inspired the authors, e.g. Towards sustainable grassland and livestock
management (Kemp and Michalk 2007), Genetic improvement of forage species
to reduce environmental impact of temperate livestock grazing systems (Abberton
et al. 2008), Proceedings of the international conference on grasses for the future
(O’Donovan and Hennessy 2010), Handbook of Plant Breeding, fodder crops and
amenity grasses (Boller et al. 2010), Producing milk from grazing to reconcile eco-
nomic and environmental performances (Peyraud et al. 2010) and Past lessons and
future prospects: plant breeding for yield and persistence in cool-temperate pastures
(Parsons et al. 2011). The reader can find a lot of quantified data in these publications.

1.1.1 Very Intensively Used Grassland in the Lowlands

In the (lowland) areas of Europe with an intensive dairy industry, the number of dairy
farms continues to decrease while numbers of cows in surviving farms are increasing
rapidly. Economic scale effects and robot milking (improving the farmer’s comfort)
are important drivers for this evolution. As these drivers most probably will persist,
this evolution is expected to continue. Although grazing may be the cheapest way to
produce milk (O’Donovan and Hennesy 2010), grazing becomes difficult with very
large herds particularly if the land around the milking parlour is restricted. Initially,
the decision to work with large herds often goes along with restricted grazing but
eventually grazing may disappear totally. The higher the numbers of dairy cows on
a farm, the higher the probability that they stay in the barn year-round. In some parts
of the world (e.g. New Zealand, The Netherlands), removable milking parlours may
1 What Global and/or European Agriculture Will Need from Grasslands . . . 5

sustain grazing even when herds become very large. The diet of cows held indoors
is a combination of grass and other forages with conserved forages being far more
important than fresh forage. Although grass remains an important component of the
diet (mainly as a provider of nitrogen and of forage structure, the latter guaranteeing
a good rumen fermentation), other forages (in many cases forage maize) become
the main diet components very often supplemented with a source of concentrated
protein. Hence, these large dairy farms need, next to the grassland area, a lot of
arable land to produce their roughage and to recycle the nutrients in the slurry.
Zero grazing can comply well with a number of sustainability indicators.
1. Harvested dry matter is higher than under grazing. If silage losses are restricted the
benefit remains. Uneven yearly distribution of grass yield becomes less important,
since the animals are mainly fed with conserved forage.
2. A high nitrogen export going along with low (grass-clover) to very low (grass
only) soil nitrate residues (Nevens and Reheul 2003a) makes zero grazing a good
system for an optimal use of slurry. A high N-input combined with early cuttings
(simulated grazing management) provides opportunities to restrict CH4 -emissions
per produced milk quantum (Ellis et al. 2008, Bannink et al. 2010, van Zijderveld
et al. 2011).
3. Nitrogen use efficiency (NUE) on the farm level can be substantially improved (a)
by modern cowsheds and covered storage of slurry, (b) by uniform and emission-
poor distribution of slurry reducing leaching and ammonia losses and (c) by the
composition of animal diets balancing energy from non-grass feed and protein
from the grass, improving the N utilization by the animal. The latter means that the
chemical composition of the grass is less important than under grazing conditions,
since excesses or shortages can be compensated by other forages.
4. Compared to continuous grazing, root depth of the grasses is on average deeper
under a cutting regime offering opportunities for a better nutrient uptake efficiency
and hence a better NUE by the grass plants (Crush et al. 2005; Abberton et al.
2008).
5. In large animal farms, most farmers live closer to their accountancy and their
animals than to their crops. One can presume that grassland management will not
(always) be the first priority of these industries. Mismanagement may deteriorate
the grassland very quickly. On the other hand, farms may pay a lot of attention
to good grassland management in order to cut feed costs.
6. According to the EU legislation on permanent grassland, farmers may avoid to
keep all their grassland longer than 5 years in order not to lose degrees of freedom
in their exploitation. Hence part of the grassland may be kept as temporary grass-
land. If managed under a high nitrogen input, it is difficult for legumes to maintain
important abundances. On the other hand, farmers may cherish the legumes in
order to save N-fertilizer costs.
6 D. Reheul et al.

What Are the Consequences of Reduced Grazing for Grassland Management,

for Grassland Breeding and for the Ecosystem Services of Grassland? Given
the intrinsic higher yield potential of early heading varieties, the attention for early
varieties may increase in zero grazing systems, provided their persistence is high.
According to Chaves et al. (2009) progress in the early varieties of Lolium perenne
L. (perennial ryegrass) was lower than in the intermediate and late heading varieties
offering opportunities for breeding, with a special emphasis on good quality. Good
quality usually is very closely connected with leafiness. Hazard et al. (2006) showed
that selection for longer leaves leads to earlier heading dates, indicating that selection
for good quality may indirectly promote earliness (Barre et al. 2009). The trait “long
leaves” has a high heritability (Cooper and Edwards 1961) and is mainly determined
by leaf elongation rate, easily detectable as quick regrowth.
Since early varieties concentrate their production early in the season, the effect
of summer droughts may be less detrimental than with intermediate or late varieties.
If zero grazing farms choose for temporary grassland, ley-arable farming offers a
number of opportunities and threats (for a review see Vertès et al. 2007) but if well
designed it may fit into a sustainable management.
In the short term, grassland sown into previous arable land significantly outyields
grassland sown into ploughed down grassland (Reheul et al. 2007), particularly under
dry conditions, most probably owing to the deeper rooting of the young grass plants.
The establishment of white clover is better in grassland sown into former arable land
and the clover tends to persist better (Reheul et al. 2007).
The rotation between grass and arable crops helps to manage weeds in the arable
phase of the cropping system.
The opening crop in the arable phase can be grown without any nitrogen fertilizer
(Nevens and Reheul 2002; Nevens and Reheul 2003b, Bommelé 2007; Reheul et al.
2007). Forage maize is an important component of ley-arable farming in large parts
of Europe. In a sustainable system, forage maize is harvested early in the autumn
offering the opportunity for a cover crop as Lolium multiflorum Lam. (Italian rye-
grass) or Secale cereale L. (winter rye) to get established well before the winter. This
way the cover crop prevents winter erosion, nutrient leaching and provides an early
cut in the next spring. The Italian ryegrass may either be ploughed down, enhancing
the soil organic matter or it may produce for the entire season, helping to overcome
risks as the success of forage maize may be jeopardized by dry springs or very wet
autumns. If the maize is harvested late, Lolium multiflorum or even Lolium perenne
may be undersown in the maize crop. Special machines are now available to sow or
drill grasses into a forage maize crop. When this is done before the canopy is closing
(maize height of approx. 40–50 cm), crop damage is minimal. Tetraploid varieties
may offer advantages owing to their early vigour, good cold tolerance and presumed
(has to be proven) deeper rooting.
Climate change is predicted to result in a higher frequency of extreme weather
conditions as hot and dry summers and wetter winters in large parts of Europe. It is
well known that ryegrasses and timothy may suffer from summer (or even spring)
drought with low yields and low quality during the dry spells. Species with a better
1 What Global and/or European Agriculture Will Need from Grasslands . . . 7

drought tolerance as Dactylis glomerata L. or Festuca arundinacea Schreb. (tall

fescue) can overcome poor performances during dry periods. The work of Pontes
et al. (2007) showed that both species can deliver as much as or even more digestible
dry matter and digestible crude protein than perennial ryegrass. Although these para-
meters may not be very relevant under grazing (animals graze -or reject- fresh grass
and not digestible matter) they may be less irrelevant for conserved forage as is the
case when ruminants stay indoors.
Quite a lot of work is currently done on fescue breeding. Mixing Lolium with Fes-
tuca may combine the advantages of both species (excellent forage quality of Lolium
spp. and e.g. good drought resistance of Festuca sp.). The mixing can be done gene-
tically in the form of Festulolium (see Eucarpia workshop of the Festulolium working
group) or mechanically by sowing mixtures of Lolium perenne and/or multiflorum
and Festuca arundinacea. While the abundance of Festulolium in a pure Festulolium
sward is not expected to change dramatically over seasons and years this may be well
the case with mixtures. Preliminary results in Belgian trials (both under grazing and
cutting) do not show important shifts in species composition although the propor-
tion of tall fescue in the harvested material is higher in early spring and during dry
summer periods. A mechanical mixture may result in a transgressive over yielding
driven by pairwise inter-specific interactions as indicated by Kirwan et al. (2007)
who compared during 3 years mixtures with four components (two grass species and
both white and red clover) in different locations across Mid and Northern Europe.
Swards were managed under a cutting regime and dressed with maximum 200 kg
N/ha/year. Preliminary results of our own cutting trials with mixtures of perennial
ryegrass, tall fescue and white clover (dressed with about 160 kg/ha N) do not in-
dicate a transgressive over yielding, probably because both perennial ryegrass and
tall fescue belong to the same group of functional types1 (Kemp and Michalk 2007)
meaning that—according to the redundancy hypothesis—their mutual replacement
has no significant impact on productivity.
The transgressive over yielding may extend into the animal, since the half life
of fescue protein in the rumen is substantially higher than that of ryegrass protein,
enhancing the probability of a better utilization of the protein by the animal (Abberton
et al. 2008).
Different breeding programmes are currently providing new varieties of tall fescue
with long and soft leaves resulting in improved palatability (Rognli et al. 2010). Many
ecotypes have a high lignin concentration in the leaves lowering the digestibility, but
owing to the high genetic variability of the species further progress is expected (De
Santis and Chiaravalle 2001). Selection for a high leaf/stem ratio is a proper way
to improve digestibility and measuring ADF and NDF are the best parameters to
quantify the progress (De Santis and Chiaravalle 2001). In the mean time, the results
of Mosimann et al. (2010), comparing mixtures in which either perennial ryegrass or
tall fescue were the dominant component, indicated a similar digestibility throughout
the year. Since tall fescue leaves have a longer life span than leaves of perennial

1
Grime et al. (1988) described Lolium perenne as a CR/CSR type, while they categorized Festuca
arundinacea as a CSR type (CSR: strategist, CR: ruderal competitor).
8 D. Reheul et al.

ryegrass (1.72 times longer, according to Lemaire et al. 2009), harvesting is quite
flexible.
Tall fescue has a stronger and deeper rooting system than ryegrasses2 (Abberton
et al. 2008, Eickmeyer 2009; Bonos 2004). This results in a better water and nutrient
use efficiency since tall fescue can retrieve water and nutrients from deeper soil
layers. Its ability to protrude compacted soils makes it more resistant to mechanical
soil compaction and allows a better water infiltration (Crush et al. 2005; Macleod
et al. 2007). Simultaneously less nutrients are expected to be leached by heavy winter
rains (Eickmeyer 2009). Compared to ryegrass, the deeper root system of fescue may
stock a higher amount of organic carbon.
Although tall fescue and Festulolium may have promising traits, evidence is
needed to show that these species perform well on the fragile sandy soils, where
much of the intensive dairy is centralized. While tall fescue is growing in roadsides
all over Europe, it is not abundant on sandy soils (were much of the animal produc-
tion in the EU lowlands is concentrated) and on soils with a low pH. This may be
an indication of poor performances/persistence on these soils. There is also a need
to find out what the effects are of the lower digestibility of fescues when they are a
component of a complex diet.
A warming up of the climate brings along new diseases and pests, advances their
outbreaks and/or enhances their frequency (Kiritani 2007; FAO 2008; Ceccarelli et al.
2010). Therefore, breeding for disease (pest) resistance will become more important
than ever, since in a sustainable agriculture the restriction of pesticide application is a
prerequisite. This is particularly true for diseases striking the grass plants during seed
production, since Mattner and Parbary (2007) showed a negative effect of a crown rust
infection of a seed crop of Lolium multiflorum in (the non diseased) post-epidemic
generation: the lower early vigour of the seedlings and poorer performances later on
(registered in pot trials) were mainly due to the smaller seed size of the diseased seed
crop.
A non-grazing management has consequences for the ecosystem quality and
ecosystem services of grassland. According to Reidsma et al. (2006) the ecosystem
quality of a region where grassland occupies a major part of the agricultural area,
can be relatively high, even if the management is very intensive. They calculated a
ecosystem quality of 20 % for intensive pastures as compared to 40 % for exten-
sive pastures, while extensive crop production has 25 % and intensive crop produc-

2
Breeding for a changing pattern of root distribution in Lolium perenne is reported by Crush et al.
(2007). They reported a wide variation in genotypes for patterns of root distribution in a full-sib
mapping population. They found no relationship between N-interception and patterns of distribution
of DM weight of roots. Genotypes reacted on moisture stress either by increased or by inhibited root
growth. Since root growth in artificial circumstances is very variable, hampering a reliable selection,
they expect much of indirect marker-assisted selection of root traits in ryegrasses. This hope seems
justified because of successes in rice (Steele et al. 2006) and maize (Ribaut and Ragot 2007). A
high root/shoot ratio does not automatically reflect a good drought tolerance. In the experiments of
Crush et al. (2005) timothy had a root/shoot ratio of 0.86 versus 0.63 for perennial ryegrass. Yet
timothy is known to have a low drought tolerance.
1 What Global and/or European Agriculture Will Need from Grasslands . . . 9

tion 10 %. Among grassland systems, the species richness is substantially lower in

cut than in grazed grassland (Smith and Rushton 1994).
Although temporary grassland is a better carbon sink than arable land, it stores
about 50 % less carbon than permanent grassland and cut grassland stores about 50 %
less carbon than grazed grassland (Mestdagh et al. 2004; Conijn 2007; Vertès et al.
2007), since a proper cutting management allows less senescent material to return
to the soil.
There seems to be a trade-off between different sustainability indicators (emissons,
carbon balances, ecosystem qualities). As a result it seems impossible to optimize all
productivities, efficiencies and eco-efficiencies as already stated by Jansén (2000).

1.1.2 Grazed Grassland

In important parts of Europe, mainly hilly, mountainous land or land with shallow
soils, grazing still is the best agricultural option for use of the land. This is reflected
in large areas of permanent grazed grassland, with a relatively low frequency of
reseeding. In order to be sustainable, grazing in the EU must comply with envi-
ronmental prescriptions as expressed in the Nitrate Directive (91/676/EEC) and the
Water Framework Directive (2000/60/EC). Hence, pasture management is pushed in
different directions: lower stocking density (where the land is cheap), less external
N inputs, restricted grazing, a combination of grazing and cutting where appropriate
and a strong reliance on biologically fixed nitrogen. Also in these areas the trend of
larger farms is striking and the evolutions toward larger farms is occurring remarkably
fast.
Low external N inputs allows legumes to persist in the grassland. The quantity of
biologically fixed nitrogen (BNF) in the grass-clover herbage can be estimated by
multiplying the white clover DM yield (expressed in ton/ha) in the herbage by 35
(BNF35) and corrected for applied mineral N (kg/ha). The total quantity of biologi-
cally fixed nitrogen in the herbage, BNF = BNF35∗1−(0.282∗N)/100 (Humphreys
et al. 2008). Total white clover BNF (including the non harvested clover DM (stub-
ble, stolons, roots) is estimated by multiplying BNF by 1.27, which brings the total
fixation at approx. 50 kg/ha per ton DM of white clover. The correction factor was
calculated by Hansen (1995), based on the work and data of Nesheim et al. (1990).
The latter applied no more than 80 kg/ha mineral N either as fertilizer N or as cattle
slurry to Swiss swards dominated by perennial ryegrass, meadow fescue and white
clover. Later publications (e.g. Humphreys et al. 2008) use the same correction factor
for much higher mineral N dressings, asssuming that the linear relationship holds
beyond the originally tested low mineral N applications. Anyway, the formula quan-
tifies common knowledge: to take maximum advantage of the biological fixation,
external mineral N-input should be low. There is ample scientific evidence that grass-
clover pastures produce almost as much DM as pastures consisting of pure grasses
dressed with 200–250 kg/ha mineral nitrogen (e.g. Peyreaud et al. 2010) provided
soils are deep and water supply in summer is sufficient. Experience on organic farms,
10 D. Reheul et al.

with no fertilizer N input, demonstrates that such grass-clover swards comply very
well with the environmental regulations.
In these circumstances, grasses with high nutrient use efficiency (NUE) are re-
quested. The trait NUE can be disentangled into a number of physiological more
precise components: NUE can be expressed as the product of the uptake efficiency
(NUptE) with the utilization efficiency (NUtE) (e.g. Gallais and Hirel 2004). NUpt
refers to the efficiency with which roots absorb nitrogen (absorbed versus supplied
nitrogen) while NUtE refers to the quantity of dry matter produced per unit N present
in the dry matter. The latter depends on the retention time and the remobilization pos-
sibilities, again depending on the leaf longevity. Experimental breeding research in
grasses has mainly focused on the NUtE at a given (mostly low) N supply (e.g. Baert
et al. 1999, 2003). The strategy to focus on NUtE gets support from other species,
since Gallais and Hirel (2004) and Schmidt (maize breeder at KWS in Germany; per-
sonal communication) indicated that under low N, NUtE was the driver for a better
NUE in (grain) maize. Despite at lot of research (e.g. the past NIMGRASS EU-
project, in which several EU grass breeders participated) and experimental breeding
work, to our knowledge only few varieties are advertised strongly to have a better N
use efficiency. However, grass breeders always have indirectly selected for a better
NUE, since at a given N supply, the most productive varieties have the best nitrogen
use efficiency. Eventually it is the (N)UE at farm level that is the most important
driver for a production system with low emissions. Hence NUE in the grass plants
should be integrated with NUE in the ruminants. If not, too many nutrients left un-
used by the animals, return to the soil or are lost in the atmosphere. In case the
animals are fed with grass exclusively (or dominantly), the balance between N and
WSC may improve the NUE in the animal and hence at farm level.
From a theoretical point of view genotypes or species with an extensive root
system and a longer life span of leaves offer the best opportunities to improve NUE.
However, genotypes with longer life spans may be more prone to leaf diseases, hence
a good resistance is a prerequisite. The same applies to varieties with long leaves.
Long leaves are advantageous in grazing since they guarantee a high supply of good
quality herbage, the high supply being necessary for a high intake as demonstrated
e.g. by the studies of Delagarde et al. (2001, 2006) and Delagarde and O’Donovan
(2005). If long leaves are the result of a high leaf elongation rate, a quick regrowth
after defoliation offers steady high herbage mass (Barre et al. 2009). As the rate of
development of foliar diseases often is refrained by high N concentrations in the
leaves, a low nitrogen use demands varieties with an excellent resistance to leaf
diseases and this trait may become more important when the climate gets warmer.
Although there is often a negative correlation between leaf length and number of
tillers and a positive correlation between numbers of tillers and persistence, grass
breeders have bred persistent varieties with long leaves.
Extensive root systems are able to restore soil structure in cases of trampling due
to adverse weather conditions. Several studies demonstrate the deeper and stronger
roots of fescues as already stated here-above and the positive effects of the association
of grasses and clovers to guarantee a good soil porosity and water infiltration (e.g.
Van Eekeren 2010).
1 What Global and/or European Agriculture Will Need from Grasslands . . . 11

Peyraud et al. (2010) give an overview of reasons why multi-species swards

under mild fertilization are the way ahead for a sustainable animal production based
on grazed grassland. The reader is referred to follow the MULTISWARD EU-project
(http://www.multisward.eu/multisward_eng/) to get an idea of existing knowledge
and ongoing research.
A recently highlighted function of (grazed) grassland is its value as a sink for
soil organic carbon (SOC). Sonneveld and Van Den Akker (2011) report values of
9–21, 7 kg/m2 in the upper 20 cm of sandy and peat soils respectively in the north of
the Netherlands. When ploughed down, the rate with which the SOC is initially lost
is approximately twice the rate of its accumulation as reported by Johnston (1986).
Indirectly this is a plea for persistent grassland and for breeding of varieties with an
excellent persistence.
There is currently a lot of work going on studying the carbon footprint of ani-
mal production systems. Although today there is no standardized methodology, it is
clear that the carbon footprint heavily depends on the farming system. There seems
to be a link with productivity: in many cases low input systems are also low out-
put systems with a high carbon footprint expressed per kg produce. Eco-efficient
production systems seem to comply best with low carbon footprints. Indeed, ac-
cording to Edwards-Jones et al. (2009), emissions by on-farm activities were by
far dominated by fertilizer-N and concentrates. Not surprisingly, these are also the
most important drivers of N-surpluses. This is again a strong argument for a limited
input of fertilizer-N and concentrates and for grassland systems heavily leaning on
biologically fixed nitrogen.
The relation between carbon sequestration by (grazed) grassland and climate
change is a much studied topic (for a review see e.g. Bartlett et al. 2008 and De Deyn
et al. 2008). Provided there is no water shortage and no shortage of minerals, a rising
temperature and a rising atmospheric CO2 concentration are expected to stimulate
the growth of the C3 forbs and grasses, both above and under the ground. More
roots, more dead material and more root exudations are expected to stimulate the
microbial web in the soil. An enhanced heterotrophic respiration may be responsible
for an initial carbon loss from the system. Enhanced mineralization of recent and
old SOC may provide more nitrogen, stimulating again growth, strengthening the
circle of accumulation and mineralization and carbon fluxes. Eventually the growing
microbial biomass may immobilize N, refraining plant growth and carbon fluxes
to the soil, except when legumes are providing extra N input. So it remains to be
seen if the net result of climate change will increase or decrease the carbon sink
in grassland. Whatever the outcome will be, the room to manipulate here is quite
small, although Dijkstra et al. (2006) and De Deyn et al. (2011) showed that species
richness (with an important role for legumes) continues to be the best guarantee for
carbon sequestration.
Grazed grassland has a high potential for biodiversity both above and under the
ground (Smith and Rushton 1994; Kemp and Michalk 2007; Van Eekeren et al.
2008, 2010 and Van Eekeren 2010). In a number of regions a reasonable yield and an
acceptable biodiversity can go hand in hand, but in quite large parts of Europe, biodi-
verse grassland systems need to be financially supported, e.g. by agro-environmental
12 D. Reheul et al.

schemes. In the absence of this support, these grasslands risk to be quickly aban-
doned, with the loss of a number of ecosystems and ecosystem services in the short
term. It remains to be seen how the economic crisis in the world will influence the
protection of these areas and their ecosystem functions.
To conclude: the semi-intensive grazed grassland of the future will have multifunc-
tional roles (Reheul et al. 2010). The swards are multi-species swards, comprising
persistent grasses and legumes with a dominant role for white clover in temperate
regions: grasses have a long growing season, long leaves, a quick regrowth, good
disease resistances, an extensive rooting system and a high NUE. The grassland and
the animal production system is managed in a way to be as eco-efficient as possible,
by applying best practices and common sense.

1.1.3 Reflections on the Paper of Parsons et al. (2011)

The paper of Parsons et al. (2011) should be compulsory reading for any (grass)
breeder. Based on a thorough analysis of past breeding work, successes and failures,
the paper partly questions if the (experimental) breeding—as it is actually conducted-
is the right way to quickly move forward, given—according to the authors—the
moderate (compared to other crops) breeding advances. The authors propose a more
academic approach of the breeding work based on a quantified definition of breed-
ing goals in clearly defined environments. They suggest focusing on specific traits,
starting from—or referring to—physiological processes in the plant, unraveling how
they are genetically regulated and interact with the environment and they propose
to find out how traits eventually may be locked into varieties. They question the
value of some experimental variety (field) trials—as they are actually designed and
conducted—and would like to focus more on a tiered approach, with an important
emphasis on the “proof of the concept”, i.e. an early testing of trait performance
rather than on variety performance. The latter is deduced from the finding that traits
may be diluted (or eventually lost) during the development of synthetic varieties and
that permanent grazed grassland can be such a complicated plant community with
interactions above and under the ground, that genetic progress may be difficult to
prove in experimental trials or in farm situations with different settings. Indirectly
they ask for more fundamental research.
Essentially they rake up an old dilemma, very nicely defined as two models by
Coors (2006). In model 1 “form follows function”, while in model 2 “function follows
form”. Model 1 means that by selection of phenotypes the breeders’goal eventually is
to change genotypes, while in model 2 one first changes genotypes in order to create
new phenotypes. Model 1 is the model that breeders are applying for over a century
now with proven success. Model 2 results from developments in molecular biology
and genetics. The transition from model 1 to model 2 seems to be happening—as
quoted by Coors (2006)—“by default, without any discussion and challenge”. Put it
in another way: it refers more or less to the confrontation between an academic view
and the view of people working in the real world, between laboratory breeding and
1 What Global and/or European Agriculture Will Need from Grasslands . . . 13

plant breeders who work “with mud and dirt and drought and wind” (quote of Prof.
T. DeJong, tree crop physiologist, UC Davis). Or to conclude with Coors (2006), “at
the end, it is the phenotype that matters”.
Some reflections on the article
1. I have once read a scientific report (but unfortunately have lost it) stating that “there
has not been a single (commercially viable) success booked in plant breeding
programmes that were driven by deliberately creating genotypes with altered
plant physiological characters”. Parsons et al. (2011) show an example of such a
failure (decreasing respiration), but they do try to explain the failure.
2. Tiered approaches are common in risk assessment (e.g. of genetically modified
plants). Some scholars argue that standardized lab tests are necessary to “prove
concepts”, while others are urgently demanding “in planta” experiments. The
reasons for the dispute are analogous to those given by Parsons: the farther away
from standardized ex situ experiments and the closer to the real complex in situ
world, the more difficult it becomes to prove anything. But in the end of the day,
it is the reactions of organisms in the real world that matters.
3. Parsons et al. (2011) emphasize the importance of “fitness of traits”, meaning that
changed traits have to sustain during the process of variety building and in the
complex communities of (grazed) grassland and management settings of animal
farms. They show by smart analysis how varieties with a higher concentration
of water soluble carbohydrates (WSC) do have a positive effect on NUE at low
nitrogen inputs but that the effects fade away at high nitrogen input. Could (F1-)
hybrids—e.g. based on CMS as proposed by Gaue and Baudis (2002)—bring
more stability? At least no loss of traits is expected during the variety construction
and we do know both from maize and cereal hybrids as well as from hybrids
grown for their vegetative parts (sugar, forage beet, a series of vegetables) that
they perform remarkably well in different environments, with in many cases, the
best yield bonuses in rather poor environments. Moreover, the idea of fitness of
traits, fits exactly in the “breeding model 1” as cited above.
4. Questioning if we really can increase the growth of perennial grasses, in particular
by altering their growth strategy, Parsons et al. (2011) “speculate with credible
evidence that our perennial grasses are holding back and not growing to the limits
of their resource supply” since they have to combine good annual growth with the
storage of reserves to guarantee persistence.Yet Chaves et al. (2009) demonstrated
that progress over the last 40 years in dry matter yield and persistence (sic) in
the short living Lolium multiflorum was very comparable to the perennial Lolium
perenne. For crown rust resistance, progress was even better in Lolium perenne
than in Lolium mulitiflorum.
5. As in many cases, the truth probably will lie in the middle. As Coors (2006) says:
we do know that the model 1 is working as proven by more than a century of
breeding; we do not know how successful model 2 will or can be, particularly
for traits regulated by QTL’s. It would be unwise to throw away a century of
experience, as it would be unwise to neglect new developments. However, in
line with the vision of Parsons et al. (2011) it is my opinion that there should be
14 D. Reheul et al.

more clarity and continuity and perseverance into the focus of the application of
new techniques or tools: this will enhance the probability to be successful or will
quickly create clarity on the (non) feasibility and the (non) practicality. I have
seen many projects focusing on molecular techniques passing by. All held bright
initial promises that became less and less brilliant the closer the project came to
an end. Once a project finishes, new projects are proposed with new and often
completely divergent promises and new focuses. There is no clear link between
the series of continuously emerging new techniques (or improvements of their
performance) and achieved results in plant breeding.

1.1.4 Conclusions

Forage grasses are expected to excel in vegetative growth with good forage qualities
during several harvests per year, to persist in these characteristics over many years
and in many different settings and yet to have a good generative growth in order to
produce enough seeds. Unlike a crop as e.g. forage maize, there is no possibility for
compensation between generative and vegetative characteristics, and unlike some
vegetables, grown for their vegetative parts, grasses can not yet take advantage of
heterosis offered by hybrids. No surprise that genetic gain is slower than in many
other crops.
There will always be funded or hyped arguments to select for extra traits. Plants
breeders are well aware that the probability to create excellent all-round varieties
is decreasing, the more traits are involved3 . I do think that it is wise to focus on
the core: producing good forage in an eco-efficient way with the application of
best practices. Furthermore, I do think that there is no need to become nervous
owing to induced hypes and/or alarming messages about dramatic evolutions in food
production and climate change. The current breeding strategies and techniques have
proven to create a steady progress and they should be continued. The introduction
of new breeding tools into the existing programmes applying recurrent selection to
create improved populations—as a base for variety development—may accelerate
the selection response provided they are well focused. Hybridization may change the
whole breeding progress, provided the created heterosis justifies higher seed costs.
As the era of plenty seems to have gone and in line with the recent SCAR-
report (Freibauer et al. 2011), the transdisciplinarity between scientific disciplines
as grass breeding, grassland management, forage use and animal sciences may be
key for speeding up the transition to sustainable grassland based production systems.
Reflecting on adjusted production systems followed by proper actions and applying
best practices in every element of the production chain can make the whole process
more sustainable. As, according to the presumed developments presented in the
SCAR report, among other things, consumer behavior is expected to change (how

3
The number of genotypes in an F2 population equals 3n with n being the number of different loci.
The greater n, the smaller the probability to find the ideal genotype.
1 What Global and/or European Agriculture Will Need from Grasslands . . . 15

difficult this will be?) and environmental externalities are expected (who knows
when?) to be internalized in markets, the demand for animal products may (locally?)
decrease4 and/or their price may rise. It is unclear today how the effect of this
evolution will affect grassland: will animal production with ruminants become even
more concentrated in the very intensive areas and take advantage of both the economy
and ecology of scale, or will we see the opposite?
If there is one particular worrying evolution, much more cumbersome than the
conceptualized rather slow progress in grass breeding, it is the growing shortage of
skilful agronomists, grassland scientists and plant breeders. We are losing a valuable
expertise and a valuable professional genetic diversity which are all sources of vital
creativity. Without these people it will be difficult to achieve any necessary transition
and we will not be able to convince society that some actual hypes drain away a lot
of energy and efforts that would be much more rewarding if they were focused on
the core business instead of circling around it. Science can change the world, but
science has to be honest. I think it is unwise to transform science into an advertising
agency, concentrating on the regular emission of new promises. It is in a series of
old values, methods and perceptions that lay many foundations of sustainability.

References

Abberton MT, Marshall AH, Humphreys MW, Macduff JH, Collins RP, Marley CL (2008) Genetic
improvement of forage species to reduce the environmental impact of temperate livestock grazing
systems. Adv Agron 98:311–355
Baert J, De Vliegher A, Reheul D, Ghesquiere A (1999) Nitrogen use efficiency of grass varieties
at high and low level of applied nitrogen. In: Proceedings of COST 814 workshop on N use
efficiency, CLO, Melle, 2–5 June 1999, 41–50
Baert J, Reheul D, Ghesquiere A (2003) Progress in breeding perennial fodder grasses 4, grass with
a higher nitren use efficiency (NUE). Czech J Genet Plant 39:68–70
BanninkA, Smits MC, Kebreab E, Mills JAN, Ellis JL Klop A, France J, Dijkstra J (2010) Simulating
the effects of grassland management and grass ensiling on methane emission from lactating cows.
J Agric Sci 148:55–72
Barre P, Moreau L, Mi F, Turner L, Gastal F, Julier B, Ghesquière M (2009) Quantitive trait loci
for leaf length in perennial ryegrass (Lolium perenne L.). Grass Forage Sci 64:310–321
Bartlett RD, Freeman C, Ostle N (2008) Microbial contributions to climate change through carbon
cycle feedbacks. ISME J 2:805–814
Bommelé L (2007) Growing potatoes and grass-clover after turned down grassland. Ph.D. Thesis,
University of Gent
Boller B, Posselt U, Veronesi F (eds) (2010) The role of forage crops in multifunctional agriculture.
In: Fodder Crops and Amenity Grasses. Handbook of Plant Breeding, 5. Springer, New York
Bonos SA, Rush D, Highnight K, Meyer WA (2004) Selection for deep root production in tall fescue
and perennial ryegrass. Crop Sci 44:1771–1776

4
In university cities as Gent and Leuven in Belgium (and most probably in other major European
cities), action groups promote one “veggie day” per week as a starter to decline meat consumption
and the consumption of animal proteins. If a large part of the population goes along with this
evolution a substantial decrease in the demand of animal products is expected, with inevitable
consequences for animal production systems and their orientation.
16 D. Reheul et al.

Ceccarelli S et al (2010) Plant breeding and climate changes. J Agric Sci 148:627–637
Chaves B, De Vliegher A, Van Waes J, Carlier L, Marynissen B (2009) Change in agronomic
performance of Lolium perenne and Lolium mulitiflorum varieties in the past 40 years based on
data from the Belgian VCU trials. Plant Breeding 128:680–690
Conijn JG (2007) Grassland resowing and grass-arable crop rotations. Report of the third and
fourth EGF-working group “grassland resowing and grass-arable rotations”. Plant Research
International, Report 148. 141 pp
Cooper JP, Edwards D (1961) The genetic control of leaf development in Lolium. I: assessment of
genetic variation. Heredity 16:63–82
Coors JG (2006) Who are plant breeders, what do they do, and why? In: Lamkey, K.R. and Lee, M
(eds) Plant breeding: the Arnel R. Hallauer International Symposium, Blackwell, 51–60
Crush JR, Easton HS, Waller JE, Hume DE, Faville MJ (2007) Genotypic variation in patterns of
root distribution, nitrate interception and response to moisture stress of a perennial ryegrass
(Lolium perenne L.) mapping population. Grass Forage Sci 62:265–273
Crush JR, Waller JE, Care DA (2005) Root distribution and nitrate interception in eleven temperate
forage species. Grass Forage Sci 60:385–392
De Deyn GB, Shiel RS, Ostle NJ, McNamara NP, Oakley S,Young I, Freeman C, Fenner N, Quirk H,
Bardgett RD (2011) Additional carbon sequestration benefits of grassland diversity restoration.
J Appl Ecol 48:600–608
De Deyn G, Cornelissen JHC, Bardgett RD (2008) Plant functional traits and soil carbon
sequestration in contrasting biomes. Ecol Lett 11:516–531
De Santis G, Chiaravalle E (2001) Heritabilities of nutritive quality factors and interrelationships
with yield in selected progenies of tall fescue. Plant Breeding 120:337–343
Delagarde R, Delaby L, Faverdin P (2006) Le calcul de ration pour vaches laitières au pâturage.
Renc Rech Ruminants 13:89–92
Delagarde R, O’Donovan M (2005) Les modèles de prevision de l’ingestion journalière d’herbe et
de la production laitière des vaches au pasturage. INRA Prod Anim 18:241–253
Delagarde R, Prache S, D’Hour P, Petit M (2001) Ingestion de l’herbe par les ruminants au
pâturage. Fourrages 166:189–212
Dijkstra FA, Hobbie SA, Reich PB (2006) Soil processes affected by sixteen grassland species
grown under different environmental conditions. Soil Sci Soc Am J 70:770–777
Edward-Jones G, Plassmann K, Harris IM (2009) Carbon footprinting of lamb and beef production
systems: insights from an empirical analysis of farms in Wales, UK. J Agric Sci 147:707–719
Eickmeyer F (2009) Klimawandel: Lösungsansätze für Futterpflanzen. Vort Pflanz 81:103–101
Ellis JL, Dijkstra J, Kebreab E, Bannink A, Odongo NE, McBride BW, France J (2008) Aspects of
rumen microbiology central to mechanistic modeling of methane production in cattle. J Agric
Sci 146:213–233
FAO (2008) Climate-related transboundary pests and diseases including relevant aquatic species.
Expert meeting, FAO, February 2008
Freibauer A, Mathijs E, Brunori G, Daminanova Z, Faroult E, Girona IGJ, O’Brien L, Treyer
S (2011) Sustainable food consumption and production in a resource constrained world. SCAR
(Standing Committee of Agricultural Research). http://ec.europa.eu/research/agriculture/scar/
pdf/scar_feg_ultimate_version.pdf. Accessed 5 Sept 2011
Gallais A, Hirel B (2004) An approach to the genetics of nitrogen use efficiency in maize. J Exp
Bot 55: 295–306
Gaue I, Baudis H (2002) European Patent Application, EP20020762392. Male sterility in grasses of
the genus Lolium. http://www.freepatentsonline.com/EP1408735.html. Accessed 5 Sept 2011
Grime JP, Hodgson JG, Hunt R (1988) Comparative plant ecology: a functional approach to common
British species. Unwin Hyman, London
Hansen S (1995) Effects on soil compaction and manure level on utilization of nitrogen in cattle
slurry. In: Samuelsen R, Solscheim B, Pithan K, Watten-Melwaer E (eds). Crop development
for the cool and wet regions of Europe—nitrogen supply and nitrogen fixation of crops for cool
and wet climates. Proceedings of the COST 814 Workshop, Tromsφ, Norway, 135–142
1 What Global and/or European Agriculture Will Need from Grasslands . . . 17

Hazard L, Betin M, Molinari N (2006) Correlated response in plant height and heading date to
selection in perennial ryegrass populations. Agron J 98:1384–1391
Humphreys J, O’Connell K, Casey IA (2008) Nitrogen flows and balances in four grass-land-based
systems of dairy production on a clay-loam soil in a moist temperate climate. Grass Forage Sci
63:467–480
Jansén J (2000) Agriculture, energy and sustainability. Case studies of a local farming community
in Sweden. Doctoral thesis, Swedish University of Agricultural Sciences, Uppsala
Johnston AE (1986) Soil organic matter, effects on soils and crops. Soil use Manag 2(3):97–105
Kemp DR, Michalk DL (2007) Sustainable grassland and livestock management. Review. J Agric
Sci 145:543–564
Kiritani K (2007) The impact of global warming and land-use change on the pest status of rice and
fruit bugs (Heteropthera) in Japan. Glob Change 13:1586–1595
Kirwan I et al (2007) Evenness drives consistent diversity effects in intensive grassland systems
across 28 European countries. J Ecol 95:530–539
Lemaire G, Da Silva SC, Agnusdei M, Wade M, Hodgson J (2009) Interactions between life span and
defoliation frequency in temperate and tropical pastures: a review. Grass Forage Sci 64:341–353
Macleod CJA, Binley A, Hawkins SL, Turner LB, Whalley WR, Haygarth PM (2007) Genetically
modified hydrographs: what can grass genetics do for temperate cathment hydrology? Hydrol
Process 21:2217–2221
Mattner SW, Parbery DG (2007) Crown rust affects plant performance and interference ability of
Italian ryegrass in the post-epidemic generation. Grass Forage Sci 62:437–444
Meerburg BG, Korevaar H, Haubenhofer DK, Nmpù-Zandstra M, Van Keulen H (2009) The
changing role of agriculture in Dutch society. J Agric Sci 147:511–521
Mestdagh I, Lootens P, Carlier L (2004) Variation in organic carbon content in Flemish grassland
soils. Grassl Sci. Eur 9:133–135
Mosimann E, Schmied R, Thuillard CP, Thomet P (2010) Production de viande sur prairies
temporaires: intérêt de la fétuque élevée. Recherche agronomique Suisse 1:194–201
Nesheim L, Boller BC, Lehmann J, Walther U (1990) The effect of nitrogen in cattle slurry and
mineral fertilizers on nitrogen fixation by white clover. Grass Forage Sci 45:91–97
Nevens F, Reheul D (2002) The nitrogen- and non-nitrogen contribution effect of ploughed grass
leys on the following arable crops: determination and optimum use. Eur J Agron 16:57–74
Nevens F, Reheul D (2003a) Effects of cutting or grazing swards on herbage yield, nitrogen uptake
and residual soil nitrate at different levels of N fertilization. Grass Forage Sci 58:431–499
Nevens F, Reheul D (2003b) Permanent grassland and 3-year leys alternating with 3 years of arable
land: 31 years of comparison. Eur J Agron 19:77–90
O’Donovan M, Hennesy D (2010) Grasses for the future. Proceedings of the interna-
tional conference Cork, Ireland, 14–15 Oct 2010. http://www.teagasc.ie/publications/2010/
46/46_GrassesForTheFutureProceedings.pdf
Parsons AJ, Edwards GR, Newton PCD, Chapman DF, Caradus JR, Rasmussen S, Rowarth, JS
(2011) Past lessons and future prospects: plant breeding for yield and persistence in cool-
temperate pastures. Grass Forage Sci 66:153–172
Peyraud JL, Van Den Pol-van Dasselaar A, Dillon P, Delaby L (2010) Producing milk from grazing
to reconcile economic and environmental performances. Grassl Sci Eur 15:865–879
Pontes LS, Carrière P, Andueza D, Louault F, Soussana JF (2007) Seasonal productivity and nutri-
tive value of temperate grasses found in semi-natural pastures in Europe: responses to cutting
frequency and N supply. Grass Forage Sci 62:485–496
Reheul D, De Cauwer B, Cougnon M (2010) The role of forage crops in multifunctional agriculture.
In: Boller B, Posselt U, Veronosi F (eds) Fodder crops and amenity grasses. Handbook of Plant
Breeding 5. Springer, New York, 1–12
Reheul D, Devliegher A, Bommelé L, Carlier L (2007) The comparison between temporary and
permanent grassland. In: Devlieger L and Carlier A (eds) Permanent and temporary grassland:
plant, environment and economy. Grassland Science in Europe 12:1–13
18 D. Reheul et al.

Reidsma P, Tekelenburg T, Van Den berg M, Alkemade R (2006) Impacts of land-use change on
biodiversity: an assessment of agricultural biodiversity in the European Union. Agric Ecosyst
Environ 114:86–102
Ribaut J-M, Ragot M (2007) Marker-assisted selection to improve drought adaptation in maize: the
backcross approach, perspectives, limitations and alternatives. J Exp Bot 58:351–360
Rognli OA, Saha MC, Bhamidimarri S, Van Der Heijden S (2010) Fescues. In: Boller B et al(ed)
Handbook of Plant Breeding, Fodder Crops and Amenity Grasses. Springer, Berlin
Smith RS, Rushton P (1994) The effects of grazing management on the vegetation of Mesotrophic
(meadow) grassland n northern England. J Appl Ecol 31:13–24
Sonneveld MPW, Van Den Akker JJH (2011) Quantification of C and N stocks in grassland topsoils
in a Dutch region dominated by dairy farming. J Agric Sci 149:63–71
Steele KA, Price AA, Shashidhar HE, Withcombe JR (2006) Marker-assisted selection to introgress
rice QTLs controlling root traits into an Indian upland rice variety. Theor Appl Genet 112:208–
221
Van Eekeren N (2010) Grassland management, soil biota and ecosystem services in sandy soils.
Ph.D. Diss, Wageningen University
Van Eekeren N, Bommelé L, Bloem J, Rutgers M, de Goede R, Reheul D, Brussaard L (2008)
Soil biological quality after 36 years of ley-arable cropping, permanent grassland and arable
cropping. Appl Soil Ecol 40:432–446
Van Eekeren N, de Boer H, Hanegraaf MC, Bokhorst D, Nierop J, Bloem T, Schouten R, de Goede
R, Brussaard L (2010) Ecosystem services in grassland associated with biotic and abiotic soil
parameters. Soil Biol Biochem 42:1491–1504
Van Zijderveld SM, Dijkstra J, Perdok HB, Newbold JR, Gerrits WJJ (2011) Dietary inclusion of
diallyl disulfide, yucca powder, calcium fumarate, an extruded linseed product, or medium-chain
fatty acids does not affect methane production in lactating dairy cows. J Dairy Sci 94:3094–3104
Vertès F, Hatch D, Velthof G, Taube F, Laurent F, Loiseau P, Recous S (2007) Short-term and
cumulative effects of grassland cultivation on nitrogen and carbon cycling in ley-arable rotations.
In: Devlieger L and Carlier A (eds) Permanent and temporary grassland: plant, environment and
economy. Grassland Science in Europe 12, 227–246
 Part II
Breeding Strategies
Chapter 2
Marker Assisted Selection Made Cheap and Easy

H. Riday

Abstract Molecular markers in perennial forages are becoming ubiquitous for

marker assisted selection (MAS). Nevertheless, widespread implementation of MAS
in perennial forage breeding programs for highly quantitative traits has not yet oc-
curred. A primary reason is likely the cost associated with genotyping plants using
molecular markers. Presented are cost-effective MAS strategies that are immediately
implementable in most diploid or polyploidy perennial forage breeding programs.
Breeding methods developed during the pre-MAS era maximize the ratio between
additive variance and the square root of the phenotypic variance (i.e. hσA ). With the
advent of molecular markers, many new breeding methods based primarily on cor-
related selection responses between molecular markers and quantitative traits have
been proposed (i.e. rσA ). MAS strategies should be considered using molecular
markers to improve hσA by enhancing traditional breeding methods. A selection
strategy pyramid is envisioned with traditional maternal halfsib selection serving
as the base followed by marker assisted paternity selection. With no additional cost
within maternal and paternal maximum linkage disequilibrium based MAS strategies
can be added. Finally, as the total cost to genotype one individual decreases, residual
linkage disequilibrium based MAS strategies can be added to the above mentioned
strategies or at some price point supplant these strategies.

2.1 Introduction

In most cases forage varieties available today perform better than varieties developed
in the past; much of this increased performance can be attributed to plant breeding.
Plant breeding has been part of the human experience since the dawn of agriculture.
Starting in the twentieth century, modern plant breeding began exploiting the theory
and knowledge of the mode of inheritance to accelerate selection gains. Breeding
methods developed during this era typically seek to maximize the ratio between ad-
ditive variance and the square root of the phenotypic variance (i.e. hσA ). With the
advent of molecular markers, many new breeding methods based primarily on corre-
lated selection responses between molecular markers and quantitative traits have been

H. Riday ()
USDA-ARS, US Dairy Forage Research Center, Madison, WI 53706, USA
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 21

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_2, © Springer Science+Business Media Dordrecht 2013
22 H. Riday

proposed (i.e. rσA ). The advantage of breeding systems built upon correlated selec-
tion responses based on molecular markers are that theoretically selection can occur
in a non-target environment due to elimination of the phenotypic variation (i.e. σP )
from the denominator of the selection gain equation and correlation values potentially
being up to one (if enough molecular markers are used).
So why have marker assisted selection (MAS) schemes as of yet not been widely
incorporated into forage breeding programs, resulting in new varieties? A primary
reason is likely the cost associated with genotyping plants using molecular markers.
Compared to animal breeding, as well as plant breeding systems involving inbred
line development, phenotyping costs in outbred forage species are comparatively
much lower. Therefore, MAS schemes are usually not cost competitive, especially
in cases where many molecular markers are required. Furthermore, due to the slow
pace of genetic gain in highly quantitative traits, such as biomass yield and persis-
tence, it is often difficult to distinguish between elite varieties. This may make it
difficult for seed companies to justify increasing costs to accelerate improvement of
these traits, as it may be more strategic to invest only enough resources in these traits
to remain competitive in the marketplace rather than to excel. Additionally, com-
petitors will almost always be only one selection cycle behind the breeder, making
extra cost recuperation for such core quantitative traits difficult, further depressing
the incentive to excel. For value-added traits in forages, which are often more ex-
pensive to phenotype, MAS schemes become more feasible, especially if premiums
for value-added traits can be obtained during seed sales. However, in markets al-
lowing transgenic traits, it is often more beneficial to pursue transgenic approaches,
which may have more dramatic phenotypic changes and which usually afford bet-
ter intellectual property protection and associated ability to recapture costs through
premiums.
End users of forages still have an immense interest in core quantitative trait im-
provement, such as for biomass yield and persistence. In the interest of end users
of forage varieties, as well as for the greater good of agriculture, efforts should be
made to improve the rate of plant breeding progress. The challenge is to make these
improvements in a cost neutral way (i.e., increased efficiency at the same price) so
that market-driven plant breeders have an incentive to adopt new methods in order
to retain competitive plant breeding programs.
A further challenge and opportunity is the reality of the diseconomy of scale in
plant breeding that is a result of the nature of selection intensity (k) (Fig. 2.1).
There is a disincentive to increase selection fractions beyond 1 in 1000 because
around this selection fraction, exponential increases in selection fractions increase
k-values at sub-linear rates. Therefore, increasing program size is not a cost effective
option; rather, selection gains at lower selection intensities need to be increased.
Traditionally, molecular markers have been incorporated in schemes to increase the
r-values in correlated selection responses in order to eliminate the need for exten-
sive phenotyping. Less attention has been given to the use of molecular markers in
increasing selection efficiency, especially in schemes that would necessitate pheno-
typing. Since forage breeding phenotyping is relatively inexpensive, it makes sense
to explore such schemes.
2 Marker Assisted Selection Made Cheap and Easy 23

Fig. 2.1 Selection intensity (k) value at various selection proportions (log scale)

2.2 Family Based Selection

In outbred forage species, maternal identity of plants is often known. Paternity,

however, is usually unknown. Riday (2011) describes the use of paternity testing
to increase selection gains in red clover, a diploid forage legume species. Since the
completion of this work, a pilot study was implemented to conduct paternity testing
in an autotetraploid alfalfa breeding program. Software was developed and paternal
identity of progeny of a 16-parent polycross were ascertained using an exclusion
analysis and 16 SSR markers (Riday 2012). These two studies establish that pater-
nity testing is now feasible in any forage species, no matter the ploidy configuration.
In most cases, paternity testing should be accomplishable using less than 30 poly-
morphic molecular markers. Increased selection gain is made by increasing parental
control and it contributes to increased efficiency in selection schemes based on khσA .
Riday (2011) describes various ways such information could be incorporated into
a breeding program. He suggests that the risk-reward consideration of using this
scheme is low while confidence in determining paternity with a set of SSR is very
high and genetic gains based on paternal halfsib selection are not affected by any
type of marker-quantitative trait linkage considerations.
An extension of this strategy could be applied to the less informative situation
where potential number and genotypes of parents of a polycross are unknown.
Progeny genotypes from a polycross would be clustered into groups based on mole-
cular markers; these groups would then be defined as pseudo-families. Breeding
values would then be assigned to pseudo-families and selection would be conducted
among the pseudo-families. Computer programs, such as ‘Structure’ (Pritchard et al.
2000), for example, could be used to cluster genotypes into groups, with genotypes
being assigned to each potential cluster at a given probability and multiple runs of
‘Structure’ used to determine the most probable number of clusters. (It should be
24 H. Riday

noted that the instructions for using ‘Structure’ say the program should not be used
on genotypes with known family structure; however since in this case some form
of clustering is desired and “true” population structure is not relevant, this viola-
tion may be acceptable). This procedure was run on 550 progeny of a 19-mother by
96-father polycross. ‘Structure’ determined that 19 clusters were the most probable
number of clusters. There was some correlation between genotype assignment to
the 19 clusters and the actual 19 maternal halfsib families of the genotypes. Between
pseudo-halfsib family variance in this case was approximately 70 % of the actual
known maternal halfsib variance. Due to violations of halfsib family inheritance of
additive genetic variance, it is unclear how much, if any, of the 70 % of the vari-
ance determined among the pseudo-halfsib families would be predictive of progeny
performance in the next generation. If further study shows that some of the between
pseudo-halfsib variance is heritable, this would be another potentially quick and easy
use of molecular markers to increase selection response.

2.3 Correlated Selection Response Based Selection

With known maternal and paternal identity of plants and with molecular marker
information collected to accomplish the paternity test, it becomes a risk free propo-
sition to pursue molecular marker-based correlated selection responses (i.e. even
if no linkage is evident between molecular markers and the quantitative traits the
molecular markers used for paternity testing). It is currently popular to investigate
and describe whole genome selection strategies (Jannink et al. 2010). This is indeed
a good strategy, but becomes riskier at lower molecular-marker-genome coverage
depending on residual linkage disequilibrium between markers and traits. Such link-
age disequilibrium has been reported to be very low in outbred forage species (Isobe
et al. 2009). Presented is a potentially less informative correlated selection response
strategy that can be used at lower molecular marker density. Rather than selecting
on residual linkage disequilibrium (i.e. whole genome selection), maximum linkage
disequilibrium is targeted instead, theoretically allowing molecular markers to be
used at much lower molecular marker density. The traditional problem with maxi-
mum linkage disequilibrium approaches is that trait marker linkage rapidly breaks
down at greater recombination distances, making breeding values assigned to spe-
cific markers unreliable. However, if the correlated selection response is restricted to
distinguishing between the two possible homologues inherited from the same halfsib
parent, targeted selection can be conducted at lower risk. Essentially this is a molec-
ular marker-correlated selection response on within halfsib family additive variance.
The limitation of this method is that some accompanying phenotypic data is nec-
essary to assign breeding values to heterozygous markers associated with common
halfsib parent-derived homologues. Furthermore, the maximum amount of additive
genetic variance that can be captured per halfsib parent is one-quarter (i.e. r1/4σA
for the maternally derived marker and r1/4σA for the paternally derived marker).
2 Marker Assisted Selection Made Cheap and Easy 25

Fig. 2.2 Selection strategy

pyramid with maternal
halfsib family selection at the
base followed by various
molecular marker assisted
selection strategies including:
paternal halfsib selection,
maximum linkage
disequilibrium based within
halfsib family selection,
residual linkage
disequilibrium based
selection, and finally epistatic
conversion

2.4 Combined Methodologies

Halfsib seed-derived maternal family information, molecular marker-derived pater-

nal halfsib information, and maximum linkage disequilibrium based within halfsib
family correlated selection response information can all be combined into a common
strategy using the four orthogonal or independent partitions of the additive genetic
variance. Strategically, the four terms can be viewed as building blocks with mater-
nal identity and accompanying breeding values forming the foundation. Additional
resources can then be used to determine paternal identity and within halfsib family
maximum linkage disequilibrium correlated selection responses (Fig. 2.2).
The advantage of this strategy is that, depending on the availability of breeding
resources and the perceived value of a particular polycross, the resources commitment
can be increased or decreased with some additional selection gains realized at each
increased level of resource commitment.
If a few molecular markers are known a priori to be very close to quantitative trait
loci or specific genes such that residual linkage disequilibrium-correlated selection
response strategies can be pursued (i.e. rσA ), then these could be added to a combined
selection strategy in combination with maternal halfsib selection, paternal halfsib
selection, and within maternal and paternal molecular marker-based halfsib selection.
Based on the cost of each strategy and amount of additive variance described for each
selection strategy, optimization equations could be written to pursue the most efficient
selection strategy on a cost and selection gain basis. Obviously as genotyping costs
decrease and molecular marker densities increase the breeding strategy would move
more and more towards whole genome selection. A critical point to consider is that
26 H. Riday

before committing from the start to a potentially high cost whole genome selection
strategy one should consider that there are many molecular marker assisted selection
strategies in conjunction with traditional family based selection strategies that can be
used in stead of or in conjunction with whole genome selection strategies that may
be more cost effective or strategic to achieve improved selection gains (Fig. 2.2).

2.5 Conclusions

A final caution needs to be discussed in relation to the use of molecular marker as-
sisted breeding strategies. Molecular markers used to describe family structure or
used to identify correlations with traits under selection are always descriptive of
genotypic potential; molecular markers never determine a plant’s genotypic poten-
tial. A plant’s genotypic potential is always determined during the random genome
reshuffling of parental genomes during reproduction (unless transgenic approaches
are taken). This means that the perfect genotype will only be observed rarely, ir-
respective of how many molecular markers are used to identify it (i.e., even if the
perfect theoretical genotype is known a priori, it still has to be created via random
mating and physically identified from among all products of random mating). The
core process of extracting DNA from a genotype is not drastically decreasing in cost.
Therefore, from a practical perspective, forage breeders will still be evaluating a lim-
ited number of selection units in their programs (i.e. 100–100,000). It is therefore
unnecessary to have an unlimited number of molecular markers for selection among
a smaller number of genotypes since only a certain amount of molecular marker
information will be necessary to rank selection units, with less information needed
for accurate ranking among fewer genotypes.
It should be apparent that in the combined molecular marker strategy described
above, the greatest amount of information per molecular marker will be obtained from
an initial set of markers. However, as the number of markers utilized is increased,
the amount of information per marker at some point will decrease. The number
of markers utilized in a selection strategy should be optimized so the maximum
amount of resources can be applied to increase the number of selection units under
evaluation. Even under the perfect correlated selection response scheme (i.e. kσA )
selection intensity and the amount of additive variance present are still paramount
drivers of selection gain.

References

Isobe S, Kölliker R, Hisano H, Sasamoto S, Wada T, Klimenko I, Okumura K, Tabata S (2009)

Construction of a consensus linkage map for red clover (Trifolium pratense L.). BMC Plant Bio
9:57
Jannink J, Lorenz AJ, Iwata H (2010) Genomic selection in plant breeding: from theory to practice.
Briefings Funct Genomics Proteomics 9:166–177
2 Marker Assisted Selection Made Cheap and Easy 27

Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus
genotype data. Genetics 155:945–959
Riday H (2011) Paternity testing a non-linkage based marker assisted selection scheme for outbred
forage species. Crop Sci 51:631–641
Chapter 3
Genome-wide SNP Marker Development
and QTL Identification for Genomic
Selection in Red Clover

S. Isobe, B. Boller, I. Klimenko, S. Kölliker, J. C. Rana, T. R. Sharma,

K. Shirasawa, H. Hirakawa, S. Sato and S. Tabata

Abstract Genomic selection (GS) has experienced remarkable advances in genome

technologies over the past few years. However, employing GS for forage breeding
has been considered difficult because forage species generally show short linkage
disequilibrium (LD) across the genome. To elongate the LD, an Advanced Intercross
Line (AIL) population was generated from crosses between six individuals origi-
nating from Switzerland, Russia and Japan. The suitability of this population was
demonstrated for GS or association analysis. For high throughput genotyping, single
nucleotide polymorphism (SNP) markers were developed by comparing transcrip-
tome sequences obtained from two red clover individuals. An Illumina Golden Gate
platform for 1,536 candidate SNPs was used for polymorphic analysis in the AIL
population. A total of 784 SNP markers were identified as polymorphic. In addition,
75 polymorphic SSR markers were used for genotyping the AIL population. Appr-
oximately 200 plants each were established in 2010 in the fields of Palampur, Moscow
region, Zurich and Chiba. Seed yields, flowering characteristics and morphological
traits were evaluated in each region. Significant QTLs and QTL interactions were
identified for the traits investigated by GMM analysis. The results suggest that the
AIL population can be used for GS and association analysis.

S. Isobe () · K. Shirasawa · H. Hirakawa · S. Sato · S. Tabata

Kazusa DNA Research Institute, 2-6-7 KazusaKamatari, Kisarazu,
Chiba, 292-0818, Japan
e-mail: [email protected]
B. Boller · S Kölliker
Agroscope FAL Reckenholz, Swiss Federal Research Station for Agroecology
and Agriculture, Reckenholzstrasse 191, 8046 Zurich, Switzerland
I. Klimenko
All-Russian Williams Fodder Crop Research Institute, 141055 Lugovaya,
Moscow Region, Russia
J. C. Rana
National Bureau of Plant Genetic Resources, Regional Station, Phagli,
Shimla 171004, Himachal Pradesh, India
T. R. Sharma
CSK Himachal Pradesh Agricultural University, Palampur 176062,
Himachal Pradesh, India

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 29

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_3, © Springer Science+Business Media Dordrecht 2013
30 S. Isobe et al.

3.1 Introduction

Genomic selection (GS) has experienced remarkable advances in genome technolo-

gies during the past few years. Because of its ability to identify all significant QTLs
(Quantitative Trait Loci), including those having small effects, as well as not re-
quiring phenotype validation of breeding populations, GS was expected to become
a major technique for molecular breeding (Heffner et al. 2009). Whenever GS is
performed, it is essential to consider the relationship between the genetic distance of
the markers and the LD (linkage disequilibrium). This is because the exact breeding
values are estimated by using DNA markers that have a strong LD with the targeted
genes (Meuwissen et al. 2001). The genetic distance showing a significant LD de-
pends upon the targeted species and the type of population. Employing GS for forage
breeding has been considered difficult because forage species generally show short
LD across the genome.
Establishing populations with family relations is nearly the only approach to
elongate the genetic distance showing significant LD. Darvashi and Spoller (1995)
proposed using AIL (Advanced Intercrossed Line) populations for more accurate
estimations with QTL mapping. AIL populations are produced by randomly and
sequentially intercrossing a population that initially originated from a cross between
two inbred lines or some variant thereof. Sequential intercrossing is often employed
in forage breeding. Therefore, AIL populations may be appropriate for the association
mapping and GS of forage species.
In this study, an AIL population was established from crosses between six in-
dividuals originating from Switzerland, Russia and Japan, and this AIL population
was demonstrated to be suitable for association analysis and future GS. In addition,
for high throughput genotyping, single nucleotide polymorphism (SNP) markers
were developed to establish a high throughput genotyping system in red clover.

3.2 Materials and Methods

3.2.1 Plant Materials

Three F1 populations were generated from crosses between ‘HR × R130’,

‘M137 × Violetta6’ and ‘M366 × M372’. The former two crosses were reported as
parental crosses for linkage map construction (Sato et al. 2005, Herrmann et al.
2008). ‘M366’ and ‘M372’ were derived from varieties bred in Switzerland. Ten F1
progeny of each of the three crosses were intercrossed in a field in Sapporo (National
Agricultural Research Center for Hokkaido Region) and Zurich. The produced seeds
from both countries were used for a second intercross in a greenhouse at Kazusa
DNA Research Institute (KDRI). The second polycrossed seeds were designated the
‘RC-AIL’ population and were used as the mapping population for this study.
3 Genome-wide SNP Marker Development and QTL Identification for Genomic . . . 31

3.2.2 Phenotype Investigation

Approximately 200 seeds of RC-AIL were sown in Chiba, Zurich, Shimla and the
Moscow Region in the autumn of 2009. The seedlings were transplanted in the
field with spaced planting. Morphological traits, flowering time and seed yield were
analysed in 2010 according to UPOV guidelines.

3.2.3 SNP Discovery

Transcriptome sequencing of the leaves of two red clover individuals, ‘HR’ and
‘R130’, were performed using a Roche 454 GS-FLX Titanium according to the stan-
dard protocol provided by Roche Diagnostics. The transcript sequences obtained in
this study and the 26,394 ESTs (Expressed Sequence Tags) of red clover published
previously (Sato et al. 2005) were assembled using the MIRA 3.0.5 program. Can-
didate SNPs were identified as base mismatches in the alignment of the transcript
sequences. Of the candidate SNPs, 1,536 SNPs were screened based on QV (Quality
value) scores and alignment coverage.

3.2.4 Genotype Investigation and Data Analysis

DNA of each plant was isolated from fresh or dried leaves. SNP marker genotyping
for each of the 65 plants grown in the four countries was performed using the Illu-
mina Golden Gate assay according to the provided protocol. In addition to SNP
markers, a total of 75 SSR (single sequence repeat) markers, mapped on the red
clover consensus linkage map (Isobe et al. 2009) at intervals of 10 cM, were used for
genotype analysis using a fragment analyser (ABI3730, Applied Biosystems). GGT
2.0 software was employed to determine the LD between two markers (van Berloo
2008). QTL identification was performed using GMM (Genotype Matrix Mapping)
version 2.1 (Isobe et al. 2007).

3.3 Results and Discussion

3.3.1 Phenotype Investigation

The RC-AIL population grew well in Chiba, Zurich and Shimla. However, plants
suffered cold temperatures in the Moscow Region, and 108 of the 200 plants died
after the first winter. A total of 28 traits were investigated in the four regions
(Table 3.1). Of the traits, three traits were investigated in all four regions, while seven,
32 S. Isobe et al.

Table 3.1 List of the investigated traits in four different regions

Traits Number of inves- Traits Number of inves-
tigated regions tigated regions
Flowering date 4 Intensity of green colour 1
Seed yield 4 Powdery mildew resistance 1
Plant length 4 Dry plant weight 1
Number of flowerheads 3 Colour of stems 1
Number of stems 3 Hairness 1
Number of internodes 3 Colour of flowers 1
Stem thickness 3 Genestin 1
Length of leaflets 3 3-Hydroxy Biochanin A 1
Width of leaflets 3 Formo Numbernetin 1
Colour of stipules 3 Biochanin A 1
Number of florets 2 Seedling vigour 1
Plant height 2 Vigour after winter 1
Leaf mark 2 Vigour in august 1
Growth habit 2 Vigour in June 1

four and 14 traits were evaluated in three, two and one region, respectively. Large
variations were observed in each investigated trait. Examples of trait distributions
are shown in Fig. 3.1.

3.3.2 SNP Discovery and Polymorphic Analysis of the RC-AIL

Population

The number of transcript sequences read by Roche 454 pyrosequencing were 499,241
and 447,259 in R130 and HR, respectively. Along with the 26,394 published ESTs,
the transcript sequences were assembled into 78,835 contigs and 535,128 singlets
using MIRA 3.0.5. The total length of the consensus sequences was 45,990,294 bp.
The average, longest and N50 sequence were 583, 7,865 and 564 bp, respectively.
By comparing the base mismatches of the aligned transcript sequences, a total of
129,019 candidate SNPs were identified between HR and R130. Of the candidate
SNPs, those with high quality were screened based on the following criteria: QV
of the consensus sequence > 50, QV of the candidate SNPs with two or greater base
types > 30, absence of SNPs on 60 bp up and downstream of the candidate SNP.
As a result, a total of 2,227 candidate SNPs were screened. Of the screened SNPs,
1,536 were validated as SNP markers in the RC-AIL population and the F1 progeny
of HR × R130. The polymorphic ratios of the tested markers were approximately the
same in the two populations: 51.0 % (784/1,536) in RC-AIL and 51.4 % (789/1,536)
in HR × R130. Because the number of overlapped polymorphic markers was 412, a
total of 1,161 SNP markers confirmed their availability. The majority of SNP markers
that showed polymorphisms were mapped onto a red clover linkage map (data not
shown).
3 Genome-wide SNP Marker Development and QTL Identification for Genomic . . . 33

Fig. 3.1 Distribution of the investigated traits in Chiba

Meanwhile, the 75 SSR markers used for polymorphic analysis identified one to
12 alleles per marker. The mean number of alleles identified for a marker was 6.5.
The genotype data derived from the 75 SSR and 784 SNP markers was used for the
LD investigation and QTL identification. The average genetic distance between two
markers was 1.01 cM.

3.3.3 Relationship Between the LD and the Genetic Distance

The relationship between the LD (D ) and the genetic distances of three different
types of red clover were investigated. The unrelated population consisted of 1,156
plants derived from 24 varieties. Details of the unrelated population were described
by Isobe et al. (2009). Linear regression was observed in a F1 mapping population
(r = −0.929) while no significant relationship was observed in an unrelated popula-
tion (r = −0.039). The intensity of the relationship between the LD and the genetic
distance in the RC-AIL population was in between that of the F1 and unrelated pop-
ulations (r = −0.24). This result suggested the possibility of QTL identification in
the RC-AIL mapping population using 1 cM interval markers (Fig. 3.2).
34 S. Isobe et al.

Fig. 3.2 Relationship between the LD (D ) and the genetic distance of two marker loci

Fig. 3.3 Significant QTLs and QTL interactions for plant length in the RC-AIL population inves-
tigated in the four regions. Large circles indicate the linkage map of red clover and small circles
show the common QTLs across four regions. Letters outside of the large circles represent marker
name identified QTLs and lines inside the large circles show significant QTL interactions under
three QTL combinations. Bar graphs indicate the distribution of phenotype; red and blue bars show
the distribution of selected QTL-allele combinations and others, respectively

3.3.4 QTL Identification

QTL identification was performed for plant length and flowering date (Fig. 3.3 and
3.4). Significant QTLs and QTL interactions were identified for the two traits. QTLs
commonly identified across the four regions were for plant length and flowering
date. However, the interactions were often region-specific. With regard to plant
length, three common QTLs were identified on LG2, LG4 and LG7. The phenotypic
3 Genome-wide SNP Marker Development and QTL Identification for Genomic . . . 35

Fig. 3.4 Significant QTLs and QTL interactions for flowering date in the RC-AIL population
investigated in the four regions. The details of this figure are the same as that of Fig. 3.3

variance explained percentage of the identified QTL interactions (three QTL combi-
nations) varied from 30–37 %. Most of the significant QTLs showed negative effects,
while interactions between LG2 and LG7 showed positive effects. With regard to
flowering time, four common QTLs were identified on LG1, LG2, LG6, and LG7.
The phenotypic variance explained percentage of the identified QTL interactions
(three QTL combinations) was estimated between 40–42 % in Shimla, Zurich and
Chiba, while the maximum value in the Moscow region was 51 %. The QTL inter-
actions identified in Shimla and Chiba showed positive effects, and those identified
in Zurich and the Moscow region showed negative effects. The large magnitude of
the percentage of phenotypic variance suggests that the RC-AIL mapping population
is appropriate for genome wide association analysis and GS with fewer than 1,000
DNA markers.

3.4 Conclusion

A total of 1,161 polymorphic SNP markers were developed in this study. The markers
will be useful tools for genetic analysis and molecular breeding in red clover. The
large phenotype variation of the RC-AIL population suggests its efficiency as a
mapping population for association analysis and as a training population for GS.
QTL identification was performed for plant height and flowering date, and signifi-
36 S. Isobe et al.

cant QTLs were identified. Of the identified QTLs, common QTLs expressed in four
different regions were observed. However, significant QTL interactions were rather
region-specific. These results suggest a possible estimation of the breeding value in
GS with genotype data of 1-cM intervals using an AIL population.

References

Darvasi A and Soller M (1995) Advanced intercross lines, an experimental population for fine
genetic population. Genetics 141:1199–1205
Heffner EL, Sorrelles ME, Jannink JL (2009) Genomic selection for crop improvement. Crop Sci
49:1–12
Herrmann D, Boller B, Studer B, Widmer F, Kölliker R (2008) Improving persistence in red clover:
Insights from QTL analysis and comparative phenotypic evaluation. Crop Sci 48:269–277
Isobe S, Kölliker R, Hisano H, Sasamoto S, Wada T, Klimenko I, Okumura K, Tabata S (2009)
Construction of a consensus linkage map for red clover (Trifolium pratense L.). BMC Plant Biol
9:57
Isobe S, Nakaya A, Tabata S (2007) Genotype matrix mapping: searching for quantitative trait loci
interactions in genetic variation in complex traits. DNA Res 14(5):217–225
Meuwissen THE, Hayes BJ and Goddard ME (2001) Prediction of total genetic value using
genome-wide dense marker maps. Genetics 157:1819–1829
Sato S, Isobe S, Asamizu E, Ohmido N, Kataoka R, Nakamura Y, Kaneko T, Sakurai N, Okumura
K, Klimenko I, Sasamoto S, Wada T, Watanabe A, Kohara M, Fujishiro T, Tabata S (2005)
Comprehensive structural analysis of the genome of red clover (Trifolium pratense L.). DNA
Res 12:301–364
Van Berloo R (2008) GGT 2.0: Versatile software for visualization and analysis of genetic Data. J
Hered 99:232–236
Chapter 4
Breeding for Resistance to Bacterial Wilt
in Ryegrass: Insights into the Genetic Control
of Plant Resistance and Pathogen Virulence

F. Wichmann, F. Widmer, F.-J. Vorhölter, B. Boller and R. Kölliker

Abstract Bacterial wilt caused by Xanthomonas translucens pv. graminis (Xtg) is a

severe disease of forage grasses leading to drastic losses in pure and mixed stands.
Italian ryegrass (Lolium multiflorum) is particularly susceptible to bacterial wilt and
breeding for resistance is the only practicable means of disease control. A detailed
understanding of the genetic control of this complex host-pathogen interaction is
indispensible for the further development of L. multiflorum cultivars with increased
resistance to bacterial wilt and to refine and optimise breeding procedures. While
several recent studies have revealed novel insights on plant resistance, little is known
about the processes involved in host-colonization and disease development. There-
fore, factors influencing pathogen virulence were investigated in Xtg using conserved
primer approaches and whole genome sequencing. Knock-out mutation of compo-
nents of the type three secretion system (T3SS), a major virulence factor in many
Xanthomonas spp., showed that the T3SS is important for Xtg virulence but not for in
planta multiplication. Analysis of the draft genome sequences revealed a substantial
number of effectors apparently characteristic for Xtg. In conlusion, our investiga-
tions on the Xtg x L. multiflorum interactions provide fundamental insights for the
development of innovative resistance breeding approaches.

4.1 Introduction

Bacterial wilt caused by Xanthomonas translucens pathovars is estimated to account

for annual forage yield losses of up to 20 % in swards (Schmidt and Nuesch 1980),
but yield losses of up to 80 % have been reported from experiments involving arti-
ficial leaf inoculation (Wang and Sletten 1995). Of the different pathovars infecting
forage grasses, X. t. pv. graminis (Xtg) is the most abundant (Egli and Schmidt 1982)
and has a relatively broad host range including several major forage grass species
such as Lolium multiflorum, L. perenne, Festuca pratensis, Phleum pratense and

R. Kölliker () · F. Wichmann · F. Widmer · B. Boller

Agroscope Reckenholz-Tänikon Research Station ART, 8046 Zurich, Switzerland
e-mail: [email protected]
F.-J. Vorhölter
CeBiTec, University of Bielefeld, Bioinformatics Facility, Bielefeld, Germany

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 37

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_4, © Springer Science+Business Media Dordrecht 2013
38 F. Wichmann et al.

Poa pratensis. Italian ryegrass (L. multiflorum) is particularly susceptible to bac-

terial wilt caused by Xtg. Breeding for resistance to bacterial wilt in ryegrasses is
usually based on artificial seedling inoculation and subsequent recurrent phenotypic
selection. Although some progress has been achieved (Boller et al. 2009), further
advances in breeding programs have often stagnated after several cycles of recurrent
selection and susceptible individuals are still observed in advanced breeding pop-
ulations. Therefore, breeding methods have to be refined and optimised to develop
L. multiflorum cultivars with increased resistance to bacterial wilt, thereby ensuring
productive and sustainable grassland agriculture. Molecular genetic tools for discov-
ering and targeting genes and alleles in germplasm collections and/or breeding lines
have the potential to enhance the efficiency of breeding programs through marker
assisted selection (MAS). Although a number of quantitative trait loci (QTL) asso-
ciated with resistance to diseases such as crown rust and bacterial wilt have been
identified in L. perenne and L. multiflorum (Dracatos et al. 2010; Studer et al. 2006),
so far no example of successful MAS has been reported for these species. This
may arise from the narrow genetic base of mapping families when compared to the
range of variation available in the breeding germplasm or the often unknown genetic
composition of pathogen populations (Dracatos et al. 2006; Kölliker et al. 2006).
Therefore, a detailed characterization of the Xtg × L. multiflorum interaction has
to take into account factors influencing plant resistance as well as factors influenc-
ing pathogen virulence. This contribution briefly reviews recent advances towards a
better understanding of plant resistance to Xtg and then reports on the investigation
of potential virulence factors by means of virulence-gene mutagenesis and whole
genome sequencing.

4.2 Genetic Control of Bacterial Wilt Resistance

in L. multiflorum

QTL analyses in a mapping population of L. multiflorum have demonstrated that

bacterial wilt resistance is controlled by one major QTL on linkage group (LG)
4 explaining between 43 and 84 % of the total phenotypic variance for resistance
(Studer et al. 2006). Phenotypic traits that result in the identification of one single
QTL explaining such a high percentage of the total observed phenotypic variance
have often been shown to be controlled by one or only a few major R-genes (Mutlu
et al. 2006). In order to investigate whether Xtg resistance was based on a major,
race-specific resistance gene, 62 selected plant genotypes were artificially inoculated
in the greenhouse with six different bacterial isolates (Wichmann et al. 2011b). Sig-
nificant differences in resistance among L. multiflorum genotypes (p < 0.001) and in
virulence among Xtg isolates (p < 0.001) were observed using the area under the dis-
ease progress curve (AUDPC). The ranking of virulence of the six bacterial isolates in
this study was for the most part congruent with the differences in virulence observed
by Kölliker et al. (2006), which had been based on three different L. multiflorum
cultivars. This demonstrates that the bacterial isolates used are reproducibly virulent
on different plant material and during different experiments. No significant genotype–
4 Breeding for Resistance to Bacterial Wilt in Ryegrass: Insights into the Genetic . . . 39

isolate interaction (p > 0.05) using linear regression modelling was observed. Simple
sequence repeat (SSR) markers were used to identify marker-resistance associations
in the same plant genotypes and using the same bacterial isolates. The SSR marker
NFA027 located on LG 5 was significantly associated with bacterial wilt resistance
across all six bacterial isolates and explained up to 37.4 % of the total variance of
AUDPC values. The major QTL on LG 4 (Studer et al. 2006) could not be confirmed
in the F2 progeny derived from the initial mapping population. This may be due
to insufficient linkage between the SSR markers used and the QTL and the small
number of parental alleles observed in the investigated individuals (Wichmann et al.
2011b). Neither the inoculation experiment nor the SSR analyses revealed major
host genotype–pathogen isolate interactions, thus suggesting that Xtg resistance, so
far observed, is effective across a broad range of different bacterial isolates and plant
genotypes.
The partial transcriptomes of two Italian ryegrass genotypes, one resistant and
one susceptible to bacterial wilt, were compared at four time points after Xtg infec-
tion (Wichmann et al. 2011a). The transcriptome analysis of the resistant genotype
revealed in total 158 genes differentially expressed after Xtg inoculation. Twenty
up-regulated genes were observed at 48 hours post infection (hpi), 52 genes were
differentially expressed 192 hpi (42 up- and 10 down-regulated), and 124 genes
were differentially expressed 288 hpi (76 up- and 48 down-regulated). No signifi-
cant differential expression was detected 8 hpi (Wichmann et al. 2011a). Recognition
of bacterial effector proteins and the induction of HR usually occur within 24 hpi
(Scheideler et al. 2002). Thus, transcriptional changes leading to a HR were either
absent or below the detection threshold of the microarray analysis. The transcrip-
tome analyses revealed a number of promising candidate genes for bacterial wilt
resistance. For example, a gene strongly up-regulated after Xtg infection was highly
similar to a gene encoding the low silicon transporter Lsi1, a protein which belongs
to a Nodulin26-like major intrinsic protein sub-family of aquaporins. Lsi1 has been
reported to be involved in silicon (Si) uptake in many plant species and Si is thought
to be essential for resistance against biotic and abiotic stress (reviewed in: Ma and
Yamaji 2008). Another promising candidate gene found to be up regulated in the re-
sistant genotype after Xtg inoculation was the germin-like protein GLP6 (Wichmann
et al. 2011a). GLP6 belongs to the oxalate oxidase (OXO)-like GLPs (Carrillo et al.
2009), which have been associated with QTL for disease resistance in rice and barley
(Ramalingam et al. 2003). Among others, these genes present promising candidates
for further characterization of Xtg resistance in L. multiflorum.

4.3 Genetic Control of Virulence in X. translucens pv. graminis

Virulence factors are produced by the pathogen to enable host colonization, to evade
or inhibit the host’s immune response and to acquire nutrients from the host. They
include secretion systems as well as secreted effector proteins. The most extensively
studied secretion system of Xanthomonas spp. is the type III secretion system (T3SS)
which is responsible for the secretion of a broad range of effectors (Cornelis 2006).
40 F. Wichmann et al.

The genes encoding the T3SS are typically localised in large gene clusters termed hrp
(hypersensitivity response and pathogenicity) genes (Arnold et al. 2003). Expression
of the hrp gene cluster results in the formation of a membrane-spanning secretion ap-
paratus, the Hrp-pilus or type III injectisome, which mediates the delivery of bacterial
effectors into the host cell (Cornelis 2006). In X. campestris pv. vesicatoria, expres-
sion of the T3SS gene cluster is induced in planta by the HrpG/HrpX two-component
regulatory system (Büttner and Bonas 2010). Approximately 20–30 effectors with
overlapping activities and diverging composition are typically secreted by one single
Xanthomonas strain (reviewed in: Büttner and Bonas 2010). These effectors fulfill
multiple functions and often are involved in effector triggered susceptibility or the
induction of HR on resistant plants (White and Yang 2009). Although some striking
similarities in virulence factors exist within the genus Xanthomonas, secretion mech-
anisms and effector repertoires vary greatly among species, pathovars and strains
(White et al. 2009). Phylogenetic analyses have shown that Xtg is quite distantly re-
lated to well characterised xanthomonads such as X. campestris pv. vesicatoria (Xca),
X. oryzae pv. oryzae (Xoo) and X. axonopodis pv. citri (Parkinson et al. 2009). It is
therefore expected, that specific and/or novel effectors and secretion pathways are
involved in the Xtg × L. multiflorum interaction. The aim of this study was to clarify
the role of the T3SS for virulence of Xtg by means of site directed mutagenesis of
the regulatory hrpG gene and to characterise T3SS components and encoded effector
proteins by means of whole genome sequencing and comparative genomics.

4.3.1 Materials and Methods

4.3.1.1 Mutagenesis of hrpG

The hrpG gene of Xtg29 (Kölliker et al. 2006) was sequenced using conserved primers
designed on the consensus sequence of publicly available sequences of the hrpG gene
of other Xanthomonas spp. Subsequent primer walking on genomic DNA resulted in
a contig of 2,045 bp length. Two DNA fragments of approximately 500 bp of each of
the two hrpG flanking regions were connected using Soeing PCR (Horton 1995). The
resulting fragment referred to as hrpG fragment was cloned into the suicide vector
pKNG101 (Kaniga et al. 1991) resulting in the plasmid pCC101 (Fig. 4.1). Electro-
competent Xtg29 were transformed with the suicide vector pCC101 and the transfor-
mants were selected with streptomycin. Double homologous recombination leading
to a hrpG deletion was induced after culturing in media devoid of streptomycin and
plating on sucrose plates. The resulting hrpG mutant was verified by PCR.
A highly susceptible L. multiflorum genotype was used for virulence screening of
the Xtg29 wildtype and the hrpG mutant. A negative control treatment consisted of
cutting the plants without inoculum. Assessment of bacterial wilt symptoms was per-
formed using four replications per genotype x treatment combination in a completely
randomized block design.
In planta multiplication was assessed by re-isolation of wildtype and mutant
strains from inoculated L. multiflorum plants grown in replicates under controlled
4 Breeding for Resistance to Bacterial Wilt in Ryegrass: Insights into the Genetic . . . 41

Fig. 4.1 Suicide vector

pCC101 derived from
pKNG191 containing the
hrpG fragment

conditions. Leaves and tillers were harvested at four different time points after in-
fection: 6 hours post infection (hpi), 4, 7 and 14 days post infection (dpi). Surface
sterilization was performed by incubating the plant material in 1 % Chloramine-T so-
lution (Honeywell Riedel de-Haën, Seelze, Germany) for 10 min. Serial dilutions of
the homogenized plant material were prepared on CircleGrow medium (MP Biomed-
icals) and bacterial cell counts per g of fresh plant material were determined after
incubation of the plates at 28 ◦ C for 7 days.

4.3.1.2 Genome sequencing

A draft sequence of the Xtg29 (Kölliker et al. 2006) genome was obtained by em-
ploying a whole genome shotgun approach by means of 454 sequencing (Roche).
For this purpose, a library of paired-end (PE) shotgun fragments was prepared from
the genomic DNA. The fragment library was sequenced using the 454 Titanium
technology. Genome data were annotated using the GenDB software (Meyer et al.
2003). DNA and protein sequences were compared with other Xanthomonas spp.
using BLASTn, BLASTx or BLASTp (Altschul et al. 1990).

4.3.2 Results and Discussion

4.3.2.1 Mutagenesis of hrpG

A strain of Xtg29 deficient of the hrpG gene was generated using double homologous
recombination after transformation with the pCC101 plasmid (Fig. 4.1) and selection
on media containing 5 % sucrose. The different recombination events were verified
42 F. Wichmann et al.

Fig. 4.2 Area under the disease progress curve (AUDPC) values of a highly susceptible L. multi-
florum genotype inoculated with the Xtg29 wildtype and the hrpG mutant. Data were collected
at 7, 14, 21 and 28 days after inoculation. Standard deviation is indicated by horizontal lines

using PCR and hrpG specific primers. In vitro growth of the hrpG mutant was not
affected by the hrpG mutation or the presence of an extrachromosmal plasmid on
GYC plates, and the colonies could not be distinguished from the wildtype strain
(data not shown).
The symptoms caused by the hrpG mutant were significantly weaker than the
symptoms caused by the wildtype Xtg29 isolate. Average disease scores ranging
from 1–9 for the plants infected with the wildtype isolate of Xanthomonas translucens
pv. graminis Xtg29 increased from 2.00 ± 0.82 at 7 days post inoculation (dpi),
to 5.00 ± 0.00 at 14 dpi, to 6.00 ± 1.15 at 21 dpi and 6.00 ± 0.82 resulting in an
AUDPC value of 112.00 ± 14.85 (Fig. 4.2). On the other hand, disease scores of
the plants infected with the hrpG mutant ranged from 1.75 ± 0.50 at 7 dpi, to
1.75 ± 0.96 at 14 dpi, 2.25 ± 0.96 at 21 dpi, and 2.75 ± 0.50 at 28 dpi, respectively,
resulting in an average AUDPC value of 49.88 ± 13.21. Control-treated plants were
described with disease scores 1.00 ± 0.00 at 7 dpi, 1.00 ± 0.00 at 14 dpi, 1.5 ± 0.58,
at 21 dpi and 2.25 ± 0.50 at 28 dpi, respectively, resulting in an average AUDPC
value of 32.28 ± 3.35.
The fact that symptom development is impaired by site-directed mutagenesis of
the hrpG gene in Xtg indicated that Xtg depends on a T3SS for pathogenicity. In order
to characterize the role of the hrpG gene for in planta multiplication of Xtg in L.
multiflorum, both the wildtype and the hrpG mutant were re-isolated from infected
plant material and population densities were determined. Population densities of the
hrpG mutant quantified in leaves at 6 h post infection were about 0.9-fold those
of the wildtype isolate Xtg29 (Fig. 4.3). At 4 days post infection (dpi), both isolates
multiplied and reached average log10 population densities per gram of fresh weight
of 8.72 for the wildtype and 8.07 for the hrpG mutant and a further increase in
population densities followed until 14 dpi. These results indicated that the hrpG gene
is not primarily responsible for survival during the early infection stages. However,
when infecting L. multiflorum plants with Xtg, the leaf tissue is wounded by cutting
with scissors. This procedure enables direct contact of the bacterial cells with the
site or tissue of multiplication i.e. the xylem. Therefore, we hypothesize that due to
4 Breeding for Resistance to Bacterial Wilt in Ryegrass: Insights into the Genetic . . . 43

Fig. 4.3 Colonization of L. multiflorum by Xtg29 wildtype and hrpG mutant 0–14 days post
infection. Population densities at 4 and 14 dpi were significantly different based on a two sided
t-test (p < 0.05)

the fact that Xtg cells gain direct access to the xylem, Xtg can make use of nutrients
available inside the host throughout the infection process.

4.3.2.2 Genome sequencing

Preliminary analysis of the data obtained for Xtg29 resulted in a draft sequence
consisting of 908 contigs with gaps between the contigs and a G + C content of
68.6 % was calculated. This G + C content is higher than in all the previously deter-
mined complete genomes of Xanthomonas spp., where the G + C content was well
conserved at about 63–65 %. The number of insertion sequence (IS) elements or IS
element fragments in the genome of Xtg29 is estimated to develop into several hun-
dred copies. High numbers of both complete and fragmentary IS elements have been
observed for all Xoo genomes so far, as well as for X. o. pv. oryzicola. Hence, this
finding relates Xtg to the rice pathogenic Xanthomonas strains, where the number of
IS elements ranges from 251 complete to 714 fragments of IS elements. Predicted
genome size of Xtg29 is 5,057,066 bp consisting of one single, circular chromosome
and no plasmids.
Despite a number of similarities to the other sequenced Xanthomonas spp. (e.g.
the gum gene cluster, EPS production and conserved hrp genes), the isolate Xtg29
was found to be different from the other sequenced Xanthomonas spp. in terms of
T3SS architecture and protein sequence identities. The two genes hrpG and hrpX
encoding the two response regulators of the T3SS were found to be localized within
the hrp gene cluster in Xtg29 which is in contrast to the other sequenced Xanthomonas
spp., where hrpG and hrpX are situated together and outside of the hrp gene cluster.
A similar architecture with the genes encoding hrpG and hrpB (which is homologous
to hrpX of Xanthomonas spp.) within the T3SS has been reported for the Ralstonia
solanacearum isolate GMI1000 for which the T3SS has been shown to be located on
the megaplasmid pGMI1000MP (Salanoubat et al. 2002). A set of homologous genes
44 F. Wichmann et al.

encoding T3SS secreted effectors that certainly is important for the interaction of Xtg
with L. multiflorum was found. These effectors may be specific for this host-pathogen
interaction and may also determine the host range of Xtg. Nevertheless, whether these
homologs are functional and/or race-specific for the isolate Xtg29 or are also found
in other Xtg isolates will require further experimental data and comparative analysis
with other Xtg strains and X.t. pathovars. From the data obtained so far, it can be
speculated that not only virulence factors and symptoms may be distinct for the
L. multiflorum-Xtg interaction, but also that resistance mechanisms may differ from
well characterised systems such as rice-Xoo and pepper-Xcv.

4.4 Conclusions

The genomic and transcriptomic approaches undertaken so far have revealed valu-
able insights into bacterial wilt resistance of L. multiflorum and Xtg virulence. Taken
together, phenotypic and molecular genetic characterization as well as the tran-
scriptome analyses indicated that Xtg resistance is conferred by various resistance
mechanisms and genes that contribute to resistance quantitatively. Two independent
analyses revealed two different genomic regions to contribute to a significant degree
to Xtg resistance. In addition, the transcriptome analyses revealed two interesting
candidate genes for Xtg resistance such as Lsi1 and GLP6. To shed light on the
L. multiflorum-Xtg interaction from a different perspective, virulence factors of Xtg
were investigated. Interactions between R. solanacearum and solanaceous crops may
serve as more suitable models due to comparative analyses that revealed a T3SS with
a genetic organization different from other sequenced Xanthomonas spp., but very
similar to the β-proteobacterial plant pathogen Ralstonia solanacearum. In contrast
to other hrpG deficient Xanthomonas strains of other species, the hrpG gene of Xtg29
is not crucial for in planta multiplication and survival, highlighting the distinctness
of the Xtg-L. multiflorum interaction. Combined, the approaches of L. multiflorum
resistance mechanisms and Xtg virulence factors have allowed to gain a more pro-
found understanding of this host-pathogen system and will enable the identification
of further candidate genes and the development of MAS tools in the future.

Acknowledgments This research was supported by the Swiss National Science Foundation (Grant
No. 31003A–112582). We thank C. Conradin for her work on hrpG mutants and S. Reinhard, S.
Kuhnen and P. Streckeisen for excellent technical support.

References

Altschul SF, Gish W, Miller W, Myers EW and Lipman DJ (1990) Basic local alignment search
tool. J Mol Biol 215:403–410
Arnold DL, Pitman A and Jackson RW (2003) Pathogenicity and other genomic islands in plant
pasthogenic bacteria. Mol Plant Pathol 4:407–420
4 Breeding for Resistance to Bacterial Wilt in Ryegrass: Insights into the Genetic . . . 45

Boller B, Peter-Schmid M, Tresch E, Tanner P and Schubiger F (2009) Ecotypes of Italian ryegrass
from Swiss permanent grassland outperform current recommended cultivars. Euphytica 170:
53–65
Büttner D, Bonas U (2010) Regulation and secretion of Xanthomonas virulence factors. FEMS
Microbiol Rev 34:107–133
Carrillo MGC, Goodwin PH, Leach JE, Leung H and Vera Cruz CM (2009) Phylogenomic rela-
tionships of rice oxalate oxidases to the cupin superfamily and their association with disease
resistance QTL. Rice 2:67–79
Cornelis GR (2006) The type III secretion injectisome. Nat Rev Microbiol 4:811–825
Dracatos PM, Cogan NOI, Keane PJ, Smith KF, Forster JW (2010) Biology and genetics of crown
rust disease in ryegrasses. Crop Sci 50:1605–1624
Dracatos PM, Dumsday JL, Olle RS, Cogan NOI, Dobrowolski MP, Fujimori M, Roderick H,
StewartAV, Smith KF, Forster JW (2006) Development and characterization of EST-SSR markers
for the crown rust pathogen of ryegrass (Puccinia coronata f.sp. lolii). Genome 49:572–583
Egli T, Schmidt D (1982) Pathogenic variation among the causal agents of baterial wilt of forage
grasses. Phytopathol Z 104:138–150
Horton RM (1995) PCR-mediated recombination and mutagenesis. Mol Biotechnol 3:93–99
Kaniga K, Delor I, Cornelis GR (1991) A wide-host-range suicide vector for improving reverse
genetics in gram-negative bacteria: inactivation of the blaA gene of Yersinia enterocolitica.
Gene 109:137–141
Kölliker R, Kraehenbuehl R, Boller B, Widmer F (2006) Genetic diversity and pathogenicity of the
grass pathogen Xanthomonas translucens pv. graminis. Syst Appl Microbiol 29:109–119
Ma JF,Yamaji N (2008) Functions and transport of silicon in plants. Cell Mol Life Sci 65:3049–3057
Meyer F, Goesmann A, McHardy AC, Bartels D, Bekel T, Clausen J, Kalinowski J, Linke B,
Rupp O, Giegerich R, Puhler A (2003) GenDB—an open source genome annotation system for
prokaryote genomes. Nucleic Acids Res 31:2187–2195
Mutlu N, Miklas P, Coyne D (2006) Resistance gene analog polymorphism (rgap) markers co-
localize with disease resistance genes and qtl in common bean. Mol Breeding 17:127
Parkinson N, Cowie C, Heeney J, Stead D (2009) Phylogenetic structure of Xanthomonas
determined by comparison of gyrB sequences. Int J Syst Evol Microbiol 59:264–274
Ramalingam J, Cruz CMV, Kukreja K, Chittoor JM, Wu JL, Lee SW, Baraoidan M, George ML,
Cohen MB, Hulbert SH, Leach JE, Leung H (2003) Candidate defense genes from rice, barley,
and maize and their association with qualitative and quantitative resistance in rice. Mol Plant
Microbe Interact 16:14–24
Salanoubat M, Genin S, Artiguenave F, Gouzy J, Mangenot S, Arlat M, Billault A, Brottier P, Camus
JC, Cattolico L, Chandler M, Choisne N, Claudel-Renard C, Cunnac S, Demange N, Gaspin C,
Lavie M, Moisan A, Robert C, Saurin W, Schiex T, Siguier P, Thebault P, Whalen M, Wincker P,
Levy M, Weissenbach J, Boucher CA (2002) Genome sequence of the plant pathogen Ralstonia
solanacearum. Nature 415:497–502
Scheideler M, Schlaich NL, Fellenberg K, Beissbarth T, Hauser NC, Vingron M, Slusarenko AJ,
Hoheisel JD (2002) Monitoring the switch from housekeeping to pathogen defense metabolism
in Arabidopsis thaliana using cDNA arrays. J Biol Chem 277:10555–10561
Schmidt D, Nuesch B (1980) Resistance to bacterial wilt (Xanthomonas graminis) increases yield
and persistency of Lolium multiflorum. EPPO B 10:335–339
Studer B, Boller B, Herrmann D, Bauer E, Posselt UK, Widmer F, Kölliker R (2006) Genetic
mapping reveals a single major QTL for bacterial wilt resistance in Italian ryegrass (Lolium
multiflorum Lam.). Theor Appl Genet 113:661–671
Wang HM, Sletten A (1995) Infection biology of bacterial wilt of forage grasses. J Phytopathol
143:141–145
White F, Potnis N, Jones J, Koebnik R (2009) The type III effectors of Xanthomonas. Mol Plant
Pathol 10:749–766
White FF, Yang B (2009) Host and pathogen factors controlling the rice-Xanthomonas oryzae
interaction. Plant Physiol 150:1677–1686
46 F. Wichmann et al.

Wichmann F, Asp T, Widmer F, Kölliker R (2011a) Transcriptional responses of Italian ryegrass

during interaction with Xanthomonas translucens pv. graminis reveal novel candidate genes for
bacterial wilt resistance. Theor Appl Genet 122:567–579
Wichmann F, Mueller-Hug B, Widmer F, Boller B, Studer B, Kölliker R (2011b) Phenotypic and
molecular characterization indicate no major race-specific interactions between Xanthomonas
translucens pv.graminis and Lolium multiflorum. Plant Pathol 60:314–324
Chapter 5
Mechanisms Utilised Within the IBERS Diploid
Lolium perenne L. Forage Grass Breeding
Programmes to Improve Rumen Nitrogen
Use Efficiency

R. C. Hayes, J. A. Lovatt and M. T. Abberton

Abstract Researchers at Institute of Biological, Environmental and Rural Sciences

(IBERS), Aberystwyth University, have completed 3 years of a 5 year LINK project,
sponsored by Defra through the Sustainable Livestock Production LINK programme
(www.greener-grasslands.ibers.aber.ac.uk) which aims to breed new forages that will
reduce the environmental footprint of livestock production. One of the key objec-
tives is to improve nitrogen use efficiency (NUE) in the rumen through the breeding
of new forage grasses to improve protein utilisation. Previously it has been shown
that feeding ryegrasses with higher water soluble carbohydrate content leads to im-
proved rumen efficiency and evidence suggests that this results in increased meat and
milk production and reduced nitrogen losses with lower ammonia and nitrous oxide
emissions through improved protein utilisation. IBERS diploid perennial ryegrass
breeding programmes involve a combination of spaced plant assessment and half-
sibling plot performance as a basis for recurrent selection over many generations,
the current focus of which is to combine increased NUE along with other desirable
agronomic traits including improved yields, increased persistency and higher dry
matter digestibility.

5.1 Introduction

Researchers at Institute of Biological, Environmental and Rural Sciences (IBERS),

Aberystwyth University, have completed 3 years of a 5 year LINK project, sponsored
by Defra and industry partners through the Sustainable Livestock Production (SLP)
LINK programme (www.greener-grasslands.ibers.aber.ac.uk). One of the key objec-
tives is to enhance nitrogen use efficiency in the rumen through the breeding of new
forage grasses to improve protein utilisation. Feeding ryegrasses with higher water
soluble carbohydrate (WSC) content has been shown to lead to improved rumen

R. C. Hayes () · J. A. Lovatt · M. T. Abberton

Public Good Plant Breeding, IBERS, Aberystwyth University,
Gogerddan, Aberystwyth SY23 3EB, UK
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 47

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_5, © Springer Science+Business Media Dordrecht 2013
48 R. C. Hayes et al.

efficiency (allowing increased microbial protein synthesis) and evidence suggests

that this results in reduced nitrogen (N) losses through lower ammonia and nitrous
oxide emissions via improved protein utilisation (Lee et al. 2003; Moorby et al. 2006)
Ruminant agriculture plays a significant role in contributing to greenhouse gas
emissions, estimated at 18 % of anthropogenic emissions worldwide (FAO 2006),
and environmental damage through N losses from excreta, both in terms of losses
to air as ammonia and nitrous oxide, and as nitrates leaching into groundwater.
Previous breeding targets at IBERS have focused on the genetic improvement of
ryegrass for traits to improve NUE in the rumen, for example ryegrass varieties that
incorporate increased dry matter digestibility (DMD) and WSC content have already
been developed and marketed. Previous studies and models have demonstrated that
these grasses supply energy to rumen microbes at the time required for microbial
protein synthesis which increases rumen NUE (Lee et al. 2002, Ellis et al. 2011).
There is a need to build on this proof of concept and breeding progress to develop
grasses that realise the full potential of increased WSC levels through an optimised
energy to protein ratio.
IBERS’ diploid perennial ryegrass breeding programmes involve comprehensive
trialling and assessment of spaced plants and half-sibling plot performance as a basis
for recurrent selection over many generations. The diploid intermediate heading date
breeding programme is now in the 14th generation of selection; some of the successful
varieties from past generations include ‘AberDart’(F7), ‘AberStar’(F9), ‘AberMagic’
(F10) and the most recent commercial variety, ‘AberGreen’ (F11). Current diploid
IBERS varieties are recognised via Recommended List trials as some of the best in
their category throughout the UK. In this paper the variety progress achieved during
the 12th generation will be discussed, in addition, predictions of performance for
the newest varieties synthesised from the 13th generation of selection in 2011 will
be reviewed.

5.2 Materials and Methods

A breeding trial (13th generation, intermediate diploid Lolium perenne L.) was un-
dertaken at IBERS, Gogerddan, Aberystwyth, Wales. The soil type was a Rheidol
series. The breeding trial was undertaken over two cutting seasons (March to Novem-
ber in 2009 and 2010). A seed rate of 30 kg/ha was used with broadcast sowing
during autumn 2008 to prepare four randomised complete block replicated plots
per half-sibling family or control variety measuring 1 × 1.2 m. Fertiliser rates of
190 kg/ha/year N, 35 kg/ha/year P, 120 kg/ha/year K and 25 kg/ha/year sulphur were
applied as a split dressing after each cut. A seven cut management was used as an in-
termediary treatment between conservation and grazing managements and cut using
a Haldrup plot harvester. Assessment of forage yield over 2 years was made, addi-
tionally, quality data including N, WSC, dry matter digestibility (DMD) and fibre
fraction, were collected in year one from cuts three, four and five and analysed using
5 Mechanisms Utilised Within the IBERS Diploid Lolium perenne L. Forage Grass . . . 49

laboratory based NIRS, 10 % of samples underwent chemical analysis to confirm

the NIRS calibrations. The breeding trial included five control varieties (cv. ‘Aber-
Dart’, ‘AberMagic’, ‘Premium’, ‘Ba14074 Syn1’ and ‘Ba14074 Syn2’) and 55, 13th
generation, intermediate heading half-sibling progeny sets, generated from crossing
240 selected 12th generation half-sibling progeny, with equal representation of 60
progeny from each of four 12th generation families.
New varieties were synthesised by selecting four mother plants based upon the
above range of desirable traits that had been identified via half-sibling progeny plot
trials. Most traits were quantitatively assessed to determine heritability scores and
significant differences. Predictions of performance for new synthetic varieties were
obtained by calculating the mean score for the four selected mother plants for each
trait from the half-sibling progeny breeding trial results. Each chosen mother plant
was cloned and an equal number of each polycrossed in pollen proof isolation cham-
bers. Seed was harvested, threshed and cleaned and the Syn1 variety seed stock
generated using equal quantities from each mother plant.
The synthetic variety ‘Ba14074’ was selected and produced during the 12th gen-
eration of selection using the same breeding method described above for the 13th
generation. Syn 1 and Syn 2 generations of ‘Ba14074’ were assessed in evaluation
trials sown in 2007 and 2008 respectively, at IBERS using an experimental design
of four randomised complete blocks against a range of commercial varieties. The
cutting regime used was five cuts in years 1 and 3 to simulate conservation man-
agement and nine cuts in year 2 for simulated grazing. Quality measurements were
taken from cuts one and two in year 1 and obtained using laboratory based NIRS,
10 % of samples underwent chemical analysis to confirm the NIRS calibrations. Fer-
tiliser rates of 165 kg/ha/year N, 30 kg/ha/year P, 105 kg/ha/year K and 20 kg/ha/year
sulphur were applied as a split dressing after each cut.
The synthetic varieties, Ba14150–152, were selected and produced from the 13th
generation of selection from the intermediate heading breeding programme using the
same methodology described above. Predictions of performance for the new varieties
were obtained as previously described.

5.3 Results

The variety, ‘Ba14074’, demonstrates the accuracy of predictions obtained from the
results of 12th generation half-sibling recurrent selection breeding plot trials as a
basis for varietal construction when compared to actual performance of the finished
variety (results for Syn 1 and Syn 2 generations did not significantly differ) in IBERS
evaluation plot trials (Fig. 5.1). The results presented in Fig. 5.1 also show the gains
in overall performance of ‘Ba14074’ (F12) compared to ‘AberDart’ (F7) for a range
of traits from IBERS evaluation trials; dry matter yield, WSC and ground cover.
For the above traits the values obtained from IBERS evaluation trials were at least
as good as or better than those previously predicted for ‘Ba14074’, predictions of
50 R. C. Hayes et al.

Fig. 5.1 Predicted and actual agronomic performance of ‘Ba14074’ in IBERS trials, performance
is presented for ‘Ba14074’ as a percentage of that achieved by ‘AberDart’ in the same breeding and
evaluation trials (‘AberDart’= 100 %)

performance for ‘Ba14074’ were obtained as described in the materials and methods
section.
WSC and DMD measurements of the 13th generation half-sibling progeny plot
breeding trials exhibited a wide range of variation (Fig. 5.2), with a number of
populations showing improvements over ‘AberMagic’, thus enabling the selection
of mother plants to synthesise varieties which are predicted to have an energy content
better optimised for ruminant digestion. Many of the mother plants observed with the
desired WSC content were also seen to exhibit many additional desirable agronomic
traits as well as variation in protein content, thus enabling the selection of three sets
of four elite mother plant clones to synthesise three new 13th generation intermediate
heading varieties (Ba14150–152).
Performance predictions were calculated for three new 13th generation synthetic
varieties as presented in Fig. 5.3, in comparison with the performance of ‘Ba14074’
(Syn 1 and Syn 2 generations) from the 13th generation breeding trial. Calculated
performance predictions were used to determine the best combination of mother
plants to produce new varieties, in this instance the varieties were all targeted at
slightly different markets and end use. ‘Ba14150’ was selected principally for very
high yield (5.2 % above ‘Ba14074’) and WSC (10.2 % above ‘Ba14074’) to be used
as a biofuel source as well as determining how far the WSC trait can be increased.
‘Ba14151’ was selected to provide excellent all round performance, with very high
ground cover (4.2 % above ‘Ba14074’), an optimised energy to protein ratio and
superior seed production potential. ‘Ba14152’ was designed to extend the grazing
5 Mechanisms Utilised Within the IBERS Diploid Lolium perenne L. Forage Grass . . . 51

Fig. 5.2 Variation in WSC

and DMD within the 13th
generation of selection of
IBERS intermediate diploid
Lolium perenne L. breeding
population compared to three
commercial control varieties.
The oval highlights
populations’ superior to
‘AberMagic’, the blue dots
represent the scores assigned
to each of 55 mother plants
from the 13th generation
half-sibling progeny breeding
plot trial

Fig. 5.3 Predicted agronomic performance of three new intermediate diploid varieties synthesised
from the 13th generation, all were designed to optimise the ratio of energy to protein. Values in paren-
thesis were used as the 100 % baseline and were obtained as a mean of Syn 1 and Syn 2 ‘Ba14074’
from the breeding trial
52 R. C. Hayes et al.

season with enhanced spring and autumn growth, giving a total yield 7.3 % greater
than seen for ‘Ba14074’, in conjunction with superb all round performance and
optimal WSC to protein content.

5.4 Discussion

During the project a number of new varieties of ryegrass were synthesised from
the intermediate heading diploid breeding programme that seek to optimise the en-
ergy to protein ratio in forage, thus enhancing rumen efficiency; this optimisation is
in conjunction with gains to yield and superior all round agronomic performance.
The three predictions of varietal performance derived from the intermediate popu-
lation illustrate the value of half-sibling recurrent selection as a breeding tool. This
methodology provides access to a genetic base that has been improved over many
generations to select from, and enables the identification of multiple desirable traits
in selected genotypes through phenotypic selection for mother plant selection and
variety synthesis.
Future breeding objectives at IBERS for forage grasses include improvements
to N and P utilisation and recycling, increasing polyunsaturated fatty acid content,
specifically that of alpha linolenic acid, improving seed yield and enhancing the
digestible to non-digestible fibre fraction. These traits will provide greater added
value to farmers and reduce the environmental impact of ruminant farming in addition
to providing all-round improved forage performance and quality.
In addition to ryegrass breeding, complementary research at IBERS has investi-
gated the variation in leaf protein content of elite genotypes of white clover. This
has led to the development of a lower protein variety. Leys of new ryegrass varieties
and the reduced protein clover are predicted to enhance rumen NUE beyond levels
currently obtainable with existing grazing options. Animal trials will explore more
definitively the relationship between the ratio of WSC and protein in the diet (at
different WSC and protein levels), excretion of urinary N, milk yield and quality
using mixtures of high WSC grass varieties and white clover varieties selected for
lower protein content.

5.5 Conclusions

Half-sibling recurrent selection has been shown to be a powerful breeder’s tool,

without which improvements seen in modern IBERS varieties would not have been
possible. IBERS’ ryegrass breeding programmes have been seen to advance impor-
tant agronomic traits and add additional benefits both to the farmer, in the form of
improved meat and milk production from grazed grass, and also to the environment
by reducing the proportion of nitrogen lost as ammonia and nitrous oxide during ru-
men digestion. The commercialisation of new IBERS’ varieties is expected to assist
farmers with maximising profits from grazed forage.
5 Mechanisms Utilised Within the IBERS Diploid Lolium perenne L. Forage Grass . . . 53

References

Ellis JL, Dijkstra J, Bannink A, Parsons AJ, Rasmussen S, Edwards GR, Kebreab E, France J (2011)
The effect of high-sugar grass on predicted nitrogen excretion and milk yield simulated using a
dynamic model. J Dairy Sci 94:3105–3118. doi:10.3168/jds.2010-4059
FAO (2006) Livestock’s long shadow: environmental issues and options. Food and Agriculture
Organisation, Rome
Lee MRF, Harris LJ, Moorby JM, Humphreys MO, Theodorou MK, MacRae JC, Scollan ND (2002)
Rumen metabolism and nitrogen flow to the small intestine in steers offered Lolium perenne
containing different levels of water-soluble carbohydrate. Anim Sci 74:587–596
Lee MRF, RJ Merry, DR Davies, JM Moorby, MO Humphreys, MK Theodorou, JC MacRae ND
Scollan. (2003). Effect of increasing availability of water-soluble carbohydrates on in vitro
rumen fermentation. Anim Feed Sci Technol 104:59–70
Mooby JM, Evans RT, Scollan ND, MacRae JC, Theodorou MK (2006) Increased concentration of
water-soluble carbohydrate in perennial ryegrass (Lolium perenne L.). Evaluation in dairy cows
in early lactation. Grass Forage Sci 61:52–59
Chapter 6
Population Genetics of the Grass
Self-incompatibility System—Practical
Implications for Grass Breeding Programmes

C. Manzanares, B. Studer, R. C. Hayes, S. Barth and D. Thorogood

Abstract Self-incompatibility (SI) is a mechanism that prevents plants from self-

pollination through molecular recognition between pollen and pistil; but neither the
genes nor the proteins involved are known. Determining haplotype diversity of the
SI locus region is an indirect way of predicting incompatibility genotype which
would enable plant breeders to develop strategies to ensure precision crossing for
exploiting heterotic combinations. In the case of ryegrass (Lolium perenne), SI is
known to be a gametophytic reaction involving two loci, S and Z. This paper reports
on an assay of marker allele diversity around the Z locus for a population produced
from twelve generations of mixed and half-sib family selection. Using conventional
genotyping of marker length polymorphisms, as well as high resolution melt (HRM)
curve analysis, the 55 plants of the population were classified into thirteen groups
according to their genotypes for each marker. HRM genotyping proved to be more
discriminating than STS markers and could enhance the precision of Z genotype
prediction. Although half-sib family selection is expected to lead to allele fixing, this
initial analysis was encouraging in that it indicated high haplotype diversity around
the Z-locus suggesting maintenance of a high degree of cross-compatibility in the
breeding population. This would be expected as SI alleles are subjected to frequency
dependent selection.

C. Manzanares () · S. Barth

Plant Biotechnology Unit, Teagasc, Oak Park, Carlow, Ireland
e-mail: [email protected]
Institute of Biological, Environmental and Rural Sciences, Aberystwyth University,
Plas Gogerddan Campus, SY23 3EB, Aberystwyth, Ceredigion, UK
School of Biosciences, University of Birmingham, B15 2TT, Edgbaston,
Birmingham, UK
R. C. Hayes · D. Thorogood
Institute of Biological, Environmental and Rural Sciences, Aberystwyth University,
Plas Gogerddan Campus, SY23 3EB, Aberystwyth, Ceredigion, UK
B. Studer
Department of Molecular Biology and Genetics, Faculty of Science
and Technology, Aarhus University, Forsøgsvej 1, 4200 Slagelse, Denmark

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 55

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_6, © Springer Science+Business Media Dordrecht 2013
56 C. Manzanares et al.

Keywords Haplotype diversity · High resolution melting · Frequency dependent

selection · Plant breeding · Lolium perenne

6.1 Introduction

Many plants possess self-incompatibility (SI) mechanisms that prevent inbreeding

by blocking fertilisation of ovules by like-pollen. These mechanisms have evolved
several times and a number of different mechanisms have been described (Franklin-
Tong 2008). Related species often possess the same mechanism, and in the grass
family, the SI system is controlled gametophytically by two complementary loci, S
and Z, which both consist of a highly diverse series of alleles. Incompatible pollen
(determined by the possession of an S-Z allele pair from the individual pollen game-
tophyte matching that of any of the allele pairs of the receptive stigma) is arrested very
quickly after alighting on the stigmatic surface and is characterised by pollen-tube
tip arrest and massive callose occlusion of the pollen grain.
Genes controlling SI are subject to negative frequency dependent selection (FDS)
where rare alleles are at a selective advantage and are maintained in plant populations.
This has led to the evolution of a highly polymorphic series of alleles. According to
theory (Wright 1939; Castric andVekemans 2004) S-alleles in single gene systems are
expected to be subject to balanced selection, where, over time, new alleles accumulate
until their frequency is equal to all other alleles in the population.
In the grasses there are only two published examples of estimates of the number
and frequency of S- and Z-alleles in populations, both having been documented in
perennial ryegrass (Lolium perenne L.). One of these is of a natural hay meadow
population (Fearon et al. 1994) and the other is of a synthetic cultivar (Devey et al.
1994). Both cases were characterised by having large numbers of both S and Z alleles
but with markedly different allele frequencies.
It is an extremely laborious process to genotype SI loci in plants possessing
multi-locus SI systems by intra-population pollination tests. But with the advent of
technology to easily and cheaply generate polymorphic molecular marker data, and
the development of physical marker maps for both S and Z loci, screening for allelic
variation at grass SI loci is made easier.
We are endeavouring to develop polymorphic molecular markers, closely linked
to S and Z that can be used to predict S and Z genotypes and enable us to study S
and Z allele diversity in Lolium perenne breeding populations. This will enable us
to gather and analyse more empirical data to help us to understand the population
genetics of SI in multi-locus systems and compare it to theoretical expectations
based on FDS. We are developing a marker screening strategy using a Lolium perenne
population produced at the Institute of Biological, Environmental and Rural Sciences,
Aberystwyth University from nine base plants over five cycles of mixed selection
followed by seven cycles of multiple half-sib family selection. Such a selection
procedure is likely to lead to significant fixation of neutral alleles. However, SI allele
diversity might be expected to be maintained as it is influenced by negative FDS.
6 Population Genetics of the Grass Self-incompatibility System—Practical . . . 57

The aim of our study, using a ‘test’ population was to be able to assess Z region
haplotype or haplotype-combination diversity and ultimately predict Z genotypes
of individuals and allele numbers and frequencies in our test population. We were
particularly intrigued by the possibility of using HRM curves to differentiate PCR
products to enhance our ability to distinguish polymorphic regions of our DNA
region of interest, both simply and cheaply.

6.2 Materials and Methods

We studied 55 inter-related plants which were the half-sib progeny of four mother
plants that were derived from twelve previous generations of mixed (first five gen-
erations) and half-sib family (next seven generations) selection. A single genotype
progenitor of a top-cross between a competitor’s commercial variety and AberDart
was added to the mother plants at the 9th and another at the 10th generation. In total,
the breeding population was derived from nine genotypes. Plant genotyping was
done using twelve markers covering a physical genomic distance of approximately
200 Kb flanking the Z locus. Two of these markers were STS markers, that showed
length polymorphisms during capillary electrophoresis and ten were PCR-amplified
markers that revealed contrasting high resolution DNA melting (HRM) curves. The
twelve markers covered a physical genomic region of 200 kb around the Z-locus of
Lolium perenne.
For the capillarity electrophoresis, PCR reactions were made on fluorescent
primers, the fluorescent products of which were detected using the ABI3130, Ap-
plied Biosystems, Foster City, CA, USA system. For HRM analysis the PCR products
are heated causing the dissociation of the double strand DNA, accompanied by a
decrease in fluorescence. The degree of fluorescence is recorded continuously over
time and relates to the rates of double strand dissociation. Fluorescence curves are
produced that vary depending on the molecular structure of the DNA fragments, in-
cluding the number of polymorphisms present in the PCR amplicon, fragment length
and base-pair constitution. Such melt curves will therefore be able to resolve any type
of polymorphism due to genotypic differences. For the HRM, PCRs were performed
according to Studer et al. (2009). In short, master mix containing the fluorescent dye
LCgreen (1X LightScanner Master Mix, Idaho Technology Inc., Salt Lake City, UT,
USA) was used.

6.3 Results

The two STS markers were genotyped based on fragment size differences revealed
using Genemapper software (example illustrated in Fig. 6.1a). With three exceptions,
genotypes could be allocated to each plant for the STS2 marker. In contrast, HRM
markers revealed a complete data set, as illustrated for the marker HRM8 in Fig. 6.1b.
58 C. Manzanares et al.

Fig. 6.1 Comparison of the result formats between STS marker (a) and HRM markers (b).
a represents the allele size and intensity of the marker STS 1 for 3 different plants of the population
using Genemapper 3.7. b represents the results given by the LightScanner® System software for
the marker HRM8. Each coloured curve represents one genotype (haplotype combination)

In order to estimate the total number of genotypes at Z, genotyping data for each
HRM and STS marker were grouped according to their melting curve profiles and
STS genotypes, respectively, and we identified a total of 13 haplotype combinations
(genotypes) (Table 6.1).
Some markers were more discriminatory than others. The most discriminatory
markers were STS2, HRM8 and HRM9 where, in combination, all the 13 genotypes
could be distinguished (Table 6.1). Moreover, one recombination event within the
200 Kb flanking the Z locus was detected in the breeding population.
6 Population Genetics of the Grass Self-incompatibility System—Practical . . . 59

Table 6.1 Classification of

the 55 plants into different HRM8 STS2 HRM9
groups according to HRM A aa A
and STS markers. We have A aa A
represented the groupings
1 A aa A
A aa A
based on HRM markers 8 and I aa A
9 and the STS2 marker. Other I aa A
markers followed similar 2 I aa A
patterns but, in almost all I aa A
cases, were less I A
discriminatory. The colours in F aa E
the table indicate the different 3 F aa E
genotype for each marker but
F aa E
H ab A
there is no relation between H ab A
colours of different markers. 4 H ab A
The capital letters H ab A
differentiate genotypes, lower G ab F
case letters indicate alleles G F
G ab F
G ab F
5 G ab F
G ab F
G ab F
G ab F
G ab F
A ac A
6 A ac A
C ad A
C ad A
7 C ad A
C ad A
C ad A
D ad B
D ad B
8 D
D ad
B
B
D ad B
A bb A
9 A bb A
E bb D
E bb D
10 E bb D
E bb D
B bd A
1 B
B
bd
bd
A
A
A dd B
A dd B
A dd B
1 A dd B
A dd B
2 A dd B
A dd C
13 A dd C
A dd C
60 C. Manzanares et al.

6.4 Discussion

In a pilot attempt to genotype the Z SI locus we investigated twelve genetic markers

covering a 200 Kb stretch of DNA around the Z locus. We estimate that, in generating
our population of 55 plants, there have been around 4,000 opportunities for recombi-
nation events to take place over the thirteen generations of selection. In independent
mapping studies with an unrelated mapping population we have screened over 5,000
plants for markers in the same region and have rarely found recombinants between
the most distant markers. Therefore we speculate that the genotypes we have iden-
tified are unlikely to have been created subsequent to a recombination event. These
genotypes therefore represent either the genotypes of the original parents of the first
cross, or genotypes that have been ingressed either deliberately (one genotype at F9
and one at F10) or by accident.
We used two STS markers that discriminated alleles based on DNA fragment
lengths. However we are aware that such markers will not reveal polymorphisms
that are based on base pair constitution (e.g. SNP variation) rather than base pair
number. The analysis of PCR products by HRM technology has a far greater dis-
criminatory power as fluorescence curves are influenced by base-pair constitution
and even epigenetic factors such as DNA methylation.
Our HRM marker genotypes matched with our STS markers in complete linkage
disequilibrium. But further, they were more discriminatory: we were able to separate
a group of plants into three genotype groupings where the two STS markers could
only distinguish two. We therefore suggest that markers based on HRM can provide
a useful tool for genotyping plants, maximising allelic diversity revealed.
With thirteen different Z genotypes revealed, we suggest that a minimum of five
different Z alleles are present. The frequencies of individual Z genotypes (and alleles)
are significantly different, most likely due to founder effects and genetic drift, yet Z
allele diversity alone (regardless of what the situation for the S locus is, and we do
not expect it to be very different!) in this population is sufficient to allay fears of a
limitation imposed on intra-population-compatibility and seed production potential.
We are in a position to predict the Z genotypes of our 55 plants but we need
to be able to empirically test these predictions by carrying out semi-in-vivo test
pollinations. However, as the SI reaction is dependent on complementary action of
the S locus, this needs to be done in combination with knowledge of the S genotypes
of our plants. We plan to use a similar screening approach to analyse S locus genotype
variation.

References

Castric, V, Vekemans X (2004) Plant self-incompatibility in natural populations: a critical

assessment of recent theoretical and empirical advances. Mol Ecol 13(10):2873–2889
Devey F, Fearon CH, Hayward MD, Lawrence MJ (1994) Self-incompatibility in ryegrass. 11.
Number and frequency of alleles in a cultivar of Lolium perenne L. Heredity 73:262–264
6 Population Genetics of the Grass Self-incompatibility System—Practical . . . 61

Fearon CH, Cornish MA, Hayward MD, Lawrence MJ (1994) Self-incompatibility in ryegrass.10.
Number and frequency of alleles in a natural-population of Lolium perenne L. Heredity 73:254–
261
Franklin-Tong VE (2008) Self-incompatibility in Flowering Plants: Evolution, Diversity, and
Mechanisms. Springer-Verlag, Cambridge
Studer B, Jensen LB, Fiil A, Asp T (2009) “Blind” mapping of genic DNA sequence polymorphisms
in Lolium perenne L. by high resolution melting curve analysis. Mol Breeding 24:191–199
Wright S (1939) The distribution of self-sterility alleles in populations. Genetics 24(4):538–552
Chapter 7
Use of Molecular Marker Information
in the Construction of Polycrosses to Enhance
Yield in a Lolium perenne Breeding Programme

A. Ghesquiere, H. Muylle and J. Baert

Abstract The aim of this study is to evaluate the use of molecular marker informa-
tion in the construction of polycrosses to enhance yield in a diploid Lolium perenne
breeding programme. The starting material used for this study consisted of 30 diploid
late perennial ryegrass plants selected from advanced breeding material. Pairwise
Jaccard similarities among these 30 genotypes were determined on the basis of 559
AFLP markers. Jaccard similarities varied between 0.334 and 0.686, with a mean
value of 0.467. Nine polycrosses (PC) of four parental plants with different levels
of genetic similarity were composed in 2007: three wide PC with low Jaccard simi-
larities between parental plants (0.334–0.447), three medium PC with intermediate
Jaccard similarities (0.405–0.540) and three narrow PC with high Jaccard similarities
(0.523–0.686). Some parental plants were used in more than one polycross. The seed
was harvested in 2008 and a progeny evaluation trial was sown in April 2009. The
dry matter yield was measured in the growing season of 2010 and will be measured
in 2011. Progenies from medium and wide polycrosses were on average significantly
higher yielding than the progenies from the narrow polycrosses (yield increase of
4.2 % and 6.1 % respectively) in 2010. There was a trend but not significant that
the yield of the progenies of the same parental plant from wider polycrosses was
higher compared to progenies from narrow polycrosses. We may conclude that the
application of molecular markers to select genetically diverse polycross parents can
result in an average yield increase.

7.1 Introduction

Perennial ryegrass is a wind-pollinated species and cultivars are usually produced

through random mating of selected parental plants resulting in population-based
synthetic cultivars. A high level of heterozygosity should be obtained to maximize
the agronomic performance. The parental plants are usually selected based on phe-
notypic characteristics determined by visual scoring of desirable traits. Molecular
markers offer powerful tools for the analysis of genetic diversity independent of the

A. Ghesquiere () · H. Muylle · J. Baert

Institute for Agricultural and Fisheries Research (ILVO), Caritasstraat 21,
9090 Melle, Belgium,
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 63

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_7, © Springer Science+Business Media Dordrecht 2013
64 A. Ghesquiere et al.

Table 7.1 Parental component number and pairwise Jaccard similarity coefficients between the 4
parental clones used in nine polycrosses (PC)
PC number PC type Comp. Number Jacc. sim. coeff.
1 2 3 4 Mean Range
1 Wide 2 3 12 15 0.386 0.343–0.435
2 Wide 3 20 26 28 0.389 0.334–0.429
3 Wide 2 11 19 28 0.410 0.388–0.447
4 Medium 10 11 29 30 0.463 0.422–0.534
5 Medium 5 8 9 12 0.463 0.405–0.502
6 Medium 13 19 24 25 0.463 0.412–0.540
7 Narrow 9 11 17 27 0.570 0.523–0.651
8 Narrow 10 14 15 16 0.581 0.549–0.686
9 Narrow 9 10 13 14 0.637 0.566–0.686

environment. The use of molecular marker information in the parental selection may
optimise the parental combinations to maximize heterosis. According to Kölliker
et al. (2005) application of molecular markers to select genetically diverse polycross
parents can result in an average yield increase. In this study molecular markers were
used to compose polycrosses with different levels of genetic diversity. The progenies
are being evaluated in a yield trial.

7.2 Material and Methods

The starting material used for this study consisted of 30 genotypes of late heading
diploid perennial ryegrass selected from advanced breeding material of the ILVO
breeding programme. Leaf material of all 30 parental plants was harvested for DNA
extraction in March 2007. The AFLP marker information was generated according
to Vandepitte et al. (2009). Each individual was profiled with six primer combina-
tions to generate polymorphic bands. For comparing pairs of individual plants, the
Jaccard similarity was calculated using information of 559 polymorphic AFLP mark-
ers. Twenty-two genotypes were retained on the basis of molecular marker diversity,
to compose nine polycrosses (PC) of four parental plants with different levels of
genetic similarity: three wide PC with low Jaccard similarities, three medium with
intermediate similarities and three narrow with a high similarity. Some parental plants
were clonally divided and used in more than one polycross. In none of the nine poly-
crosses half-sibs were represented. Table 7.1 shows the parental clone numbers and
Jaccard similarities between the clones.
Halfsib seeds were harvested on each parental genotype in July 2008. For yield de-
termination a progeny evaluation trial was established in April 2009 in a randomized
block design with three replicates. Each of the halfsib families was sown.
The standard varieties in the trial were Aberavon, Cancan and Melways. The trial
was cut five times in 2010 and the DMY was measured. The DMY will be determined
again in 2011.
7 Use of Molecular Marker Information in the Construction of Polycrosses to . . . 65

Fig. 7.1 Frequency distribution of the mean Jaccard similarity coefficient of all possible polycrosses
of four parental clones selected from the 30 original genotypes

7.3 Results and Discussion

7.3.1 Genetic Similarity

Pairwise Jaccard similarities among the 30 genotypes were determined on the basis of
559 AFLP markers. Jaccard similarities varied between 0.334 and 0.686, with a mean
value of 0.467. The mean Jaccard similarity for all possible combinations of 4 parental
clones was calculated and varied between 0.386 and 0.637. This molecular marker
data was used to construct the nine polycrosses described in Table 7.1. Figure 7.1
shows the frequency distribution of the mean Jaccard similarity coefficient of all
possible polycrosses of four parental clones.

7.3.2 Dry Matter Yield

On average, progenies from medium and wide polycrosses were significantly

(P > 0.05) higher yielding than the progenies of the narrow polycrosses in 2010.
There was a yield increase of 4.2 % and 6.1 % respectively. Figure 7.2 shows the
total DMY in 2010 of the nine polycrosses. The wide PC2 and PC3 and the medium
PC4 were the best performing polycrosses in the trial and were significantly higher
yielding than the narrow polycrosses. The narrow PC 9 had the lowest DMY of all
polycrosses.
Some parental clones were used in more than one polycross (Table 7.1). Seven of
these clones were included in polycrosses with a different type of molecular diversity
(wide, medium, narrow). There was a trend but not significant that the yield of the
progenies of the same parental plant from wider polycrosses was higher compared
66 A. Ghesquiere et al.

Fig. 7.2 Total dry matter yield (relative to the mean of the total DMY of the standard varieties)
of the nine polycrosses composed with different levels of genetic similarity: wide, medium and
narrow and of the standard varieties. Error bars indicate the highest and the lowest DMY of the
four halfsibs of the PC and the highest and the lowest DMY of the three standard varieties

Table 7.2 Differences of the total dry matter yield (relative to the mean of the total dry matter
yield of the standard varieties) of the progenies of the same parental plant used in different types
of polycrosses
Clone number Wide PC Medium PC Narrow PC
9 −6.8 −8.9
10 +3.8 −0.2
11 +0.9 +1.0 −2.3
12 −1.6 −2.8
13 −6.4 −10.7
15 −3.3 −5.6
19 −2.4 −4.8
Average −1.6 −2.7 −5.5

to progenies from narrow polycrosses. Table 7.2 shows the differences of the DMY
of the progenies of the seven clones that were used in different types of polycrosses.
On average the DMY of progenies of the parental clones used in wider polycrosses
was 3.9 % higher than when used in narrow polycrosses. Clone 10 is performing
better than most of the other clones. Clone 10 is being used in PC 4, 8 and 9. The
high variation in total DMY between the half sibs in PC 8 and 9 (Fig. 7.2) can be
explained by the good general combining ability of clone 10. In PC 9 the progenies
of clone 10 are significantly higher yielding than the other half sibs in the PC. Clones
7 Use of Molecular Marker Information in the Construction of Polycrosses to . . . 67

with a very good general combining ability can result in high performing progenies
even when they are used in narrow polycrosses.

7.4 Conclusion

The total dry matter yield of the first production year was higher for the wide and
medium polycrosses when compared to those of the narrow polycrosses. If the results
of 2011 confirm the first year results, we may conclude that molecular markers can
be used for the selection of genetically diverse polycross parents in the breeding
programme of perennial ryegrass. This can result in an increase of dry matter yield
of the synthetic cultivar without compromising phenotypic uniformity.

References

Kölliker R, Boller B, Widmer F (2005) Marker assisted polycross breeding to increase diversity and
yield in perennial ryegrass (Lolium perenne L.). Euphytica 146:55–65
Vandepitte K, Roldan-Ruiz I, Leus L, Jacquemyn H, Honnay O (2009) Canopy closure shapes clonal
diversity and fine-scale genetic structure in the dioecious understorey perennial Mercurialis
perennis. J Ecol 97:404–414
Chapter 8
An Analysis of Chromosome Pairing Behaviour
in Newly Synthesized Alfalfa Tetraploids
by Means of SSR Markers

D. Rosellini, N. Ferradini, S. Allegrucci, A. Nicolia and F. Veronesi

Abstract A duplication of a species’chromosomes results in the formation of a poly-

ploid with polysomic inheritance, or autopolyploid, while the union of the genomes
of different species results in the formation of a polyploid with disomic inheri-
tance, or allopolyploid. Cultivated alfalfa shows polysomic (tetrasomic) inheritance;
however, no information of chromosome pairing behaviour is available for newly
tetraploidized M. sativa. We are studying two tetraploid plants obtained by bilateral
sexual polyploidization, that is, by crossing a diploid Medicago sativa subsp. falcata
plant that produces 2n eggs (PG-F9) with a 2x Medicago sativa. subsp. coerulea x
falcata plant that produces 2n pollen (12P). We are employing SSR markers to inves-
tigate the chromosome pairing behaviour of these two plants. They were crossed with
an unrelated tetraploid pollen donor, and parental SSR allele segregation patterns are
examined in the two progenies. Our results so far, indicate that random pairing, and
consequently tetrasomic inheritance, is the rule in newly tetraploidized M. sativa.

8.1 Introduction

Polyploidization is an increase in genome number and occurs in nature as the conse-

quence of the union of 2n gametes (sexual polyploidization), or as the consequence of
somatic genome duplications (somatic polyploidization). A duplication of a species’
chromosomes results in the formation of a polyploid with polysomic inheritance,
or autopolyploid, while the union of the genomes of different species results in the
formation of a polyploid with disomic inheritance, or allopolyploid. Some poly-
ploids have both modes of inheritance (segmental allopolyploids). Cultivated alfalfa
(Medicago sativa L. 2n = 4x = 32) shows polysomic (tetrasomic) inheritance (Quiros
1982; Julier et al. 2003).
Cultivated tetraploid (4x) alfalfa likely derives from wild diploid (2x) M. sativa
through sexual polyploidization. Whether a newly formed polyploid behaves like
a polysomic or a disomic polyploid depends on the genotypes involved. When the
union of 2n gametes occurs between genetically distant types, the two homologous

D. Rosellini () · N. Ferradini · S. Allegrucci · A. Nicolia · F. Veronesi

Department of Applied Biology, University of Perugia, Borgo XX Giugno,
74, 06121 Perugia, Italia
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 69

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_8, © Springer Science+Business Media Dordrecht 2013
70 D. Rosellini et al.

Table 8.1 Informative SSR markers

Marker Chromo- Primer sequences Annealing
some
FMT13 1 GATGAGAAAATGAAAAGAAC CAAAAACTCACTCTAACACAC 50
MTIC451 2 GGACAAAATTGGAAGAAAAA AATTACGTTTGTTTGGATGC 55
MTIC332 4 CCCTGGGTTTTTGATCCAG GGTCATACGAGCTCCTCCAT 60

chromosome from one parent may pair preferentially (disomic behaviour). On the
contrary, if the 2n gametes come from similar genotypes the four ‘homologs’ tend
to pair at random (polysomic behaviour); pairing behaviour may also vary among
chromosomes.
We are studying two tetraploid (4x) plants obtained by crossing a diploid (2x) M.
sativa subsp. falcata plant that produces 2n eggs (PG-F9) with a 2x M. sativa. subsp.
coerulea x falcata plant that produces 2n pollen (12P). We are employing previously
mapped SSR markers to investigate the chromosome pairing behaviour of these two
plants.

8.2 Materials and Methods

PG-F9 and 12P were hand crossed without emasculation and the progeny obtained
screened for the 4x state by root tip chromosome count. Two 4x plants, named
S29 and S48, derived by bilateral sexual polyploidization, were randomly selected
and crossed without emasculation with a plant from the Italian variety Classe (4x)
used as pollen donor. Progeny seeds were germinated on Petri plates and the plants
reared in a greenhouse. The genomic DNA was extracted (GenElute Plant Genomic
DNA Miniprep Kit—SIGMA) from 50 plants of the S29 x Classe and S48 x Classe
progenies, and from the three parental plants (PG-F9, 12P and Classe).
Eighteen SSR markers were selected from the published literature (Diwan et al.
2000; Julier et al. 2003; Sledge et al. 2005; Mun et al. 2006), on the basis of chro-
mosome location. SSR amplification was performed as follows: buffer 1X, MgCl2
1,5 mM, dNTP 0,2 mM, primer FOR/REV 0,5 μM, Taq (SIGMA) 1U, genomic DNA
30 ng, in 20 μl final volume. PCR cycling was 94 ◦ C 3 min, 40 cycles at 94 ◦ C
30 s, Ta ◦ C 30 s, 72 ◦ C 30 s (See Table 8.1 for the temperatures of annealing, Ta).
After screening in agarose, fluorescinated primers were used for amplification of the
three parental plants and the two progenies (43 plants each) with 8 selected primer
pairs, and capillary electrophoresis was performed (3130x Genetic Analyzer, Applied
Biosystem). Electrophoretic data were analyzed using the software Gene Mapper 4
(Applied Biosystem). SSR allele segregation was compared with those expected
with tetrasomic or disomic inheritance and X2 values calculated. Bivalent pairing is
assumed in this work (Ma et al. 2002), so the occurrence of double reduction was
not considered; in any case, it was found to be limited (Julier et al. 2003).
8 An Analysis of Chromosome Pairing Behaviour in Newly Synthesized . . . 71

Table 8.2 Allelic patterns of three SSR markers in the 2x and 4x plants studied. Numbers are allele
sizes in bp
SSR Markers Alleles Plants
PG-F9 (2x) 12P (2x) Classe (4x) S29 (4x BSP) S48 (4x BSP)
FMT13 1 144 144 144 144
2 147 147 147 147
3 152 152 152
4 156
5 166
MTIC332 1 110
2 123
3 127 127 127 127
4 128 128 128
5 129 129
6 136
7 138 138 138
MTIC451 1 124 124
2 130 130 130
3 134
4 136 136 136
5 138 138 138 138

8.3 Results and Discussion

Five to eight alleles per marker were counted for the 8 selected markers; the two
diploid parents were found to be genetically different, as demonstrated by the low
number of common alleles (3 in 25, data not sown). Informative polymorphisms were
found so far for only 3 of the 8 selected markers, distributed on 3 chromosomes. Two
were useful for S29 and three for S48 (Table 8.2).
Considering, for example, the PG-F9 alleles 1 and 4 of marker MTIC451 in the
plant S48 (Table 8.2), a gametic segregation 1/6(M1 M4):2/6(M1 -):2/6(- M4):1/6(- -)
is expected with complete random pairing, whereas a segregation 1/2(M1 -):1/2(M4 -)
is expected with complete preferential pairing (Fig. 8.1). Two allelic configurations
of the S29 and S48 4x plants were useful for estimating chromosome pairing be-
haviour: (1) two PG-F9-specific alleles, or (2) one homozygous PG-F9-specific
allele (Table 8.2). Alleles of 12P were not informative, because they were shared
with Classe, with only three exceptions. This was not completely unexpected, be-
cause 2x M. coerulea (parent of 12P) is the prevalent wild ancestor of cultivated M.
sativa.
Three PG-F9 and two 12P alleles were not transmitted to S29; two PG-F9 and two
12P alleles were not transmitted to S48 (examples in Table 8.2, markers MTIC332
and MTIC 451). This is likely due to homozygous gamete formation, that occurs
when the mechanism of 2n gamete formation is second division restitution (SDR)
without crossing-over between the centromere and the locus, or when, in case of
first division restitution, a crossing-over occurs (Bingham 1980). In PG-F9, the SDR
mechanism of 2n gamete formation has been observed (Tavoletti 1994).
72 D. Rosellini et al.

Fig. 8.1 Scheme of the formation of a tetraploid through sexual polyploidization, and of segregation
of two alleles from one parent (disregarding those from the other parent) in the tetraploid meiosis
under complete preferential or random pairing models

Complete preferential pairing did not occur, as demonstrated by the presence of

all 4 gamete types for markers MTIC332 and MTIC451 (Table 8.3). X2 calcula-
tions indicate tetrasomic inheritance in 4 out of 5 cases; only chromosome 4 in the
S48 progeny appeared to deviate from complete random pairing, due to lower than
expected ‘M4M7’ and higher than expected ‘- -’ gamete types (marker MTIC332,
Table 8.3). However, this result derives from segregation distortion, not from prefer-
ential pairing. The disomic inheritance hypothesis could only be tested by X2 for the
FMT13 marker, because two gamete types are expected, but at different frequencies
between the two pairing behaviours. The X2 values were highly significant, thus
rejecting the disomic hypothesis.
Significant preferential chromosome pairing was evidenced in a mapping study
performed using a population obtained by crossing M. falcata × M. sativa tetraploid
genotypes (Ma et al. 2002). Our results involve less distant, but still different geno-
types; however, maybe surprisingly, our data are more in line with those of Julier
et al. (2003) that did not support preferential pairing in cultivated M. sativa.
8 An Analysis of Chromosome Pairing Behaviour in Newly Synthesized . . . 73

Table 8.3 X2 analysis of disomic vs tetrasomic segregation in the progenies of two newly tetraploi-
dized plants
Plant, Chromosome, Gamete Numbers of progenies X2a
Marker types
Observed Expected Expected Disomic Tetraso-
disomic tetrasomic mic

S29, Chromosome 1, FMT13 M3M3, M3 - 38 21 35 27.52 ** 1,54

-- 4 21 7
S29, Chromosome 4, MTIC332 M4M7 4 0 7
M4 - 17 21 14 NT 1.93
- M7 14 21 14
-- 7 0 7
S48, Chromosome 1, FMT13 M3M3, M3 - 34 21 35 16.10** 0,17
-- 8 21 7
S48, Chromosome 2, MTIC451 M1M4 4 0 7 NT 1,57
M1 - 15 21 14
- M4 15 21 14
-- 8 0 7
S48, Chromosome 4, MTIC332 M4M7 2 0 7.2 NT 8,49*
M4 - 14 21.5 14.3
- M7 14 21.5 14.3
-- 13 0 7.2
NT non testable by X2 because of 0 expectation for some gamete types
*significant at P ≤ 0.05; **significant at P ≤ 0.01
X for 1 df and P = 0.05 is 3.84; X2 for 3 df and P = 0.05 is 7.81
a 2

8.4 Conclusions

Considerable genetic differences between the two diploid parental genomes of the
two newly tetraploidized plants studied here can be assumed based on genealogy and
were confirmed by SSR allelic diversity. Despite this, we obtained evidence that, once
merged in a new tetraploid individual, the four homologous chromosomes pair at
random, and consequently the new tetraploids originated by sexual polyploidization
have tetrasomic inheritance. This indicates that, in alfalfa, chromosome pairing is
not influenced significantly by genetic differences between subspecies.
Alfalfa belongs to the so-called M. sativa complex that includes 2x M.coerulea,
2x and 4x M.sativa ssp. falcata, and 4x M. sativa ssp. glutinosa. Our data suggest
that, even at the very moment of tetraploidy acquisition by sexual polyploidization,
genomic divergence within this complex does not result in significant levels of prefer-
ential chromosome pairing and disomic inheritance. We are completing this analysis
with markers for all the chromosomes.

Acknowledgments Funding was provided by the Italian Ministry of University and Science,
project: Sexual polyploidization in alfalfa: insight into gene expression changes in newly
synthesized tetraploids, PRIN 2008 (prot. 2008AEAXRK_001, PI Fabio Veronesi).
74 D. Rosellini et al.

References

Bingham ET (1980) Maximizing heterozygosity in autotetraploids. In: Lewis WH (ed) Poliploidy—

Biological Relevance. Plenum Press, New York, pp 471–489
Diwan N, Bouton JH, Kochert G, Cregan PB (2000) Mapping of simple sequence repeat (SSR)
DNA markers in diploid and tetraploid alfalfa. Theor Appl Genet 101:165–172
Julier B, Flajoulot S, Barre P, Cardinet G, Santoni S, Huguet T, Huyghe C (2003) Construction of
two genetic linkage maps in cultivated tetraploid alfalfa (Medicago sativa) using microsatellite
and AFLP markers. BMC Plant Biol 3:9
Ma CX, Casella G, Shen ZJ, Osborn TC, Wu RL (2002) A unified framework for mapping quanti-
tative trait loci in bivalent tetraploids using single-dose restriction fragments: A case study from
alfalfa. Genome Res 12:1974–1981
Mun J-H, Kim D-J, Choi H-K, Gish J, Debellé F, Mudge J, Denny R, Endré G, Saurat O, Dudez
A-M, Kiss GB, Roe B,Young ND, Cook DR (2006) Distribution of microsatellites in the genome
of Medicago truncatula: A resource of genetic markers that integrate genetic and physical maps.
Genetics 172:2541–2555
Quiros C (1982) Tetrasomic segregation for multiple alleles in alfalfa. Genetics 101:117–127
Sledge MK, Ray IM, Jiang G (2005) An expressed sequence tag SSR map of tetraploid alfalfa
(Medicago sativa L.). Theor Appl Genet 111:980–992
Tavoletti S (1994) Cytological mechanisms of 2n eggs formation in a diploid genotype of Medicago
sativa subsp. falcata. Euphytica 75:1–8
Chapter 9
Genome Constitution in Selected and Unselected
Plants of F2 –F4 Generations Derived
from an Allotetraploid Festuca pratensis ×
Lolium perenne Hybrid

Z. Zwierzykowski, T. Ksia˛żczyk, M. Taciak, E. Zwierzykowska,

N. Jones and A. Kosmala

Abstract The objective of this work was to assess the genomic constitution and
intergeneric recombination in three successive unselected generations, (F2 −F4 ), de-
rived from an intergeneric hybrid between Festuca pratensis Huds. (2n = 4x = 28)
and Lolium perenne L. (2n = 4x = 28). Examination based on genomic in situ
hybridization analyses of randomly chosen plants in each generation indicated pro-
gressive changes in genome balance towards that of Lolium. The dominance of
Lolium chromatin over Festuca likely resulted from extensive recombination be-
tween chromosomes of the parental genomes, together with substitutions of whole
Festuca chromosomes by whole Lolium chromosomes. The total number of Lolium
chromosomes increased from generation to generation. The number of recombi-
nant chromosomes, and recombination breakpoints per genotype, also increased in
successive generations, but their number was higher for Festuca than for Lolium.
The patterns of genome constitutions and recombination were similar to those we
observed in selected generations (F2−F4 breeding populations) developed from the
same F1 hybrid plants.

Keywords Allotetraploid · Genome constitution · GISH · Festulolium · Festuca

pratensis × Lolium perenne · Recombination

Z. Zwierzykowski () · T. Ksia˛żczyk · M. Taciak · E. Zwierzykowska · A. Kosmala

Institute of Plant Genetics of the Polish Academy of Sciences,
Strzeszyńska 34, 60–479 Poznań, Poland
e-mail: [email protected]
N. Jones
Organization Institute of Biological, Environmental and Rural Sciences,
Aberystwyth University, Aberystwyth, Edward Lywyd Building,
Penglais Campus, SY23 3DA, UK

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 75

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_9, © Springer Science+Business Media Dordrecht 2013
76 Z. Zwierzykowski et al.

9.1 Introduction

Festulolium hybrid cultivars that combine high yield and quality of ryegrasses
(Lolium multiflorum and L. perenne) with the persistence, and abiotic and biotic
stress resistance of fescues (Festuca pratensis and F. arundinacea), are considered
ideal components of agricultural and turf-grass systems (Ghesquière et al. 2010).
The development of cultivars derived from synthetic amphiploids (alloploids) has
been limited in the past because homoeologous chromosome pairing can lead to
aneuploidy and genetic instability in advanced generations. Relatively low seed fer-
tility, and its high level of variation, observed in advanced breeding populations of
Festulolium, are reasons why such amphiploids cannot be included successfully in
breeding programmes. Recently, cytogenetic investigations of genome architecture
of Festulolium cultivars, using genomic in situ hybridization (GISH), have demon-
strated extensive homeologous recombination in the following populations developed
from two allotetraploid (2n = 4x = 28) hybrids—F. pratensis × L. multiflorum and
F. pratensis × L. perenne (Zwierzykowski et al. 1998, 2006, 2011; Canter et al.
1999; Kopecký et al. 2006). In our previous research, we studied the genome con-
stitutions and dynamics of intergeneric recombination in breeding populations of
F2 −F8 generations developed from F. pratensis (4x) × L. perenne (4x) hybrids
(Zwierzykowski et al. 2006, 2011). Extensive recombination between chromosomes
of the two parental genomes, as well as substitution of whole Festuca chromosomes
by whole Lolium chromosomes, was observed in these progeny. In this work, we
present preliminary results on genome constitution and recombination in unselected
materials of F2−F4 generations derived from the tetraploid F. pratensis (4x) × L.
perenne (4x) hybrid, and compare the results with those obtained in breeding mate-
rials of F2 −F4 generations (developed from the same F1 hybrid plants), selected for
agronomic traits, including vigor, winter hardiness, drought tolerance and fertility
(Zwierzykowski et al. 2006).

9.2 Materials and Methods

9.2.1 Plant Materials

Tetraploid hybrids of F. pratensis (Fp) × L. perenne (Lp) (2n = 4x = 28) were gen-
erated by intercrossing autotetraploid forms of both species (Zwierzykowski et al.
2006). Five F1 partially male and female fertile hybrids were intercrossed under con-
trolled conditions, and F2 progeny were produced. In each of the F2 −F4 generations
studied, 150 randomly collected genotypes were grown in control conditions in a
glasshouse. Thirty randomly chosen plants, taken from each generation, were used
for cytogenetic analysis.
9 Genome Constitution in Selected and Unselected Plants F2 −F4 Generations . . . 77

9.2.2 Genomic in Situ Hybridization (GISH)

To determine the number and genomic structure of chromosomes, root-tip ‘spreads’

were prepared according to Zwierzykowski et al. (1998), and GISH was performed
according to Zwierzykowski et al. (2006). To discriminate between Lp and Fp chro-
mosomes, total genomic DNA of Lp cv. Solen was used as a probe and labeled with
digoxigenin-11-dUTP by a standard commercially available ‘nick translation’ pro-
tocol. Total genomic DNA of Fp cv. Westa was used as a blocker. The cytogenetic
analyses were performed to characterize: (i) the total number of chromosomes, (ii)
the number of parental chromosomes, (iii) the number of recombinant chromosomes,
and (iv) the number of recombination breakpoints.

9.3 Results and Discussion

In unselected plants of three successive generations of open pollination, progressive

changes in genome balance in favour of the ‘dominant’ the Lolium genome were ob-
served (Table 9.1). This genomic dominance of the Lolium chromatin resulted from
extensive recombination between chromosomes of the parental genomes, and from
a substitution of whole Festuca chromosomes by whole Lolium chromosomes. The
total number of Lolium chromosomes increased from a mean 14.17 in the F2 to 15.27
in the F4 , and the total number of Festuca chromosomes decreased correspondingly
from a mean of 13.66 to a value of 12.70. The number of recombinant chromosomes
and recombination breakpoints per genotype also increased from generation to gen-
eration, although the respective values of both characters were higher for Festuca
(1.20–4.47 and 2.60–6.77) than for Lolium (0.90–2.10 and 0.90–2.41).
The patterns of genome constitutions and intergeneric recombination were similar
to those we observed in F2 –F4 breeding populations, for example, the total number
of Lolium chromosomes increased from 14.36 in the F2 to 15.81 in the F4 , and the
number of recombination breakpoints per genotype also increased from generation
to generation, although the respective values of this character were higher for Festuca
(1.14–5.19) than for Lolium (0.68–4.04) (Table 9.1; Zwierzykowski et al. 2006).
The reasons for the dominance of Lolium over Festuca are not understood, al-
though it appears to operate in the same way in both the selected and unselected
populations. There is no evidence that gametic competition, pollination effects or
selection for vigour in the early stages of seedling growth can be the causative agents.
We are therefore left with a single line of conjecture that the process might involve dif-
ferences in centromere organization and competitiveness (Jones and Hegerty 2009).
Does centromere drive operate here, possibly at female meiosis, in a way we have
hitherto not even suspected; and are there any differences in the CENH3 genes,
or loading of the centromere-specific CENH3 histone variant which could underlie
such a process? Loading differences are known to be at the basis of uniparental
chromosome elimination in hybrids of Hordeum vulgare × H. bulbosum, rather than
silencing of CENH3 genes (Sanie et al. 2011). We have no answers to such questions
 78

Table 9.1 GISH analysis in selected and unselected plants of the F2 −F4 generations obtained from F. pratensis (4x) × L. perenne (4x) hybrids.
Data of selected plants from the F2 −F4 generations were published previously. (Zwierzykowski et al. 2006)

Generation Lolium chromosomes Festuca chromosomes

Total no. No. of No. of Total no. No. of No. of
of chromo- recombinant breakpoints of chromo- recombinant breakpoints
somes chromosomes somes chromosomes
Mean Range Mean Range Mean Range Mean Range Mean Range Mean Range
F2 Selected 14.36 14–16 0.68 0–4 0.68 0.4 13.57 12–14 0.86 0–4 1.14 0–6
Unselected 14.17 12–17 0.90 0–2 0.90 0–2 13.66 11–15 1.20 0–3 2.60 1–6
F3 Selected 14.77 13–17 2.00 0–5 2.23 0–6 13.27 11–15 2.85 1–5 3.77 1–7
Unselected 14.60 13–17 1.80 0–4 2.00 0–5 13.33 10–15 3.33 1–5 4.80 1–11
F4 Selected 15.81 13–20 3.46 1–5 4.04 2–6 12.12 8–15 3.89 2–6 5.19 2–10
Unselected 15.27 12–18 2.10 1–5 2.41 1–6 12.70 10–16 4.47 2–6 6.77 4–11
Z. Zwierzykowski et al.
9 Genome Constitution in Selected and Unselected Plants F2 −F4 Generations . . . 79

in Festulolium, neither do we know if the drift involves Lolium chromosomes in a

non-specific way. In any event the mechanism appears to be progressive by small
incremental shifts at each generation.

9.4 Conclusions

In each of the F2 −F4 generations of tetraploid hybrids of F. pratensis × L. perenne

(2n = 4x = 28) GISH analysis showed that the balance of chromatin changed.
There was a progressive shift in the balance of chromatin in favour of Lolium, and a
corresponding decline in that of Festuca. This dominance of Lolium chromatin over
Festuca appears to result from extensive recombination between chromosomes of
the parental genomes, together with substitutions of whole Festuca chromosomes
by whole Lolium chromosomes. The outcome of the experiments mirror that found
in previous work based on selected breeding materials of F2 − F4 generations. No
mechanism for this process of chromatin drift has yet been identified, but it is hy-
pothesized that it may be based on the differential activity of the centromeres of the
two genera concerned.

Acknowledgments This work was supported by the Polish Ministry of Science and Higher
Education (Grant no. N N310 090736).

References

Canter PH, Pašakinskienė I, Jones RN, Humphreys MW (1999) Chromosome substitutions and
recombination in the amphiploid Lolium perenne × Festuca pratensis cv Prior (2n = 4x = 28).
Theor Appl Genet 98:809–814
Ghesquière M, Humphreys M, Zwierzykowski Z (2010) Festulolium. In: Boller B, Posselt UK,
Veronesi F (eds) Fodder Crops and Amenity Grasses, Series: Handbook of Plant Breeding, Vol
5. Springer, Heidelberg, pp 288–311
Jones RN, Hegerty M (2009) Order out of chaos in the hybrid plant nucleus. Cytogenet Genome
Res 126:376–389
Kopecký D, Loureiro J, Zwierzykowski Z, Ghesquière M, Doležel J (2006) Genome constitution and
evolution in Lolium × Festuca hybrid cultivars (Festulolium). Theor Appl Genet 113:731–742
Sanei M, Pickering R, Kumke K, Nasuda S, Houben A (2011) Loss of centromeric histone H3
(CENH3) from centromeres precedes uniparental chromosome elimination in interspecific
barley hybrids. Proc Natl Acad Sci USA 108:E498–E505
Zwierzykowski Z, Tayyar R, Brunell M, Lukaszewski AJ (1998) Genome recombination in in-
tergeneric hybrids between tetraploid Festuca pratensis and Lolium multiflorum. J Hered
89:324–328
Zwierzykowski Z, Kosmala A, Zwierzykowska E, Jones N, Jokś W, Bocianowski J (2006) Genome
balance in six successive generations of the allotetraploid Festuca pratensis × Lolium perenne.
Theor Appl Genet 113:539–547
Zwierzykowski Z, Zwierzykowska E, Taciak M, Kosmala A, Jones RN, Zwierzykowski W,
Ksia˛żczyk T, Krajewski P (2011) Genomic structure and fertility in advanced breeding popula-
tions derived from an allotetraploid Festuca pratensis × Lolium perenne cross. Plant Breeding
130:476–480
Chapter 10
Estimation of Temporal Allele Frequency
Changes in Ryegrass Populations Selected
for Axillary Tiller Development

G. Brazauskas, I. Pašakinskienė and T. Lübberstedt

Abstract The effects of selection on allele frequency at five genes with puta-
tive effect on shoot morphology were examined in two perennial ryegrass (Lolium
perenne L.) populations undergoing recurrent selection. A synthetic C0 population
was created by intercrossing five unrelated ryegrass genotypes within the EU FP5
project ‘GRASP’. Two rounds of selection (both positive and negative) for axillary
tiller formation were performed, leading to selected populations C+ −
1 and C1 , respec-
tively. The mean number of axillary tillers per plant was 2.18, 3.90 and 0.22 for
C 0 , C+ −
1 and C1 , respectively. Five ryegrass genes putatively involved in the control
of plant architecture and hormone response were SNP genotyped in all three pop-
ulations. A test of selective neutrality (Waples’ test), which tests the hypothesis of
genetic drift versus selection, was applied. This test indicated selection for the gene
LpIAA1 in C− 1 , where allele frequency changes could not be explained by genetic
drift alone (p < 0.05). LpIAA1 belongs to a large family of genes, called Aux/IAA,
which comprises genes that are auxin-regulated and were shown to control shoot
morphology in Arabidopsis and rice.

10.1 Introduction

Shoot morphology as one of the main components of plant architecture plays an

important role in fodder grasses. The architecture of the shoot system affects the
light harvesting potential of plants, the synchrony of flowering and seed set, and

G. Brazauskas ()
Lithuanian Research Centre for Agriculture and Forestry, Institute of Agriculture,
Instituto a. 1, Akademija, 58344 Kedainiai, Lithuania
e-mail: [email protected]
I. Pašakinskienė
Faculty of Natural Sciences, Vilnius University,
M. K. Ciurlionio 21/27, 03101 Vilnius, Lithuania
T. Lübberstedt
Department of Agronomy, Iowa State University,
1204 Agronomy Hall, 50011 Ames, Iowa, USA

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 81

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_10, © Springer Science+Business Media Dordrecht 2013
82 G. Brazauskas et al.

ultimately the reproductive success of a plant. Growth and development in grasses

has been described by Langer (1979). Briefly, perennial ryegrass plants consist of
numerous basal shoots or tillers. During vegetative growth, each tiller is made up of
several leaves on a short stem. An axillary shoot meristem is present in the axil of
each leaf, and its outgrowth results in formation of a new daughter tiller. Daughter
tillers develop their own root systems and become independent from their mother
tiller allowing clonal growth as a means of vegetative reproduction. Leaf and tiller
production continues until plants switch to reproductive development in response
to specific environmental cues. Perennial forages have opposing selective pressures
placed upon them during plant breeding. Seed production practices favour genotypes
with high numbers of reproductive tillers, and good seed yields. However, nutritional
quality is reduced during flowering (Jung et al. 1996), so the end-user, the farmer,
requires varieties with fewer reproductive tillers. Effective regulation of resource
partitioning between vegetative and reproductive growth is important in perennial
ryegrass and other perennial forage grasses.
Selection mapping (SM) refers to a range of approaches that identify alleles, loci,
and epistatic interactions using populations that have been subjected to iterative cy-
cles of recombination and selection (Wisser et al. 2008). The effects of selection
can be seen as differences in allele frequency, diversity, and/or patterns of recom-
bination, through comparisons of temporally or spatially defined subpopulations. A
fundamental challenge in SM, however, is to differentiate the effects of selection
from those of genetic drift. Selection increases the frequencies of favourable alleles
while genetic drift is a random change in allele frequencies due to small population
size. A loss of favourable alleles due to random genetic drift leads to reduction in
genetic variance and, thus, limits future selection response (Guzman and Lamkey
1999). The assessment of the effects of random genetic drift and selection is impor-
tant for designing efficient recurrent selection (RS) programs. Standard statistical
tests (e.g. χ 2 , G tests) for assessing significant changes in allele frequencies ne-
glect the effects of random genetic drift and are, therefore, not appropriate for the
analysis of changes in allele frequencies in RS with finite population size. In con-
trast, Waples provided a test statistic for monitoring allele frequency changes, which
takes into account the increased variance in allele frequencies between generations
caused by random genetic drift. Up to now, only a limited number of studies utilized
Waples’ neutrality test (Waples 1989) in plant breeding with all instances coming
from studies on maize (Labate et al. 1999; Pinto et al. 2003; Coque and Gallais
2006; Falke et al. 2007; Wisser et al. 2008). Hence, a detailed evaluation of al-
lele frequency changes in populations undergoing RS in other plant species is still
lacking.
In the present study, we evaluated the selection response of two perennial ryegrass
populations after one cycle of recurrent selection on the basis of SNP markers in five
candidate gene regions. Our objective was to investigate allele frequency changes in
candidate regions due to the effects of random genetic drift and selection.
10 Estimation of Temporal Allele Frequency Changes in Ryegrass . . . 83

10.2 Materials and Methods

Five of the 20 perennial ryegrass (LTS) genotypes, studied in the EU FP5 project
‘GRASP’ (Posselt et al. 2006), were crossed in all possible pairwise combinations
including reciprocal crosses. Two of the five genotypes, LTS15 and LTS16, are eco-
types, two, LTS03 and LTS04, are parent genotypes of VrnA mapping population
(Jensen et al. 2005) and one, LTS11, is a ryegrass mutant expressing enhanced axil-
lary tillering (Pašakinskienė 2005). Equal amounts of seed from each combination
were sown in the greenhouse (single plant per pot) to form the synthetic C0 popu-
lation. 340 C0 plants were evaluated for axillary tiller development by counting the
average number of axillary tillers per 20 primary tillers at the beginning of flowering.
The 34 plants with the highest and the lowest average number of axillary tillers per
plant were selected to produce selected subpopulations C0 S+ and C0 S− , respectively.
These selected plants were intercrossed in isolated greenhouses by open pollination
to produce populations C+ −
1 and C1 , respectively. In order to synchronize flowering all
◦
plants were vernalized at 6 C with 8 h photoperiod per day for 100 days. Following
the same selection procedure as in C0 34 plants were selected in each C+ −
1 and C1 .
+ −
The selected plants from C1 and C1 along with 34 random plants from C0 R and five
LTS genotypes, were used for DNA extractions from young leaf tissue following the
CTAB protocol.
Five ryegrass genes with putative effects on shoot morphology and hormone re-
sponse were chosen as candidate genes for genotyping. Details on primer design and
SNP detection for these genes in 20 LTS genotpes were published earlier (Brazauskas
et al. 2010). The SNP selection procedure consisted of several steps. First, a minimum
number of SNP subsets was determined for each candidate gene to distinguish all
haplotypes present among the parental genotypes using software PolyMin (Frei et al.
2009). Second, the best combination of SNPs was determined to meet Sequenom
MassARRAY sequence requirements, namely sufficient sequence conservation next
to each SNP to enable diagnostic primer annealing. Third, duplicate SNPs were
selected, where available, for genotyping error control and rigorous haplotype call-
ing. SNP genotyping was performed using the Sequenom MassARRAY (Maldi-TOF
mass spectrometry) system at the Centre for Integrative Genomics (CIGENE, Aas,
Norway). SNP genotyping results were used to calculate haplotype frequencies in
each of the four samples, LTS, C0 R, C1 S+ and C1 S− . A test of selective neutrality
(Waples’ test) was applied. It tests, whether the observed variation in allele frequency
between two samples taken at different times can be explained as a sample drawn
from a population of size Ne (effective population size) that has undergone t gener-
ations of genetic drift. The test statistic is distributed as a chi-square (Waples 1989)
and is calculated as χ 2 = (x − y)2 /var(x − y), where x equals the estimated allele fre-
quency in an initial sample, y equals the estimated allele frequency in a subsequent
sample and var(x − y) equals the variance of this difference. The derivation of the
variance in (x − y) is explained by Waples (1989) in detail. We assumed that the
effective population size is equal to twice the number of selected plants (2 N = 68)
in a population of allogamous diploid genotypes, where migration was controlled by
84 G. Brazauskas et al.

Table 10.1 Average number of axillary tillers per plant in C0 , C+ −

1 , C1 and respective selected
subpopulations
Cycle Total population Subpopulation S+ Subpopulation S−
a
N Ax. tiller N Ax. tiller N Ax. tiller
numberb number number
C0 340 2.18 (0.10–4.35) 34 3.41 (3.00–4.35) 34 1.83 (0.10–2.42)
C+
1 340 2.46 (1.25–6.20) 34 3.90 (3.00–6.20) – –
C−
1 340 1.98 (0.00–3.85) – – 34 0.22 (0.00–0.45)
a
N, number of genotypes
b
Average axillary tiller number per plant and it’s range in parenthesis

isolation and mutation could be ignored (Labate et al. 1999). Bonferroni correction
was applied to account for multiple testing.

10.3 Results and Discussion

Plants in C0 expressed substantial variation for axillary tiller formation with the mean
number of axillary tillers per plant ranging from 0.10 to 4.35, and an overall average
of 2.18 (Table 10.1). A reciprocal selection at 10 % intensity (34 selected genotypes
from a total of 340) resulted in selected subpopulations C0 S+ and C0 S− with an
average number of 3.41 and 1.83 axillary tillers per plant, respectively. Substan-
tially higher selection differential (D = 1.23) was observed for positive selection as
opposed to that for negative selection (D = −0.35). The selected genotypes were
intercrossed to produce C+ −
1 and C1 of 340 plants each. Despite substantially higher
selection differential the resulting selection response was ony slightly higher for pos-
itive selection (R = 0.28) in comparison to negative selection (R = −0.20) and the
average number of axillary tillers reached 2.46 and 1.98 in C+ −
1 and C1 , respectively.
A second round of reciprocal selection was performed with the same intensity as
during the first cycle (10 %, 34 selected genotypes), and resulted in two selected
subpopulations C1 S+ and C1 S− . Here, a higher selection differential was observed
for negative selection (D = −1.76) in comparison to positive selection (D = 1.44).
The average number of axillary tillers per plant was 3.90 and 0.22 in C1 S+ and C1 S− ,
respectively.
SNP genotyping with 25 SNPs in five genes (Table 10.2) was performed on a
random subsample of 34 C0 genotypes (C0 R), two selected subpopulations C1 S+ and
C1 S− , and the five parental genotypes (LTS). Nineteen haplotypes were identified in
five genes in total. Five haplotypes per gene were detected in LpRUB1 and LpBRI1
while LpIAA1, LpSHOOT1 and LpTB1 had only three haplotypes per gene among
the five parental genotypes. The loci under investigation were dispersed over linkage
groups 2, 3, 4 and 5 (Table 10.2).
All 19 haplotypes detected in parental genotypes were also present in C0 R and no
novel haplotypes were detected. Furthermore, a minor haplotype frequency change
10 Estimation of Temporal Allele Frequency Changes in Ryegrass . . . 85

Table 10.2 GeneBank accession numbers, putative functions, linkage groups (LG), number of
SNPs genotyped and haplotypes detected for five perennial ryegrass genes
Gene GeneBank Putative function LG No of SNPs No. of haplotypes
LpIAA1 GU987119 Auxin signaling 4 5 3
LpRUB1 GU987120 Auxin signaling 5 5 5
LpBRI1 GU987121 Brassinosteroid signaling 3 6 5
LpSHOOT1 GU987122 Outgrowth of axillary buds 4 7 3
LpTB1 GU987123 Outgrowth of axillary buds 2 2 3

Table 10.3 Waples’ testa for temporal changes in allele frequency between Cycle 0 and Cycle 1 for
both positive and negative selection
Gene N(C0 R)b N(C1 S− ) χ 2 (C1 S− ) N(C1 S+ ) χ 2 (C1 S+ ) dfc
∗
LpIAA1 34 34 10.30 34 1.25 2
LpRUB1 34 34 8.34 34 6.64 4
LpBRI1 34 34 8.39 34 2.68 4
LpSHOOT1 34 34 0.42 34 0.03 2
LpTB1 34 34 1.84 34 0.13 2
∗
Significant at p < 0.05
a
Based on Ne = 68
b
N equals sample size at the respective cycle
c
df, degrees of freedom

was observed from LTS to C0 . This change was expected and could be attributed
to both genetic drift and sampling variance in C0 due to the low number (N = 34)
of genotypes in C0 R. After two cycles of selection, two haplotypes (10.5 %) were
lost in C1 S+ and one haplotype (5.3 %) was eliminated in C1 S− . All three lost
haplotypes had low frequencies (0.02–0.10) in both LTS and C0 R. An increase in the
number of haplotypes belonging to the extreme classes of frequencies were earlier
reported (Pinto et al. 2003). This is a feature of a dispersive process in which the
intermediate allele frequencies tend towards the limits of zero (lost) or 1 (fixation). No
fixed haplotypes were detected in our data set while the highest haplotype frequency
reached 0.74 (LpTB1).
Waples’ test (1989) was applied to each locus to detect if any of the observed
changes in allele (haplotype) frequency between C0 and C1 within populations were
greater than those expected by drift alone. Loci were tested for statistically significant
changes in allele frequency assuming an effective population size of Ne = 2 N = 68,
where N is the number of selected lines (N = 34). No additional significant tests
would be obtained using any smaller value of Ne because such tests would be more
conservative. The Waples’ test indicated selection for the gene LpIAA1 in C− 1 , where
allele frequency changes could not be explained by genetic drift alone (p < 0.05)
(Table 10.3).
The recurrent selection process is aimed at the enrichment of rare alleles. This
selection response was earlier observed in a complex maize population selected for
quantitative disease resistance for four cycles (Wisser et al. 2008). Authors specu-
late that loci under selection are more likely to be detected through the increases
86 G. Brazauskas et al.

in frequency of rare alleles at linked marker loci than through frequency shifts of
more common alleles. The more frequent alleles at marker loci are more likely to be
associated with multiple alleles at the genes under selection, including favourable
and unfavourable alleles, thus the higher-frequency alleles would change in both
directions under selection. However, in our study, allele frequency changes were es-
timated within candidate genes with putative effects on axillary tiller formation. This
implies that a tight linkage with alleles under selection should be expected and both
higher and lower frequency alleles should be under direct selection response. This
was observed in our data set where low-frequency alleles showed only small fluc-
tuations in their frequencies while all three haplotypes at non-neutral locus LpIAA1
had intermediate frequencies (p = 0.21 − 0.43) in C0 .
The effective size of a population, Ne , determines the rate of change in the com-
position of a population caused by genetic drift and is crucial in determining the
level of variability in a population, and the effectiveness of selection relative to drift.
The probability of obtaining a significant Waples’ test statistic strongly depends on
the size of the effective population. Smaller Ne leads to larger variance in allele
frequencies and a higher percentage of statistically significant tests. However, this
effect is of lower magnitude in the early generations and the ratio of S/Ne , where S
is the number of sampled individuals, becomes more important (Waples 1989). We
assumed that effective population size in our data set is equal twice the number of
selected lines (Ne = 2 N = 68) as the bottleneck generation of 34 selected lines will
create the strongest effect on genetic drift. Other factors, such as migration, non-
random mating or variation in offspring number were controlled by the experimental
design. Several methods exist for empirical Ne estimation from marker data, includ-
ing inbreeding and variance effective population sizes (Charlesworth 2009). Labate
et al. (1999) estimated effective population sizes based on temporal changes in allele
frequencies between C0 and C12 for two maize populations undergoing reciprocal
recurrent selection. Authors indicate that directional selection deflates effective pop-
ulation size where loci under selection yield Ne estimates closer to Ne = 0.5 N in
contrast to that of neutral loci with Ne = 2 N.
LpIAA1 is orthologous to AUXIN RESISTANT 5 (AXR5), a member of a large
family of genes, called Aux/IAA, which comprises genes that are auxin-regulated
(Reed 2001). AXR5 was isolated from a mutant of Arabidopsis called axr5-1 (Yang
et al. 2004). Mutant plants are resistant to auxin and display a variety of auxin-related
growth defects including defects in root and shoot tropisms. More specifically, there
are fewer lateral branches on the primary inflorescence of arx5-1 plants, but more
inflorescence branches growing from rosette. More recently an ortholog of AXR5,
called OsIAA1, was also characterized in rice (Song et al. 2009). The OsIAA1-
overexpression transgenic plants showed distinctive morphological changes such as
decreased plant hight and loose plant architecture. Results from our study as well
as mutant studies from Arabidopsis and rice indicate that LpIAA1 is a promising
candidate for shoot morphology control and could be used for the development of
functional markers for marker assisted breding in perennial ryegrass.
10 Estimation of Temporal Allele Frequency Changes in Ryegrass . . . 87

Acknowledgments This study was funded by the EU framework V project GRASP (QLRT-2001-
00862) and partly supported by the Fulbright Scholarship (Grantee ID 68433937).

References

Brazauskas G, Pašakinskienė I, Asp T, Lübberstedt T (2010) Nucleotide diversity and linkage

disequilibrium in five Lolium perenne genes with putative role in shoot morphology. Plant Sci
179:194–201. doi:10.1016/j.plantsci.2010.04.016
Charlesworth B (2009) Effective population size and patterns of molecular evolution and variation.
Nat Rev Genet 10:195–205. doi:10.1038/nrg2526
Coque M, Gallais A (2006) Genomic regions involved in response to grain yield selection at high
and low nitrogen fertilization in maize. Theor Appl Genet 112:1205–1220. doi:10.1007/s00122-
006-0222-5
Falke KC, Flachenecker C, Melchinger AE, Piepho HP, Maurer HP, Frish M (2007) Tempo-
ral changes in allele frequencies in two European F2 flint maize populations under modified
recurrent full-sib selection. Theor Appl Genet 117:765–776. doi:10.1007/s00122-006-0443-7
Frei U, Wollenweber B, Lübberstedt T (2009) „PolyMin“: Software for identification of the min-
imum number of polymorphisms required for haplotype and genotype differentiation. BMC
Bioinformatics 10:176. doi:10.1186/1471-2105-10-176
Guzman PS, Lamkey KR (1999) Predicted gains from recurrent selection in the BS11 maize
population. Maydica 44:93–99
Jensen LB, Andersen JR, Frei U, Xing Y, Taylor C, Holm PB, Lübberstedt T (2005) QTL mapping
of vernalization response in perennial ryegrass (Lolium perenne L.) reveals co-location with an
orthologue of wheat VRN1. Theor Appl Genet 110:527–536. doi:10.1007/s00122-004-1865-8
Jung GA, Van Wijk AJP, Hunt WF, Watson CE (1996) Ryegrasses. In: Moser LE, Buxton DR,
Casler MD (eds) Cool-Season Forage Grasses. Madison, Wisconsin
Labate JA, Lamkey KR, Lee M, Woodman WL (1999) Temporal changes in allele frequen-
cies in two reciprocally selected maize populations. Theor Appl Genet 99:1166–1178.
doi:10.1007/s001220051321
Langer RHM (1979) How grasses grow. Edward Arnold Publishers Limited, London
Pašakinskienė I (2005) Creeping stem and barren inflorescence: a new mutation in Lolium perenne
‘VIROIZ’. In: Zwierzykowski Z, Kosmala S (eds) Recent advances in genetics and breeding of
the grasses. Institute of Plant Genetics, Poznan 209–213
Pinto LR, Vieira MLC, de Souza CL, de Souza AP (2003) Reciprocal recurrent selection effects
on the genetic structure of tropical maize populations assessed at microsatellite loci. Genet Mol
Biol 26:355–364
Posselt UK, Barre P, Brazauskas G, Turner LB (2006) Comparative analysis of genetic similarity
between perennial ryegrass genotypes investigated with AFLPs, SSRs, RAPDs and ISSRs.
Czech J Genet Plant Breed 42:87–94
Reed JW (2001) Roles and activities of Aux/IAA proteins in Arabidopsis. Trends Plant Sci 6:420–
425. doi:10.1016/S1360-1385(01)02042-8
Song Y, You J, Xiong L (2009) Characterization of OsIAA1 gene, a member of rice Aux/IAA family
involved in auxin and brassinosteroid hormone responses and plant morphogenesis. Plant Mol
Biol 70:297–309. doi:10.1007/s11103-009-9474-1
Waples RS (1989) Temporal variation in allele frequencies: testing the right hypothesis. Evolution
43:1236–1251
Wisser RJ, Murray SC, Kolkman JM, Ceballos H, Nelson RJ (2008) Selection mapping of
loci for quantitative disease resistance in a diverse maize population. Genetics 180:583–599.
doi:10.1534/genetics.108.090118
Yang X, Lee S, So J, Dharmasiri S, Dharmasiri N, Ge L, Jensen C, Hangarter R, Hobbie L, Estelle M
(2004) The IAA1 protein is encoded by AXR5 and is substrate of SCFTIR1 . Plant J 40:772–782.
doi:10.1111/j.1365-313X.2004.02254.x
Chapter 11
Understanding the Genetic Basis for Slow
Plant-Mediated Proteolysis in Festulolium
Hybrids

S. A. O’Donovan, A. H. Kingston-Smith and M. W. Humphreys

Abstract Inefficiencies in the rumen associated with excessive rates of degradation

of forage protein have meant that resource capture is poor compared to the potential
gains achievable from use of new grass cultivars. Increased rumen efficiency through
improved utilization of feed protein will decrease emissions of environmentally dam-
aging wastes and greenhouse gas emissions. New interspecific forage grass hybrids
(Festulolium) derived from hybridizing ryegrass species Lolium perenne or Lolium
multiflorum with the related fescue species Festuca arundinacea var glaucescens,
native to mountain pastures in Mediterranean regions provide an opportunity to en-
hance ruminant nutrition and provide an environmental safeguard. Under rumen
conditions, the fescue protein is degraded at a significantly slower rate than that of
ryegrass which should lead to increased efficiency of feed N-utilization and aid live-
stock gain whilst also reducing N losses that contribute to greenhouse gas emissions.
Procedures are underway to determine the necessary genome composition required
to optimize expression of the fescue trait in a ryegrass genetic background and to de-
termine the most effective plant breeding strategy to ensure that this is achieved. It is
hypothesized that mechanisms evolved in the fescue species necessary for adaptations
to high temperatures in Mediterranean locations are also functional and relevant for
protein protection in stress conditions in the rumen. Combining these with ryegrass-
derived traits for forage yield and quality should assist strategies for more sustainable
livestock agriculture at a time of climate change.

11.1 Introduction

Food security for an increasing world population is threatened by increased summer

droughts and high temperatures. Providing food security is not only a problem in
developing countries but is also re-emerging as a problem for developed countries
and is creating an increasing pressure on the planet’s resources (Oweis and Peden
2008). Livestock agriculture is a major provider of food security in the UK, but due
to its large land area usage and the inefficiencies of ruminant nutrition, contributes
significantly to emissions of environmentally harmful wastes including greenhouse

S. A. O’Donovan () · A. H. Kingston-Smith · M. W. Humphreys

IBERS, Aberystwyth University, SY23 3EB, Aberystwyth, Ceredigion, Wales
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 89

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_11, © Springer Science+Business Media Dordrecht 2013
90 S. A. O’Donovan et al.

gasses (methane and nitrous oxide) and ammonia. Grassland accounts for over 70 %
of crude protein (CP) consumed by ruminants in the UK compared with alternative
sources that include cereals (14 %), oilseeds (10 %) and legumes (peas and beans
< 5 %) (Wilkins and Jones 2000). Ruminant digestion can successfully convert fi-
brous biomass such as grass, which is in abundance around the world, to produce
high quality protein sources fit for human consumption and thereby circumvent
use of alternative CP sources that have in the past incurred health scares such as
BSE. However, the advantages of grassland agriculture must be weighed against the
environmentally damaging concomitant greenhouse production (Moss et al. 2000).
Poor Protein Capture Poor protein capture by livestock from the large amounts of
plant biomass that is consumed has serious implications for livestock agriculture’s
sustainability. There is a poor return with only 30 % of nitrogen ingested being
converted into animal protein products milk and meat. Ruminants are inefficient
utilisers of grass protein due in part to the impacts of rumen stresses that induce plant–
mediated proteolysis (Kingston-Smith et al. 2010). Plant biomass eaten as forage is
exposed in the rumen to hostile conditions including darkness, no oxygen, high
temperatures (39 ◦ C) and microbes that in unison cause the plant cells to initiate an
early regulated cell death (senescence). This results in poor utilization of grass-based
nutrients and the creation of N waste in the forms of urea and ammonia (Kingston-
Smith et al. 2008).
Improvement of Ruminant Dietary Intake for Future Agricultural Sustainability
The cause of poor utilization of dietary protein and loss of N from the rumen has
been attributed to the ‘Asynchrony Hypothesis’ with regard to the imbalance be-
tween energy and protein supply and demand by the rumen microbial populations.
Asynchrony occurs because ruminants grazing on fresh forages supply the rumen
microbial populations with excess RDP (Rumen Degradable Protein) relative to fer-
mentable energy; this is because of rapid availability of plant proteins due to their
degradation by plant and microbial enzymes (Nolan and Dobos 2005; Moorby et al.
2008). This loss of non-protein nitrogen (NPN) limits ruminal microbial growth. It
is the post-ruminal flow of microbial protein that contributes directly to amino acid
absorption and host animal tissue production.
New Models for the Reduction of Environmentally Damaging Wastes Caused by
Livestock Agriculture and Improving Ruminant Nutrition Nitrogen and energy use
efficiency in ruminant digestion has been addressed using two approaches involving
alternative plant breeding techniques: Matching the availability of energy to protein
intake, and reducing protein degradability to synchronize with energy availability
(Moorby et al. 2008).
Energy-use-efficiency and reduction of waste N is achieved by increasing the
fermentable energy available to the rumen microbial populations from freshly in-
gested forage (Humphreys 1989; Miller et al. 2001) and in conserved forages such
as silage (Merry et al. 2006). Practically, this matching of energy to protein intake has
been addressed by using improved perennial ryegrasses (Lolium perenne) known as
high-sugar grasses (HSG’s). These have been bred for increased water soluble car-
bohydrate (WSC) content for improved forage nutritional value for grazing animals.
11 Understanding the Genetic Basis for Slow Plant-Mediated Proteolysis . . . 91

In the second approach, the synchronization of reduced protein degradability to

energy availability was addressed by (Kingston-Smith and Theodorou 2000) who
challenged the traditional view that breakdown of ingested forage in the rumen was
a process mediated by rumen micro-organism communities. The authors have shown
that plant-mediated proteolysis plays a major role in the rapid break down of freshly
ingested forage protein and that fresh living forage grasses ingested by grazing rumi-
nant animals are still mostly intact when they enter the rumen (Kingston-Smith and
Theodorou 2000). The hostile rumen environment of high temperatures and anaer-
obic conditions (39 ◦ C, no O2, ), has been shown to trigger living cells in freshly
ingested forage to release enzymes which could be part of a defense mechanism
referred to as programmed cell death (PCD) (Kingston-Smith et al. 2005).
Plant Breeding Solutions to Achieve more Efficient Forage for Livestock Agriculture
In vitro comparisons of protein stability from grass accessions demonstrated differen-
tial sensitivity of the forage protein to rumen conditions. Ryegrasses are particularly
susceptible to proteolysis under rumen conditions (Beha et al. 2002). However, pro-
tein from the fescue species Festuca arundinacea var. glaucescens (Boiss.) was
found to be more resilient given identical stress conditions (Shaw 2006) so that
the rate of plant-mediated proteolysis was considerably slower for F. arundinacea
var glaucescens compared to Lolium. Specifically, the mean protein half-life of
L. perenne was 4.33 h, in L. multiflorum this was 2.06 h, whereas the mean protein
half-life under rumen-simulated conditions for F. arundinacea var glaucescens was
15.83 h. The enhanced rumen-stress tolerance of the fescue species may have arisen
from its natural adaptations to high temperatures and water limiting conditions which
it encounters regularly in its natural habitat in Mediterranean grasslands (Shaw 2006).
Therefore, combining the desirable and complementary characters of ryegrass (high
digestibility and yield) and Festuca arundinacea var glaucescens provides opportu-
nities to both improve ruminant nutrition and increase resilience of the crop to the
onset of increased summer stresses thereby helping sustain forage production and
achieving yield potential.
Shaw (2006), attempted to transfer the protein-protective mechanism of F. arun-
dinacea var glaucescens into L. multiflorum through the use of a backcross breeding
programme and introgression-mapping technologies, returning over two backcross
generations to a diploid L. multiflorum genome with some genotypes having one or
more trans-located Festuca chromosome segments.
It is likely that the mechanisms for protein protection in the fescue species are
complex and governed by a suite of interacting genes (quantitative trait loci; QTL)
of greater or lesser effect. To monitor transfers of F. arundinacea var glaucescens
genes into Lolium, Shaw (2006) employed fescue-specific amplified fragment length
polymorphism (AFLP) markers. The same and other marker technologies and a near-
identical breeding programme was employed successfully at the same time to transfer
a QTL for drought and heat-tolerance from Festuca into Lolium, thereby confirm-
ing the efficacy of the technique used, at least for the traits studied (Humphreys
et al. 2005). However, Shaw (2006) during her marker-assisted backcrossing pro-
gramme whilst although able to demonstrate the successful transfer of a number of
92 S. A. O’Donovan et al.

F. arundinacea var glaucescens chromosome segments onto L. multiflorum chro-

mosomes, the heritability of the Festuca trait for protein protection over generations
was inconclusive. The significant reduction in the expression of the Festuca trait
over generations led to questioning of the suitability of the introgression-mapping
approach for transfers of a functional protein protection mechanism from Festuca
into Lolium.
However, as well as the successful use of Festuca gene introgression for improved
drought tolerance from F. arundinacea var glaucescens into Lolium reported by
Humphreys et al. (2005), and also from its close relative F. arundinacea (Humphreys
and Thomas 1993; Humphreys and Pasakinskiene 1996), there are several examples
of the successful use of introgression-mapping for transfers of both simple and com-
plex traits in the Lolium-Festuca genome complex (Moore et al. 2005; Armstead et al.
2006; Kosmala et al. 2006; Gronnerod et al. 2004). Therefore, there is some cause for
optimism that the introgression-breeding approach may be suitable for gene trans-
fers for improved protein protection for ryegrass. The very high frequency of gene
exchange at meiosis in Lolium-Festuca grass hybrids provides access throughout the
Festuca genome to genes for all traits of potential value to Lolium (Humphreys et al.
2003).
Alm et al. (2011), conducted a QTL study in Festuca pratensis to locate genes
associated with resistance to cold and drought stress and found a major QTL for
resistance to severe drought stress on chromosome 3 that spanned the entire linkage
group. Another study conducted by Turner et al. (2008) located QTL for drought
resistance in Lolium perenne and found no association with chromosome 3. The
outcome would suggest that Festuca chromosome 3 would be a good source of
genetic variants for improved drought resistance in Lolium. This has proved to be
the case with Humphreys and Pasakinskiene (1996) and Humphreys et al. (2005)
identifying Festuca translocations derived from drought-tolerant fescue species onto
Lolium chromosome 3 that led to significant improvements in resistance to severe
drought stress. In field experiments in France, breeders’ lines derived from these
introgression lines proved as effective under drought stress as certain interspecific
hybrids that had the entire Festuca genome present (Humphreys et al. 2011).
The alternative to introgression as a route at combining in a single plant genome
the positive attributes of Lolium and Festuca species is the use of amphiploidy where
entire Lolium and Festuca genomes are combined. The majority of Festulolium cul-
tivars marketed currently have been produced by this method (Ghesquière et al.
2010). The stability of such hybrids would be enhanced given a functional chromo-
some pairing regulator that can provide for strict homologous chromosome pairing
and disomic inheritance. Such a system is present in polyploid fescue species such
as Festuca arundinacea var glaucescens (Jauhar 1993). New procedures in devel-
opment at IBERS provide opportunities to exploit a functional fescue chromosome
pairing regulator and thereby provide opportunity for the effective use of amphiploidy
as a means to harness the protein protection mechanism present in this fescue species.
Alternative breeding approaches underway at IBERS that employ either introgression
or amphiploidy should answer which strategy is the more appropriate and in future
11 Understanding the Genetic Basis for Slow Plant-Mediated Proteolysis . . . 93

provide options to the livestock industry of use of grass cultivars that will improve
ruminant nutrition through mitigating plant-mediated proteolysis in the rumen.
The assessment of the alternative breeding approaches is underway as part of a
new PhD programme funded by EBLEX and HCC. The breeding approach must
both optimize the expression of the Festuca arundinacea var glaucescens trait
and maintain the positive attributes of Lolium. The programme will test the sug-
gestion that protein stability in the rumen would be increased by increasing the
Festuca:Lolium genome dosage in Festulolium hybrids. Hybrid combinations hav-
ing different genome dosage will be compared and the variation between and within
genome combinations for expression of the protein protection mechanism deter-
mined. The findings will be compared with the alternative introgression breeding
approach where Festuca-derived genes (or regulatory regions) responsible for pro-
tein protection will be sought that may have previously been lost inadvertently by
Shaw (2006) during her introgression-mapping backcross breeding programme. The
current studies should provide an answer to whether introgression-mapping is appro-
priate for transfer of the traits that confer reduced plant-mediated proteolysis. The
PhD project will also attempt to identify the mechanism responsible for protein pro-
tection in F.arundinacea var. glaucescens. Taken together, the Mediterranean origins
of the fescue species where it is exposed regularly to heat stress and the observation
by Shaw (2006) that the expression of the thermo-tolerance conferring heat shock
protein (HSPs) (Vierling 1991); Howarth and Ougham (1993), HSP70 was immedi-
ate on exposure to rumen conditions in F. arundinacea var. glaucescens but delayed
by up to five hours in L. multiflorum), it is plausible to suggest that the fescue HSP
provides at least one potential mechanism to explain the greater resistance of the Fes-
tuca species compared with Lolium. If the hypothesis is proven correct, then gene
introgression would seem a feasible route to achieve expression of the fescue trait
following transfer to ryegrass. A successful outcome will provide a co-adaptive trait
that will serve to improve field survival of Lolium during hot summers, and will also
enhance rumen-use-efficiency to aid food security and to combat climate change.

Acknowledgments This research is sponsored by EBLEX and HCC.

References

Alm V, Busso CS, Ergon Å, Rudi H, Larsen A, Humphreys MW, Rognli OA (2011) A QTL analysis
and comparative genetic mapping of frost tolerance, winter survival and drought tolerance in
meadow fescue (Festuca pratensis Huds.). Theor Appl Genet 123(3):369–382
Armstead IP, Harper JA, Turner LB, Skot L, King IP, Humphreys MO, Morgan WG, Thomas HM,
Roderick HW (2006) Introgression of crown rust (Puccinia coronata) resistance from meadow
fescue (Festuca pratensis) into Italian ryegrass (Lolium multiflorum): genetic mapping and
identification of associated molecular markers. Plant Pathol 55(1):62–67
Beha EM, Theodorou MK, Thomas BJ, Kingston-Smith AH (2002) Grass cells ingested by rumi-
nants undergo autolysis which differs from senescence: implications for grass breeding targets
and livestock production. Plant Cell Environ 25(10):1299–1312
94 S. A. O’Donovan et al.

Ghesquière M, Humphreys MW, Zwierzykowski Z (2010) Festulolium. In: Boller B, Posselt

UK, Vernonesi F (eds) Fodder crops and amenity grasses. Springer, Heidelberg, pp 293–316.
doi:10.1007/978-1-4419–0760-8_12
Gronnerod S, Fjellheim S, Humphreys MW, Ostrem L, Canter PH, Greig Z, Jorgensen O, Larsen
A, Rognli OA (2004) Application of AFLP and GISH techniques for identification of Festuca
chromosome segments conferring winter hardiness in a Lolium perenne × Festuca pratensis
population developments in plant breeding. Proceedings 3rd international symposium on molec-
ular breeding of forage and turf. Dallas, Texas, USA, pp 18–22, May 2003 (Hopkins A, Wang
Z-Y, Mian R, Sledge M, Barker RE, ISBN:1402018673, 11, pp 81–86)
Howarth C, Ougham H (1993) Gene expression under temperature stress. New Phytol 125(1):1–26
Humphreys J, Harper JA, Armstead IP, Humphreys MW (2005) Introgression-mapping of genes
for drought resistance transferred from Festuca arundinacea var. glaucescens into Lolium
multiflorum. Theor Appl Genet 110(3):579–587
Humphreys MO (1989) Water-soluble carbohydrates in perennial ryegrass breeding. Grass Forage
Sci 44(2):237–244
Humphreys MW, Pasakinskiene I (1996) Chromosome painting to locate genes for drought
resistance transferred from Festuca arundinacea into Lolium multiflorum. Heredity 77:530–534
Humphreys MW, Thomas H (1993) Improved drought resistance in introgression lines derived from
Lolium multiflorum × Festuca arundinacea hybrids. Plant Breeding 111:155–161
Humphreys MW, Canter PJ, Thomas HM (2003) Advances in introgression technologies for
precision breeding within the Lolium-Festuca complex. Ann of Appl Biol 143:1–10
Humphreys MW, MacLeod CJA, Whalley WR, Turner LB, Farrell MS, Ghesquiere M, Haygarth PM
(2011) Designing grass cultivars for droughts and floods. In: “Breeding strategies for sustainable
forage and turf grass improvement”. Proceedings 29th Eucarpia Fodder Crops and Amenity
Section Meeting 4th–8th Sept 2011, Dublin Ireland
Jauhar PP (1993) Cytogenetics of the Festuca-Lolium complex. Relevance to breeding. In:
Monographs on theoretical and applied genetics 18. Springer, Heidelberg, ISBN 0-387-52113-5
Kingston-Smith A, Theodorou M (2000) Post-ingestion metabolism of fresh forage. New Phytol
148(1):37–55
Kingston-Smith A, Merry R, Leemans D, Thomas H, Theodorou M (2005) Evidence in support of
a role for plant-mediated proteolysis in the rumens of grazing animals. Brit J Nutr 93(1):73–79
Kingston-Smith A, Davies T, Edwards J, Theodorou M (2008) From plants to animals; the role of
plant cell death in ruminant herbivores. J Exp Bot 59(3):521
Kingston-Smith AH, Edwards JE, Huws SA, Kim EJ, Abberton M (2010) Plant-based strate-
gies towards minimising “livestocks” long shadow’. Proceedings of the Nutrition Society
2000(69):1–8
Kosmala A, Zwierzykowski Z, Gasior D, Rapacz M, Zwierzykowska E, Humphreys MW (2006)
GISH/FISH mapping of genes for freezing tolerance transferred from Festuca pratensis into
Lolium multiflorum. Heredity 96(3):243–251
Merry R, Lee M, Davies D, Dewhurst R, Moorby J, Scollan N, Theodorou M (2006) Effects of
high-sugar ryegrass silage and mixtures with red clover silage on ruminant digestion. 1. In vitro
and in vivo studies of nitrogen utilization. J Anim Sci 84(11):3049
Miller L, Moorby J, Davies D, Humphreys M, Scollan N, Macrae J, Theodorou M (2001) Increased
concentration of water-soluble carbohydrate in perennial ryegrass (Lolium perenne L.): milk
production from late-lactation dairy cows. Grass Forage Sci 56(4):383–394
Moorby JM, Kingston-Smith AH, Abberton MT, Humphreys MO, Theodorou MK (2008) Im-
provement of forages to increase the efficiency of nitrogen and energy use by ruminants. In
42nd university of Nottingham feed conference, University of Nottingham
Moore BJ, Donnison IS, Harper JA, Armstead IP, King J, Thomas H, Jones RN, Jones TH, Thomas
HM, Morgan WG, Thomas A, Ougham HJ, Huang L, Fentem PA, Roberts LA, King IP (2005)
Molecular tagging of a senescence gene by introgression mapping of a mutant stay-green locus
from Festuca pratensis. New Phytol 165(3):801–806
11 Understanding the Genetic Basis for Slow Plant-Mediated Proteolysis . . . 95

Moss AR, Jouany J-P, Newbold J (2000) Methane production by ruminants: its contribution to
global warming. Ann Zootech 49:231–253
Nolan JV, Dobos RC (2005) Nitrogen transactions in ruminants. In: Dijkstra J, Forbes JM, France
J (eds) Quantitative aspects of ruminant digestion and metabolism, 2nd edn. CABI (Pub),
Cambridge (ISBN 0-85199-814-3)
Oweis T, Peden D (2008) Water and livestock. In: Rowlinson P, Steele M, NefzaouiA (eds) Livestock
and global climate change 2008. Cambridge University Press (Pub), pp 19–20 (ISBN 978-0-
906562-62-8)
Shaw RK (2006) Effects of gene transfer from Festuca to Lolium on plant-mediated proteolysis.
In: IBERS: Animal and Microbial Sciences, Vol PhD: Aberystwyth University
Turner LB, Cairns A, Armstead IP, Thomas H, Humphreys MW, Humphreys MO (2008) Does
fructan have a functional role in physiological traits? Investigation by quantitative trait locus
mapping. New Phytol 179(3):765–775
Vierling E (1991) The roles of heat shock proteins in plants. Annu Rev Plant Biol 42(1):579–620
Wilkins R, Jones R (2000) Alternative home-grown protein sources for ruminants in the United
Kingdom. Anim Feed Sci Tech 85(1–2):23–32
Chapter 12
Chromosomal Rearrangements in BC1 Progeny
Obtained from Crosses of Tetraploid
F. pratensis × L. perenne Hybrids
with Tetraploid L. perenne

T. Ksia˛żczyk, Z. Zwierzykowski and E. Zwierzykowska

Abstract Physical mapping of genes responsible for quality traits requires well
established cytogenetic maps and chromosome identification. Genomic in situ
hybridization (GISH) allows to discriminate parental genomes and to track re-
combination between them. Fluorescence in situ hybridization (FISH) with 5S and
18S–5.8S–25S (35S) rDNA probes provides chromosomal landmarks, and allows to
detect chromosome re-arrangements; a characteristic rDNA position provides use-
ful markers for chromosome identification. The aim of this study was to assess the
genomic constitution and chromosome arrangements in BC1 progeny obtained from
crosses of tetraploid (2n=4x=28) F. pratensis × L. perenne hybrids with tetraploid
(2n=4x=28) L. perenne. GISH examination in BC1 progeny showed variability in
respect to somatic Lolium and Festuca chromosome number, as well as Lolium and
Festuca recombinant chromosomes and homoeologous recombination breakpoints.
FISH experiments showed various numbers of both rDNA loci (3–5 sites for 5S rDNA
and 10–13 sites for 35S rDNA). Lolium chromosome 3 and Festuca chromosomes 2
and 3 were also involved in recombination showing rearrangements.

Keywords Chromosomal rearrangements · Festuca pratensis · FISH · GISH ·

Introgression · Lolium perenne

12.1 Introduction

In introgression breeding programs, intergeneric recombination can be used to trans-

fer abiotic and biotic stress resistance traits from Festuca species into Lolium species.
Meadow fescue (F. pratensis Huds.; Fp) and perennial ryegrass (L. perenne L.; Lp)
can be hybridized at various ploidy levels, producing diploid, triploid and tetraploid

T. Ksia˛żczyk () · Z. Zwierzykowski · E. Zwierzykowska

Institute of Plant Genetics of the Polish Academy of Sciences,
Strzeszyńska 34, 60–479, Poznań, Poland
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 97

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_12, © Springer Science+Business Media Dordrecht 2013
98 T. Ksia˛żczyk et al.

intergeneric hybrids, in which homoeologous chromosomes pair and recombine

freely (King et al. 1998; Zwierzykowski et al. 1998, 2006; Kopecký et al. 2009).
GISH is the method of choice in the Lolium-Festuca complex for analysis of the
parental genomes and detection of the exchanges between constituent genomes (King
et al. 1998; Zwierzykowski et al. 1998, 1999, 2006; Kopecký et al. 2006). More pre-
cise analysis of a genome constitution and identification of chromosomes in various
amphiploid and introgression forms of Festulolium can also be obtained through the
use of combined FISH/GISH (Kosmala et al. 2006; Ksia˛żczyk et al. 2010; Harper
et al. 2011). This method utilizes in situ hybridization with total genomic DNA as
a probe to distinguish genomes, and with chromosome specific DNA probes (e.g.
rDNA probes) to identify pairs of mitotic chromosomes or to visualize pairing of
homoeologues at meiosis, as well as to identify segments carrying desirable genes of
one species when introgressed into another (Kosmala et al. 2006; Kopecký et al. 2008;
Ksia˛żczyk et al. 2010). The application of GISH/FISH techniques in F1 hybrids of
Fp×Lp allowed the identification of Lp chromosome 3 (acc. to Thomas 1981), Fp
chromosomes 2 and 3 (acc. to Thomas 1981), and this approach revealed variation
in their numbers which can be easily tracked in Festulolium hybrids (Ksia˛żczyk
et al. 2010). The present study aimed at characterizing the genome constitution
of tetraploid introgression forms, derived from crosses of tetraploid F1 hybrids of
F. pratensis×L. perenne into tetraploid L. perenne, and to describe (i) somatic Lp and
Fp chromosome number, (ii) number of complete and recombinant Lp and Fp chro-
mosomes, (iii) number of homoeologous recombination breakpoints in both species,
and also (iv) to identify the number and chromosomal sites of rDNA sequences.

12.2 Materials and Methods

Tetraploid (2n=4x=28) F1 hybrids of F. pratensis (4x)×L. perenne (4x), genera-

ted by intercrossing autotetraploid forms of both species (Zwierzykowski et al. 2006),
were backcrossed with L. perenne ‘Solen’ (4x) to produce tetraploid BC1 progeny.
Forty randomly chosen BC1 plants were used for cytogenetic analyses. The cultivars
of L. perenne (4x) and F. pratensis (4x) and intergeneric tetraploid F1 hybrids of F.
pratensis (4x)×L. perenne (4x) were previously studied by GISH/FISH (Ksia˛żczyk
et al. 2010), and were selected as control plants in our analyses.
Chromosome preparations were made from a single root per plant of the BC1
hybrids, according to Ksia˛żczyk et al. (2010). GISH was performed according
to Zwierzykowski et al. (2006), using the total genomic DNA of Lp ‘Solen’ and
Fp ‘Westa’ as a probe and block, respectively. FISH was performed as described
by Ksia˛żczyk et al. (2010), using the ribosomal DNA probes 5S rDNA and 25S
rDNA (the latter being used for detection of plant 35S rDNA loci). The Lp and
Fp chromosomes identified by rDNA-FISH were numbered according to Thomas
(1981).
12 Chromosomal Rearrangements in BC1 Progeny Obtained from Crosses . . . 99

Fig. 12.1 GISH a (22Lp (2R) + 6 Fp (1R)) and FISH b analyses of the same somatic metaphase
chromosome spread of a BC1 plant. GISH image a was created after FISH using total genomic DNA
from L. perenne as a probe labelled with digoxigenin and detected by anti-digoxigenin conjugated
with FITC (green/yellow), with blocking genomic DNA of F. pratensis (orange/red); chromosomes
were counterstained with propidium iodide. FISH image b was created using as a probe (i) 5S rDNA
labelled with rhodamine (red) and (ii) 25S rDNA labelled with digoxigenin and detected by anti-
digoxigenin conjugated with FITC (green); chromosomes were counterstained with DAPI (blue).
GISH and FISH images are supplemented by the white arrows indicating Lolium/Festuca recombi-
nant chromosomes (R) and by the white lines with intervals indicating recombination breakpoints.
The nomenclature of rDNA-bearing chromosomes (Arabic numerals) follows Thomas (1981).
Uppercase letters denote the genomic origin of tagged chromosomes. Scale bars represent 5 μm

12.3 Results and Discussion

The chromosome number of a total of 40 BC1 plants ranged from 25 to 29. Among all
BC1 plants, aneuploid cells were found to be in majority (60 %) in relation to expected
tetraploid cells (40 %). The number of complete Lp chromosomes ranged from 18 to
22 (mean 20.7), while the number of complete Fp chromosomes ranged from 5 to 8
(mean 6.67). The number of Lp recombined chromosomes ranged from 0 to 2 (mean
0.7), while the number of Fp recombinant ones ranged from 0 to 3 (mean 0.75). 22.5 %
of BC1 plants had no Lp and Fp recombined chromosomes, and the number of non-
recombinant chromosomes of Lp ranged from 19 to 21 (mean 20.4), while the number
of non-recombinant ones of Fp ranged from 6 to 7 (mean 6.77). Among all karyotyped
BC1 progeny, there were 58 recombined Lp/Fp chromosomes, of which 55 had one
arm recombined (Fig. 12.1) and 3 had both arms recombined (2 Lp chromosomes and
100 T. Ksia˛żczyk et al.

1 Fp chromosome). Of a total of 65 recombination breakpoints observed, 58 chromo-

somes had single recombination breakpoint (Fig. 12.1a), 2 had double recombination
breakpoints, and 1 had triple recombination breakpoints. The mean number of
recombination breakpoints was similar for both Lp (0.8) and Fp (0.82) chromosomes.
FISH with both rDNA probes (Fig. 12.1b) showed inter-individual variation in
number and chromosomal location of 5S and 35S ribosomal DNA sites in the BC1
plants. The number of 5S rDNA sites ranged from 3 to 5, while the number of 35S
rDNA sites ranged from 10 to 13. The number of Lp homologues of chromosome
3 also varied and ranged from 2 to 4. Variation in chromosome patterns of rDNA
loci is common, and has been previously observed in diploid and tetraploid Lp and
Fp cultivars and triploid and tetraploid F1 Festuca × Lolium hybrids (Ksia˛żczyk
et al. 2010). As was expected in the BC1 , three Lp and one Fp large 5S rDNA sites
were interstitially located, while the additional small 5S rDNA locus was always
found in a distal region of one unknown Fp chromosome. The 35S rDNA sites
were always located at the secondary constrictions on the Lp chromosomes (1, 2
and 3), as well as proximally located close to the centromere (7), while in the Fp
chromosome (2) they were located at the secondary constriction. The rDNA-FISH
revealed that Lp chromosome 3 and Fp chromosomes 2 and 3 were also involved
in recombination as well as showing rearrangements. Among 58 recombined Lp/Fp
chromosomes, 15 Lp and 8 Fp were rDNA-bearing, of which 5 Lp had both 5S and
35S rDNA loci (chromosome 3), 10 had Lp 35S rDNA locus (chromosomes 1, 2 or
7) (Fig. 12.1b), 5 had Fp 35S rDNA locus (chromosome 2), 2 had Fp large 5S rDNA
locus (chromosome 3), and 1 had Fp small 5S rDNA locus (unknown chromosome).

12.4 Conclusions

BC1 progeny studied by GISH showed variability in respect of Lp and Fp somatic

chromosome number, recombined chromosomes and homoeologous recombination
breakpoints. Variation in the number of 5S and 35S rDNA sites in BC1 plants indi-
cated structural karyotype heterozygosity, due to intraspecific rDNA loci variation as
was observed in the parents previously studied. The combined use of GISH and FISH,
together with two rDNA probes, enabled the identification of Lp and Fp recombi-
nant rDNA-bearing chromosomes, and this technique is therefore suitable for rapid
identification of the modified Lp chromosome (3) and Fp chromosomes (2 and 3).
The FISH/GISH technique can now be used for monitoring Lp and Fp chromosome
changes, although a larger number of chromosome-specific FISH probes are needed
to identify all Lp and Fp chromosomes.

Acknowledgments We are grateful to Mr. Włodzimierz Zwierzykowski (Institute of Plant Genetics

PAS, Poznań) for excellent technical assistance. We also thank Prof. Neil Jones (Aberystwyth
University, UK) for English revision of the manuscript.
12 Chromosomal Rearrangements in BC1 Progeny Obtained from Crosses . . . 101

References

Harper J, Armstead I, Thomas A, James C, Gasior D, Bisaga M, Roberts L, King I, King J (2011)
Alien Introgression in the grasses Lolium perenne (perennial ryegrass) and Festuca pratensis
(meadow fescue): the development of seven monosomic substitution lines and their molecular
and cytological characterization. Ann Bot 107(8):1313–1321
King IP, Morgan WG, Armstead IP, Harper JA, Hayward MD, Bollard A, Nash JV, Forster JW,
Thomas HM (1998) Introgression mapping in the grasses. I. Introgression of Festuca pratensis
chromosomes and chromosome segments into Lolium perenne. Heredity 81:462–467
Kopecký D, Loureiro J, Zwierzykowski Z, Ghesquière M, Doležel J (2006) Genome constitution and
evolution in Lolium × Festuca hybrid cultivars (Festulolium). Theor Appl Genet 113:731–742
Kopecký D, Lukaszewski AJ, Doležel J (2008) Cytogenetics of Festulolium (Festuca × Lolium
hybrids). Cytogenet Genome Res 120:370–383
Kopecký D, BartoŠ J, Zwierzykowski Z, Doležel J (2009) Chromosome pairing of individual
genomes in tall fescue (Festuca arundinacea Schreb.), its progenitors, and hybrids with Italian
ryegrass (Lolium multiflorum Lam.). Cytogenet Genome Res 124:170–178
Kosmala A, Zwierzykowski Z, Ga˛sior D, Rapacz M, Zwierzykowska E, Humphreys MW (2006)
GISH/FISH mapping of genes for freezing tolerance transferred from Festuca pratensis to
Lolium multiflorum. Heredity 96:243–251
Ksia˛żczyk T, Taciak M, Zwierzykowski Z (2010) Variability of ribosomal DNA sites in Festuca
pratensis, Lolium perenne, and their intergeneric hybrids, revealed by FISH and GISH. J Appl
Genet 51(4):449–460
Thomas HM (1981) The Giemsa C-band karyotypes of six Lolium species. Heredity 46:263–267
Zwierzykowski Z, Tayyar R, Brunell M, Lukaszewski AJ (1998) Genome recombination in in-
tergeneric hybrids between tetraploid Festuca pratensis and Lolium multiflorum. J Heredity
89(4):324–328
Zwierzykowski Z, Lukaszewski AJ, Naganowska B, Lesniewska A (1999) The pattern of homoeol-
ogous recombination in triploid hybrids of Lolium multiflorum with Festuca pratensis. Genome
42(4):720–726
Zwierzykowski Z, Kosmala A, Zwierzykowska E, Jones N, Jokś W, Bocianowski J (2006) Genome
balance in six successive generations of the allotetraploid Festuca pratensis and Lolium perenne.
Theor Appl Genet 113(3):539–547
 Part III
Novel Emerging Tools
Chapter 13
Establishing Chromosome Genomics in Forage
and Turf Grasses

D. Kopecký, J. Číhalíková, J. Kopecká, J. Vrána, M. Havránková, Š. Stočes,

J. Bartoš, H. Šimková, J. Šafář, M. Kubaláková, P. Navrátil and J. Doležel

Abstract Analyses of large genomes are hampered by high proportions of repetitive

DNA, that make the assembly of short sequence reads difficult. This is also the case
for meadow fescue (Festuca pratensis Huds.), one of predominant grass species in
temperate and Northern regions with the genome size estimated at 1C = 3,175 Mbp.
This species is known for its ability to survive under freezing conditions and it
has been used widely in intergeneric hybridization with various ryegrass species
to produce superior Festulolium cultivars. Here we describe attempts to dissect
the meadow fescue’s genome into smaller fractions—individual chromosomes and
groups of chromosomes. Following the methods of flow cytogenetics developed for
legumes and cereals, we have developed a chromosome sorting protocol for grasses
and currently we are able to sort F. pratensis chromosome 4 (the largest in the
genome) and two groups of three chromosomes each: 2, 3, 7 and 1, 5, 6. As the first
step we sequenced chromosome 4 by Illumina with 50x coverage and assembled
low copy and genic regions. This facilitated detailed comparative analysis with se-
quenced genomes of rice, Brachypodium and sorghum and provided the first insight
into the genome composition of this species. The possibility to purify chromosome
4 opens the way for a more efficient analysis of genetic loci on this chromosome that
control important agronomic traits, such as freezing tolerance. Moreover, purified
chromosomes are excellent templates for PCR screening as well as cytogenetic and
physical mapping.

13.1 Introduction

Ryegrasses and fescues are among the predominant forage and turf species in
temperate climates. Ryegrass species are known for their high yield, palatability,
digestibility, rapid establishment, favorable nutrient characteristics, dark green color
and uniformity of turf. However, they are susceptible to stress conditions such as

D. Kopecký () · J. Číhalíková · J. Kopecká · J. Vrána · M. Havránková · Š. Stočes · J. Bartoš ·

H. Šimková · J. Šafář · M. Kubaláková · P. Navrátil · J. Doležel
Institute of Experimental Botany, Centre of the Region Haná for Biotechnological and Agricultural
Research, Sokolovská 6, 77200 Olomouc, Czech Republic
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 105

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_13, © Springer Science+Business Media Dordrecht 2013
106 D. Kopecký et al.

winter freezing, and heat and drought during summer. Fescues have complemen-
tary characteristics to those of ryegrasses. Therefore, they are widely combined in
mixtures. They have also become subjects of intergeneric hybridization. Tens of
x Festulolium cultivars with various genomic constitutions have been released and
are widely used in forage and turf seed industry in Europe (Kopecký et al. 2006).
Despite their agronomic importance, the progress in genetics, genomics and bioin-
formatics of grasses is far behind that of cereals. Among others, such studies are
complicated by the outcrossing nature of these species, population-based breeding
and frequent aneuploidy (Kopecky et al. 2005). Moreover, genomes of species within
Festuca-Lolium are large and complex. The genome of meadow fescue was estimated
at 1C = 3175 Mbp based on flow-cytometry, similar to that of human (Doležel et al.
2003; Kopecký et al. 2010). Ryegrass genome is just a little smaller (1C = 2623 Mbp
of L. perenne and 1C = 2567 Mbp of L. multiflorum). Because of the complexity and
size of genomes (full of repeats), any procedural or technical simplification would
be highly welcome in genomic studies.

13.2 Chromosome Genomics

Analysis of a complex genome where a majority of DNA is present as repeats can be

simplified by several approaches. Several methods were used to avoid sequencing of
the repetitive parts of genomes. Sequencing of cDNAs to generate ESTs (Expressed
Sequence Tags) is among the most successful. However, ESTs fail to sample rare or
conditional transcripts (Martienssen et al. 2004) and other methods were proposed
to target the gene space in large and complex genomes (Cot fractionation and methyl
filtration) (Peterson et al. 2002; Rabinowicz et al. 1999). Unfortunately, these two
methods did not provide the expected improvements.
Another alternative to overcome the complexity of large genomes is to dissect
them to smaller parts and sequencing these parts individually. Working with natu-
rally uniform and independent units—chromosomes—is perhaps the most powerful
approach. In hexaploid wheat, individual chromosomes represent only 1–3 % of
the entire genome (Doležel et al. 2009) and even diploid species’ chromosomes
(as meadow fescue) dissect the 3.2 Gbp genome into 373–543 Mbp units, each
representing 11.7–17.1 % of the whole (Kopecký et al. 2010).
There are two ways to isolate individual chromosomes: microdissection and flow-
sorting. Microdissection enables the isolation of any chromosome or a chromosome
segment. However, the total yield is limited to only a few copies of a particular
chromosome (Zhou and Hu 2007). Flow cytometric sorting relies on differences in
chromosome size. It can generate samples of unlimited copy numbers of specific
chromosomes with purity usually exceeding 90 %. Moreover, DNA of sorted chro-
mosomes is suitable for further molecular analysis (Doležel et al. 2004). The output of
flow cytometry analysis is a histogram of relative chromosome fluorescence intensity
(reflecting chromosome size), which is called a flow karyotype (Fig. 13.1). Ideally,
each chromosome is represented by a single peak on a flow karyotype. However,
13 Establishing Chromosome Genomics in Forage and Turf Grasses 107

Fig. 13.1 Flow karyotyping

in F. pratensis (2n = 2x = 14).
The flow karyotype consists
of two peaks representing
groups of chromosomes and
one peak representing
chromosome 4F. Peak 1
represents small
chromosomes 1F, 5F and 6F
and Peak 2 consists of
chromosomes 2F, 3F and 7F.
X axis relative DAPI
fluorescence intensity, Y axis
number of events

chromosome size similarities frequently result in the formation of composite peaks

formed by mixtures of two or more chromosomes. Doležel et al. (2009) proposed
that there has to be at least 10 % difference in chromosome size to generate a sep-
arate peak on a flow karyotype. Unfortunately, in most plant species chromosomes
are not that different. For example, flow cytometry analysis of hexaploid wheat (21
chromosome pairs) produced only four peaks, and only one of those contained a
single chromosome, 3B (Vrána et al. 2000). All other chromosomes were present in
composite peaks and could not be sorted into uniform fractions. Similarly, in bar-
ley only chromosome 1H, and in rye only chromosome 1R can be sorted into pure
fractions (Suchánková et al. 2006; Kubaláková et al. 2003). However, the plasticity
of plant genomes (especially in polyploids) makes it possible to develop special cy-
togenetic stocks with reconstructed karyotypes. Telosomic lines made it possible to
sort individual chromosome arms in wheat; in instances where a telocentric overlaps
in size with smaller complete chromosomes, isochromosomes provide a solution.
Similarly, wheat-rye and wheat-barley addition lines enabled sorting of individual
chromosomes or chromosome arms of barley and rye (Doležel et al. 2009). Nowa-
days, all wheat, barley and rye chromosomes/chromosome arms can be isolated in
pure fractions and are being used for physical mapping, sequencing and map-based
cloning.

13.3 Karyotypes of Grasses (Fescues and Ryegrasses)

Meadow fescue (Festuca pratensis Huds.) is a diploid species with seven chromo-
some pairs. The largest is metacentric chromosome 4, followed by another metacen-
tric chromosome 3 (the one with a prominent secondary constriction), submetacentric
chromosome 2 and metacentric chromosome 7. The smallest are submetacen-
tric chromosomes 5, 6 and 1. Individual lengths (as determined at C-metaphase) vary
108 D. Kopecký et al.

from 4.67 μm to 6.79 μm (Kopecký et al. 2010). Italian ryegrass (Lolium multiflo-
rum Lam.) has a similar karyotype, with chromosome length ranging from 3.37
μm (chromosome 1) to 5.33 μm of chromosome 4 (Kopecký et al. 2010). The
largest is the metacentric chromosome 4. Similar in length are submetacentric chro-
mosome 2 and metacentric chromosomes 3 and 7. These three chromosomes carry
secondary constrictions. The smallest are submetacentric chromosomes 5, 6 and 1.
Perennial ryegrass (Lolium perenne L.) has a very similar karyotype. On the other
hand, the karyotype of hexaploid tall fescue (F. arundinacea Schreb.) consists of 21
chromosome pairs of similar size.
The variation in length of diploid grass species’ chromosomes was promising
for separation (and sorting) of individual chromosomes using flow cytometry. The
above mentioned over 10 % difference in chromosome size allowed us to predict that
chromosome 4 would form separate peak on the flow karyotype and generate a pure
sample, while the remaining six chromosomes would form two composite peaks.

13.4 Flow Cytometry and Chromosome Genomics in Grasses

After overcoming problems with the synchronization of cell cycle in such puny little
root tips of germinating seed of meadow fescue, we were able to isolate and stain
mitotic chromosomes for a flow cytometry analysis. As predicted, the flow karyotype
of meadow fescue consists of three peaks (Fig. 13.1). The first (composite) peak rep-
resents three smallest chromosomes—1F, 5F and 6F. Another three chromosomes
(2F, 3F and 7F) form the composite second peak. The third peak represents chromo-
some 4F. Thus, this chromosome is the only one which can be directly sorted from
the standard karyotype. Nowadays, we routinely sort chromosome 4F with purity
exceeding 92 %.
The two composite peaks of the meadow fescue flow karyotype may be parti-
tioned into sections, as done in wheat, theoretically enabling sorting of particular
chromosomes with a reasonable purity. Moreover, special cytogenetic stocks could
be used to sort specific chromosome constructs, such as those used in cereals. The
difference in chromosome size between L. multiflorum and F. pratensis warrants a
prediction that chromosomes 2F, 3F and 7F can be sorted from disomic substitution
lines, where two homologous chromosomes of tetraploid L. multiflorum are substi-
tuted by their homoeologues from F. pratensis (Kopecký et al. 2008a). We do have
such lines available for every chromosome of F. pratensis.

13.5 Applications for Sorted Grass Chromosomes

Sorted chromosomes can be used for a wide range of applications in genomics,

genetics and cytogenetics.
13 Establishing Chromosome Genomics in Forage and Turf Grasses 109

13.5.1 Bacterial Artificial Chromosome (BAC) Libraries

The development of a chromosome specific BAC library is one of the most attractive
uses of flow-sorted chromosomes. The strategy of fingerprinting a well defined part
of the genome (particularly single chromosomes/chromosome arms) and sequencing
clone by clone is certainly more precise and efficient than the whole genome sequenc-
ing, especially in all cases where high proportions of repeats make contig assembly a
challenging proposition. Chromosome specific BAC libraries are attractive resources
for physical mapping by ordering BAC clones according to their fingerprint patterns.
Such libraries have been developed for most of wheat chromosomes/chromosome
arms and the short arm of rye chromosome 1 (1RS) (Šimková et al. 2008; Šafář et al.
2010). The utility of chromosome specific BAC libraries was demonstrated by the
construction of a physical map of chromosome 3B of wheat (Paux et al. 2008).
Until now, only the whole genome BAC libraries have been available for species
of the Festuca—Lolium complex. Donnison et al. (2005) and Farrar et al. (2007)
developed BAC libraries for F. pratensis and L. perenne, respectively. A partial BAC
library for F. pratensis was developed and used for cytogenetic mapping by Kopecký
et al. (2008b, 2010).
Successful sorting of chromosome 4 of Festuca (and presumably of Lolium) opens
the way for the development of a BAC library specific for this chromosome, BAC
libraries for chromosomes 2F, 3F and 7F could be potentially created using single
chromosome substitution lines. Given the above statement on partitioning of com-
posite peaks, we can assume that enriched BAC libraries for other chromosomes
(with purity of ∼65 %) could be developed as well.

13.5.2 Development of Molecular Markers

Chromosome specific BAC libraries are a valuable source of molecular markers. By

BAC End Sequencing (BES), Paux et al. (2008) generated over 700 ISBP (Insertion
Site-Based Polymorphisms) markers for chromosome 3B, which are already used
in breeding programs. Similarly, using BES, Bartoš et al. (2008) and Kofler et al.
(2008) developed ISBP and SSR markers, respectively, specific for the short arm of
chromosome 1R of rye. The advantage of chromosome-based approaches for marker
development is that the chromosomal specificity of markers can be verified by PCR
on flow-sorted chromosomes serving as templates.
Genetic maps are available for both ryegrass species (L. multiflorum and
L. perenne), as well as for tall and meadow fescues. However, the genetic map of
meadow fescue comprises only 466 markers (Alm et al. 2003). The development of
the DArTFest array (Kopecký et al. 2009) enriched the genetic map by additional al-
most 150 DArT markers (Bartoš et al. 2011), however, the number of markers is still
limiting for fine genetic mapping and positional cloning of agronomically important
genes. Thus, a low number of molecular markers for meadow fescue calls for the
110 D. Kopecký et al.

use of chromosome-based approach. Besides the development of markers by BAC-

end sequencing, the example of wheat chromosome 3B shows that a combination of
flow sorting and the DArT technology is fully capable of yielding high numbers of
chromosome specific DArT markers (Wenzl et al. 2010).

13.5.3 Cytogenetic Mapping Using Sorted Chromosomes

Sorted chromosomes can be used as an alternative template for cytogenetic map-

ping, instead of traditional cytological preps squashed or dropped on a microscopic
slide. The advantage of this approach is the purity of a chromosome fraction on a
slide. Moreover, dropped individual chromosomes solve the interference problem
brought about by the presence of cytoplasm and cell wall residues always present on
squashed preparations. Spatial resolution and sensitivity of the technique can also be
significantly improved by using super stretched sorted chromosomes (Valárik et al.
2004).
We used sorted chromosomes as a template for cytogenetic mapping of various
DNA sequences in F. pratensis. Localization of several clones from a partial BAC
library identified all seven chromosomes of F. pratensis. The detailed cytogenetic
mapping of various sequences (BAC clones and repeats) on chromosome 4 showed
the effectiveness of this method. In total, we localized five BAC clones and 13
repeats (centromeric and telomeric repeats and 11 microsatellites). Surprisingly, the
telomeric repeat originating from wheat produced signals not only in the telomeric
regions of both chromosome arms, but also in a proximal region of the long arm (4FL).
Similarly, additional signals were also detected in intercalary regions of chromosome
arms 2FL, 5FL and 6FS (where S and L denote short and long arms, respectively).
These signals might represent positions of ancient structural rearrangements. Signals
in the centromeric/pericentromeric region was also detected by FISH with BAC
clones 1F21, 1G18 (additional signals on both arms), 2B14 (additional signal on the
short arm), 2D4 (additional signal on the long arm) and 2N9 (additional signals on
both arms). Trials with 11 microsatellites (motifs CA, GC, TA, CAG, CAT, CGG,
GAA, GAC, GAG, TAA and TAC) on chromosome 4F did not provide satisfactory
results. Only the microsatellite with the CAG motif provided a signal in the distal part
of the short arm of 4F. No other microsatellites were detected on this chromosome.

13.5.4 Sequencing of Individual Chromosomes

Sequencing of BAC ends from a chromosome specific BAC library provides an

insight into the molecular organization of individual chromosomes. By sequencing
ends of almost 11,000 BAC clones of wheat chromosome 3B-specific library, Paux
et al. (2008) has shown that 86 % of sequences were repeats and only 1.2 % came from
coding regions. The balance were sequences of unknown role or function. Similar
13 Establishing Chromosome Genomics in Forage and Turf Grasses 111

results were obtained by sequencing ends of rye 1RS specific BAC clones (Bartoš
et al. 2008).
Nowadays, the dramatic reduction in cost and increase in effectiveness of the
Next Generation Sequencing enables sequencing of entire chromosomes (and even
genomes) in a reasonable time and for a reasonable cost (Doležel et al. 2009). Mayer
et al. (2011) sequenced individual chromosomes of barley using the Roche 454
technology with 1.28–2.76x coverage. This enables the prediction of total number of
barley genes (∼32,000) and provided a tool for fine comparative studies with model
species: rice, sorghum and Brachypodium. Moreover, a virtual gene order has been
designed for all seven barley chromosomes.
We sequenced chromosome 4F by Illumina HiSeq2000 with the coverage of over
50x. Using the same approach of comparative analysis as Mayer et al. (2011) and
Wicker et al. (2011), we were able to identify collinear regions of F. pratensis
chromosome 4 in genomes of barley and three model species: rice, sorghum and
Brachypodium.

Acknowledgments Authors would like to thank Prof. Adam J. Lukaszewski for critical reading and
valuable comments.This work has been supported by the Czech Science Foundation (grant award
P501/11/0504) and by the Ministry of Education, Youth and Sports of the Czech Republic and the
European Regional Development Fund (Operational Programme Research and Development for
Innovations No. CZ.1.05/2.1.00/01.0007).

References

Alm V, Fang C, Busso CS, Devos KM, Vollan K, Grieg Z, Rognli OA (2003) A linkage map of
meadow fescue (Festuca pratensis Huds.). Theor Appl Genet 108:25–40
Bartoš J, Paux E, Kofler R, Havránková M, Kopecký D, Suchánková P, Šafář J, Šimková H, Town
CD, Lelley T, Feuillet C, Doležel J (2008) A first survey of the rye (Secale cereale) genome
composition through BAC end sequencing of the short arm of chromosome 1R. BMC Plant Biol
8:95
Bartoš J, Sandve SR, Kölliker R, Kopecký D, Christelová P, Stočes S, Østrem L, Larsen A, Kilian
A, Rognli OA and Doležel J (2011) Genetic mapping of DArT markers in the Festuca-Lolium
complex and their use in freezing tolerance association analysis. Theor Appl Genet 122:1133–
1147
Doležel J, Bartoš J, Voglmayr H, Greilhuber J (2003) Nuclear DNA content and genome size of
trout and human. Cytometry 51:127–128
Doležel J, Kubaláková M, Vrána J, Bartoš J (2004) Flow cytogenetics. In: Goodman RM (ed)
Encyclopedia of plant and crop science, Marcel Dekker, Inc., New York, pp 460–463
Doležel J, Šimková H, Kubaláková M, Šafář J, Suchánková P, Číhalíková J, Bartoš J, Valárik M
(2009) Chromosome Genomics in the Triticeae. In: Feuillet C, Muehlbauer GJ (eds) Genetics
and genomics of the triticeae, plant genetics and genomics: crops and models 7. Springer,
Heidelberg, pp 285–316
Donnison IS, O’Sullivan DM, Thomas A, Canter P, Moore B, Armstead I, Thomas H, Edwards
KJ, King IP (2005) Construction of a Festuca pratensis BAC library for map-based cloning in
festulolium substitution lines. Theor Appl Genet 110:846–851
Farrar K, Asp T, Lubberstedt T, Xu M, Thomas AM, Christiansen C, Humphreys MO, Donnison IS
(2007) Construction of two Lolium perenne BAC libraries and identification of BACs containing
candidate genes for disease resistance and forage quality. Mol Breeding 19:15–23
112 D. Kopecký et al.

Kofler R, Bartoš J, Gong L, Stift G, Suchánková P, Šimková H, Berenyi M, Burg K, Doležel J,

Lelley T (2008) Development of microsatellite markers specific for the short arm of rye (Secale
cereale L.) chromosome 1. Theor Appl Genet 117:915–926
Kopecky D, Lukaszewski AJ, Gibeault V (2005) Reduction of ploidy level by androgenesis in
intergeneric Lolium-Festuca hybrids for turf grass breeding. Crop Sci 45:274–281
Kopecký D, Loureiro J, Zwierzykowski Z, Ghesquiere M, Dolezel J (2006) Genome constitution and
evolution in Lolium × Festuca hybrid cultivars (Festulolium). Theor Appl Genet 113:731–742
Kopecký D, Lukaszewski AJ, Doležel J (2008a) Meiotic behaviour of individual chromosomes of
Festuca pratensis in tetraploid Lolium multiflorum. Chromosome Res 16:987–998
Kopecký D, Lukaszewski AJ, Doležel J (2008b) Cytogenetics of Festulolium (Festuca × Lolium
hybrids). Cytogenet Genome Res 120:370–383
Kopecký D, Bartoš J, Lukaszewski AJ, Baird JH, Černoch V, Kölliker R, Rognli OA, Blois H, Caig
V, Lübberstedt T, Studer B, Shaw P, Doležel J, Kilian A (2009) Development and mapping of
DArT markers within the Festuca–Lolium complex. BMC Genomics 10:473
Kopecký D, Havránková M, Loureiro J, Castro S, Lukaszewski AJ, Bartoš J, Kopecká J, Doležel
J (2010) Physical distribution of homoeologous recombination in individual chromosomes of
Lolium multiflorum/Festuca pratensis substitutions. Cytogenet Genome Res 129:162–172
Kubaláková M, Valárik M, Bartoš J, Vrána J, Číhalíková J, Molnár-Lang M, Doležel J (2003)
Analysis and sorting of rye (Secale cereale L.) chromosomes using flow cytometry. Genome
46:893–905
Martienssen RA, Rabinowicz PD, O’Shaughnessy A, McCombie WR (2004) Sequencing the maize
genome. Curr Opin in Plant Biol 7:102–107
Mayer KFX, Martis M, Hedley PE, Šimková H, Liu H, Morris JA, Steuernagel B, Taudien S,
Roessner S, Gundlach H, Kubaláková M, Suchánková P, Murat F, Felder M, Nussbaumer T,
Graner A, Salse J, Endo T, Sakai H, Tanaka T, Itoh T, Sato K, Platzer M, Matsumoto T, Scholz
U, Doležel J, Waugh R, Stein N (2011) Unlocking the Barley Genome by Chromosomal and
Comparative Genomics. Plant Cell 23:1249–1263
Paux E, Sourdille P, Salse J, Saintenac C, Choulet F, Leroy P, Korol A, Michalak M, Kianian S,
Spielmeyer W, Lagudah E, Somers D, Kilian A, Alaux M, Vautrin S, Berges H, Eversole K,
Appels R, Šafář J, Šimková H, Doležel J, Bernard M, Feuillet C (2008) A physical map of the
1-gigabase bread wheat chromosome 3B. Sci 322:101–104
Peterson DG, Wessler SR, PatersonAH (2002) Efficient capture of unique sequences from eukaryotic
genomes. Trends Genet 18:547–550
Rabinowicz PD, Schutz K, Dedhia N, Yordan C, Pernell LD, Stein L, McCombie WR, Martienssen
RA (1999) Differential methylation of genes and retrotransposons facilitates shotgun sequencing
of the maize genome. Nat Genet 23:305–308
Šafář J, Šimková H, Kubaláková M, Číhalíková J, Suchánková P, Bartoš J, Doležel J (2010) De-
velopment of chromosome-specific BAC resources for genomics of bread wheat. Cytogenet
Genome Res 129:211–223
Šimková H, Šafář J, Suchánková P, Kovářová P, Bartoš J, Kubaláková M, Janda J, Číhalíková J,
Mago R, Lelley T, Doležel J (2008) A novel resource for genomics of Triticeae: BAC library
specific for the short arm of rye (Secale cereale L.) chromosome 1R (1RS). BMC Genomics
9:237
Suchánková P, Kubaláková M, Kovářová P, Bartoš J, Číhalíková J, Molnár-Lang M, Endo TR,
Doležel J (2006) Dissection of the nuclear genome of barley by chromosome flow sorting.
Theor Appl Genet 113:651–659
Valárik M, Bartoš J, Kovářová P, Kubaláková M, de Jong JH, Doležel J (2004) High-resolution
FISH on super-stretched flow-sorted plant chromosomes. Plant J 37:940–950
Vrána J, Kubaláková M, Šimková H, Číhalíková J, Lysák MA, Doležel J (2000) Flow-sorting of
mitotic chromosomes in common wheat (Triticum aestivum L.). Genetics 156:2033–2041
Wenzl P, Suchánková P, Carling J, Šimková H, Huttner E, Kubaláková M, Sourdille P, Paul E,
Feuillet C, Kilian A, Doležel J (2010) Isolated chromosomes as a new and efficient source of
DArT markers for the saturation of genetic maps. Theor Appl Genet 121:465–474
13 Establishing Chromosome Genomics in Forage and Turf Grasses 113

Wicker T, Mayer KFX, Gundlach H, Martis M, Steuernagel B, Scholz U, Šimková H, Kubaláková

M, Choulet F, Taudien S, Platzer M, Feuillet C, Fahima T, Budak H, Doležel J, Keller B, Stein N
(2011) Frequent gene movement and pseudogene evolution is common to the large and complex
genomes of wheat, barley, and their relatives. Plant Cell 23:1706–1718.
Zhou RN and Hu ZM (2007) The development of chromosome microdissection and microcloning
technique and its applications in genomic research. Curr Genomics 8:67–72
Chapter 14
DArTFest DNA Array—Applications
and Perspectives for Grass Genetics,
Genomics and Breeding

D. Kopecký, J. Bartoš, A. J. Lukaszewski, J. H. Baird, S. R. Sandve,

O. A. Rognli, R. Kölliker, S. L. Byrne, C. Tomaszewski, S. Barth, A. Kilian,
V. Černoch, M. Klíma, P. Azhaguvel, M. Saha and J. Doležel

Abstract DArTFest is a DNA microarray developed for the Festuca-Lolium com-

plex, consisting of 7,680 probes. As it offers high-throughput and low-cost screening
of thousands of genomic loci, it has been successfully used in a number of projects.
These include the analysis of interspecific, intraspecific and intravarietal genetic
variation, genetic mapping, assigning markers to chromosomal segments, compar-
ative analysis of the Festuca and Lolium genomes against model plant species rice
and Brachypodium, an analysis of genomic constitution of Festulolium cultivars and

D. Kopecký () · J. Bartoš · J. Doležel

Institute of Experimental Botany, Centre of the Region Haná for Biotechnological and Agricultural
Research, Sokolovská 6, 77200 Olomouc, Czech Republic
e-mail: [email protected]
A. J. Lukaszewski · J. H. Baird
Department of Botany and Plant Sciences, University of California, 92521 Riverside, USA
S. R. Sandve · O. A. Rognli
Department of Plant and Environmental Sciences, Norwegian University of Life Sciences,
P.O. Box 5003, 1432, Aas, Norway
R. Kölliker
Agroscope Reckenholz Tanikon Research Station ART, Reckenholzstr 191, 8046 Zurich,
Switzerland
S. L. Byrne
Department of Genetics and Biotechnology, Aarhus University, Forsoegsvej 1, 4200
Slagelse, Denmark
C. Tomaszewski · S. Barth
Teagasc Crops, Environment and Land Use Programme, Oak Park Research, Carlow, Ireland
A. Kilian
Diversity Arrays Technology, 1 Wilf Crane Crescent, 2600 Yarralumla, Australia
V. Černoch · M. Klíma
DLF – Trifolium Hladké Životice, Fulnecká 95, 74247 Hladké Životice, Czech Republic
P. Azhaguvel · M. Saha
Forage Improvement Division, The Samuel Roberts Noble Foundation, Inc, 2510 Sam
Noble Parkway, 73401 Ardmore, OK, USA

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 115

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_14, © Springer Science+Business Media Dordrecht 2013
116 D. Kopecký et al.

QTL mapping of agronomically important traits such as resistance against crown

rust and frost tolerance. In order to facilitate the use of DArT markers linked to traits
of interests, a large number of them has been already sequenced and conversion to
PCR-based markers is in progress.

14.1 The DArTFest Array

The need for a high-throughput genotyping system has led to the development of
various DNA arrays and chips that can be used to screen thousands or hundreds
of thousands genomic loci in a single pass. However, most of them require the
availability of extensive sequence information in the target organism. In contrast,
Diversity Arrays Technology (DArT) is a microarray hybridization based technique
that permits simultaneous screening thousands of polymorphic loci without any prior
sequence information. DArT is high-throughput, low-cost, quick and reproducible.
The DArTFest is the first DArT (Diversity Array Technology) array developed for
forage and turf grasses (Kopecký et al. 2009).
For the development of the DArTFest array, 40 accessions each of L. perenne L.,
L. multiflorum Lam., F. pratensis Huds. and F. arundinacea Schreb, plus all seven
available accessions of F. glaucescens Boiss. were chosen to discover the maximum
genetic variability within the Lolium-Festuca complex and included ecotypes, cul-
tivars and parents of mapping populations. For each of the five species tested, we
developed a library of DArT clones using the PstI/TaqI method of complexity reduc-
tion. The DArTFest array consists of 7,680 probes derived from these methyl-filtered
genomic representations (Kopecký et al. 2009). This contribution reviews the uses of
the DArTFest array and highlights the contribution of this platform to grass genetics,
genomics and breeding.

14.2 Genetic Diversity

The DArT technology was originally developed for fast and low-cost screening of
genetic variability. In line with this, the DArTFest array was initially used for the
evaluation of genetic variability in five species of Festuca-Lolium complex. As ex-
pected, the accessions (ecotypes, cultivars and pre-breeding genotypes) belonging to
individual species always clustered together (Kopecký et al. 2009). Similarly, Pratley
et al. (in preparation) evaluated genetic variability among and within three ryegrass
species (L. perenne, L. multiflorum and L. rigidum) and tall fescue. The observed
intraspecific and interspecific variability was high enough to prevent identification
of individual cultivars, but it was possible to separate different species. On the other
hand, individual plants of five different Festulolium cultivars always clustered based
on their identity (Kopecký et al. 2011). Baird et al. (2011) used DArTFest array
for evaluation of genetic variability in commercial turf-type tall fescues. The anal-
ysis revealed extremely low genetic variability among cultivars being released and
14 DArTFest DNA Array—Applications and Perspectives for Grass Genetics . . . 117

propagated for the US seed market. This may be caused by either a severe genetic
bottleneck during conversion to turf germplasm and/or extensive sharing of breeding
material.

14.3 Genetic and Physical Mapping

DArT markers were used to enrich the existing genetic maps of three grass species.
Out of 2,761 probes, which scored positively in L. multiflorum, 529 DArT markers
were placed on its genetic map (Bartoš et al. 2011). This significantly enriched
original genetic map generated with a total 352 other markers (Studer et al. 2006).
In L. perenne, 297 DArT markers were genetically mapped with a total map length
of 966 cM (Tomaszewski et al. 2011). The average spacing between markers was
reduced from 7.5 cM in the original map (Anhalt et al. 2008) to 1.54 cM. The genetic
map of F. pratensis was enriched by 149 DArT markers to a total 736 cM (Bartoš
et al. 2011). In hexaploid tall fescue (F. arundinacea var. genuina Schreb.), a limited
number of polymorphic DArT markers resulted 115 and 88 mapped DArT markers on
male (1221.9 cM) and female SSR integrated genetic map (1194.3 cM), respectively,
in a pseudo F1 test cross progenies of B400 × W279 (Azhaguvel et al. unpublished).
This is in agreement with findings of much lower genetic variability in fescues
as compared to ryegrass species (Kopecký et al. 2009). DArT markers are also
being used for development of physical map of L. perenne (Ian Armstead, personal
comm.).

14.4 Markers for Introgression Mapping

Using seven substitution lines of tetraploid L. multiflorum, each carrying one chro-
mosome of F. pratensis (Kopecký et al. 2010), we were able to anchor 160 DArT
markers to individual F. pratensis chromosomes. The anchoring could potentially
be made at sub-chromosomal level using recombinant lines, where markers would
be anchored to small chromosome segments—bins. Chromosome substitution lines
could also be developed for Lolium chromosomes in Festuca genome and anchor
DArT markers to Lolium chromosome bins.

14.5 Identification of Hybrids and Their Genomic Constitution

DArTFest array was used for discrimination of parental species in interspecific hy-
brids (Kopecký et al. 2011). Hybrid origin was proved in all plants of five Festuca ×
Lolium cultivars. Each cultivar was represented by 20 plants. The analysis of ge-
nomic constitution revealed differences among cultivars from intermediate types to
introgression forms. Interestingly, plants belonging to individual cultivars always
118 D. Kopecký et al.

formed tight clusters. Such sensitivity and resolution, which was never achieved
before, calls for broader use of DArTFest array, which was confirmed to be suitable
for low-cost and rapid characterization and protection of existing and newly released
Festulolium cultivars.

14.6 Association Mapping and Marker Assisted Selection

Marker Assisted Selection (MAS) is said to make the breeding process more efficient
and faster. However, its efficient application requires markers linked to traits of
interest. Bartoš et al. (2011) were able to identify 96 DArT markers with significantly
different distribution between high freezing tolerant and low freezing tolerant plants,
but only five of these markers could be mapped on the F. pratensis genetic map.
Colinearity with sequenced Brachypodium genome indicated the localization of a
freezing tolerance QTL in the central part of chromosome 4 and a distal part of the
short arm of chromosome 2 in Lolium and Festuca. Similarly, Tomaszewski et al.
(2011) were able to identify QTLs for crown rust resistance in linkage groups 2, 3,
4 and 7. The same approach is in progress for identification of QTLs for snow mold
and drought tolerance.

14.7 Sequence Analysis of DArT Markers

Sequencing 620 genetically mapped DArT markers provided first insights into the
origin and the nature of markers on DArTFest array (Bartoš et al. 2011). Among them,
398 (64.2 %) were singletons. The remaining 222 markers were redundant and as-
signed to 90 marker bins, which consisted of two to six markers. Blast search against
an in-house built composite plant repeat database indicated that only 44 (7.1 %) of
DArT markers contained repetitive elements. These observations support the no-
tion that DArT markers represent low-copy genomic regions (Wenzl et al. 2006).
Moreover, blast searches against non-redundant protein sequences and expressed
sequences revealed that a majority of DArT markers were potential gene-derived
sequences (Bartoš et al. 2011). The availability of sequences from DArT markers
provides an opportunity to convert them to PCR-based markers for future application
in breeding and gene discovery.

Acknowledgments This work has been supported by the Ministry of Education, Youth and Sports
of the Czech Republic and the European Regional Development Fund (Operational Programme
Research and Development for Innovations No. CZ.1.05/2.1.00/01.0007) and by the Ministry of
Agriculture of the Czech Republic (grant award NAZV QH71267).
14 DArTFest DNA Array—Applications and Perspectives for Grass Genetics . . . 119

References

Anhalt UCM, Heslop-Harrison P, Byrne S, Guillard A, Barth S (2008) Segregation distortion in

Lolium: evidence for genetic effects. Theor Appl Genet 117:297–306
Baird JH, Kopecký D, Lukaszewski AJ, Green RL, Bartoš J, Doležel J (2011) Genetic diversity of
turf-type tall fescue using Diversity Arrays technology. Crop Science 52:408–412
Bartoš J, Sandve SR, Kölliker R, Kopecký D, Christelová P, Stočes S, Østrem L, Larsen A, Kilian A,
Rognli OA, Doležel J (2011) Genetic mapping of DArT markers in the Festuca-Lolium complex
and their use in freezing tolerance association analysis. Theor Appl Genet 122:1133–1147
Kopecký D, Bartoš J, Lukaszewski AJ, Baird JH, Černoch V, Kölliker R, Rognli OA, Blois H, Caig
V, Lübberstedt T, Studer B, Shaw P, Doležel J, Kilian A (2009) Development and mapping of
DArT markers within the Festuca–Lolium complex. BMC Genomics 10:473
Kopecký D, Havránková M, Loureiro J, Castro S, Lukaszewski AJ, Bartoš J, Kopecká J, Doležel
J (2010) Physical distribution of homoeologous recombination in individual chromosomes of
Lolium multiflorum/Festuca pratensis substitutions. Cytogen Genome Res 129:162–172
Kopecký D, Bartoš J, Christelová P, Černoch V, Kilian A, Doležel J (2011) Genomic constitution
of Festuca × Lolium hybrids revealed by the DArTFest array. Theor Appl Genet 122:355–363
Studer B, Boller B, Herrmann D, Bauer E, Posselt UK, Widmer F, Kölliker R (2006) Genetic
mapping reveals a single major QTL for bacterial wilt resistance in Italian ryegrass (Lolium
multiflorum Lam.). Theor Appl Genet 113:661–671
Tomaszewski C, Byrne S, Foito A, Kildea S, Kopecky D, Dolezel J, Heslop-Harrison JS, Stewart
D and Barth S (2011) Genetic linkage mapping in an F2 perennial ryegrass population using
DArT markers. Plant Breeding 131:345–349
Wenzl P, Li H, Carling J, Zhou M, Raman H, Paul E, Hearnden P, Maier C, Xia L, Caig V, Ovesná
J, Cakir M, Poulsen D, Wang J, Raman R, Smith KP, Muehlbauer GJ, Chalmers KJ, Kleinhofs
A, Huttner E, Kilian A (2006) A high-density consensus map of barley linking DArT markers
to SSR, RFLP and STS loci and agricultural traits. BMC Genomics 7:206
Chapter 15
Using DArT Markers in Festuca × Lolium
Breeding

M. Ghesquière, J.-L. Durand, T. Bourgoin, E. Huttner and A. Kilian

Abstract DArT technology was applied to a tetraploid (2n = 4x = 28) L. multi-

florum × F. glaucescens hybrid lineage over ten generations. By selecting 151
Lolium- and 210 Festuca-specific DArT markers among the 3,884 ones devel-
oped by Kopecký et al. (Development and mapping of DArT markers within the
Festuca–Lolium complex, BMC Genomics 10:473, 2009), it is shown that DArT
polymorphism is well consistent with the history of the plant material and the events
in relation with interspecific hybridisation: amphiploidisation, introgression into 4x-
and 2x-L. multiflorum, reduction of effective size, chromosome mapping as well as
parent-specific response to summer water deficit. In this respect, frequency of Fes-
tuca markers within a 4x-BC1 population was found to have increased on average by
2.4 % among plants having survived in sward after summer while frequency de-
creased by 10.2 % after three generations of seed multiplication from the initial
polycross of BC1 parents.

Keywords Molecular markers · Interspecific hybridisation · Marker-assisted

introgression · Drought tolerance

15.1 Introduction

Since 2004, Festulolium definition covers any hybrid variety between Festuca sp.
and Lolium sp. in agreement with EU directive (2004/55/EC). If both L. multiflorum
and L. perenne have been involved at about the same extent in Festuca × Lolium
hybridisation, only F. pratensis has been widely used as Festuca parent so far. In

M. Ghesquière () · J.-L. Durand

French National Institute for Agricultural Research, INRA, Unit of Multidisciplinary Research
on Pastures and Forage Crops, 86600 Lusignan, BP6, France
e-mail: [email protected]
T. Bourgoin
Platform for Plant Technology Innovation, AgriObtentions, 86600 Lusignan, BP6, France
E. Huttner · A. Kilian
Diversity Array Technology Pty Ltd, 1 Wilf Crane Cr., PO Box 7141, 2600 Yarralumla,
ACT, Australia

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 121

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_15, © Springer Science+Business Media Dordrecht 2013
122 M. Ghesquière et al.

particular, F. glaucescens (2n = 4x = 28) is not so well documented although,

as F. pratensis, it is a progenitor of the modern hexaploid tall fescue, F. arundi-
nacea (2n = 6x = 42). In amphiploid Festulolium hybrids, F. glaucescens enables
more homologous chromosome paring and more stability over generation compar-
atively to amphiploids derived from F. pratensis. Better persistency provided by
F. glaucescens relatively to F. pratensis was also established from recent registration
of L. multiflorum × F. glaucescens amphiploid cultivars on the French National list
(Ghesquière et al. 2010). Deep water extraction in summer from long-living rooting
system was demonstrated to be a key-function in Festuca for improving summer
drought tolerance in Lolium sp. (Durand et al. 2007).
From an evolutionary point of view, all Festulolium varieties including amphiploid
forms from F. pratensis have a marked shift towards Lolium genome, which is even
more enhanced following backcross into Lolium sp., further breeding and/or seed
multiplication (Canter et al. 1999; Kopecký et al. 2006; Zwierzykowski et al. 2006).
Although genome in-situ hybridisation (GISH) method is effective to describe evo-
lution at this scale, a more high-throughput and accurate technology is lacking for a
quantitative survey of genetic changes over Festulolium breeding. DArT technology
based on DNA restriction and hybridisation on a micro-array revealed to fulfil these
requirements of robust system for broad genome coverage (Jaccoud et al. 2001). It
was first applied within the Festuca/Lolium complex of species by Kopecký et al.
(2009) with successful application to genotyping of Festulolium hybrids (Kopecký
et al. 2011) and to mapping in L. multiflorum and F. pratensis (Bartoš et al. 2011).
This paper reports quantitative genome changes in a Festulolium hybrid plant
material derived from F. glaucescens. This was achieved by sampling plant individ-
uals over a ten generations L. multiflorum × F. glaucescens hybrid lineage ranging
from amphiploid forms up to 4x- and 2x-introgression forms into L. multiflorum,
and by using 2149 DArT markers. The assay also included the evaluation of DArT
marker response to drought by sampling, after summer, surviving plants within a
dense sward of an introgression Festulolium population.

15.2 Materials and Methods

A 94-DNA sample microplate was elaborated from one individual of F. arundinacea

(2n = 6x = 42) as control, 3 F1-hybrids between L. multiflorum and F. glaucescens
(2n = 4x = 28) and the ten parents of the amphiploid Festulolium cultivar Lueur
of F4 generation. Furthermore, 73 individuals deriving from one generation of back-
crossing (BC1) amphiploid hybrids into 4x-L. multiflorum were added. They included
nine out the ten parents of the cultivar F99.4 and 64 individuals of the commercial seed
generation, i.e. following three generations of seed multiplication the initial poly-
cross. Thirty-three individuals among the 64 ones were randomly sampled within a
population whose plants were continuously maintained in pots after germination and
hence, never knew sward conditions as well as limiting resources. By contrast, 31
15 Using DArT Markers in Festuca × Lolium Breeding 123

individuals came from a plot broadcast sown in spring 2005 which went through a se-
vere summer water-deficit before sampling in September 2006. The last 7 individuals
were diploid (2n = 2x = 14) after two generations of seed multiplication of a BC4
introgression mapping population of F. glaucescens chromosome 5 (SAGES 2004).
Thus, the whole lineage encompassed ten generations, all deriving from the same
original F. glaucescens source, either of 4x- or 2x-ploidy level and of amphiploid or
introgression nature.
Ten μl of purified DNA at a concentration of 100 ng/μl each sample was pro-
vided at Diversity Array Technology Pty Ltd (DArT P/L, Yarralumla, Australia).
All individuals were genotype-profiled by using the same procedure described by
Kopecký et al. (2009) and a set of 2,149 DNA probes, of which 1,667 belonged to
the initial DNA library of 3,884 polymorphic clones. This library was developed
from 39 individuals of Festuca arundinacea, 40 and 7 individuals of F. pratensis and
F. glaucescens (respectively) and 40 individuals of each of the two diploid Lolium
species, L. perenne and L. multiflorum. Although unbalanced in terms of number
of individuals, all species were expected to have equally contributed to the final
DArT array of 3,884 polymorphic clones by the initial production of 1,536 restricted
DNA-clones each species.
1,667 markers among the 2,149 assayed had individual score published
(http://www.biomedcentral.com/content/supplementary/1471-2164-10-473S1.xls)
which enabled to compute marker frequency within each species. A marker was
hold as specific of any species when its frequency was less than 0.01 in the 4
other species. Fa-Fg markers, i.e. Festuca-specific markers but not of F. pratensis
origin, were defined by merging scores from F. arundinacea and F. glaucescens
due to unbalanced plant sampling. Ambiguous DArT scores were considered as
missing data and were not further taken into account in computing. Five hundred
and eleven markers among 1,667 ones were thus found to be species-specific: 205
Fa-Fg markers, 151 Lm-markers, 126 Fp-markers and 29 Lp-markers. Note that
species-specific markers so defined may have mean frequency not necessarily closed
to 1 within the species they are expected to be specific; mean frequency of Fa-Fg
markers and Lm-markers was 0.910 and 0.740 (respectively) in the initial sampling
of the DArTFest array. The rate of Festuca marker introgression over the Lm × Fg
lineage was computed as: Fa-Fg score/205/(Fa-Fg score/205 + Lm score/151). A
restricted set of 139 Fa-Fg markers and 122 Lm-markers was further defined on the
basis to be present in the 9 parents who were available in the initial polycross of the
Festulolium cvF99.4. Changes from polycross to commercial seed generation were
especially surveyed at those markers. In respect with mapping, Bartoš et al. (2011)
mapped 530 DArT markers in L. multiflorum. The present assay of 2,149 markers
included 262 of those mapped markers, of which 29 belonged to the 151 Lm-specific
markers previously identified.
Multiple correspondence analysis (MCA) was performed by using the Corresp
procedure of SAS/STAT software (release 8.1 for 320 SunOS; SAS Institute Inc.,
Cary, NC, USA) to point out marker associations between individuals at any gener-
ation or plant sampling. Markers of extremely high or low frequency were used only
as supplementary variables in MCA. The response of plant sampling in sward vs
124 M. Ghesquière et al.

Table 15.1 Number of polymorphic DArT markers and Festuca rate of introgression over
L. multiflorum × F. glaucescens lineage. Selection of 205 Fa-Fg and 151 Lm-specific markers was
performed out 1667 DArT markers; mean and standard deviation of frequency over markers refer
to the original sampling of F. arundinacea + F. glaucescens and L. multiflorum used to produce
the original DNA clone library (Kopecký et al. 2009). Number of lost DArT markers from the 122
Fa-Fg and 139 Lm markers initially present in the parent polycross of the cv F99.4 in brackets
Fa-Fg markers Lm markers Rate of Fa-Fg introgression
Total number 205 151 –
Mean frequency 0.910 0.740 –
Standard deviation 0.190 0.226 –
F. arundinacea 173 30 0.809
F1 hybrids 180 115 0.536
Cv Lueur parent polycross 197 129 0.529
Cv F99.4 parent polycross 122 139 0.393
Cv F99.4 seed
Control 90 (32) 136 (3) 0.345
After summer 104 (18) 137 (2) 0.403
Introgression mapping population 10 (112) 118 (21) 0.086

control was tested through X2 statistics at 1 df over all markers and for each marker
individually; Fisher’s exact test was performed when X2 requirement was not met,
e.g. class size higher than 5.

15.3 Results

15.3.1 Rate of Festuca-Marker Introgression Over Generation

One-hundred and seventy-three Fa-Fg markers gave a positive score (84.4 %) onto the
F. arundinacea control, that is closely to the mean frequency (0.910) of those markers
within the initial sampling of F. arundinacea-F. glaucescens (Table 15.1). On the
other hand, 30 Lm-markers among 151 (19.9 %) gave unexpected positive response.
Among F1 hybrids and parent polycross of the Festulolium cv Lueur, the Festuca rate
of introgression was estimated to be 0.536 and 0.529 (respectively), i.e. not far from
the 1:1 Festuca:Lolium genome balance expected from strict amphiploid hybrids. By
contrast, the rate of Fa-Fg markers strongly decreased to 0.393 in the cv F99.4 parent
polycross, that is after four polycrossing generations following the initial BC1. Since
DArT markers are of dominant inheritance, only one generation of back-crossing
amphiploid hybrids into L. multiflorum would have not made the rate of Festuca
introgression to depart from the 0.5 rate expected under strict disomic inheritance.
In the commercial seed generation of the cv F99.4, the rate of introgression also
contrasted according to the origin of the markers and sampling within population.
Only 3 and 2 Lm-specific makers were lost in the seed control and in the sample of
plants collected after summer (respectively) while the corresponding loss of Fa-Fg
15 Using DArT Markers in Festuca × Lolium Breeding 125

Table 15.2 Change of mean frequency at the 122 Fa-Fg and 139 Lm DArT markers initially present
in the parent polycross of the introgression cv F99.4. Allele (+) mean frequency estimated from
polymorphic markers within population only, assuming gene equilibrium and dominant inheritance
of the markers. Standard deviation of mean frequency in brackets
Fa-Fg markers Lm-markers
Mean Allele (+) Mean Allele (+)
marker frequency mean frequency marker frequency mean frequency
Cv Lueur parent
polycross 0.674 (0.007) – 0.449 (0.009) –
Cv F99.4 parent
polycross 0.267 (0.006) – 0.575 (0.010) –
Cv F99.4 seed
Control 0.153 (0.003) 0.0647 (0.0007) 0.567 (0.004) 0.2546 (0.0015)
After summer 0.177 (0.003) 0.0647 (0.0007) 0.570 (0.004) 0.2514 (0.0015)
Introgression map- 0.018 (0.002) 0.1154 (0.0050) 0.494 (0.011) 0.4150 (0.0068)
ping population

markers reached 32 and 18 only (respectively). The decrease was even more enhanced
as well as unbalanced in the BC4 introgression mapping population by the loss of 21
Lm-markers against 112 Fa-Fg markers. All indicates that intense selection against
Festuca genome occurs following primary introgression in L. multiflorum but that
sampling in the droughted sward enabled to partly reverse the rate of introgression
towards its value in the initial polycross.

15.3.2 Response to Summer Drought

Mean frequency score of the 139 Lm-markers slightly increased from 0.449 in the
cv Lueur parent polycross to about 0.57 in the cv F99.4 whatever generation. (Ta-
ble 15.2). On the contrary, mean frequency of Fa-Fg markers decreased from 0.674
to 0.267 following back-crossing. However, Fa-Fg marker mean frequency was
found significantly higher among the plants sampled after summer (0.177 vs 0.153;
P < 0.0037) while that of Lm-markers remained almost perfectly stable on average
(0.570 vs 0.567; P < 0.7972).
Eight contrasts of frequency between sample in sward and seed control were
significant (P < 0.05, one-tailed X2 or Fisher’s exact test) out 132 Lm-markers against
9 significant contrasts out 110 Fa-Fg markers. Global error I of significant individual
tests was estimated to be <0.067 for Fa-Fg markers and <0.267 for Lm-markers.
Furthermore, the 8 significant Lm-marker contrasts divided into 4 negative against 4
positive ones while all the 9 Fa-Fg marker contrasts but one were positive, suggesting
that Festuca markers tended overall to greater positive unidirectional response than
Lolium markers. Unfortunately, no Lm-marker of significant response was found
already mapped by Bartoš et al. (2011).
126 M. Ghesquière et al.

0.25

2nd Factor
0.20

0.15 Initial parent

Polycross
After summer
0.10 drought

0.05

0.00

-0.05 Control

-0.10

-0.15

1st Factor
-0.20
-0.20 -0.10 0.00 0.10 0.20 0.30 0.40 0.50

Fig. 15.1 Multiple correspondence analysis plotting on the two first factors (13.6 % and 9.4 % of
total inertia, respectively) the coordinates of nine BC1 parents of the Festulolium cv F99.4 (+)
and 64 individuals of commercial seed generation: 33 individuals as control (◦) vs 31 individuals
sampled in sward after summer (•). Individual scores at 139 Fa-Fg and 122 Lm DArT markers of
which, 21 and 40 markers (respectively) used as supplementary variables. Arrows indicate contrary
variation of Fa-Fg marker frequency through three generations of seed multiplication vs response
to summer drought within commercial seed generation

Comparison of mean frequency with rate of introgression indicates that apparent

increase of Fa-Fg markers by sampling within the cv F99.4 resulted from retain-
ing numerous polymorphic Fa-Fg markers rather than increasing marker frequency.
Multiple correspondence analysis illustrates how sampling plants after summer has
made to reverse the steady loss of Festuca markers over three generations of seed
multiplication (Fig. 15.1).
Estimate of allele frequency at polymorphic Fa-Fg markers, assuming gene equi-
librium within population, was close to 0.065 in the cv F99.4 against 0.115 in the
diploid introgression mapping population, i.e. far, in both cases, from the 0.25 fre-
quency expected from BC1 introgression and no selection occurring (Table 15.2).
Within the tetraploid cv F99.4, this suggests that introgressed Fa-Fg markers should
be predominantly present as simplex genotype (LLLF) rather than as genotype of
higher Festuca dosage (duplex LLFF).

15.4 Discussion

The DArT markers which were experimented enabled to quantify genome changes
across a diversified L. multiflorum × F. glaucescens lineage. It agreed well with the
global scheme of Festulolium variety evolution (Ghesquière et al. 2010) based on
15 Using DArT Markers in Festuca × Lolium Breeding 127

GISH assessment (Kopecký et al. 2006; Zwierzykowski et al. 2006). In this model,
only L. multiflorum × F. glaucescens amphiploids had evolution rate considerably
slow down in comparison with amphiploids from F. pratensis. Higher preferential
homologous chromosome pairing and hence, disomic inheritance in L. multiflorum
× F. glaucescens amphiploids effectively prevent that Festuca genome is little by lit-
tle eliminated over generations. However, rapid loss of Festuca chromosomes occurs
following backcrossing into Lolium sp. Using isozyme loci and 4x-BC2 introgression
populations (Ghesquière et al. 2000), the rate of Festuca introgression was found
not to decrease so strongly as from DArT makers. Possibly, stronger interspecific
linkage disequilibrium in BC1 than in BC2 population enhanced higher selection
rate against Festuca loci. Within 2x recombinant populations, similar frequency and
rate of transmission over generations was found among homeoalleles at Pgi-2 locus,
whatever they come from F. pratensis or F. glaucescens (Humphreys and Ghesquière
1994). Frequency of DArT markers after two generations of seed multiplication a
F. glaucescens chromosome 5 introgression population indicated rather stronger se-
lection. As isozyme loci (e.g. Pgi-2) were not located on chromosome 5, it is possible
that transmission rate varies according to Festuca chromosomes. Since tetraploid
F. glaucescens is already of amphiploid nature, transmission and recombination rate
into Lolium sp. may also differ between homologous chromosome pairs.
Response to natural selection feeds thinking about breeding for abiotic tolerance
by using a Festuca × Lolium hybridization approach. Few evidences have been re-
ported in literature. True amphiploid forms are only rarely found in nature; however,
it was reported that 3x-hybrids can be continuously produced from unreduced ga-
metes of diploid L. perenne and F. pratensis and that, depending on environmental
conditions, selection makes 2x-introgression forms to develop either of predominant
Lolium or Festuca genome (e.g. Humphreys and Harper 2008; Casler et al. 2002,
2009). Under artificial conditions, survival in rain-out shelters was enhanced within
BC2 populations from F. arundinacea × L. multiflorum hybrids (Humphreys 1989;
Humphreys and Thomas 1993). Extending this introgression scheme to indepen-
dent plant material, it was found that F. pratensis chromosome 3 and F. glaucescens
chromosome 5 carry favourable genes of drought tolerance (SAGES 2004), which
was latter confirmed following a QTL approach within F. pratensis (Alm et al. 2011)
and, in the field, by 2x-recombinant populations close to finished variety (Humphreys
et al. 2011).
We showed herein that natural selection similarly operated within a sward of a
segregating 4x-BC1 population in the span of few months of strong water deficit.
However, the positive response of Festuca markers (+2.4 %) appeared to not balance
the loss at each generation of seed multiplication (−3.4 %). Consequently, breeding
for Festuca traits within introgression populations may be not fully effective unless
it is early assisted by genotyping at a whole genome scale as provided by DArTs.

Acknowledgments The authors thank Denise Cadier, Françoise Durand and Claudine Largeaud
for help and careful management of all living plants and DNA materials required for this study.
128 M. Ghesquière et al.

References

Alm V, Busso CS, Ergon Å, Rudi H, Larsen A, Humphreys MW, Rognli OA (2011) A QTL analysis
and comparative genetic mapping of frost tolerance, winter survival and drought tolerance in
meadow fescue (Festuca pratensis Huds.). Theor Appl Genet 123(3):369–382
Bartoš J, Sandve SR, Kölliker R, Kopecký D, Christelová P, Stočes S, Østrem L, Larsen A, Kilian A,
Rognli OA, Doležel J (2011) Genetic mapping of DArT markers in the Festuca-Lolium complex
and their use in freezing tolerance association analysis. Theor Appl Genet 122:1133–1147.
doi: 10.1007/s00122-010-1518-z
Canter PH, Pašakinskiene I, Jones N, Humphreys MW (1999) Chromosome substitutions and
recombination in the amphiploid Lolium perenne × Festuca pratensis cv Prior (2n = 4x = 28).
Theor Appl Genet 98:809–814
Casler MD, Peterson PR, Hoffman LD, Ehlke NJ, Brummer EC, Hansen JL, Mlynarek MJ, Sulc MR,
Henning JC, Undersander DJ, Pitts PG, Bilkey PC, Rose-Fricker CA (2002) Natural selection
for survival improves freezing tolerance, forage yield and persistency of Festulolium. Crop Sci
42:1421–1426
Casler MD, Van Santen E, Humphreys MW, Yamada T, Tamura K, Ellison NW, Jackson RD,
Undersander DJ, Gildersleeve R, Stendal C, Reiter M, Opitz C (2009) Remnant oak savanna
acts as refugium for meadow fescue introduced during nineteenth century human migration in
the USA. In: Yamada T, Spangenberg G (eds) 5th International Molecular Breeding of Forage
and Turf (2007), Sapporo, Japan, Springer, Heidelberg
Durand JL, Bariac T, Ghesquière M, Brion P, Richard P, Humphreys MW, Zwierzykowski Z (2007)
Ranking of the depth of water extraction by individual grass plants using natural 18O isotope
abundance. Environ Exper Bot 60(1):137–144
Ghesquière M, Barre P, Marhadour S, Kerlan MC (2000) Estimation of introgression rate of a fescue
isozymic marker into tetraploid Italian ryegrass at early generations of backcross. Euphytica
114:223–231
Ghesquière M, Humphreys MW, Zwierzykowski Z (2010) Festulolium. In: Boller B, Posselt UK,
Vernonesi F (eds) Fodder crops and amenity grasses, handbook of plant breeding, Springer
Science + Business Media LLC, Heidelberg, 293–316. doi: 10.1007/978-1-4419–0760-8_12
Humphreys MW (1989) The controlled introgression of Festuca arundinacea genes into Lolium
multiflorum. Euphytica 42:105–116
Humphreys MW, Ghesquière M (1994) Assessing success in gene transfer between Lolium
multiflorum and Festuca arundinacea. Euphytica 77:283–289
Humphreys MW, Harper J (2008) Festulolium loliaceum, an understudied natural UK grass hybrid
species that may provide benefits to UK grasslands withstanding the onsets of climate change.
Crop wild relative 6:7–9
Humphreys MW, Thomas H (1993) Improved drought resistance in introgression lines derived from
Lolium multiflorum × Festuca arundinacea hybrids. Plant Breed 111:151–161
Humphreys M, Macleod K, Whalley R, Turner L, Farrel M, Haygarth P (2011) A multifunc-
tional grass cultivar for agriculture and flood control. In: proceedings of the 29th eucarpia
fodder crops and amenity grasses section meeting, Dublin, Ireland, 4–8 Sept. 2011, Springer
Science + Business Media B.V. (these proceedings)
Jaccoud D, Peng K, Feinstein D, Kilian A (2001) Diversity arrays: a solid-state technology for
sequence information independent genotyping. Nucleic Acids Res 29:E25
Kopecký D, Loureiro J, Zwierzykowski Z, Ghesquière M, Doležel J (2006) Genome constitution and
evolution in Lolium × Festuca hybrid cultivars (Festulolium). Theor Appl Genet 113:731–742
Kopecký D, Bartoš J, Lukaszewski AJ, Baird JH, Černoch V, Kölliker R, Rognli OA, Blois H, Caig
V, Lübberstedt T, Studer B, Shaw P, Doležel J, Kilian A (2009) Development and mapping of
DArT markers within the Festuca - Lolium complex. BMC Genomics 10:473. doi:10.1186/1471-
2164-10–473. http://www.biomedcentral.com/1471-2164/10/473
15 Using DArT Markers in Festuca × Lolium Breeding 129

Kopecký D, Bartoš J, Christelová P, Černoch V, Kilian A, Doležel J (2011) Genomic constitution

of Festuca × Lolium hybrids revealed by the DArTFest array. Theor Appl Genet 122:355–363.
doi: 10.1007/s00122–010-1451-1
SAGES (2004) Sustainable Grasslands Withstanding Environmental Stresses, 5th PCRDT shared
cost project QLK5-CT-2000-00764 for 2001–2003, Key-Action 5.1.1. Sustainable Agriculture,
Technological Implementation Plan, European Commission, Belgium, p 56, http://www.sages-
eu.co.uk/
Zwierzykowski Z Kosmala A, Zwierzykowska E, Jones N, Joks W, Bocianowski J (2006) Genome
balance in six successive generations of the allotetraploid Festuca pratensis × Lolium perenne.
Theor Appl Genet 113:539–547
Chapter 16
Development of an SNP Identification Pipeline
for Highly Heterozygous Crops

T. Ruttink, L. Sterck, E. Vermeulen, A. Rohde and I. Roldán-Ruiz

Abstract Next Generation Sequencing technologies significantly advance the devel-

opment of molecular markers for molecular breeding. Dedicated NGS data-analysis
procedures must be developed for de novo reference assembly and SNP discovery
for crop species without a reference genome sequence. In outcrossing fodder crops,
the high degree of polymorphism hampers de novo assembly, contig clustering, read
mapping, and SNP discovery. Using selected candidate genes as case studies, we
illustrate the reconstruction of a reference transcript sequence from RNA-seq data
from multiple genotypes, we validate de novo transcript assembly by Sanger se-
quencing, and analyse how read mapping and SNP discovery parameters determine
sensitivity and specificity during SNP discovery. Thus, we propose a general strategy
to construct a non-redundant reference transcriptome for crops without a sequenced
genome, using predicted proteins from a closely related model species as a guidance
for clustering and annotation. This reference transcriptome is required for candidate
gene discovery and exome-wide identification of polymorphisms.

Keywords Lolium perenne · Next generation sequencing · Gene discovery · SNP

identification · Candidate gene

16.1 Introduction

The genomes of outcrossing fodder crops are characterized by a high degree of het-
erozygosity and heterogeneity within populations and cultivars. This high level of ge-
netic diversity can be exploited for the development of molecular markers for linkage

T. Ruttink () · A. Rohde · I. Roldán-Ruiz

Institute for Agricultural and Fisheries Research (ILVO), Caritasstraat 21, 9090 Melle, Belgium
e-mail: [email protected]
L. Sterck
Department of Plant Systems Biology, VIB, Technologiepark 927, 9052 Gent, Belgium
Department of Plant Biotechnology and Bioinformatics, Gent University,
Technologiepark 927, 9052 Gent, Belgium
E. Vermeulen
Faculty of Applied Engineering Sciences, University College Ghent., Schoonmeersstraat 52,
9000 Gent, Belgium

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 131

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_16, © Springer Science+Business Media Dordrecht 2013
132 T. Ruttink et al.

map construction, association genetics or genomic selection. Next generation se-

quencing (NGS) technologies significantly advance the process of gene discovery
and marker development (Shendure and Ji 2008; Varshney et al. 2009). Here, we
propose and validate bioinformatics procedures to identify DNA-polymorphisms in
candidate genes in Lolium perenne. First, because a fully annotated reference genome
sequence is not yet available for L. perenne, we performed de novo assembly of Il-
lumina RNA-seq data from 14 independent L. perenne genotypes, which can be
mined for candidate genes. Second, we selected orthologs of TB1 and four candidate
genes in the MAX/RMS/DAD pathway that act in the strigolactone biosynthesis and
response pathways, which control branching and plant architecture (Leyser 2009).
These candidate genes were PCR-amplified from the 14 genotypes used for NGS se-
quencing, cloned, and Sanger sequenced to validate de novo transcriptome assembly.
Third, these Sanger sequence data were used to analyse the genetic diversity, and to
evaluate parameters for read mapping and SNP discovery. Based on these findings,
we propose a general strategy for de novo assembly of a reference transcriptome and
subsequent SNP discovery in highly polymorphic crop species.

16.2 Material and Methods

Ilumina GAII transcriptome sequencing Fourteen L. perenne genotypes repre-

senting different architectural types were selected from an association mapping
population of 602 genotypes. Two independent samples of basal stem tissue con-
taining axillary meristems were collected from each genotype. After RNA extraction
(RNeasy including DNase treatment; Qiagen) and quality control (Agilent 2100 Bio-
analyzer), 5 μg of total RNA from the replicated samples were pooled per genotype,
and used for non-normalised, indexed mRNA-Seq library preparation (mRNA-Seq
Sample Prep Kit; Illumina, protocol version 1004898 Rev. D, September 2009; and
Multiplexing Sample Preparation Oliogonucleotide Kit; Illumina). The concentra-
tions (Qubit fluorometer; Invitrogen), the insert size (range between 100–200 bp)
and purity of the libraries were determined (Agilent 2100 Bioanalyzer). Equimolar
amounts of two libraries were pooled per lane of a paired-end flow cell (v4). Clus-
tering was conducted on an Illumina Cluster Station followed by 76 + 7 + 76 cycle
sequencing on an Illumina Genome Analyzer (version IIx, SCS 2.6.26/RTA 1.6.47.1)
using the Genomic DNA Sequencing Primer, the Index Seq Primer, the Multiplex-
ing Rd2 Seq Primer, and clustering and sequencing kits (SBS v4; Illumina). About
15–30 million paired-end reads were obtained per genotype. After trimming for low
quality bases (limit 0.01; minimal length 25 bp; CLCbio), de novo assembly was
performed with the CLCbio Genomics Workbench (v4.0.3) using default parameters,
obtaining so-called CLC-contigs. A second round of clustering of the CLC-contigs
was performed with CAP3 (Huang and Madan 1999) with default settings using
in-house developed Perl scripts, yielding so-called CAP3-contigs.
PCR amplification, cloning, and Sanger sequencing of candidate genes For
each candidate gene, primers were designed with the Primer3 software in SNP-free
16 Development of an SNP Identification Pipeline for Highly Heterozygous Crops 133

regions of the consensus sequence obtained after the second round of clustering with
CAP3. Fragments of candidate genes were PCR-amplified from genomic DNA and
cloned into the TOPO TA cloning vector (Invitrogen). For each genotype, six colony-
PCR products per amplified fragment were verified by agarose gelelectrophoresis.
Of these, four colony-PCR products per genotype were randomly taken, purified,
and Sanger-sequenced on both strands.
Identification of polymorphisms We determined the degree of polymorphism in
the candidate genes in two independent ways. First, the quality-trimmed 76 bp
paired-end reads were mapped to the full-length consensus sequence for each gene
to discover SNPs (referred to as read mapping SNPs or ‘RM-SNPs’). NGS read
mapping parameters were: 90 % length fraction, 95 % similarity, in/del cost = 3,
mismatch cost = 2. SNP discovery parameters were: window length = 11, max gap
or mismatch count = 7, min average qual = 10, min central qual = 15, min read
depth = 15, min SNP frequency = 14 %, 10 % or 7 % (≈4, 3 or 2 out of 28 alleles).
Second, SNPs were identified by comparison of all Sanger sequences (referred to
as ‘Sanger SNPs’). SNPs were classified based on their occurrence in all sequenced
alleles (maximal 28): occurring once (unique), or occurring at least twice (Sanger
2+ SNPs). We then used the Sanger SNPs to confirm the RM-SNPs, and determined
the number of false negative and false positive RM-SNPs (Table 16.2). For LpMAX2,
we could only PCR-amplify fragments from 4 genotypes representing one single al-
lele, suggesting some hidden SNPs in the primer binding region (data not shown).
So, we aligned and compared all LpMAX2 CLC-contigs instead, to confirm RM-
SNPs. Likewise, the LpMAX4 CLC-contigs and RM-SNPs suggested the presence
of a divergent allele in genotypes Lp16 and Lp20. However, this allele could not be
cloned and sequenced, probably due to hidden SNPs that hamper PCR-amplification
of this allele from those genotypes.

16.3 Results and Discussion

De novo assembly of transcript sequences and identification of candidate

genes The RNA-seq data from 14 individual genotypes were assembled separately
to avoid mixing highly polymorphic allelic sequences during de novo assembly.
The de novo assembler of the CLCbio Genomics Workbench reconstructed between
49,000 and 79,849 CLC-contigs per genotype.
Brachypodium distachyon is the most closely related species to L. perenne with
a fully annotated genome sequence (International Brachypodium Initiative 2010).
Therefore, we used B. distachyon as an intermediate for the identification of or-
thologs of candidate genes (MAX1, MAX2, MAX3, MAX4 from Arabidopsis thaliana
and TB1, a TCP-family transcription factor from Zea mays), using Plaza 2.0 for phy-
logenetic analysis of gene families (Proost et al. 2009). We selected all CLC-contigs
(representing the different alleles in 14 genotypes) with a significant similarity
(tBLASTn) to the candidate B. distachyon proteins, and clustered them with CAP3
as a ‘second’ assembly step. While the CLC-contigs of individual genotypes may
134 T. Ruttink et al.

LpMAX2 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600

5’ UTR exon 1 3’ UTR

Lp03

Lp07

Lp09

Lp10

Lp11

Lp12

Lp13

Lp14

Lp16

Lp19

Lp20

Lp23

Lp25

Lp27

All Lps
together

all SNPs
2+ SNPs
RM SNPs
14%
10%
7%

Fig. 16.1 Reconstruction of reference transcripts from RNA-seq data and identification of
SNPs in LpMAX2. Exon (yellow arrow), 5 and 3 untranslated region (UTR; red arrow)
and coordinates (in basepairs) are indicated on the reference sequence. CLC-contigs of dif-
ferent genotypes (Lp03-Lp27; coloured by alternating green or black lines) are aligned to
the reference. SNP positions (short vertical lines) indicate different allelic variants. Contigs
obtained by de novo assembly of all the reads from 14 genotypes together are indicated for
comparison (All Lps together). SNPs identified by comparing contigs are indicated if they
occur at least once (all SNPs, black line) or in multiple contigs (2+ SNPs, blue line). SNPs
identified by mapping RNA-seq 76 bp paired-end reads to the common reference (RM-
SNPs; using a SNP frequency of either 14 %, 10 % or 7 % of the reads) are indicated on red lines

contain only partial or fragmented sequence for the candidate genes, it was possible
to reconstruct near full-length transcripts by aligning the available fragments from
multiple genotypes (Fig. 16.1).
The CAP3 contig assembly typically yielded a few CAP3-contigs per candidate
gene. Because allelic as well as paralogous CLC-contigs are selected via BLAST,
CAP3-contigs derived from paralogs must be filtered out after the CAP3 clustering.
To distinguish between alleles and paralogs, we compared CAP3-contigs (BLASTx)
against all B. distachyon proteins and retained only those with a best BLAST hit with
the original candidate B. distachyon gene. In addition, we performed phylogenetic
analysis with the orthologous gene families defined in Plaza 2.0. Finally, we identified
a single ortholog for LpMAX2, LpMAX3, and LpMAX4, two MAX1 orthologs, named
LpMAX1-1 and LpMAX1-2, and a LpTB1 transcript that is identical to LpTB1 isolated
by Brazauskas et al. (2010).
Construction of a non-redundant reference transcriptome sequence Applica-
tion of RNA-seq for gene discovery, gene expression analysis, and SNP discovery
16 Development of an SNP Identification Pipeline for Highly Heterozygous Crops 135

requires the construction of a non-redundant reference sequence for read mapping.

This reference should contain a single representative (consensus) sequence of each
locus. Otherwise, the estimated number of genes per family is wrong, and reads are
divided between redundant reference contigs covering the same locus, leading to
bias in expression levels and failure to detect polymorphisms.
During de novo assembly, regions with few polymorphisms are collapsed into
a single reference sequence, representing the consensus of the alleles. In contrast,
short regions of high polymorphism density tend to block contig extension, resulting
in neighboring, non-overlapping contigs. In addition, allelic sequences with long
stretches of high polymorphism density are considered as distinct sequences during
de novo assembly and are consequently reconstructed as two independent contigs,
despite representing the same locus. These issues hamper de novo assembly of highly
polymorphic genome sequences (Zharkikh et al. 2008, Miller et al. 2010, Velasco
et al. 2010, Donmez and Brudno 2011). Similarly, de novo assembly of transcript
sequences in highly polymorphic species typically leads to the assembly of multiple
(on average 2 to 4 in our data set) CLC-contigs per transcript in each L. perenne
genotype, hence a highly redundant reference transcriptome. The number and po-
sition of contig breakpoints depends on the specific combination of allele-pairs in
a given genotype. Therefore, allelic fragments obtained from multiple independent
genotypes will overlap at least partially and can be aligned in a second step using
clustering algorithms such as CAP3 (Fig. 16.1).
Paradoxically, combining RNA-seq data from multiple genotypes before the de
novo assembly only makes the fragmentation issue worse by increasing the degree of
polymorphism. For example, de novo assembly of reads from all genotypes together,
yields eight contigs for LpMAX2. These are fragmented and partially redundant, and
cover less than the total length of the transcript (Fig. 16.1). In contrast, combining
contigs from multiple genotypes after the de novo assembly actually helps solve
this fragmentation issue. Obtaining more sequence data from a single genotype
can not solve the fragmentation problem, because it is caused by the degree of
polymorphism between alleles, not by low sequence depth itself. Our results show
that it is more effective to first assemble transcriptomes from multiple genotypes in
parallel (k-mer based de novo assembly), followed by a second clustering step using
a global sequence alignment algorithm (e.g. CAP3), than to increase the tolerance
to polymorphisms during primary de novo assembly.
Validation of de novo transcript assembly by comparison to Sanger sequencing
of candidate genes The accuracy of the transcriptome sequence and subsequent
SNP identification depends on the accuracy of the de novo assembly. We compared
CLC-contigs to individually cloned and Sanger-sequenced alleles of the respective
genotypes. The number of nucleotide substitutions reflects the putative de novo
assembly error rate. In a total aligned sequence length of 22,284 bp, 102 nucleotides
(0.46 %) were different between the cloned fragments and the de novo assembled
sequences in the respective genotypes (Table 16.1). In the case of LpMAX4, it is
possible that failure to PCR-amplify a divergent allele from genotypes Lp16 and
Lp20 precluded confirmation of CLC-contigs, leading to an inflated estimation of
136 T. Ruttink et al.

Table 16.1 Validation of de novo transcript assembly by comparison to Sanger sequencing of

candidate genes
Gene Average read depth Total length of Number of Mismatches (%)
(number of reads/ aligned mismatches
position/genotype) sequence (bp)
LpMAX1-1 6.1 (2.5–14.1) 8,999 37 0.41
LpMAX2 28.9 (9.4–46.6) 3,643 8 0.22
LpMAX3 4.3 (2.4–6.6) 2,623 14 0.53
LpMAX4 8.6 (3.8–12.0) 3,758 39 1.04
LpTB1 4.6 (2.6–8.8) 3,261 4 0.12
Total 10.5 22,284 102 0.46

the error rate. In general, these five candidate genes had a relatively low read depth
per genotype in our NGS data (Table 16.1), and we expect that de novo assembly of
genes with a higher read depth will be even more accurate. In any case, the high level
of accuracy (>99.5 %) reflects the reliability of the de novo assembly, and confirms
that the procedure described here can be used for gene discovery and identification
of polymorphisms.
Genetic diversity in candidate genes Using the read mapping and SNP identifi-
cation parameters described in the material and methods section, we detected 157
RM-SNPs that could be confirmed by Sanger sequencing, and only two false pos-
itive RM-SNPs. The SNP density in these candidate genes as estimated by Sanger
sequencing of 10–14 genotypes ranges between 3–4 SNPs per 100 bp for the 2+ SNPs,
and can reach up to more than 8 SNPs per 100 bp including unique alleles. The 18
false positive RM-SNPs of LpMAX4 listed in Table 16.2 are probably caused by
failure of PCR-amplification of the corresponding alleles from genotypes Lp16 and
Lp20. So, these parameters can be used to generate a transcriptome-wide high quality
RM-SNP set for the development of gene-specific molecular markers.
We also detected a high percentage of false negative RM-SNPs: up to one-third
of the 2+ SNPs and almost all of the unique SNPs in the Sanger datasets. Two factors
may cause significant underestimation of the polymorphism level. First, a substantial
fraction of the diversity may reside in a large number of ‘unique’ alleles per gene, and
these fall below the detection limit. Lowering the SNP frequency threshold (from
14 % to 10 % or 7 %) may allow detecting more SNPs, including unique alleles,
as demonstrated for LpMAX1 and LpMAX2, but also increases the number of false
positive RM-SNPs. In the case of LpMAX1, disregarding low frequency SNPs implies
excluding as much as five unique alleles from 14 genotypes (data not shown), and
about half the number of SNPs, thus representing a substantial fraction of the genetic
diversity.
Second, a high local density of SNPs (i.e. read mapping mismatches) precludes
reads from divergent alleles to be correctly placed onto the reference, thus eliminat-
ing them from SNP discovery. In the case of LpMAX2, 36 of the 96 2+ SNPs were not
detected in the RM-SNP set. These are mostly located in regions with high polymor-
phism density. In LpMAX3, we identified only two distinct alleles in 11 genotypes,
16 Development of an SNP Identification Pipeline for Highly Heterozygous Crops 137

Table 16.2 Validation of read mapping SNPs by comparison to Sanger sequencing of candidate
genes
Gene SNP-type Sanger/CLC- False Total SNP density
Length (bp) contig SNPs positive (SNP/100 bp)
Read depth unique 2+ RM SNP 2+ Total
LpMAX1 Sanger 58 51 109 3.1 6.7
1637 bp RM 14 % 0 27 27
70.5x RM 10 % 4 33 1 38
RM 7 % 8 38 7 53
LpMAX2 CLC-contigs 81 99 180 4.4 8.6
2187 bp RM 14 % 10 63 1 74
354.0x RM 10 % 14 78 4 96
RM 7 % 16 84 9 109
LpMAX3 Sanger 34 34 3.8 3.8
897 bp RM allele 1 12 12
24.3x RM consensus 21 21
LpMAX4 Sanger 2 31 33 2.1 2.2
1488 bp RM 14 % 27 2 29
38.8x RM 10 % 27 7 34
RM 7 % 27 11 38
LpTB1 Sanger 2 2 0.2 0.2
972 bp RM 14 % 2 2
51.2x RM 10 % 2 2
RM 7 % 2 1 3

which differ at 34 SNPs in 897 bp (Table 16.2). Only 12 RM-SNPs are detected using
allele ‘1’ as reference for read mapping and SNP discovery, because reads derived
from allele ‘2’ are not mapped onto the reference due to high numbers of mismatches
(polymorphisms). However, if the two allelic sequences are merged into a consensus
sequence and then used as reference, 21 SNPs of a total of 34 Sanger 2+ SNPs are
correctly detected in the RM-SNP set (Table 16.2). Clearly, this effect is independent
of the read depth or the frequency of the alleles in the population, but leads to a strong
local underestimation of allelic diversity in regions with the highest polymorphism
density. These data illustrate that the threshold level of sequence similarity used
for read mapping must be adjusted to the expected number of polymorphisms or,
vice versa, that the reference sequence must be adjusted to optimally facilitate read
mapping.
So, the selection of the reference sequence is critical. Importantly, a common ref-
erence sequence can be constructed ‘on the fly’ by CAP3-clustering of all alleles in
the sample pool and using the resulting consensus sequence as representative allele
(e.g., LpMAX3). This sequence may not exist as a ‘true’ allele in the population, but
as a consensus it should have the minimal number of mismatches to all allelic variants
and thus provides a more balanced approach to determine the mismatch threshold
for read mapping within that specific sample pool. Using any of the existing alleles
(either from within the sample pool, or from a fully sequenced reference genotype
outside the current sample pool) creates more bias. All reads originating from this
138 T. Ruttink et al.

allele would have a perfect match while all other reads would have increasing mis-
match penalties depending on the genetic distance between alleles, hence inferring
combined mismatch treshold/reference-dependent bias.
An alternative to short read mapping, is to use the CLC-contigs from de novo
assembly for SNP identification or confirmation, as was demonstrated for LpMAX2.
Other alternatives are to perform read mapping per genotype if sufficient read depth
is available, or to perform pairwise comparisons of all genotypes.
False negative (hidden) SNPs interfere with the design and performance of SNP
marker assays. For the design of primer extension assays, or probe-based genotyping
assays, typically the 50 basepair sequences flanking the selected SNP should not
contain neighboring SNPs. With an average density of 3–4 2+ SNPs per 100 bp, it may
prove difficult to find such SNP-free regions. In addition, a relatively high percentage
of false negative (hidden) SNPs flanking known markers can cause failure of the SNP
assay. Therefore, suppressing false negative SNP identification improves the SNP
assay conversion rate and SNP calling rate for genotyping. Especially in species with
such high degree of polymorphism, Genotyping by Sequencing (GBS) may prove to
be a valuable alternative to probe-based or primer-based SNP-genotyping assays.

16.4 Conclusion

De novo assembly algorithms were first developed for species with a relatively low
degree of polymorphism, such as naturally inbreeding Arabidopsis thaliana, but are
less suitable for highly polymorphic species. They tolerate only a low frequency
of polymorphisms, as to resolve highly repetitive regions for reference genome
assembly. As a consequence, de novo assembly introduces contig breakpoints at
highly polymorphic regions in a genotype-dependent manner. We propose a two-
step assembly strategy that resolves transcript fragmentation and simultaneously
collapses separately assembled allelic sequences (obtained either within a genotype
or across multiple genotypes) into a non-redundant consensus transcriptome required
for read mapping, expression analysis and SNP discovery. We propose that a complete
transcriptome can be reconstructed in L. perenne, by first performing de novo tran-
scriptome assembly of multiple genotypes, then systematically performing BLAST
searches with all proteins of a reference species, followed by CAP3 clustering of the
CLC-contigs from different libraries (genotypes, treatments, and/or tissues), and fil-
tering CAP3-contigs based on best reciprocal BLAST hits. Analysis of the LpMAX2
and LpMAX3 genes illustrates how the choice of the reference sequence influences the
number of identified RM-SNPs, and why constructing a common reference sequence
is a useful strategy to suppress false negative SNPs, while maintaining stringency
for read mapping.
Based on these case studies, we optimised parameters for de novo assembly, con-
tig clustering, and SNP identification. These will be integrated into a bioinformatics
pipeline to reconstruct a reference L. perenne transcriptome, using the predicted
16 Development of an SNP Identification Pipeline for Highly Heterozygous Crops 139

protein set of a closely related species, such as B. distachyon, as guidance for clus-
tering and annotation. This database can then easily be searched for orthologs of
candidate genes and includes allelic variants when multiple genotypes have been se-
quenced in parallel. Thus, read mapping onto this reference transcriptome facilitates
the identification of polymorphisms in a transcriptome-wide sequence set for crop
species without the need for a fully sequenced genome. This bioinformatics pipeline
can be easily adapted to other highly heterozygous non-model species through the
use of precomputed orthologous protein sets derived from a number of model species.
In conclusion, the SNP discovery is greatly enhanced through NGS and can generate
more markers at lower cost for application in molecular breeding.

Acknowledgments This research was supported by the Agency for the promotion of Innovation
by Science and Technology (IWT) Flanders (project LO-080510) and by the Research Foundation
Flanders (FWO Grant K208710N). RNA-seq data was generated in collaboration with Torben Asp,
Jakob Hedegaard, and Christian Bendixen at Aarhus University, Denmark. We would like to thank
Sabine van Glabeke and Nancy Mergan for excellent technical support.

References

Brazauskas G, Pasakinskiene I, Asp T, Thomas Lübberstedt T (2010) Nucleotide diversity and

linkage disequilibrium in five lolium perenne genes with putative role in shoot morphology.
Plant Sci 179(3):194–201. doi:10.1016/j.plantsci.2010.04.016
Donmez N, Brudno M (2011) Hapsembler: an assembler for highly polymorphic genomes
(2011). In: Research in computational molecular biology—15th annual international conference,
RECOMB 2011, Vancouver, BC, Canada. Proceedings 2011. doi:10.1007/978-3-642-20036-6
Huang X, Madan A (1999) CAP3: a DNA sequence assembly program. Genome Res 9:868–877
International Brachypodium Initiative (2010) Genome sequencing and analysis of the model grass
brachypodium distachyon. Nature 463:763–768. doi:10.1038/nature08747
Leyser O (2009) The control of shoot branching: an example of plant information processing. Plant
Cell Environ 32:694–703. doi:10.1111/j.1365-3040.2009.01930.x
Miller JR, Koren S, Sutton G (2010) Assembly algorithms for next-generation sequencing data.
Genomics 95:315–327. doi:10.1016/j.ygeno.2010.03.001
Proost S, Van Bel M, Sterck L, Billiau K, Van Parys T, Van de Peer Y, Vandepoele K (2009)
PLAZA: a comparative genomics resource to study gene and genome evolution in plants. Plant
Cell 21:3718–3731. doi:10.1105/tpc.109.071506
Shendure J, Ji H (2008) Next-generation DNA sequencing. Nat Biotechnol. doi:10.1038/nbt1486
Varshney RK, Nayak SN, May GD, Jackson SA (2009). Next-generation sequencing technolo-
gies and their implications for crop genetics and breeding. Trends Biotechnol 27(9):522–530.
doi:10.1016/j.tibtech.2009.05.006
Velasco R et al. (2010) The genome of the domesticated apple (Malus x domestica borkh.). Nat
Genet. doi:10.1038/ng.654
Zharkikh A et al. (2008) Sequencing of highly heterozygous genome of Vitis vinifera L. cv pinot
noir: problems and solutions. J Biotechnol 136:38–43
Chapter 17
First Insights into the Mitochondrial Genome
of Perennial Ryegrass (Lolium perenne)

K. Diekmann, T. R. Hodkinson, K. H. Wolfe and S. Barth

Abstract Plant mitochondrial genomes encode the majority of genes that are in-
volved in aerobic respiration which provides energy for the plant cell. To date only
32 plant mitochondrial genomes have been sequenced completely and only one from
the Pooideae subfamily to which Lolium belongs. We aimed to sequence the complete
mitochondrial genome of perennial ryegrass to assess its variation in comparison to
other grasses and to provide insights into agronomically important traits such as cy-
toplasmic male sterility. We found all 33 known plant mitochondria protein-coding
genes, three ribosomal RNA and 20 transfer RNA genes. However, plant mitochon-
drial genomes are complex in their structure and exhibit a low degree of conservation
in relation to other mitochondrial genomes. Thus, a complete assembly of the peren-
nial ryegrass mitochondrial genome was not possible due to the lack of a suitable
reference genome and the high variability. However, we reduced the large number
of contigs to a set of 43 that contained either protein-coding genes or consisted of
more than 10 kb. Based on this data set we found evidence for intracellular gene
transfer events from the chloroplast genome to the mitochondrial genome and pos-
sibly duplicated gene exons. We estimated the size of the mitochondrial genome of
perennial ryegrass to be approximately 565 kb.

Keywords Lolium perenne · Perennial ryegrass · Mitochondrial genome

K. Diekmann () · S. Barth

Teagasc Crops Research Centre Oak Park, Carlow, Ireland,
e-mail: [email protected]
School of Natural Sciences, Trinity College Dublin, Dublin 2, Ireland
T. R. Hodkinson
School of Natural Sciences, Trinity College Dublin, Dublin 2, Ireland
K. H. Wolfe
Smurfit Institute of Genetics, Trinity College Dublin, Dublin 2, Ireland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 141

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_17, © Springer Science+Business Media Dordrecht 2013
142 K. Diekmann et al.

17.1 Introduction

Mitochondria can be found in fungal, animal and plant cells. They are commonly
known as the power plants of the cell because they are the location of the aero-
bic respiration and thus involved in major metabolic processes. Mitochondria are
believed to have endosymbiotic origins. They very likely derive from the rickettsial
subdivision of α-proteobacteria. The most mitochondrial-like eubacterial genome se-
quenced belongs to the α-proteobacterium Rickettsia prawazekii (Gray et al. 1999).
It is believed that the ancestral mitochondrial genome either lost or transferred the
majority of its genes to the nucleus. Nowadays only 50–60 genes remain within plant
mitochondrial genomes. From a plant breeding perspective mitochondrial genomes
are nevertheless interesting because mitochondria are, in general maternally inher-
ited and mutations within the plant mitochondrial genome can confer cytoplasmic
male sterility (CMS) which is a valuable tool in hybrid plant breeding. CMS has also
great potential in Lolium perenne breeding. Hence gaining information about the
L. perenne mitochondrial genome was the main objective of the present study.
To date, more than 2000 animal mitochondrial genomes have been sequenced
and made publicly available. In contrast only 32 plant mitochondrial genomes are
publicly available (Fig. 17.1). The first plant mitochondrial genome sequence was
published in 1993 (Marchantia polymorpha). In 1997, the mitochondrial genome
sequence of the model plant Arabidopsis thaliana became available. The first grass
mitochondrial genome sequence, Triticum aestivum, was published in 2005. Dif-
ferences between the number of publicly available animal and plant mitochondrial
genome sequences are very likely due to the complex nature of plant mitochondrial
genomes. While animal mitochondrial genomes are rather conserved in their genome
size (varying from 10−50kb, ∼16kb in average), plant mitochondrial genomes show
a huge variation. Although not sequenced yet, the Brassica hirta mitochondrial
genome is expected to be the smallest of the higher land plants with approximately
208 kb. Cucumis melo, the musk melon, on the other hand is estimated to contain
the biggest mitochondrial genome. Its size is currently estimated to exceed 2,400 kb.
This huge variation in genome size can not only be detected between unrelated plant
species, but also within plant families. In the Cucurbitaceae family, for example,
the genome size varies enourmously (∼380 kb in water melon, over ∼1000 kb in
courgette, and more than 2,400 kb in musk melon. High variation in the genome
size can also be detected within the grass family. Currently ten grass mitochondrial
genome sequences are published (Fig. 17.2). So far Triticum aestivum contains the
smallest mitochondrial genome (∼452 kb) and Tripsacum dactyloides the biggest
(∼704 kb).
Plant mitochondrial genome size can be affected by several factors. One factor
is the occurrence of repetitive regions that vary in size and number from one plant
species to another. Horizontal and intracellular gene transfer processes also contribute
to the genome size variation. In general, intracellular gene transfer events involve
transfer from the chloroplast genome to the mitochondrial genome. However, while
transferred ribosomal genes often remain functional in the mitochondria, protein-
coding genes often become dysfunctional due to shifts in the reading frame. The
17 First Insights into the Mitochondrial Genome of Perennial Ryegrass (Lolium perenne) 143

Fig. 17.1 Mitochondrial genome size of higher land plants. Highlighted are all sequenced members
of the Poaceae family

Fig. 17.2 Number of publicly available mitochondrial genome sequences from moss and higher
plant species in October 2011. (www.ncbi.nlm.nih.gov)
144 K. Diekmann et al.

biggest variation in genome size is very likely caused by the huge variation in the
intergenic spacer regions. In all plant mitochondrial genomes the number of genes
is similar. But the proportion of coding region relative to the total mitochondrial
genome size varies from 3 % in the biggest so far sequenced plant mitochondrial
genome (courgette) to 22 % in the smallest (rapeseed). Hence, the bigger a plant
mitochondrial genome is the larger are the intergenic spacer regions. In addition,
these intergenic spacer regions demonstrate a huge sequence variation. The complex
nature of plant mitochondrial genomes needed to be considered when sequencing
the mitochondrial genome of L. perenne.

17.2 Material and Methods

For sequencing the L. perenne mitochondrial genome, leaves of 14 days old etiolated
seedlings from cultivar Shandon were harvested and the DNA isolated following a
protocol of Kiang et al. (1993). Unfortunately only small amounts of DNA were
obtained by this method and hence a whole genome amplification step had to be in-
corporated to obtain enough DNA for sequencing. To account for the repetitive and
large non-conserved regions of the plant mitochondrial genome, a hybrid sequencing
approach was carried out consisting of conventional Sanger sequencing with an insert
size of 2.5 kb and GS FLX (454) sequencing with a read file length of around 500 bp.
Based on already available grass mitochondrial sequencing information, the genome
size of L. perenne was expected to be approximately 500 kb. Assuming no contam-
ination of the isolated DNA, read files obtained by Sanger sequencing covered the
genome around five times and 454 sequencing around 200 times. Sequencing was
outsourced to the sequencing company GATC.

17.3 Results

Sanger files were firstly assembled using the program Phrap (http://www.phrap.org/),
and a hybrid assembly approach using the Newbler software (454 Life Sciences,
Branford, Connecticut, USA) was used to assemble read files derived from both se-
quencing approaches. However, both assemblies resulted in large numbers of contigs
(Phrap: 227, Newbler: > 4000). Another assembly carried out in Lasergene (DNAs-
tar, Inc., Madison, Wisconsin) resulted in even more contigs (> 8000). At this stage a
complete assembly of the L. perenne mitochondrial genome did not seem possible.
However, in order to reduce the amount of contigs and end up with a draft assembly
it was hypothesized that contigs based on fewer than 100 reads are very likely based
on low quality read files or possible sample contamination, while contigs based on
more than 100 reads most likely contain the valuable L. perenne mitochondrial se-
quence information. Thus, a final draft assembly based on 43 contigs consisting of
a total of 564,927 bp with 33 protein coding genes, three ribosomal RNA genes and
20 transfer RNA genes (14 single copy) was obtained.
17 First Insights into the Mitochondrial Genome of Perennial Ryegrass (Lolium perenne) 145

Comparative analyses of the draft assembled L. perenne mitochondrial genome

to published grass mitochondrial genomes confirmed earlier observations that mito-
chondrial coding regions are highly conserved, however gene order (Ogihara et al.
2005) and intergenic spacer regions (Sugiyama et al. 2005) were highly variable. A
rearrangement study revealed only few regions with conserved gene order across all
grass mitochondrial genomes currently available. A comparison of the L. perenne
chloroplast genome to the draft assembly of the L. perenne mitochondrial genome
revealed at least three chloroplast regions that were copied to the mitochondrial
genome. Transferred tRNA genes are expected to be functional. However, protein-
coding genes are generally dysfunctional due to frameshifts. Some evidence for
potential horizontal gene transfer events were also detected. The abuscular mycor-
rhiza Glomus intraradices lives in close symbiosis to many land plants. It makes
soil derived nutrients, especially phosphorus, available to plants and benefits from
the photosynthetically fixed carbon. This symbiosis is expected to be as old as the
earliest land plants (Karandashov and Bucher 2005). The mitochondrial genome of
G. intraradices is publicly available and was compared to the mitochondrial se-
quences of L. perenne. Thus a 305 bp fragment was detected that showed a 76 %
similarity to the L. perenne mitochondrial genome. This fragment can also be de-
tected in other land plant species and could be derived from a horizontal gene transfer
event.

17.4 Conclusion

The mitochondrial genome of Lolium perenne was sequenced and a draft assembly
established. First analyses provided insights into the organization of the L. perenne
mitochondrial genome. Evidence for intracellular and horizontal gene transfer events
was also detected. The availability of L. perenne mitochondrial genome sequences
enables the investigation of CMS in L. perenne and closely related grass species in
the future and provides a new resource for the development of molecular markers.

References

Gray MW, Burger G, Lang BF (1999) Mitochondrial evolution. Science 283(5407):1476–1481

Karandashov V, Bucher M (2005) Symbiotic phosphate transport in arbuscular mycorrhizas. Trends
Plant Sci 10(1):22–29
Kiang A-S, Connolly V, McConnell DJ, Kavanagh TA (1993) Cytoplasmic male sterility (CMS) in
Lolium perenne L.: 1. Development of a diagnostic probe for the male-sterile cytoplasm. Theor
Appl Genet 86(6):781–787
Ogihara Y, Yamazaki Y, Murai K, Kanno A, Terachi T, Shiina T, Miyashita N, Nasuda S, Nakamura
C, Mori N, Takumi S, Murata M, Futo S, Tsunewaki K (2005) Structural dynamics of cereal mi-
tochondrial genomes as revealed by complete nucleotide sequencing of the wheat mitochondrial
genome. Nucleic Acids Res 33(19):6235–6250
Sugiyama Y, Watase Y, Nagase M, Makita N, Yagura S, Hirai A, Sugiura M (2005) The com-
plete nucleotide sequence and multipartite organization of the tobacco mitochondrial genome:
comparative analysis of mitochondrial genomes in higher plants. Mol Genet Genomics
272:603–615
Chapter 18
Quantifying Early Vigour and Ground Cover
using Digital Image Analysis

M. Cougnon, J. Verhelst, K. De Dauw and D. Reheul

Abstract Early vigour, or the fast leaf area development by an establishing crop
is important in many breeding programs and is currently evaluated by subjective
rating. The objective of this study was to test whether analysis of digital images
can quantify ground cover in a fast and repeatable way in two trials where ground
cover evaluations were important. Both in a greenhouse trial comparing the early
vigour of different varieties of tall fescue (Festuca arundinacea Schreb.) as in a field
trial comparing different varieties of rye (Secale cereale L.), Italian ryegrass (Lolium
multiflorum L.) and lopsided oat (Avena strigosa Schreb.) for use as cover crops, we
took pictures on regular intervals. In both trials, parameters that allowed accurate
discrimination between pixels that represented bare soil and pixels that represented
soil covered with plants were easily found. The Hue dimension of the Hue Saturation
Brightness colour space was the parameter with the largest discriminating power
between ground and plant covered pixels. Both in the field as in the greenhouse,
there were significant differences in ground cover between the varieties. We found a
good regression between ground cover and biomass production in the early growth
stage of the cover crops in the field trial, until the vegetation reached a soil cover of ca.
50 % and a corresponding biomass of ca. 500 kg dry matter/ha (R2 = 88 %, p value =
<0.001). In later stages, correlation between ground cover and biomass was weak
due to the presence of erect growing genotypes, with few tillers and a lower ground
cover but with a good aboveground biomass production. We conclude that image
analysis has a good potential to quantify early ground cover and early aboveground
biomass production as it can work both fast and accurate in the field.

18.1 Introduction

Plant breeders routinely use ground cover as a parameter to measure early vigour or
density of a grass sward. Trained plant breeders use discontinuous scores to assess it
qualitatively. The use of digital image analysis allows quantification of the covered
ground area in a repeatable way.

M. Cougnon () · J. Verhelst · K. De Dauw · D. Reheul

Faculty of Bioscience Engineering, Department of Plant Production,
Ghent University, Ghent, Belgium
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 147

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_18, © Springer Science+Business Media Dordrecht 2013
148 M. Cougnon et al.

The method to quantify ground cover is basically simple. A software programme

counts the number of green pixels in a digital image of the crop and calculates the
proportion of these green pixels in the whole image. The operator has to set limits
to define which pixels are green, and thus covered by plant material, and which are
not.
Richardson et al. (2001) used digital image analysis and a commercial software
package to quantify ground cover in a sward of turfgrass. Image analysis produced a
much lower variance than the subjective scores resulting in more significant differ-
ences. A macro was developed (Karcher and Richardson 2005) to perform ground
cover analysis on batches of pictures, minimizing the time needed to analyse a high
number of similar pictures.
Also Lock et al. (2004) tested the accuracy of digital image processing for quan-
tification of ground cover in turf grass. They improved robustness by using geometric
classifiers in addition to the hue, saturation and lightness (HSL) colour space. Image
analysis for ground cover quantification was successfully used in a non-grass crops
e.g. by Purcell (2000) and by Behrens and Diepenbrock (2006).
The first objective of this study was to test whether analysis of digital images
can quantify ground cover in a repeatable way in practical trials where ground cover
evaluations are important. Secondly, correlation between ground cover and biomass
production was investigated.

18.2 Material and Methods

Two trials in which ground cover and early vigour were important, were selected for
testing digital image analysis in practise.
The first trial compared the early vigour of eleven different varieties of tall fescue
(Festuca arundinacea Schreb.) and perennial ryegrass (Lolium perenne L.) in flats
measuring ca. 0.2 m2 in the greenhouse (De Dauw 2009). The grass was planted
in the flats and not sown in order to have an equal number of uniformly spaced
plants in each flat. On three occasions in the year 2009 (early summer, late summer,
autumn) we planted three flats per variety and took pictures of the flats on regular
time intervals. For picture taking each flat was placed under a camera (CANON eos
50 D with lens EF-S 17–85 mm) which was mounted on a stand with a horizontal
arm 75 cm above the flats in a room with artificial light. No flash was used to take
the pictures, shutter speed, aperture and white balance were equal for all pictures
taken in the three experiments. When the canopy of the flats was completely closed
the experiments were stopped and the aboveground dry matter (DM) yield of each
flat was determined.
The second trial was a field trial comparing six varieties of rye (Secale cereale
L.) and one variety of both Italian ryegrass (Lolium multiflorum L.) and lopsided
oat (Avena strigosa Schreb.) for use as cover crops with two different sowing dates
(22/09/2010 and 14/10/2010) in the autumn of 2010 (Verhelst 2011). The trial was
sown at a density of 300 viable seeds m−2 for rye and oat, and 2000 viable seeds m−2
18 Quantifying Early Vigour and Ground Cover using Digital Image Analysis 149

for ryegrass. The trial was organized as a split plot design with three replicates. The
main factor was sowing date and the subfactor was variety/species. Individual plot
size was 30 m2 . Between 24/10/2010 and 7/2/2011 each plot was photographed with
an interval of ca. 20 days, using a portable stand with a horizontal arm on which
a camera (FUJIFILM FinePix F470) was mounted, 1.5 m above the ground. As
we worked with natural light in the field trial, camera setting were different on each
occasion we took pictures. We used the setting that the camera selected automatically
for the given circumstances. Flash was never used. A picture covered a ground surface
of approx. 2.5 m2 . On two occasions (20/1/2011 and 21/2/2011) both ground cover
and aboveground biomass of the plots were determined. Biomass production was
determined, by cutting all plantlets just above the ground, on a surface of 1 m2 per
plot.
Image analysis was done with Image J, a public domain, Java-based image
processing program (http://rsbweb.nih.gov/ij/). Using the colour thresholder plu-
gin (http://www.dentistry.bham.ac.uk/landinig/software/software.html) the original
colour images were transformed into binary images where all pixels covered by plant
material in the original image (green) had value 1 and all other pixels (ground, litter)
had value 0. Pictures were processed in the Hue (H), Saturation (S), Brightness (B)
colour space. In this colour space, each pixel in a picture has a value for each of the
three dimensions ranging from 0–255. Pixels with different colours can be divided
in groups by putting limits on the values of H, S and B. These limits are commonly
called thresholds. Threshold values for H, S and B allowing to discriminate between
green pixels and other pixels, were searched by trial and error on a random subset
of all pictures taken on the same occasion (called “batch” hereafter). Once the right
threshold values were found for a particular batch, it was analysed in an automated
way using a macro. Inputs for this macro were a batch of pictures and threshold
values; output was the ground cover for each picture in the batch.
Weeds were present in all plots of the field trial. Distinction between weeds and
the crop of interest was made, based on the different size and shape (circularity) of
small weeds in the pictures. Mostly, weeds were smaller than the sown rye/grass
plants. In addition, the rye and grass plantlets occurred as a continuous line in the
picture as they were sown in rows, whereas the weeds were mostly growing between
the rows. Dicot weeds had a rather round shape (or higher circularity) compared to
the sown grass and rye. Values for size and shape that distinguished crop and weeds
in the images were also determined by trial and error.

18.3 Results and Discussion

In both trials, discrimination between ground and plant material was better in the
Hue, Saturation and Brightness (HSB) colour space than in the Red, Green and Blue
(RGB) colour space. The Hue dimension of the HSV colour space was the parameter
with the largest discriminating power.
150 M. Cougnon et al.

As the pictures were taken under controlled, artificial light in the greenhouse trial,
the same threshold values worked for most the batches taken in the different seasons.
When the canopy of the flats was nearly closed (ground cover > 80 %), the threshold
values needed to be altered slightly in order to ensure a good discrimination between
leaves and ground.
In the field trial, light was an uncontrollable factor. Therefore new threshold values
had to be searched for each batch of pictures taken. Cloudy days with no direct
sunlight were preferred for picture taking in the field, as shadow caused problems in
the image analysis. The presence of weeds was another factor of difference between
the greenhouse trial and the field trial. The most abundant species were Stellaria
media L. and Poa annua L. In the early growth stages, the distinction between weeds
and crops based on size and circularity worked well (Fig. 18.1), and most of the
weeds could be removed. In later stages we were no longer possible to distinguish
between weeds and crops. This was not considered as problematic, as the rye and
grass were far more competitive than the weeds. Only for oat, the least vigorous
species, weeds were clearly present and thus contributing to the ground cover.
In both trials, significant differences in ground cover between the varieties/species
were found. In the field trial, significant differences in ground cover were found until
the best varieties reached a soil cover of around 80 %. This stage occurred around
80 days after sowing for the fastest growing varieties sown on the early sowing
date (21/09/2010). Soil cover by oat decreased after day 100 due to frost damage
(Fig. 18.2a). For the later sowing date (14/10/2010) significant differences in soil
cover between varieties persisted until the end of the trial (Fig. 18.2b), and the soil
cover at the end of the trial was below 50 %.
Similar results were obtained in the greenhouse trial (Fig. 18.3). Both in the trials
sown in early summer and late summer, differences between the varieties were largest
on the moment that the fastest growing varieties reached 80 % ground cover. In the
very early stages when ground cover is very low and in the later stages of growth,
when the canopy is nearly closed, differences in ground cover between the varieties
were small and insignificant. Once a ground cover of 80 % was reached, the increase
in ground cover decreased. Newly formed leaves overlap with the older leaves, and
the plants start to grow higher with the result that the difference between the fast
developing varieties and the slower developing varieties gradually disappeared.
In the field trial, a good relationship was found between biomass production and
ground cover for the plots that were sown on 14/10/2010, but not for the plots sown
on 21/9/2010 (Fig. 18.4a). A linear model was fitted to the data of the plots sown on
14/10/2010. The relationship found (y = −16.1 + 1.32x, R2 = 0.88, p = 0.0000)
indicates that for a ground cover up to 50 %, an increase of 1 % in ground cover,
corresponds with a biomass increase of 13 kg DM.ha−1 (Fig. 18.4b). In later growth
stages, as it was the case for the early sown cover crops, the relationship between
ground cover and aboveground biomass production is lost due to differences in
growth habit between the different species/varieties. Some (tetraploid) rye varieties
combined both an erect growing with few tillers resulting in a rather low ground cover
and a very good biomass production. Italian ryegrass on the other hand, combined
a good ground cover with a low biomass production. Once a biomass of 150 g.m−2
18 Quantifying Early Vigour and Ground Cover using Digital Image Analysis 151

Fig. 18.1 Above: picture

taken on 24/10/2010 of a rye
plot sown on 22/09/2010;
Middle: the same picture as
above after thresholding,
threshold values: Hue:
36–115, Saturation: 0–255,
Brightness: 109–255. Soil
cover = 20.3 %; Under: same
picture as in the middle, after
removing particles with a
size < 250 pixels and a
circularity > 0.2. Soil
cover = 18.6 %
 152 M. Cougnon et al.

50
40
80
Ground cover (%)

Ground cover (%)

30
60

20
40

10
Rye var3 Rye var3
Rye var1 Rye var1
20

Oat Oat

0
40 60 80 100 20 40 60 80 100

a Days after sowing (d) b Days after sowing (d)

Fig. 18.2 Mean ground cover for a vigorous (var3) and a slow (var1) growing rye variety and
lopsided oat sown on 22/9/2010 (a) and 14/10/2010 (b) error bars indicate standard deviation (sd)
100

100
80

80
Ground cover (%)+-sd

Ground cover (%)+-sd

60

60
40

40

Fa var1 Fa var1
20

20

Lp var1 Lp var1
Fa var2 Fa var2
Fa var3 Fa var3
0
0

10 15 20 25 30 5 10 15 20 25

a Days after planting (d) b Days after planting (d)

Fig. 18.3 Mean ground cover for three varieties of Festuca arundinacea (Fa) and a variety of Lolium
perenne (Lp) planted in early summer (a) and in late summer (b) in a greenhouse trial. Error bars
indicate standard deviation (sd)
100 150 200 250

40
Biomass (g DM/m²)

Biomass (g DM/m²)

y=-16.1+1.32x
30

As
Rye var1
Rye var2
20

Lm
Rye var3
50

Rye var4
10

early sown Rye var5

late sown Rye var6
0

20 40 60 80 15 20 25 30 35 40 45

a Ground cover (%) b Ground cover (%)

Fig. 18.4 Relationship of biomass production and ground cover for cover crops in a field trial. a
Contrast between the early sown (21/9/2010) and late sown (14/10/2010) cover crops. b Regression
for the late sown cover crops
18 Quantifying Early Vigour and Ground Cover using Digital Image Analysis 153

was reached, the ground cover of all varieties/species, productive or not, was situated
between 75 and 90 %.
In the greenhouse trial, biomass production was determined at the end of the ex-
periments. No relationship was found between ground cover and biomass production
in none of the three experiments. At the end of the experiments, ground cover was
nearly complete for all grass varieties resulting in a low variation for ground cover
and a much higher variation in biomass production. Moreover and similar to the field
trial, genotypes with contrasting growth habits (erect, few tiller vs. prostrate, much
tillers) confounded the results at the end of the experiment.

18.4 Conclusion

As in both trials, ANOVA resulted in significant differences between the vari-

eties/species, the method can be considered as repeatable. Correlation between
ground cover and above ground biomass production was good until a certain level
of ground cover was reached. In the field experiment we studied, this level was at
50 % ground cover. Beyond that level, correlation was lost due to the differences in
growth habit of different genotypes.
In the trails we analysed, image analysis proved suitable for quantification of
ground cover by crops. Ground cover gave indirectly information about biomass
production, but this relationship was restricted only to the early growth stages.

References

Behrens T, Diepenbrock W (2006) Using digital image analysis to describe canopies of winter
oilseed rape (Brassica napus L.) during vegetative developmental stages. J Agron Crop Sci
192:295–302
De Dauw K (2009) Breeding of tall fescue (Festuca arundinacea Schreb.). Master thesis Ghent
University, pp 62 (In Dutch)
Karcher DE, Richardson MD (2005) Batch analysis of digital images to evaluate turfgrass
characteristics. Crop Sci 45:1536–1539
Lock R, Rademacher I, Nonn H, Künbauch W (2004) Methods of digital image processing to
quantify ground cover of turf grass. In: Lüscher A (ed) Grassland science in Europe, Vol 9.
Swiss Grassland Federation, Switserland, pp 790–792
Purcell LC (2000) Soybean canopy coverage and light interception measurements using digital
imagery. Crop Sci 40:834–837
Richardson MD, Karcher DE, Purcell LC (2001) Quantifying turfgrass cover using digital image
analysis. Crop Sci 41:1884–1888
Verhelst J (2011) Winterrye and Italian ryegrass as cover crops. Master thesis Ghent University,
pp 74 (In Dutch)
Chapter 19
Expression of the Lolium perenne Terminal
Flower 1 Gene in Alfalfa and Tobacco

N. Ferradini, A. Nicolia, F. Veronesi and D. Rosellini

Abstract The Terminal Flower 1 gene of Lolium perenne (LpTFL1) was overex-
pressed in Festuca rubra and Arabidopsis thaliana, and a delay or even the complete
suppression of flowering was obtained. We have evaluated the LpTFL1 GENE as a
possible candidate to delay or prevent the flower transition process in alfalfa. This
may be useful in forage crops to lengthen the vegetative phase, and in transgenic
crops to control gene flow. Alfalfa was transformed via Agrobacterium tumefaciens
using the binary vector pCAMBIA3300-LpTFL1 (kindly provided by C. S. Jensen),
in which the LpTFL1 gene was under the control of the Zea mais Ubiquitin promoter.
To ensure a high level of expression of the gene, in a second construct, the CaMV
35S dual-enhancer promoter was used. RT PCR analysis confirmed the expression
of LpTFL1 in several transgenic alfalfa plants. These were phenotypically normal
throughout the growth cycle, flowering was unaffected, and the plants set seed nor-
mally; the same was true for tobacco, that was transformed with the same constructs.
Our results indicate that LpTFL1 cannot be used for flowering repression in alfalfa.

19.1 Introduction

In forage crops, like alfalfa, the delay or the suppression of flowering may have a
positive effects on forage quality. In addition, the manipulation of flowering transi-
tion could allow to prevent gene flow and dispersal of transgenes from genetically
engineered crops. Genetic and molecular analyses have shown that several genes are
involved in the control of the switch from vegetative to reproductive growth.
The Arabidopsis Terminal Flower 1 (TFL1) gene is responsible for the main-
tenance of inflorescence meristem identity. It encodes a putative flowering signal
molecule, and its mutation results in the precocious conversion of the inflorescence
meristem into a flower meristem, that forms a terminal flower (Bradley et al. 1997;
Ohshima et al. 1997). Constitutive expression in Arabidopsis produces plants with
large rosettes, long, branched stem, delayed or, in extreme cases, lack of flower

D. Rosellini () · N. Ferradini · A. Nicolia · F. Veronesi

Department of Applied Biology, University of Perugia, Borgo XX Giugno 74,
06121 Perugia, Italia
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 155

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_19, © Springer Science+Business Media Dordrecht 2013
156 N. Ferradini et al.

Fig. 19.1 pCAMBIA-

LpTFL1 and pCAMBIA-
2300-LpTFL1 binary vectors

differentiation (Ratcliffe et al. 1998). The orthologous genes were investigated to

various extent in several species (Danilevskaya et al. 2010 and references therein).
The Terminal Flower 1 gene of Lolium perenne (LpTFL1) plays a primary role
in controlling the process of meristem differentiation and the transition from the
vegetative to reproductive phase. It was overexpressed in A. thaliana and Festuca
rubra (Jensen et al. 2001), obtaining a delay or even the complete suppression of
flowering. We have evaluated the LpTFL1 GENE as a possible candidate to delay or
prevent the flower transition process in alfalfa. We also tested the effect of LpTFL1
expression in tobacco.

19.2 Materials and Methods

The Medicago sativa genotype RSY1, from the RegenS-Y germplasm and the Nico-
tiana tabacum variety Petit Havanna were used in this work. The two binary vectors
used for alfalfa and tobacco transformation are shown in Fig. 19.1.
The pCAMBIA 3300-LpTFL1 vector (kindly provided by C.S. Jensen, DLF Tri-
folium, DK) contains the 528 bp L. perenne Terminal Flower 1 (LpTFL1) gene
under the control of the maize Ubiquitin promoter (ZmUBI) and its first intron, and
of the Nos terminator. The selectable marker gene bar, that confers resistance to
phosphinothricin, is under the control of the CaMV 35S promoter and terminator.
The coding sequence of LpTFL1 gene was amplified by PCR from pCAMBIA3300-
LpTFL1 with the primers 5 TGCTTTCCTCAGCGGATCCATGTCTAGGTCTGT
and 5 GGTTATACTAGTTCAGCGCCTCCTGGCAGCAGT, which contain the
BbvCI and SpeI restriction sites, respectively (bold). The 541 bp amplicon was
digested with the same enzymes and ligated into the corresponding sites of the
pCAMBIA2300-d35S-NosT vector between the dual CaMV35S promoter and the
Nos terminator, thus obtaining the pCAMBIA2300-LpTFL1 transformation vector
(Fig. 19.1). It contains the NptII selectable marker gene.
19 Expression of the Lolium perenne Terminal Flower 1 Gene in Alfalfa and Tobacco 157

The two vectors were separately introduced into A. tumefaciens, strain AGL1,
by electroporation and used in the transformation experiments. Alfalfa trans-
formation and regeneration was carried out as described by Ferradini et al.
(2011), whereas tobacco transformation and regeneration were performed by the
standard leaf discs method. In all tissue culture phases, except rooting, the
transgenic plants were selected on 7.5 mg l−1 (alfalfa) or 10 mg l−1 (tobacco)
phosphinothricin (pCAMBIA3300-LpTFL1 transformation) or 25 mg l−1 kanamycin
(pCAMBIA2300-LpTFL1 transformation). Cefotaxime (400 mg l−1 ) was added in
all media to prevent Agrobacterium growth.
Alfalfa green somatic embryos and tobacco green shoots obtained in selective
conditions were cultured into fully grown plants.
The genomic DNA was extracted from young, fully expanded leaves of the pu-
tative transgenic plants (GeneElute Plant Genomic DNA Miniprep Kit, Sigma).
PCR analysis was performed to detect the presence of LpTFL1 (alfalfa primers:
NosT-LpTFL1-For: GCGGGACTCTAATCATAAAAACC and LpTFL1-Rev: ACT-
GTATCTGTGCCTTCCTTCAG, expected amplicon 299 bp; tobacco primers:
LpTFL1-For: CAATGACACGACCAACAATAAGAG and ZmUBI-PROM-Rev:
TAATAAATAGACACCCCCTCCACA, expected amplicon 1184 bp). All PCR am-
plifications were performed using: 1x Buffer, 1.5 mM MgCl2 , 0.2 mM dNTP, 0.4
μM primers, 1U Taq (Sigma).
Total RNA was extracted (SpectrumTM Plant Total RNA kit-Sigma) from 100 mg
leaf tessues in pre-blooming phase of the PCR positive plants and purified from resid-
ual genomic DNA using the On-column DNASE I Digest Set (Sigma). The complete
DNA degradation was evaluated by PCR using the specific primer pair Actin-For 400
(CCACCACTCAATACGATGTTTCCA) and Actin-Rev 400 (CTAACGCACAC-
CTTCCCGTAT). cDNA was synthesized using M-MuLV Reverse Transcription Kit
(Diatheva) and used as template for amplification with the primers UBI int-For (TG-
GCAGCAGTCTCTCTCTGAC) and LpTFL1-Rev (see above). The actin cDNA was
amplified as a control with the primers indicated above.

19.3 Results

In the alfalfa transformation experiment with the construct pCAMBIA3300-LpTFL1,

29 putative transgenic somatic embryos were regenerated with phosphinothricin se-
lection, and 22 were cultured into plants (Fig. 19.2; Table 19.1). Seventeen plants
were PCR-positive (not shown).
RT-PCR confirmed the presence of the transcript in 8 of the 17 putative LpTFL1
plants; the level of expression showed some variation (Fig. 19.2). All LpTFL1-
expressing alfalfa plants were phenotypically indistinguishable from non-transgenic
controls. The pattern and timing of flowering was normal and fertile flowers were
produced (not shown). Seed set in transgenic plants ranged between 0.17 and 3.76
seeds per floret, whereas controls produced 3.06 and 3.77 seeds per floret. The
variation in seed set was within a normal range, and was not related to the expression
of LpTFL1.
158 N. Ferradini et al.

Fig. 19.2 Example of RT-PCR results in putative transgenic alfalfa plants: a amplification of the
LpTFL1 cDNA. b amplification of the actin cDNA (control). First lanes on the left (a and b):
GeneRuler 100 bp Plus DNA Ladder (Fermentas) CT : cDNA of a non-transgenic plant; 2, 4, 5, 6,
7, 9: cDNA of transgenic plants; BRT : no reverse transcriptase; C+ : pCAMBIA 3300-LpTFL1 (a)
or alfalfa genomic DNA (b); B: water

Table 19.1 Results of alfalfa transformation

Vector Leaf explants Somatic embryos Converted embryos RT-PCR-positive
plants
pCAMBIA3300-LpTFL1 184 169 22 8
pCAMBIA2300-LpTFL1 100 67 17 9

Table 19.2 Results of tobacco transformation

Vector Leaf explants Shoots Converted shoots PCR-positive plants
pCAMBIA3300-LpTFL1 140 53 23 12
pCAMBIA2300-LpTFL1 40 18 12 7

In the transformation experiment with pCAMBIA2300-LpTFL1, 67 kanamycin

resistant embryos were regenerated, 17 were converted to plants, and 9 were PCR
ad RT-PCR positive (Table 19.1). They were also phenotypically normal throughout
development and flowering was not different from non transgenic control plants.
In tobacco, the transformation experiments produced 12 PCR-positive plants
with the pCAMBIA3300-LpTFL1 construct and 7 PCR-positive plants with the
pCAMBIA2300-LpTFL1 construct (Table 19.2).
RT-PCR analysis of the transgenic plants confirmed the presence of the transcript
in 7 of the 12 pCAMBIA3300-LpTFL1 plants (Fig. 19.3). The pCAMBIA2300-
LpTFL1 plants have not yet been tested for transcript presence. None of the plants
showed phenotypic differences from the control plants. All the LpTFL1 tobacco
plants produced flowers and capsules with viable seeds.

19.4 Discussion

We have evaluated the LpTFL1 gene of perennial ryegrass as a possible candidate in

controlling the flowering process in alfalfa by overexpressing it with two constitutive
promoters, ZmUBI and CaMV35S. RT-PCR confirmed the expression of LpTFL1
19 Expression of the Lolium perenne Terminal Flower 1 Gene in Alfalfa and Tobacco 159

Fig. 19.3 Example of RT-PCR results in tobacco putative transgenic plants. CT : cDNA of a non-
transgenic plant; 1, 4, 9, 17, 24, 27: cDNA of putative transgenic plants; BRT : no RT enzyme; C+ :
pCAMBIA 3300-LpTFL1; B: water. Last lane: GeneRuler 100 bp Plus DNA Ladder (Fermentas)

gene in 17 plants. Our results show that the overexpression of the LpTFL1 gene in
alfalfa did not alter the architecture of the plants, nor it affected flowering and seed
set in any visible way. The same was true for tobacco, a phylogenetically distant
species.
On the contrary, LpTFL1 overexpression in species as diverse as A. thaliana,
L. perenne and F. rubra, resulted in a flowering delay or even suppression (Jensen et al.
2001, 2004). In Arabidopsis, for example, the overexpression of LpTFL1 produced
dramatic alterations in both the vegetative and reproductive phases: a significant
increase in the number of nodes before and after the floral transition phase and a
delay or absence of flowering (Jensen et al. 2001).
In our case, the absence of a phenotype both when the gene was driven by the
ZmUBI and by the CaMV 35S promoter, suggests that the lack of impact on pheno-
type is not due to a low level of expression; in fact, the Dual 35S promoter is known
to confer high constitutive expression in alfalfa (Samac et al. 2004). This will be
ascertained by further molecular analyses.
Our negative result may be explained by the taxonomic distance between perennial
ryegrass and the two recipient species, M. sativa and N. tabacum, even though
LpTFL1 did affect the phenotype in A. thaliana, also a taxonomically distant species.
Similar to our result, overexpression of Arabidopsis TFL1 had no effect on tobacco
flower development, whereas overexpression of CEN, the Antirrhinum TFL1 putative
ortholog, caused delayed and indeterminate flowering: this indicates that sequence
divergence in the TFL proteins among species underlies functional diversification.
Comparison between LpTFL1, the tobacco homologous proteins CET2 and CET4,
and M. truncatula TFL1-a revealed 71 and 72 % similarity, respectively. However, it
has been observed that overall sequence similarity does not always imply functional
similarity in this gene family (Yoo et al. 2010), while a single amino acid change in
the TFL gene can turn the encoded protein from a flowering repressor to a flowering
activator (Hanzawa et al. 2005).
Genes controlling flowering in legumes have been investigated (Hecht et al. 2005,
2011), but not yet in alfalfa. Even though the flowering pathways may differ between
annual and perennial Medicago species, the genome of M. truncatula is a valuable
resource for isolating alfalfa flowering genes.

Acknowledgments Funding was provided by the Italian Ministry of University and Science,
project: Impact of genetic engineering on the alfalfa genome and strategies to reduce it, PRIN
2007 (prot. 2007NEF8ZK 003, PI Daniele Rosellini).
160 N. Ferradini et al.

References

Bradley D, Ratcliffe O, Vincent C, Carpenter R, Coen E (1997) Inflorescence commitment and

architecture in Arabidopsis. Science 275:80–83
Danilevskaya ON, Meng X, Ananiev EV (2010) Concerted modification of flowering time and
inflorescence architecture by ectopic expression of TFL1-like genes in maize. Plant Physiol
153:238–251
Ferradini N, Nicolia A, Capomaccio S, Veronesi F, Rosellini D (2011) Assessment of sim-
ple marker-free genetic transformation techniques in alfalfa. Plant Cell Rep. 30:1991–2000.
doi:10.1007/s00299-011-1107-x
Hanzawa Y, Money T, Bradley D (2005) A single amino acid converts a repressor to an activator of
flowering. PNAS 102:7748–7753
Hecht V, Foucher F, Ferrándiz C, Macknight R, Navarro C, Morin J, Vardy ME, Ellis N, Beltrán JP,
Rameau C, Weller JL (2005) Conservation of arabidopsis flowering genes in model legumes.
Plant Physiol 137:1420–1434
Hecht V, Laurie RE, Vander Schoor JK, Ridge S, Knowles CL, Liew LC, Sussmilch FC, Murfet
IC, Macknight RC, Weller JL (2011) The pea GIGAS gene is a FLOWERING LOCUS T
homolog necessary for graft-transmissible specification of flowering but not for responsiveness
to photoperiod. Plant Cell 23:147–161
Jensen CS, Salchert K, Nielsen KK (2001) A Terminal Flower 1-like gene from perennial ryegrass
involved in floral transition and axillary meristem identity. Plant Physiol 125:1517–1528
Jensen CS, Salchert K, Gao C, Andersen C, Didion T, Nielsen KK (2004) Floral inhibition in red
fescue (Festuca rubra L.) through expression of a heterologous flowering repressor from Lolium.
Mol Breeding 13:37–48
Ohshima S, Murata M, Sakamoto W, Ogura Y, Motoyoshi F (1997) Cloning and molecular analysis
of the Arabidopsis gene Terminal Flower 1. Mol Gen Genet 254:186–194
Ratcliffe OJ, Amaya I, Vincent CA, Rothstein S, Carpenter R, Coen ES, Bradley DJ (1998) A
common mechanism controls the life cycle and architecture of plants. Development 125:1609–
1615
Samac DA, Tesfaye M, Dornbusch M, Saruul P, Temple SJ (2004) A comparison of constitutive
promoters for expression of transgenes in alfalfa (Medicago sativa). Transgenic Res 13:349–361
Yoo SJ, Chung KS, Jung SH, Yoo SY, Lee JS, Ahn JH (2010) BROTHER OF FT AND TFL1
(BFT) has TFL1-like activity and functions redundantly with TFL1 in inflorescence meristem
development in Arabidopsis. Plant J 63:241–253
Chapter 20
Morphological and Molecular Characterization
of Branching in Red Clover (Trifolium pratense)

A. Van Minnebruggen, I. Roldan-Ruiz, J. Van Dingenen,

E. Van Bockstaele, A. Rohde and G. Cnops

Abstract Clover is an essential element of sustainable grasslands. Clover reduces

the need for nitrogen fertilizer and results in improved nutritional value of grasslands.
Plant architecture, which is under genetic and environmental control, may have a
strong influence on traits such as forage yield, re-growth capacity, seed yield and
persistence in red clover. The genetic aspect of branching has been widely studied in
model plants but has received little attention in red clover. Our present aim is to trans-
late the knowledge regarding genes involved in bud outgrowth from model plants to
red clover. Branching was studied in two environments during one growing season in
clonal replicates of two genotypes with contrasting architecture, a highly branched
and prostrate genotype (Crossway_2), and a poorly branched and erect genotype
(Diplomat_8). The number of nodes and the quantity as well as the position of bud
outgrowth into branches differed greatly between genotypes and were similar across
both environments. The influence of auxin and strigolactone on bud outgrowth was
investigated by applying these hormones to isolated single node segments. Further-
more, genes from the strigolactone pathway were isolated from red clover and their
expression was studied in various tissues of the two genotypes.

20.1 Introduction

Red clover can be used in monoculture or in mixed grasslands, and in conventional

as well as organic farming. In mixtures, the addition of red clover reduces the need
for nitrogen fertilization and positively affects milk production due to its high palata-
bility and associated high intake rate. (Taylor and Quesenberry 1996; Bertilsson and
Murphy 2003). However, red clover has a low persistence and disappears from grass-
lands after two to three seasons. Plant architecture is investigated for its contribution
to persistence in this species, as it affects yield, tolerance to defoliation and re-growth
capacity.

G. Cnops () · A. V. Minnebruggen · I. Roldan-Ruiz · J. V. Dingenen · A. Rohde

Institute for Agricultural and Fisheries Research (ILVO),
Burg. Van Gansberghelaan 96, 9820 Merelbeke, Belgium
e-mail: [email protected]
E. V. Bockstaele
BUrg. Van Gansberghelaan 96, 9820 Merelbeke, Belgium

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 161

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_20, © Springer Science+Business Media Dordrecht 2013
162 A. V. Minnebruggen et al.

Plant architecture is determined by the degree of branching, internode elongation,

the determinacy of the shoot organ size and growth habit. Buds are formed in the axils
of the leaves and may either develop into branches or stay dormant. The processes that
control axillary bud formation and outgrowth are under genetic and environmental
control (e.g., photoperiod, plant density, defoliation), and influence architectural
differences between individual plants. The genetic aspect of plant architecture has
been demonstrated in model species to be regulated by genes involved in hormone
biosynthesis, signaling or response (McSteen and Leyser 2005; Ongaro and Leyser
2008, Ferguson and Beveridge 2009). Little is known, however, on the role of these
different aspects in red clover.
Due to its obligate outcrossing, red clover populations are genetically very di-
verse and cultivars also remain heterogeneous (Dias et al. 2008). In a previous study
(Cnops et al. 2010), we described the diversity of architectural types in 15 culti-
vars and six landraces. From that collection of plants, two genotypes (Crossway_2
and Diplomat_8) with extreme branching phenotypes were chosen for detailed mor-
phological, physiological and molecular analyses. Crossway_2 is a highly branched
genotype derived from a prostrate cultivar with improved persistence under grazing
from New Zealand (Rumball et al. 2003). Diplomat_8 is a poorly branched, erect
genotype selected within a French cultivar mainly used for mowing purposes. Here,
we present a detailed description of the architecture of these contrasting genotypes,
their response to hormones determining branching, and the expression of genes in-
volved in the strigolactone biosynthesis, signaling and response pathways. Together,
the data elucidate some of the genetic differences underlying architecture in red
clover.

20.2 Material and Methods

In a first step, the two studied genotypes were clonally propagated. Cuttings con-
taining one node, preferentially with an outgrowing bud, were rooted in soil under
moist conditions. After rooting, the cuttings were transplanted to 10 L pots. For
experiments carried out in the growth chamber (20 ◦ C; 16 h light), six clones of each
genotype were investigated. For experiments carried out in open air, 12 clones of
each genotype were used. The plants were regularly watered and no fertilization was
applied.
Analysis of Branching Patterns The number of nodes and branches and the flowering
time were recorded in the growth chamber and in open air. Plants in the growth
chamber were analyzed weekly during a period of three months; plants grown in
open air were analyzed once every two weeks (March till July 2010).
Response to Hormone Treatment Per treatment 25–30 stem fragments, containing
arrested buds derived from the nodes closest to the crown (node1 or node2), were
isolated from elongated branches of non-flowering plants of both genotypes grown
20 Morphological and Molecular Characterization of Branching in Red Clover . . . 163

under controlled conditions. The stem fragments were placed on water solidified
with 0.7 % plant tissue culture agar MC29, supplemented with 20 μM GR24 or
1 μM NAA. After three days the culture medium was refreshed. The stems were
photographed immediately after isolation and three, five and seven days later using
a stereomicroscope. Lengths were measured using the software programme ImageJ
(http://rsbweb.nih.gov/ij). Data were analyzed using ANOVA with post-hoc tests
Tukey HSD or Games-Howell (P < 0.05) (SPSS).
Patterns of Gene Expression Tissue from first and second internode and node1 (start-
ing from the crown) was isolated from plants grown in controlled conditions. Total
RNA was extracted using the Qiagen RNeasy Mini extraction kit. A DNAse treat-
ment was done using the DNA-free kit of Ambion, and was followed by a LiCl
precipitation. cDNAs were prepared using the Superscript III First-Strand Synthesis
Supermix kit of Invitrogen according to the manufacturer’s instructions. Q-PCR was
performed with SYBR Green I Master mix and was run on a LightCycler 480 Real-
time PCR system. Per genotype and per tissue five biological repeats were analyzed.
The expression of the genes TpMAX2, TpMAX4 and TpBRC1, genes involved in
the strigolactone (SL) pathway (MAX4 = SL biosynthesis gene; MAX2 = SL signal
transduction gene and BRC1 = SL response gene) were determined. The values were
normalized using TpACTIN. One of the internode1 tissues was used as a reference
for the calculation of relative expression values as calculated by the lightcycler480
software (1.5.0SP4).

20.3 Results and Discussion

Analysis of Branching Pattern Red clover forms a crown at soil level that consists
of a main axis (MA) and lateral branches. When the plants are not mown, the MA
in most genotypes remains vegetative during the first growing season. First-order
branches are formed from the axillary meristems (buds) in the leaf axils of the MA.
In leaf axils of first-order branches, axillary buds are formed that can develop into
second-order branches. Second-order branches can produce higher order branches
(Taylor and Quesenberry 1996; Cnops et al. 2010).
The number and position of nodes and branches were recorded at regular time
intervals in clonal replicates of Crossway_2 and Diplomat_8 grown in the growth
chamber and in open air. In Fig. 20.1, the temporal dynamics of the number of nodes
and branches in each genotype is shown for growth room conditions. Crossway_2
plants had from the start of the experiment more nodes and branches. The differ-
ences between the two genotypes were evident already during the vegetative phase
and increased during further development. At 198 ◦ C.days (mid vegetative phase),
Crossway_2 had produced three times more nodes and 5.5 times more branches than
Diplomat_8. At flowering time (751 ◦ C.days), 4.5 times more nodes and 6.5 times
more branches had been produced in Crossway_2. Four weeks later, Crossway_2 still
164 A. V. Minnebruggen et al.

Fig. 20.1 Differences in branching complexity between Crossway_2 and Diplomat_8. The total
number of nodes and branches is shown for plants grown in two different environments: growth
room, (top) open air bottom. The average value obtained for six clones (top) or 12 clones (bottom)
is represented; error bars are standard errors. Flowering time is indicated with an arrowhead

had 5.7 times more nodes and 5.9 times more branches than Diplomat_8 (Fig. 20.1).
The differences found were due to the production of a much higher number of nodes
and branches at all hierarchical orders, but were especially pronounced from the
second order onwards in Crossway_2 compared to Diplomat_8. These differences
were not due to a shift in development since both genotypes had similar flowering
times under growth room conditions.
We determined the bud outgrowing potential (BOP) (calculated as number of
branches/number of nodes) in both genotypes to see if the higher branching capacity
in Crossway_2 was due to an higher node initiation rate or to a higher rate of bud
outgrowth. The BOP was higher in Crossway_2 than in Diplomat_8 for all time
points except for the last two, for which a BOP of 20 % was estimated for both
genotypes (Fig. 20.1). This suggests that the differences in branching between these
two genotypes are not only due to a higher node initiation rate in Crossway_2 but
also due to a higher bud outgrowth capacity.
In open air, both genotypes flowered later (747 ◦ C.days) than in the growth
chamber (656 ◦ C.days), but similar results were obtained for plants grown in both
environments. The number of nodes, branches and the BOP was in Crossway_2 at all
time points higher than in Diplomat_8 (Fig. 20.1). The number of nodes and branches
in Crossway_2 plants grown in open air was higher compared to plants kept in the
growth chamber. However, the BOP at flowering time was in Crossway_2 compa-
rable for both environments. Diplomat_8 produced at flowering a similar number
of nodes and branches in both environments, but the BOP was slightly lower in the
growth chamber than in open air.
Response to Hormone Treatment To investigate if the genotypes differed in their
response to strigolactones (SL), stem fragments with single nodes were isolated
20 Morphological and Molecular Characterization of Branching in Red Clover . . . 165

Fig. 20.2 Influence of

hormone treatment on bud
outgrowth of nodes on
isolated stem fragments.
Relative length increment
[(length day 7–length day
0)/length day 0] at each time
point was determined as the
mean of 25 nodes. Bars are
standard errors.
CW2 = Crossway_2;
D8 = Diplomat_8

according to Chatfield et al. (2000) and grown in petri dishes containing medium
with or without the synthetic SL GR24 or NAA (Fig. 20.2). The addition of 1 μM
NAA reduced bud growth almost completely. Relative bud growth was significantly
reduced in both genotypes seven days after the addition of GR24. These data suggest
that both genotypes react similarly to external application of auxin and SL and that
they probably do not differ in their signalization. Differences in endogenous hormone
concentrations will be measured in future experiments.
Patterns of Gene Expression Red clover sequences of MAX4 (SL biosynthesis),
MAX2 (SL signaling) and BRANCHED1 (SL response) were PCR-amplified using
degenerated primers and cloned. The expression of these three genes was analyzed in
the two genotypes in the first and second internode and the first node of outgrowing
branches (Fig. 20.3). Strikingly, the expression of, TpMAX2 and TpBRC1 is higher
in the tested internode and node tissues of Crossway_2 than in those of Diplomat_8.
A similar expression pattern was observed for TpMAX4 in internode1. Based on
the phenotypes of Arabidopsis mutants, high MAX2, MAX4 and BRANCHED1 are
expected to be associated with less branching phenotypes, which is not the case in
the studied (dormant) tissues of red clover. These results need to be challenged in
further experiments addressing more tissues and additional genes; for example the
single node system is an ideal system to follow gene expression during bud outgrowth
under well-defined conditions.

20.4 Conclusion

The two genotypes described in this paper are a good model to study architectural
differences in red clover. We established an assay for bud growth in isolated stem
fragments, which gave us a powerful tool to study the influence of hormones on
bud outgrowth under controlled conditions and to look for individual differences in
166 A. V. Minnebruggen et al.

Fig. 20.3 Expression analysis

of strigolacton related genes
in red clover. Relative
expression pattern of genes in
the first and second internode
and node1 red clover tissues.
Data are the average of 5
biological repeats with
standard errors. The values
were normalized according to
TpACTIN

red clover gene expression during this process. Further research with the expression
pattern of a broader collection of genes and re-growth under mowing conditions
is required to fully understand the architectural differences observed in red clover
genotypes. The knowledge obtained from the molecular, physiological and molecular
study will be applied to a broader collection of genotypes representing a wide variety
20 Morphological and Molecular Characterization of Branching in Red Clover . . . 167

of architectural characteristics. Using allele mining, we hope to identify valuable

genotypes for introduction into breeding applications.

References

Bertilsson J, Murphy M (2003) Effect of feeding clover silages on feed intake, milk production and
digestion in dairy cows. Grass Forage Sci 58:309–322
Chatfield SP, Stirnberg P, Forde BG, Leyser O (2000) The hormonal regulation of axillary bud
growth in Arabidopsis. Plant J 24:159–169
Cnops G, Rohde A, Saracutu O, Malengier M, Roldán-Ruiz I (2010) Morphological and molecular
diversity of branching in red clover (Trifolium pratense). In: Huyghe C (ed) Sustainable Use of
Genetic Diversity in Forage and Turf Breeding. Springer Netherlands, pp 73–77
Dias PMB, Julier B, Sampoux JP, Barre P, Dall’Agnol M (2008) Genetic diversity in red clover
(Trifolium pratense L.) revealed by morphological and microsatellite (SSR) markers. Euphytica
160:189–205
Ferguson BJ, Beveridge CA (2009) Roles for auxin, cytokinin, and strigolactone in regulating shoot
branching. Plant Physiol 149:1929–1944
McSteen P, Leyser O (2005) Shoot branching. Annu Rev Plant Biol 56:353–374
Ongaro V, Leyser O (2008) Hormonal control of shoot branching. J Exp Bot 59:67–74
Rumball W, KEeogh RG, Miller JE (2003) Crossway and Grasslands Broadway red clovers
(Trifolium pratense L.). New Zealand J Agric Res 46:57–59
Taylor NL, Quesenberry KH (1996) Red clover science. Kluwer Academic Publishers, Dordrecht
 Part IV
Breeding Towards Breeding Objectives
Chapter 21
Designing Grass Cultivars for Droughts
and Floods

M. W. Humphreys, C. J. A. Macleod, W. R. Whalley, L. B. Turner,

M. S. Farrell, M. Ghesquière and P. M. Haygarth

Abstract Temperate productive grasslands are often located in areas of high rainfall
prone to flooding, but even here moderate summer droughts occur with regular-
ity causing significant yield reductions. Grasslands capable of resisting both water
excess and deficit are required. Alternative breeding technologies are employed to
combine as Festulolium cultivars the desirable traits of Lolium and Festuca species,
and also through their enhanced root systems, improve soil structures and hydrology.
An amphiploid L. perenne × F. pratensis cultivar can significantly reduce rainfall
runoff compared to either its parental species. Evidence suggests this was due to
an initial intensive root growth followed by extensive root senescence. This appears
to alter soil structure and increase soil porosity and moisture retention providing
an ecosystem service by both combating run-off subsequent to heavy rainfall and
increasing soil water supply during dry periods.
In a second programme aimed at improving drought resistance in Lolium, genes
for drought resistance were transferred from Festuca arundinacea var glaucescens.
These significantly increased water-use-efficiency and forage yield of Lolium under
soil water deficit conditions with no compromise to forage quality.

M. W. Humphreys () · L. B. Turner · M. S. Farrell

IBERS, Aberystwyth University, Gogerddan,
Aberystwyth SY23 3EB, Wales, UK
e-mail: [email protected]
C. J. A. Macleod
The James Hutton Research Institute, Craigiebuckler,
Aberdeen AB15 8QH, Scotland, UK
W. R. Whalley
Rothamsted Research, West Common, Harpenden
AL5 2JQ, Hertfordshire, UK
M. Ghesquière
INRA, Unit of Multidisciplinary Research on Pastures
and Forage Crops, BP6, 86600 Lusignan, France
P. M. Haygarth
The Lancaster Environment Centre, Lancaster University,
Lancaster LA1 4YQ, Lancaster, UK

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 171

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_21, © Springer Science+Business Media Dordrecht 2013
172 M. W. Humphreys et al.

21.1 Introduction

The Lolium-Festuca complex comprises diverse, heterogeneous, and primarily out-

breeding species that hybridise naturally having genomes that are functionally
interchangeable following chromosome exchange and their intrinsic propensity for
extensive interspecific chromosome pairing and recombination (Humphreys et al.
2003). As a consequence, genome combinations may be constructed capable of
providing a virtually unlimited supply of genetic variation and a large range of
contrasting grass phenotypes relevant to traits such as ontogeny, persistency, stress
tolerance, water and nutrient uptake and use, and agronomic value available to
plant breeders ready for evaluation for their suitability for incorporation in crop
improvement programmes (Ghesquière et al. 2010).
It is accepted widely that the diploid ryegrasses Lolium perenne and L. multiflo-
rum in terms of their rapid and prolonged seasonal growth that provides large foliar
yields and forage with high nutritive value and palatability are the most suitable grass
species for sustainable livestock agriculture. However, they lack generally the stress
resistance and water- and nutrient-use-efficiency common to many related fescue
(Festuca) species and ecotypes (Ghesquière et al. 2010). Their evolution, specia-
tion and subsequent widespread dispersal provides the opportunity for identifying
genotypes that confer the specific adaptations necessary for growth in the diverse
edaphic and climatic conditions that span globally most if not all temperate grass-
lands (Humphreys et al. 2011). From early in forage grass cultivar development,
the potential and desirability of combining the complementary characters found in
Lolium and Festuca species within stable hybrid forms has been recognised and over
the years a range of intergeneric cultivars has been developed (Humphreys et al. 2006;
Ghesquière et al. 2010). However, only recently following a change to EU legisla-
tion has it been possible to develop commercially as cultivars under their own unique
cultivar category; Festulolium, any Lolium × Festuca species hybrid combination
with the provision this was obtained by conventional plant breeding technologies.
Two breeding approaches are employed, amphiploidy where entire Lolium and
Festuca genomes are combined or introgression-breeding approaches where selected
traits are transferred from a donor species, usually Festuca with little disruption to
the host genome (Thomas et al. 2003). The former amphiploid approach has been
used more widely and various cultivars have been marketed, particularly in Cen-
tral Europe. The development of genome sequencing and marker-assisted breeding
technologies makes introgression and precision-led gene transfer another option, as
evidence from many sources would indicate that alien Festuca genes when trans-
ferred onto Lolium chromosomes function effectively and are transmitted normally
in accordance with Mendelian expectations and in many cases without incurring
linkage drag (Humphreys et al. 2005).
The current work describes the use of both breeding approaches to develop Fes-
tulolium cultivars capable of providing both sustainable crop production and an
environmental service during a time of climate change.
21 Designing Grass Cultivars for Droughts and Floods 173

21.2 Materials and Methods

21.2.1 Root Ontogeny and its Impact on Soil Structure

and Hydrology

Details of this project, SuperGraSS funded by a BBSRC are described elsewhere

(Macleod et al. 2007; Gregory et al. 2010; Humphreys et al. 2010).
The species and species hybrids used comprise six cultivars They were: Lolium
perenne, (Lp) cv AberStar, Lolium multiflorum, (Lm) cv AberEpic, Festuca
pratensis, (Fp) cv Bf993 (all 2n = 2x = 14), Festuca arundinacea (Fa) cv Dovey
(2n = 6x = 42), and Festulolium cultivars Lp × Fp cv Prior, and Lm × Fa cv 99-1
(both 2n = 4x = 28). Five individual tillers of equal size of each species were estab-
lished in 1 m deep pipes. A total of eight identical replicates were established for each
plant genotype. The pipes were lined with 153 mm wide 250 gauge lay-flat polythene
tubing (LBS Horticulture, Colne, Lancs. UK) heat-sealed at the bottom and punched
with four drainage holes. A 3 cm layer of horticultural grit was added to the base to
maintain good drainage and the polythene tubing was filled with Humax John Innes
No3 potting compost with wetting agent. Five tillers were planted in each pipe in
September 2006. The pipes were held in frames in a block design and maintained in
a heated (minimum temperature 8 ◦ C), unlit glass house throughout the year. Water
was applied evenly and this was confirmed with rain gauges placed within the blocks.
Below ground growth was investigated in mid-April and again in mid-September
in 2007 and 2008. The polythene sleeve was slid out of the pipe to enable non-
destructive observations of roots around the sides of the soil column. The maximum
depth at which roots were visible was recorded. The column of potting compost was
marked into 10 cm sections and root density was scored for each section. Root system
size was the sum of scores for all the sections, and root profile was the regression
coefficient for a line fitted through the root scores from the lowest section with roots
to the top section. These were calculated for the total and new root system. Root
persistency between spring and autumn in years 1 and 2 was also recorded.
The same six grass cultivars were established and maintained in the field as in the
root-pipe experiment over an identical time-frame and with the same cutting regime
and fertiliser applications. The cultivars were grown as four replicated 10 × 3 m
hydrologically-isolated plots, with rainfall and run-off measured over the course
of 31 rainfall events that subsequent to an initial six months establishment phase,
spanned all changes in grass growth over the two year field study (as in Gregory et al.
2010).

21.2.2 Development of Lolium Cultivars with Enhanced

Water-use-efficiency and Drought Resistance

Drought resistant L. multiflorum lines were developed by transfer of genes for drought
resistance from a Mediterranean fescue, F. arundinacea var glaucescens onto a ter-
174 M. W. Humphreys et al.

minal location on chromosome three of ryegrass without further disruption to the

Lolium genome (Humphreys et al. 2005). Subsequently in a DefraLINK programme
LK0688 using a sequence-tagged-site (STS) marker described by Humphreys et al.
(2005), the fescue genes were transferred into a breeders’ line producing population
Bb2540. In a field trial with three replications, performance over two consecu-
tive severe summer droughts (2009–2010) and consistent high temperatures (circa
38 ◦ C day/25 ◦ C night) was assessed at INRA, Lusignan, France, and also un-
der artificial drought conditions in a glasshouse at IBERS over one summer over a
three consecutive-month period with no irrigation. Recovery following irrigation was
evaluated at both locations. For the artificial drought assessment at IBERS, 3 × 1 m2
plots, were sown of Bb2540 at 3.3 g/m2 and compared with a L. multiflorum control
cv Atalja sown, established, and maintained as with Bb2540. All glasshouse plots
were maintained at field capacity until start of the drought (15-06-10) with soil water
content monitored twice weekly in two locations in each drought bin at four 10 cm
soil depths (0–40 cms). The methodologies used in the drought experiments were
standard IBERS protocols (e.g. Alm et al. 2011) with the incorporation of the soil
water measures. These were carried out using a Delta-T PR2 soil moisture sensor,
data captured by a Delta-T HH2 Moisture Meter (Delta-T Devices, Ltd., Cambridge,
UK). Forage yield was measured over four cuts prior to the start of drought, two
cuts during drought, followed by a final cut on 08-10-10 4 weeks into the recovery
phase subsequent to completion of the drought treatment when soil water content was
returned to field capacity. Water-use-efficiency (WUE) was calculated as g DM/unit
H2 O consumed.
Forage yield was measured in 3 × 10 × 3 m plots at INRA Lusignan over two
cuts in 2009 and three cuts in 2010 and compared between various Festulolium intro-
gression populations and amphiploid cultivars, the control cultivar L. multiflorum cv
Abys, hybrid ryegrass cultivars, Delicial and Lifema, and a Festulolium amphiploid
cultivar Leuer that contained entire genomes of F. arundinacea var glaucescens.
Regrowth and persistency was scored (0–5 scale) on 01-09-10 after two consecutive
dry summers.
Statistical analyses were performed on all according to standard procedures within
GenStat® Version 11.1 (Payne et al. 2008).

21.3 Results

21.3.1 Root Ontogeny and its Impact on Soil Structure

and Hydrology

21.3.1.1 Year 1

Species’ cultivars and hybrids differed significantly (P < 0.001) for annual forage
production, and spring and autumn root depth, root size, new root production and
21 Designing Grass Cultivars for Droughts and Floods 175

root profile. Root longevity between spring and autumn also differed significantly
(P = 0.009). For forage production, Lm cv AberEpic was most productive followed
by Lp × Fp cv Prior, Fa cv Dovey and Lp cv AberStar. The most substantial root sys-
tem was developed by Lp × Fp cv Prior followed by Fa cv Dovey, Lm cv AberEpic
and Lp cv AberStar.

21.3.1.2 Year 2

Significant differences (P < 0.001) in forage production between cultivars persisted

during Year 2 with Lp cv AberStar most productive largely due to its high tiller
number. Fa cv Dovey was also high yielding but this was due to leaf size rather than
tiller number. The two Festulolium cultivars Lp × Fp cv Prior and Lm × Fg cv 99-1
were ranked intermediate. Root depth in spring and autumn differed significantly
between genotypes (P < 0.001), but inYear 2, well developed root systems and growth
were particularly evident in Lp cv AberStar. However, the extensive root system
present in the Lolium cultivars, which was superior to the Festulolium cultivars in
the spring of Year 2, was due primarily to development of new roots within 20 cms
of the soil surface. Root persistency between spring and autumn, as Year 1, differed
amongst genotypes (P < 0.05), having increased in Fp cv Bf993 by 44 %, but having
decreased in Lolium spp. especially Lm cv AberEpic. Reduced root persistency was
particularly evident in Lp × Fp cv Prior (circa 50 %) and this was due mainly to
extensive senescence of its well developed root system, particularly at depth during
the summer of Year 2. In spring Year 2, the root profile of the two Festulolium
cultivars Prior and 99-1 was superior to the other plant species. This changed during
the summer months ofYear 2 with significant reductions in root growth and increased
root turn-over especially at depth in the soil column, reflected by both root profile
and deepest root measurements. By autumn Year 2, most new root growth in the
Festulolium cultivars was confined to the upper soil regions. At this time, overall
root growth was similar within cultivars of the same species, but differences in root
distribution were evident.

21.3.2 Development of Lolium Cultivars with Enhanced

Water-use-efficiency and Drought Resistance

The two year INRA field trial was an effective measure of the drought resistance
of population Bb2540 having two consecutive prolonged summer droughts (no rains
throughout August) with accompanying high temperatures (circa 37 ◦ C). With a total
yield over three cuts in 2010 (Year 2 of the field, is considered a better indicator than
Year 1 of crop performance under drought stress), diploid L. multiflorum Bb2540
yielded 9.23 TDM/ha which was not significantly different (P < 0.05) from that gen-
erated by amphiploid hybrid ryegrass cultivars, Delicial nor Lifema. Interestingly,
there was also no significant difference (P < 0.05) in total yield of Bb2540 compared
176 M. W. Humphreys et al.

with the amphiploid Festulolium cultivar Leuer (8.63 TDM/ha) which was selected in
France for its persistency under drought. The cultivar comprised two entire genomes
of F. glaucescens var arundinacea compared to the single translocated Festuca chro-
mosome 3 segment from the same species present in Bb2540. There was indeed no
difference in total yield between the three amphiploid cultivars and Bb2540 over
the entire two Year field trial. The subsequent persistency scores (0–5 scale) taken
on 1st September 2010 indicated a greater persistency for the Festulolium cv Leuer
(P < 0.05) over Bb2540 and the amphiploid hybrid ryegrasses which all shared a com-
mon persistency (score = 3). These were all significantly more persistent (P < 0.05)
than the control Lolium variety Lm cv Abys (2x) (score = 1).
The forage quality of Bb2540 (with two other IBERS Festulolium introgression
lines included in the INRA field trial) measured as %water-soluble-carbohydrate
(WSC) over all three cuts in 2010 was significantly greater (P < 0.05) than all other
cultivars used in the field trial (including Lm cv Abys) at all three forage cuts. The
same outcome was obtained in field trials (2009 and 2010) at IBERS where Bb2540
was the highest ranked cultivar for %WSC (over three cuts/yr) being significantly
higher (P < 0.05) than UK Nationally Listed cultivar Lm cv Ligrande. %WSC of
Bb2540 was not significantly different to that of other Nationally Listed Lolium
cultivars; Lm cultivars Alamo (in 2010), and DaVinci, and Belluna (in both 2009
and 2010).
At IBERS, in a simulated drought experiment in a purpose-built glasshouse, water-
use-efficiency was calculated during a three month period with no irrigation and also
during a subsequent one month recovery phase when water was returned sufficient to
restore soil water to field capacity. 3 × 1 m2 replicated swards of Bb2540 (and two
alternative introgression populations) and also as control, Lm cv Atalja were sown,
established and compared for their forage yields taken over four cuts between March
and June 2010 prior to the start of the drought treatment. Over these cuts and given
an optimal water supply (field-capacity) Bb2540 out-yielded the control Lm cultivar
Atalja (P < 0.001), particularly in those cuts made early in March and June. During
the drought treatment, cuts were made in July 4 weeks (moderate drought) and in
September 12 weeks (severe drought) and water-use-efficiency (WUE) calculated as
Total Dry Matter (TDM)/unit H2 O consumed. The drought treatment had a highly
significant impact on forage yield in all grass populations tested. However, total yield
from Bb2540 over the two cuts during the drought treatment was twice that of Atalja
with an 88 % improvement in WUE (P < 0.001). Water uptake by Bb2540 overall was
1 % higher than by Atalja, but this was not statistically significant suggesting that
it was more the water utilisation capabilities of Bb2540 rather than more efficient
water-uptake that gave Bb2540 an adaptive advantage. Over all glasshouse plots, soil
water content declined rapidly from an initial field capacity of 38 % to a 10 % level
that was maintained consistently during the latter half of the drought treatment (circa
6 weeks) when foliar growth had declined significantly (P < 0.001). On re-watering
during the one month recovery period, field capacity at 38 % and optimal water
supply was once again restored. The yield of Bb2540 after 1 month recovery was
> 100 % greater that of the control, Lm cv Atalja.
21 Designing Grass Cultivars for Droughts and Floods 177

21.4 Discussion

Given a changing climate with grasslands being subjected at increasing frequen-

cies to episodes of either extreme rainfall or persistent drought, there is a pressing
requirement for grass breeders and geneticists to respond to ensure a maintained
food security by increasing the resilience of currently productive cultivars, espe-
cially Lolium perenne and L. multiflorum. Hybridisation with Festuca species to
produce Festulolium cultivars provides us that opportunity and as shown here also
provides us with additional important examples of an ecosystem service. Two al-
ternative approaches to Festulolium breeding are described, the use of amphiploidy
that generated Festulolium loliaceum cv Prior, where entire genomes of Lolium and
Festuca were combined, and the alternative more targeted gene introgression ap-
proach that generated diploid L. multiflorum population Bb2540 that contained a
single translocated Festuca-derived genome sequence in an otherwise intact Lolium
genome (Humphreys et al. 2005).
Gregory et al. (2010) and Humphreys et al. (2010) describe a part of the multi-
discipline BBSRC-funded programme SuperGraSS. Here it is demonstrated how the
initial extensive root growth and the subsequent senescence of Festulolium loliaceum
cv Prior, particularly at depth was likely to have had a major impact on soil structure
and hydrology and led to the reduced rainfall run-off reported in the SuperGraSS
programme. In brief, it was shown in the BBSRC-funded project that Prior had a
significant impact on plot scale rainfall runoff compared to either its parental species.
Over a two-year period, Prior (4x) reduced runoff by 51 % compared to the UK Na-
tional Listed L. perenne cv AberStar, and by 43 % compared to Festuca pratensis cv.
Bf993, a progenitor of cultivar S215 the fescue parent of Prior. A number of previous
laboratory studies have described how roots change soil hydraulic properties (e.g.,
Whalley et al. 2005). These reports have demonstrated how changes to the water re-
lease characteristics of soils tend to be associated with either an increased number of
larger pores in the rhizosphere or an increase in water repellence. Root activity tends
to increase the number of large pores in the soil, and as it is known that certain fescue
species such as Festuca arundinacea var glaucescens produce larger deeper rooting
systems than ryegrass (Durand et al. 2007), this it is thought will contribute to their
proven overall greater drought resistance both by their capability to access water
deep in the soil and by their ability to restructure soil to aid in its water retention.
Humphreys et al. (2005) described the introgression breeding programme that led
to the transfer of genes for drought resistance from F. arundinacea var glaucescens
into diploid L. multiflorum. The Festuca genes were transferred onto a distal location
on the satellite region of Lolium chromosome 3. A major QTL for resistance to severe
drought stress has been identified on linkage group 3 in F. pratensis that would in-
clude the F. arundinacea var glaucescens translocation (Alm et al. 2011) The Festuca
translocation had been transferred into a breeders’ line (Bb2540) using a sequence-
tagged-site (STS) marker described by Humphreys et al. (2005) and the heritability
and continued function of the Festuca-derived drought resistance genes confirmed in
178 M. W. Humphreys et al.

a previous Defra-funded programme (Humphreys and Humphreys 2007). To the au-

thors knowledge the report here is the first ever detailed study of the potential impact
on forage yield and WUE to be achieved by any forage grass by the incorporation of
a single alien chromosome segment from a wild-related species. A highly significant
increase (P < 0.001) in WUE of 88 % in Bb2540 compared with the control Lolium
cultivar Atalja demonstrates very well the possible impact and advantage such a cul-
tivar might have in drought-prone grasslands. In its natural habitat, F. arundinacea
var glaucescens undergoes a phase of summer quiescence (Humphreys et al. 1997)
which would be a highly unsuitable agronomic trait for livestock agriculture in Eu-
ropean grasslands, where continued grassland growth and production in summer is
considered an essential requirement. The high yields of Bb2540 both in irrigated
and in droughted conditions are proof of its excellent agronomic performance. In
field trials at IBERS and at INRA, Lusignan, where summer droughts are a major
constraint to grassland persistency and productivity, Bb2540 performed consistently
well, outperforming most other cultivars tested. Over 2 years, the cultivar had pro-
ductivity consistent with the amphiploid cultivar Leuer which utilized entire genomes
of F. glaucescens to confer its drought resistance traits. No convincing evidence was
found that the drought resistance achieved in Bb2540 was due to the presence of
a superior root system and water uptake to that found in the Lolium control which
was evidence that it resulted from a more effective retention and use of the water it
obtained.
A major advantage of the introgression-breeding approach was the ability to se-
lect for improved drought resistance without compromise to forage quality. WSC
measures in Bb2540 both at IBERS and INRA were consistently high and equivalent
to, or better than, current UK Nationally Listed L. multiflorum cultivars.
The Festuca genes from chromosome three are currently being transferred by
marker-assisted breeding into L. perenne where their impact on WUE will be as-
sessed. For future grassland agriculture in the UK, it will be the impact that Festuca
genes for drought resistance can bring into L. perenne, the agricultural grass species
of choice, that which will determine their ultimate value and use in sustaining future
grassland production in a time of climate change.
Although not apparently relevant to Bb2540, the generally superior root systems
of Festuca species compared with those of Lolium offer a range of potential uses
in regard to environmental safeguards. We cite here using as example Festulolium
loliaceum cv Prior, the potential use of Festulolium hybrids to modify soil structures
to encourage improved rainfall retention and reduce flooding and run-off of soil
nutrients. The substantial root growth and turn-over exhibited in the cultivar Prior will
also likely support effective C-sequestration properties. Large Festuca roots will also
stabilize soils, and safeguard against soil erosion. They will be more effective than
those of Lolium for penetration of hard compacted soils. There is also accumulating
evidence to indicate roots of Festuca species are more efficient in nutrient and water
capture.
To conclude, given positive support from all the relevant stakeholders, commercial
use of Festulolium cultivars will provide new opportunities for European grasslands
21 Designing Grass Cultivars for Droughts and Floods 179

where they should demonstrate a truly multifunctional value sustaining both pro-
ductive and persistent grassland agriculture at a time of climate change whilst also
providing additional environmental safeguards.

References

Alm V, Busso CS, Ergon Å, Rudi H, Larsen A, Humphreys MW, Rognli OA (2011) A QTL analysis
and comparative genetic mapping of frost tolerance, winter survival and drought tolerance in
meadow fescue (Festuca pratensis Huds.). Theor Appl Genet 123(3):369–382
Durand JL, Bariac T, Ghesquière M, Brion P, Richard P, Humphreys MW, Zwierzykowski Z (2007)
Ranking of the depth of water extraction by individual grass plants using natural 18O isotope
abundance. Environ Exp Bot 60(1):137–144
Ghesquière M, Humphreys MW, Zwierzykowski Z (2010) Festulolium. In: Boller B, Posselt UK,
Veronesi F (eds) Fodder crops and amenity grasses. Springer Science + Business Media, New
York, pp 293–316 (doi: 10.1007/978-1-4419-0760-8_12)
Gregory AS, Webster CP, Watts CW, Whalley WR, Macleod CJA, Joynes A, Papadopoulos A,
Haygarth PM, Binley A, Humphreys MW, Turner LB, Skøt L (2010) Soil management and grass
species effects on the hydraulic properties of shrinking soils. Soil Sci Soc Am J 74:753–761
Humphreys MW, Humphreys MO (2007) Genetic analysis and selection of grass traits using DNA
marker technology for sustainable grassland management. Final Report Defra LS3607
Humphreys MW, Thomas HM, Harper JA, Morgan WG, JamesA, GhamariZareA, Thomas H (1997)
Dissecting drought—and cold-tolerance traits in the Lolium-Festuca complex by introgression
mapping. New Phytol 137(1):55–60
Humphreys MW, Canter PJ, Thomas HM (2003) Advances in introgression technologies for
precision breeding within the Lolium-Festuca complex. Ann Appl Biol 143:1–10
Humphreys J, Harper JA, Armstead IP, Humphreys MW (2005) Introgression-mapping of genes
for drought resistance transferred from festuca arundinacea var. Glaucescens into lolium
multiflorum. Theor Appl Genet 110(3):579–587
Humphreys MW, Yadav RS, Cairns AJ, Turner LB, Humphreys J, Skøt L (2006) A changing climate
for grassland research. New Phytol 169:9–26
Humphreys MW, Turner LB, O’Donovan SA, Macleod CJA, Whalley WR, Haygarth PM (2010)
Grass root turn-over for improved soil hydrology to combat flooding. Proceedings of the
European Grassland Federation Conference 2010, Kiel, Germany, 30th Aug–2nd Sept 2010
Humphreys MW, Marshall AH, Collins RP, Abberton MT (2011) Exploiting genetic and phenotypic
plant diversity in grasslands. In: Lemaire G, Hodgson J, Chabbi A (eds) Grassland Productivity
and Ecosystem Services. CABI Publishing, Chapter 16, pp 148–157
Macleod CJA, Binley A, Clark LJ, Hawkins SL, Humphreys MW, Turner LB, Whalley WR, Hay-
garth PM (2007) Genetically modified hydrographs: what can grass genetics do for temperate
catchment hydrology? Hydrol Process 21(16):2217–2221
Payne RW, Murray DM, Harding SA, Baird DB, Soutar DM (2008) GenStat® for Windows® 11th
edition, introduction. Hemel Hempstead: VSN International
Thomas HM, Morgan WG, Humphreys MW (2003) Designing grasses with a future—combining
the attributes of Lolium and Festuca. Euphytica 133:19–26
Whalley WR, Riseley B, Leeds-Harrison PB, Bird NRA, Leech PK, Adderley WP (2005) Structural
differences between bulk and rhizosphere soil. Eur J Soil Sci 56(3):353–360
Chapter 22
Variation and Heritability of α-Linolenic Acid
Content and Rumen Escape Protein Fraction
in Fodder Grass and Clover

J. Baert, M. Vandewalle, J. De Riek, J. De Boever, V. Fievez and C. Van Waes

Abstract α-Linolenic acid (C18:3) in forages enhances ω-3 fatty acid and conju-
gated linoleic acid content in milk and meat of ruminants with beneficial effects on
human health. An increase of the fraction of rumen escape protein (REP) in grass
and clover may reduce nitrogen losses by cattle. To determine the variation of C18:3
and REP, field plot trials under conservation management were set up. The trials in-
cluded cultivars of perennial ryegrass, Italian ryegrass, meadow fescue, tall fescue,
timothy, cocksfoot and red and white clover. For the grasses we found the highest
C18:3 content in timothy and the lowest in Italian ryegrass. Tetraploid ryegrass va-
rieties had on average a higher C18:3 content than diploid varieties. The linolenic
acid content in white clover was higher than in red clover and the content in both
clovers was higher than in the grasses. For all species the C18:3 content was highly
positively correlated with the protein content. Among the grasses the REP was high-
est in cocksfoot and tall fescue and lowest in perennial ryegrass. Diploid ryegrass
cultivars had a higher REP fraction than tetraploid cultivars. The REP in red clover
was higher than in white clover and the REP in both clovers was lower than in the
grasses. REP was highly negatively correlated with dry matter digestibility and the
content of digestible protein. For the estimation of the heritability we determined the
C18:3 content and REP of 300 single plants of each of the species perennial ryegrass,
tall fescue and red and white clover, grown in pots. We carried out positive and neg-
ative selections for both parameters in the four species. The results of their offspring
suggested a moderate to good heritability of both parameters and opportunities for
breeding. Improved grass/clover mixtures may be an important source of C18:3 in
ruminant feeding. However because of the negative correlation between REP and

J. Baert () · M. Vandewalle · J. D. Riek

Institute for Agricultural and Fisheries research (ILVO),
Plant Sciences Unit, Caritasstraat 21, Melle 9090, Belgium
e-mail: [email protected]
J. D. Boever · C. Van Waes
Institute for Agricultural and Fisheries research (ILVO),
Animal Sciences Unit, Scheldeweg 68, Melle 9090, Belgium
V. Fievez
University of Ghent, Animal Production Unit,
Proefhoevestraat 10, Melle 9090, Belgium

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 181

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_22, © Springer Science+Business Media Dordrecht 2013
182 J. Baert et al.

the digestible protein content, breeding for improved nitrogen efficiency in cattle by
increasing the fraction of rumen escape protein in the forage is not obvious.

Keywords Grass · Clover · Linolenic acid · Rumen escape protein · Variation ·

Heritability

22.1 Introduction

The nutrition of cattle may affect the quality of the meat and milk as well as their
environmental impact. In order to breed grass and clover varieties that meet current
sustainable requirements we studied the variation and heritability of two quality
parameters in grass and clover: the content of α-linolenic acid (C18:3) and the
fraction of rumen escape protein (REP).
α-Linolenic acid is the major fatty acid in grass and clover. This poly-unsaturated
ω-3 fatty acid and the derived conjugated linoleic acid (CLA) in ruminant milk and
meat may be beneficial to human health (prevention of cardiovascular diseases and
cancer).
Protein of grass and clover is extensively degraded in the rumen. The degraded
protein is partly transformed by the rumen bacteria into useful microbial protein
and partly lost in the urine to the environment causing ammonia volatilization and
nitrate leaching. The fraction of the protein that escapes rumen degradation may be
digested in the intestine. The levels of excreted nitrogen may be reduced by increasing
the availability of energy for rumen fermentation (e.g. water soluble carbohydrate
content, Humphreys et al. 2010) or by enhancing the REP fraction of the forage
(Tamminga et al. 1994).
Using fast and reliable screening procedures we determined the variation of the
C18:3 content and REP of grass and clover species and varieties. Within a few
species we carried out a divergent selection for both parameters and estimated their
heritability.

22.2 Material and Methods

To study the variation of C18:3 and REP of the grasses a strip plot experiment in three
replicates with two levels of nitrogen application rate (350 and 230 kg N/ha/year)
and 24 varieties belonging to six species was sown in September 2006 in Merelbeke
(Belgium). The variation of the clover was tested in a random block design with four
varieties of red and three varieties of white clover. In 2007 and 2008 the grass and
clover plots (size: 8.1 m2 ) were cut five times a year. The species and varieties in the
trials are presented in Table 22.1.
The heritability of C18:3 and REP was estimated for the species: perennial rye-
grass (diploid), tall fescue, red clover (diploid) and white clover. In 2008, 300 single
plants (12 cultivars × 25 plants per cultivar) from each of the four species were
22 Variation and Heritability of α-Linolenic Acid Content and Rumen Escape . . . 183

Table 22.1 Species and varieties in variation trials

Species Variety Ploidy Type
Lolium perenne Rebecca 2x Early heading
Indiana 2x Early heading
Aberdart 2x Intermediate heading
Premium 2x Intermediate heading
Barnhem 2x Late heading
Merks 2x Late heading
Merlinda 4x Early heading
Lacerta 4x Early heading
Aberglyn 4x Intermediate heading
Graciosa 4x Intermediate heading
Abercraigs 4x Late heading
Floris 4x Late heading
Phleum pratense Comer 6x Hay type
Motim 6x Pasture type
Festuca pratensis Merifest 2x
Pradel 2x
Festuca arundinacea Barolex 6x
Barelite 6x
Bariane 6x
Dactylis glomerata Barmoral 4x
Lolium multiflorum Bellem 2x
Davinci 2x
Gemini 4x
Barmega 4x
Trifolium pratense Merula 2x
Lemmon 2x
Milvus 2x
Larus 4x
Trifolium repens Barblanca 4x Large leaves
Riesling 4x Intermediate leaves
Barbian 4x Small leaves

grown in 12 L pots. Two cuts of the sowing year were analyzed for the determination
of the C18:3 content and the REP. For each species the 30 plants with the highest
and lowest C18:3 content or REP were selected. In total 16 isolated polycrosses with
these plants were installed. In 2009 we harvested the seeds of each of the divergent
selection groups. These progenies (300 plants per selection group) were again sown
in 12 L pots in 2010 together with seeds of three of the original cultivars as reference.
Two cuts were harvested in 2010 and analyzed for C18:3 and REP. The heritability
of both parameters for each species was calculated as: (the difference between the
divergent progeny groups)/(difference between the divergent parental groups).
In both the variation and heritability study we determined at each cut the fresh
yield and the content of dry matter (DM), crude protein (CP), water soluble carbo-
hydrates (WSC), neutral detergent fiber (NDF), acid detergent fiber (ADF), C18:3
and the in vitro dry matter digestibility with cellulase (DMD). Except for yield all
these parameters were estimated by near infrared reflectance spectroscopy (NIRS)
on dried samples of the grass and clover. NIRS calibrations of C18:3 were developed
 184 J. Baert et al.

12

10
C18:3 content (g/kg DM)

0
Barmega

Bariane

Graciosa
Barolex
Barelite

Rebecca
Abercraigs

Barmoral
Premium
Barnhem
Aberdart

Merifest

Aberglyn

Merlinda
Lacerta
Indiana
Davinci
Gemini
Bellem

Comer

Mom
Pradel

Merks

Floris
Fig. 22.1 C18:3 content (average ± s.e. of 2 years, 5 cuts a year, 2 N rates) of 24 varieties belonging
to the grass species perennial ryegrass (green), Italian ryegrass (blue), tall ferscue (red), meadow
fescue (brown), timothy (yellow) and cocksfoot (orange). (dark: tetraploid, light: diploid)

for grass and clover separately. The REP was estimated by means of a multiple re-
gression equation based on DMD, DM and ADF with a determination coefficient
of 84 % and a residual error of 3.1 %-units (Vandewalle et al. 2010). To derive this
prediction equation REP was determined on a limited sample collection of 56 fresh
grass and clover samples through nylon bag incubations in the rumen of fistulated
cows according to the CVB protocol (CVB 2003), whereas corresponding dried and
ground samples were chemically analyzed. The content of protein digestible in the
intestine (PDI) was estimated by an equation including DMD, CP, DM and NDF also
derived from the limited sample set.

22.3 Results

22.3.1 Variation of C18:3

For the grasses there were significant effects of the harvest year, the cuts within each
year, the N fertilization level, the species and the varieties on the C18:3 content.
At the high N application rate the C18:3 content was 0.4 g/kg DM higher than at
the low rate. The mean C18:3 content of the varieties ranged from 7.1–10.1 g/kg
DM (Fig. 22.1) with an average of 8.5 g/kg DM. The C18:3 content of timothy was
significantly higher and the C18:3 content of Italian ryegrass significantly lower than
of the other species. For both ryegrass species tetraploid varieties had on average a
higher C18:3 content than the diploid varieties.
22 Variation and Heritability of α-Linolenic Acid Content and Rumen Escape . . . 185

16
C18:3 content (g / kg DM) 14
12
10
8
6
4
2
0

Lemmon

Barblanca
Riesling

Barbian
Merula
Milvus

Larus

Fig. 22.2 C18:3 content (average ± s.e. of two years, five cuts a year) of seven clover varieties
belonging to white clover (white) and red clover (black: tetraploid, grey: diploid)

The C18:3 content of the clover (mean: 13.3 g/kg DM) was much higher than that
of the grasses. White clover had a significantly higher C18:3 content (14.0 g/kg DM)
than red clover (12.8 g/kg DM) (Fig. 22.2). There were no significant differences
among varieties within the species. For both the grasses and the clovers the C18:3
content was highly positively correlated with the protein content.

22.3.2 Variation of REP

For the grasses, REP was significantly affected by harvest year, cuts within each
year, species and variety. The N application rate had no significant effect. The mean
REP of the varieties ranged from 38.0–44.6 % (Fig. 22.3) with an average of 41.2 %.
The REP of cocksfoot and tall fescue was significantly higher and that of perennial
ryegrass significantly lower than the other species. For both ryegrass species diploid
varieties had a higher REP than the tetraploid varieties.
The REP of the clover (mean: 34.2 %) was much lower than the REP of the grasses.
Red clover had a significantly higher REP (36.2 %) than white clover (31.7 %)
(Fig. 22.4). There were almost no differences between varieties within the species.

22.3.3 Heritability of C18:3 and REP

Table 22.2 shows the C18:3 content and REP of the positive and negative selections
that were made for both parameters of each species and of their offspring. Based on
these divergent selection the heritability was calculated (Table 22.3).
 186 J. Baert et al.

46
Rumen Escape Protein (%)

44

42

40

38

36

34
Lacerta

Davinci
Mom
Bariane
Barolex
Barelite
Comer
Graciosa
Floris
Merlinda
Aberdart
Barnhem
Merks
Barmega

Barmoral
Indiana
Premium
Gemini
Pradel
Bellem
Merifest
Rebecca
Aberglyn
Abercraigs

Fig. 22.3 REP (average ± s.e. of two years, five cuts a year, 2 N rates) of 24 varieties belonging
to the grass species perennial ryegrass (green), Italian ryegrass (blue), tall ferscue (red), meadow
fescue (brown), timothy (yellow) and cocksfoot (orange). (dark: tetraploid, light: diploid)

39
Rumen Escape Protein (%)

37
35
33
31
29
27
25
Barblanca

Barbian

Riesling

Merula

Lemmon

Milvus
Larus

Fig. 22.4 REP (average ± s.e. of 2 years, 5 cuts a year) of 7 clover varieties belonging to white
clover (white) and red clover (black: tetraploid, grey: diploid)

By selecting the 10 % best plants the progress of the positive selection group of the
parents was higher for C18:3 than for REP because of the higher variation in C18:3.
Selection for higher REP could not be done without decreasing protein digestible in
the intestine.
Differences between the positive and negative groups were still observed in the
offspring but to a lower extent than in the parents. The heritability of C18:3 and REP
22 Variation and Heritability of α-Linolenic Acid Content and Rumen Escape . . . 187

Table 22.2 C18:3 content, REP and PDI (in % of the mean of 3 reference varieties) of the positive
(+) and negative (−) selections of the parents and offspring of 2 grass and 2 clover species
Generation (year) Species C18:3 REP (PDI)
− + − +
Parents (2008) Tall fescue 77 125 91 (104) 110 (96)
Perennial ryegrass 76 127 92 (103) 109 (96)
White clover 84 117 93 (107) 107 (99)
Red clover 81 121 91 (100) 110 (99)
Offspring (2010) Tall fescue 96 115 95 (104) 113 (91)
Perennial ryegrass 72 112 96 (105) 107 (99)
White clover 94 104 97 (105) 104 (100)
Red clover 86 102 100 (96) 104 (93)

Table 22.3 Heritability of

Species C18:3 REP
C18:3 content and REP for
2 grass and 2 clover species Tall fescue 0.39 0.96
Perennial ryegrass 0.79 0.64
White clover 0.28 0.45
Red clover 0.41 0.25

for clover was on average lower than for the grasses. Perennial ryegrass showed a
good heritability for C18:3 content and tall fescue for REP.

22.4 Discussion

Although the α-linolenic acid content in dried grass and clover is low, there is a
significant variation between and within species with a good heritability. Elgersma
et al. (2003) also found consistent differences in the C18:3 content among cultivars
of perennial ryegrass throughout the season. Fresh perennial ryegrass may contain
about 15 g C18:3/kg DM. Thanks to the variation and heritability of C18:3, a variety
with a 10 % higher C18:3 content may be bred. This variety may produce 15 kg
C18:3 more per ha (let dry matter yield be 10 t/ha/year). The intake of 10 kg dry matter
of this variety offers the cow 165 g C18:3 which is similar to 1.4 kg of an enriched
concentrate (12 % C18:3) based on linseed. This is about the quantity of concentrate
that is supplemented in winter to cows for the production of high ω-3 milk. Because
of the higher C8:3 content in white clover, mixtures of improved grass and clover
varieties may replace even higher quantities of the linseed concentrates.
Due to the strong negative correlation between REP and digestibility, breeding
for higher REP doesn’t hold many prospects. Lower digestibility may reduce micro-
bial protein production. Among the grasses tall fescue has a high REP with a high
heritability. Some small progress in REP should be possible without loss of PDI.
Clovers have a high CP content with a much lower REP than grasses. In this respect
tall fescue may very well complement red clover as a high yielding mixture for the
production of home grown protein roughage under conservation management.
188 J. Baert et al.

Acknowledgments This research is supported by the IWT-Flanders—contractual research in

agriculture (No. 50639).

References

CVB (2003) Protocol voor in situ pensincubatie: bepaling van afbraaksnelheid en uitwasbare fracties
van eiwit, zetmeel, celwanden en organische restfractie. Centraal Veevoeder Bureau, Lelystad,
p 14
Elgersma A, Ellen G, van der Horst H, Muuse BG, Boer H, Tamminga S (2003) Influence of cultivar
and cutting date on the fatty acid composition of perennial ryegrass (Lolium perenne L.). Grass
Forage Sci 58:323–331
Humphreys M, Feuerstein U, Vandewalle M, Baert J (2010) Ryegrasses. In: Boller B, Posselt UK,
Veronesi F (ed) Handbook of plant breeding 5, fodder crops and amenity grasses. Springer, New
York
Tamminga S, Van Straalen WM, Subnel APJ, Meijer RGM, Steg A, Wever CJG, Blok MC (1994)
The Dutch protein evaluation system: the DVE/OEB-system. Livest Prod Sci 40:139–155
Vandewalle M, Van Ranst G, Fievez V, De Boever J, Van Waes C, De Riek J, Baert J (2010) Variability
of the rumen escape protein and fatty acid composition of grass and clover species and cultivars.
In: Huyghe C (ed) Sustainable use of genetic diversity in forage and turf breeding. Springer,
New York
Chapter 23
Similarities and Differences in Leaf Proteome
Response to Cold Acclimation Between
Festuca pratensis and Lolium perenne

A. Kosmala, A. Bocian, M. Rapacz, B. Jurczyk, Ł. Marczak

and Z. Zwierzykowski

Abstract Lolium perenne is used extensively as a forage and turf grass due to
its high nutritive values, rapid establishment and persistence. Festuca pratensis,
a species closely related to L. perenne, has lower nutritive values but is one of the
most winter-hardy species within the Lolium-Festuca complex. Frost tolerance is the
main component of winter-hardiness and the plant species growing in temperate re-
gions can acquire it through exposure to low, non-lethal temperatures, a phenomenon
known as cold acclimation. Herein, we review our recent results of two proteomic
projects, focused on F. pratensis and L. perenne. Each project involved the compar-
ison of leaf protein accumulation profiles during cold acclimation between plants
with different levels of frost tolerance by the use of two-dimensional electrophore-
sis and identification of differentially accumulated proteins by mass spectrometry.
In the present paper similarities and differences in leaf proteome response to cold
acclimation between F. pratensis and L. perenne are summarized.

Keywords Cold acclimation · Festuca pratensis · Frost tolerance · Lolium

perenne · Proteome

Abbreviations
CA Cold acclimation
HFT High frost tolerant
LFT Low frost tolerant

A. Kosmala () · A. Bocian · Z. Zwierzykowski

Institute of Plant Genetics, Polish Academy of Sciences, Strzeszyńska 34,
60-479 Poznań, Poland
e-mail: [email protected]
M. Rapacz · B. Jurczyk
Department of Plant Physiology, University of Agriculture in Kraków,
Podlużna 3, 30-239 Kraków, Poland
Ł. Marczak
Institute of Bioorganic Chemistry, Polish Academy of Sciences,
Noskowskiego 12/14, 61-704 Poznań, Poland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 189

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_23, © Springer Science+Business Media Dordrecht 2013
190 A. Kosmala et al.

MS Mass spectrometry
PPFD Photosynthetic photon flux density
2-DE Two-dimensional electrophoresis
TEL50 The temperature at which 50 % of total electrolytes were released from leaf
tissues
%Vol Relative volumes of spots

23.1 Introduction

Perennial ryegrass (Lolium perenne L.) is a high quality forage and turf grass species.
However, its ability to perform in harsh winter climates is limited. Meadow fescue
(Festuca pratensis Huds.), a species closely related to L. perenne, has lower nutritive
values but is one of the most frost tolerant species within the Lolium-Festuca complex.
Frost tolerance is the main component of winter-hardiness and one of the best indi-
cators of plant ability to survive harsh winter conditions. The plant species growing
in temperate regions can acquire frost tolerance through exposure to low, non-lethal
temperatures, a phenomenon known as cold acclimation (CA). This process is asso-
ciated with altered expression patterns of a specific set of genes (Thomashow 1999;
Sandve et al. 2011).
The analysis of the cellular proteome complement during CA could help to un-
derstand the mechanisms involved in cell responses to low temperature stresses.
The tools most frequently used for visualizing protein components of protein sam-
ples and their identifications are two-dimensional electrophoresis (2-DE) and mass
spectrometry (MS), respectively.
Herein, we review the results derived from two proteomic projects, one focused
on the proteome of F. pratensis (Kosmala et al. 2009), other on the proteome of
L. perenne (Bocian et al. 2011). In each project, the comparison of leaf protein ac-
cumulation profiles during CA between plants with different levels of frost tolerance
within each species was carried out. The 2-DE technique was adopted, followed
by the identification of proteins which were accumulated differentially between the
selected plants by MS.
Here, we summarize and emphasize the similarities and differences in leaf
proteome response to CA between F. pratensis and L. perenne.

23.2 Dynamics of Frost Tolerance Level During Cold

Acclimation of F. pratensis and L. perenne

High frost tolerant (HFT) and low frost tolerant (LFT) genotypes within each species
were selected from populations of F. pratensis cv. Skra and L. perenne cv. Arka. The
way of selection was similar for both species and involved a two-step procedure.
Pre-hardening was performed during seven days at 12 ◦ C, 8/16 h photoperiod, and
23 Similarities and Differences in Leaf Proteome Response to Cold Acclimation . . . 191

Fig. 23.1 Changes in the a -4

level of frost tolerance of high -6
frost tolerant () and low -8
frost tolerant () genotypes -10
during cold acclimation of -12

T EL50
Festuca pratensis (a), Lolium -14
perenne (b). T EL50 means the -16
temperature at which 50 % of -18
total electrolytes were -20
released from leaf tissues. -22
Values of T EL50 and least -24

significant differences for 0 5 10 15 20

P = 0.05 were calculated on cold acclimation [days]

the basis of freezing tests
made in two replicates at five
b -4
freezing temperatures: −3,
-6
−6, −9, −12 and −15 ◦ C
-8

-10
T EL50

-12
-14

-16
-18

-20
0 5 10 15 20
cold acclimation [days]

200 μmol m−2 s−1 photosynthetic photon flux density (PPFD) and CA during 21 days
at 4/2 ◦ C, 10/14 h photoperiod, and 200 μmol m−2 s−1 PPFD. The 1st step of plant
selection involved the estimation of the regrowth after freezing at −8 ◦ C, −11 ◦ C
and −14 ◦ C; the 2nd step, involved the estimation of the TEL50 values (temperature
causing a 50 % electrolyte leakage) one day before CA and after 2, 8, 26 h, and 3, 5,
7, 14 and 21 days of CA. Two genotypes from each species with the extreme values
of TEL50 after 21 days of CA, a low frost tolerant (LFT) and a high frost tolerant
(HFT) plant were chosen for further proteomic research. In Fig. 23.1 we present the
physiological traits demonstrating the level of frost tolerance during CA for HFT
and LFT genotypes of both F. pratensis (a) and L. perenne (b).
The comparison of the diagrams revealed differences in the dynamics of frost
tolerance during CA of the analyzed species. Differences in the values of TEL50
between HFT and LFT genotypes of F. pratensis appeared after the 5th day, and
increased continuously until the 21st day of CA. The HFT (TEL50 = −21.4 ◦ C) and
LFT (TEL50 = −15.9 ◦ C) F. pratensis plants differed by as much as −5.5 ◦ C in
their frost tolerance after 21 days of CA. On the other hand, CA between the 7th
and 21st day only slightly (statistically not significantly) increased frost tolerance
in L. perenne. In fact, the differences in frost tolerance between the analyzed plants
appeared on the 3rd day of CA and frost tolerance reached its final levels on the
7th day of CA. The HFT (TEL50 = −15.13 ◦ C) and LFT (TEL50 = −11.42 ◦ C)
L. perenne plants differed by as much as −3.7 ◦ C in their frost tolerance after 21
days of CA.
192 A. Kosmala et al.

RuBisCO activase beta

0,3

0,2

% Vol 0,1

0
before 2h 8h 2d 3d 5d 7d 14 d 21 d
CA
time of CA
a

phosphoglycerate kinase
0,5
0,4
% Vol

0,3
0,2
0,1
0
before 2h 8h 2d 3d 5d 7d 14 d 21 d
CA

time of CA
b

RuBisCO activase beta

0,7
0,6
0,5
% Vol

0,4
0,3
0,2
0,1
0
before 2h 8h 2d 3d 5d 7d 14 d 21 d
CA
time of CA
C
Fig. 23.2 Comparisons of the accumulation profiles of RuBisCO activase beta a, c, d and phos-
phoglycerate kinase b, e during cold acclimation (CA) of high frost tolerant (- – -) and low frost
tolerant (—) genotypes of Festuca pratensis a, b and Lolium perenne c, d, e. %Vol means relative
volumes of spots. Three biological replicates were used to calculate mean %Vol. The trend curves
are shown

23.3 Leaf Proteome Response to Cold Acclimation in

F. pratensis and L. perenne

2-DE was performed on the leaf samples harvested from three clones of each selected
genotype (three biological replicates) one day before CA and after 2, 8, 26 h, and
3, 5, 7, 14 and 21 days of CA. The proteomic experiments were carried out in the
same conditions for both species. Following electrophoresis the gels were stained
23 Similarities and Differences in Leaf Proteome Response to Cold Acclimation . . . 193

RuBisCO activase beta

2,5

% Vol
1,5

0,5

0
before 2h 8h 2d 3d 5d 7d 14 d 21 d
CA
time of CA
d

phosphoglycerate kinase

0,6
% Vol

0,4

0,2

0
before 2h 8h 2d 3d 5d 7d 14 d 21 d
CA
time of CA
e
Fig. 23.2 (Continued)

with colloidal Coomassie Brilliant Blue G-250. Spot detection and image analyses
(normalization, spot matching, protein accumulation analyses, statistics) were per-
formed with Image Master 2-D Platinum software (GE Healthcare). Relative volumes
(%Vol) of spots were used to create the detailed protein accumulation profiles for
HFT and LFT genotypes of each species during CA. The protein abundance for each
spot at a corresponding time point before and during CA between HFT and LFT
plant was compared. The spots with at least 1.5-fold differences (P ≤ 0.05) in pro-
tein abundance or spots present only in one genotype (HFT or LFT) and absent in the
other genotype at a minimum of one time point of CA, were subjected to MS anal-
yses and protein identification. Fragmentation spectra and/or peptide masses were
measured using a Matrix Assisted Laser Desorption/Ionization Time of Flight MS or
electrospray ionization-MS/MS tandem MS. The data were exported to the Mascot
software for MSDB, NCBI or SwissProt database search (www.matrixscience.com).
The detail proteomic results can be found in our recent papers (Kosmala et al.
2009; Bocian et al. 2011). Herein, we indicate only the general similarities and
differences in leaf proteome response to CA, which could be noticed between
F. pratensis and L. perenne.
First of all, the time points of CA at which the differences in protein accumulation
between HFT and LFT genotypes were most frequent, were different for F. pratensis
and L. perenne, and involved the 26th hour of CA in F. pratensis and both the 5th and
194 A. Kosmala et al.

the 7th day of CA in L. perenne. Interestingly, as mentioned earlier, frost tolerance

reached its maximum levels in L. perenne genotypes exactly on the 7th day of CA
(Fig. 23.1b). Similar numbers of differentially accumulated proteins were observed:
41 for F. pratensis and 42 for L. perenne. However, the proportions of those proteins
to the total protein numbers detected on 2-D maps were different for the two species,
5.1 % (41/800) and 7.2 % (42/580) for F. pratensis and L. perenne, respectively.
Generally, different proteins were found to be differently accumulated between LFT
and HFT L. perenne, compared to LFT and HFT F. pratensis genotypes; the only
exceptions were, chloroplastic phosphoglycerate kinase and RuBisCO activase beta,
common for both species (Fig. 23.2a–23.2e). However, in both F. pratensis and
L. perenne, chloroplast proteins, including proteins directly involved in the process
of photosynthesis, were the major group selected as differentially accumulated dur-
ing CA between genotypes differing in frost tolerance levels. On the other hand,
chloroplast proteins are a major component of all the proteins found in leaves.
Both phosphoglycerate kinase and RuBisCO activase beta are involved in the
reactions of the dark phase of photosynthesis. In the case of F. pratensis, a decreased
accumulation of RuBisCO activase during the last two weeks of CA in the LFT
genotype (Fig. 23.2a) and an increased accumulation of phosphoglycerate kinase
during the whole period of CA in the HFT genotype (Fig. 23.2b) were observed. In
the case of L. perenne, two forms of RuBisCO activase were observed, one showed
higher abundance in the HFT genotype during the whole process of CA (Fig. 23.2d),
and the other decreased accumulation level after 21 days of CA in the LFT genotype
(Fig. 23.2c). Lower accumulation level of phosphoglycerate kinase was observed at
most time points of CA also in LFT plant (Fig. 23.2e). The accumulation patterns
of both enzymes observed in F. pratensis and L. perenne may suggest that HFT
genotypes are more efficient in photosynthetic carbon metabolism. The increasing
rate of this metabolism could be one of the mechanisms protecting plants against
photoinhibition of photosynthesis. It was demonstrated that plants more tolerant to
cold-induced photoinhibition are often also more frost tolerant (Rapacz et al. 2007).
However, at this stage of the research this hypothesis is speculative.

23.4 Conclusions

Further research, involving other methods of differential protein quantification (in-

stead of 2-DE), e.g. isobaric tags for relative and absolute quantification (iTRAQ),
should be applied to better dissect the proteome dynamics of F. pratensis and
L. perenne during cold conditions. Phosphoglycerate kinase and RuBisCO activase
beta were found to be differentially accumulated during CA between HFT and LFT
genotypes, both in F. pratensis and L. perenne, and this prompts further study on
these proteins.

Acknowledgments The research was carried out in the frame of the projects of Polish Ministry of
Science and Higher Education (Nos. PBZ-MNiSW-2/3/2006/21 and 2 P06A 044 30).
23 Similarities and Differences in Leaf Proteome Response to Cold Acclimation . . . 195

References

Bocian A, Kosmala A, Rapacz M, Jurczyk B, Marczak Ł, Zwierzykowski Z (2011) Differences in

leaf proteome response to cold acclimation between Lolium perenne plants with distinct levels
of frost tolerance. J Plant Physiol 168:1271–1279
Kosmala A, Bocian A, Rapacz M, Jurczyk B, Zwierzykowski Z (2009) Identification of leaf pro-
teins differentially accumulated during cold acclimation between Festuca pratensis plants with
distinct levels of frost tolerance. J Exp Bot 60:3595–3609
Rapacz M, Gasior D, Kosmala A, Zwierzykowski Z, Humphreys MW (2007) The role of photosyn-
thetic apparatus in cold acclimation of Lolium multiflorum. Characteristics of novel genotypes
low-sensitive to PSII over-reduction. Acta Physiol Plant 29:309–316
Sandve SR, Kosmala A, Rudi H, Fjellheim S, Rapacz M, Yamada T, Rognli OA (2011) Molecular
mechanisms underlying frost tolerance in perennial grasses adapted to cold climates. Plant Sci
180:69–77
Thomashow MF (1999) Plant cold acclimation: freezing tolerance genes and regulatory mecha-
nisms. Annu Rev Plant Physiol Plant Mol Biol 50:571–599
Chapter 24
Multi-population QTL Detection for Flowering
Time, Stem Elongation and Quality Traits
in Medicago truncatula

L. del Carmen Lagunes Espinoza, T. Huguet and B. Julier

Abstract Medicago truncatula, as a model species, is useful to study the genetic

control of flowering time, stem elongation and quality (protein content and di-
gestibility) in legume crops. These traits were measured in four mapping populations
originating from five parental lines. Single and multi-population quantitative trait
locus (QTL) detections were carried out. A large variation was observed within
populations and transgressive segregations were noted. On average, genotypes with
long primary branches and a high branch elongation rate showed an early flowering
date and a long main stem and had lower digestibility and protein content. Ninety
QTLs for morphogenesis and 27 QTL for quality were identified and localized over
all eight chromosomes. All QTLs for quality traits were located in genomics regions
showing QTL for morphogenesis, suggesting common regulation and/or colocation
of genes responsible for their variation. Using genomic resources publicly available,
a list of candidate genes that could control variation for aerial morphogenesis and
quality was established.

Keywords Morphogenesis · Heritability · Model species · Legume · Gene

B. Julier () · L. C. Lagunes Espinoza

Unité de Recherche Pluridisciplinaire Prairies et Plantes Fourragères, INRA,
UR 4, Le Chêne, RD 150, BP 80006, 86600 Lusignan, France
e-mail: [email protected]
L. C. Lagunes Espinoza
Colegio de Postgraduados (COLPOS), Campus Tabasco, Periférico Carlos A. Molina s/n,
86500 H. Cárdenas, Tabasco, Mexico
T. Huguet
UMR5245 EcoLab (Laboratoire Ecologie Fonctionnelle et Environnement), ENSAT,
Université de Toulouse, INP, UPS, 18 chemin de Borderouge,
31326 Castanet-Tolosan, France
EcoLab, CNRS, 31326 Castanet-Tolosan, France

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 197

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_24, © Springer Science+Business Media Dordrecht 2013
198 L. C. Lagunes Espinoza et al.

24.1 Introduction

Vegetative growth, flowering date and forage quality are major breeding traits in
forage legumes. They are known to have a quantitative inheritance so the analysis
of their genetic determinism is quite complex in alfalfa (Medicago sativa) or clovers
(Trifolium sp.). With the assumption that genes or genomic regions responsible for
trait variation in a model species will also contribute to explain genetic variation in
a related crop species, we have studied genetic determinism of breeding traits in the
model species Medicago truncatula. In this diploid and autogamous species, recom-
binant inbred lines (RILs) mapping populations, markers and genomic sequences
are available to detect QTLs and to identify candidate genes that may explain trait
variation.
Four mapping populations were studied for stem elongation, flowering date,
energy value and protein content. QTL were detected in each cross and a
multi-population QTL detection was carried out.

24.2 Material and Methods

Four RIL populations of M. truncatula involving a total of five parental lines were
studied: LR4 (Julier et al. 2007), LR5 (Ameline-Torregrosa et al. 2008), LR1 (Pierre
et al. 2008) and LR6 (Fig. 24.1). Each RIL population was composed of 173 to
233 lines. The populations were analysed for aerial morphogenetic traits and quality
traits in experiments conducted in 2002, 2003, 2004 (for LR4), spring and autumn
2005 (for LR1 and LR5), 2007 and 2008 (for LR6). Plants of the RILs populations
and the five parental lines were grown in individual pots in a greenhouse at INRA
Lusignan (France). All lines of LR4 in 2002 and 2003 were repeated three times but,
in order to maximise the number of RILs under study, there was no repetition for
LR4 in 2004, LR1, LR5, LR6, except for 15 RILs randomly taken and five parental
lines. These 20 lines were repeated three times. Each repetition was composed of one
plant. The flowering time (FT) was recorded when a plant had one open flower on
a primary branch, and transformed in degree-days with a temperature basis of 0 ◦ C.
The length of the first two emerging primary branches (LPB) was measured twice
a week during the growth period. When all the plants of the trial had flowered, the
length of the main stem (LMS) was measured. The plants were harvested, dried and
grounded. Energy value, evaluated through dry matter enzymatic digestibility (DIG)
and protein content (PC) were measured using NIRS equations. The curve of branch
elongation as a function of sums of degree-days showed a linear phase. For each
RIL, branch elongation rate (BER) was calculated as the slope of this linear phase.
Framework maps of each cross included about 60 SSR markers and covered about
600 cM. QTL mapping was performed using QTL Cartographer (Basten et al. 1994;
Basten et al. 2002) with the composite interval mapping (CIM) procedure. The
threshold for adding a QTL, determined at 5 % risk by a permutation test method
(1000 replications), was set to 11.33 (LOD ≥ 2.46). The limits of the confidence
24 Multi-population QTL Detection for Flowering Time, Stem Elongation . . . 199

Fig. 24.1 Crossing design LR5

LR6
between five M. truncatula A20 Jemalong-6 F83005.5
parents to obtain the four
mapping populations LR1,
LR4, LR5 and LR6. The LR4
arrows indicate the direction LR1
of the cross, from the male to DZA315.26 DZA45.5
the female parent

interval of QTL position were estimated at the positions where the LOD value drop-
off was equal to one (Lander and Botstein 1989).
To better estimate the position of the QTLs that were common to different pop-
ulations and their effects, a multi-population QTL analysis was carried out with the
MCQTL software package (Jourjon et al. 2005). First a consensus genetic map was
built from the genetic maps of LR1, LR4, LR5 and LR6 by using BioMercator soft-
ware (Arcade et al. 2004). Each QTL position and its support interval were projected
on the consensus map. In a second step, adjusted means of aerial morphogenetic
and quality traits per RIL were calculated. Adjusted means (Blanc et al. 2006) and
the consensus map were used to launch the multi-population QTL analysis with the
“connected” option.
The version 3.5 of M. truncatula genome sequence (http://www.medicagoha-
pmap.org/?genome) was used to detect the BACs that belong to the support interval
of the major QTLs. Among the genes annotated on the BACs, we searched out those
involved in plant growth or quality.

24.3 Results and Discussion

Significant variation among lines was observed in each population for all traits. In
most cases, transgressions towards lower and higher values than those of parents
were observed.
Positive and significant correlations were observed between BER and LPB in all
seasons; negative and significant correlations were observed between FT and LMS
and between FT and LPB except for LR5 in autumn 2005. These correlations indicate
that on average, genotypes with long primary branches and a high branch elongation
rate showed an early flowering date and a long main stem. Correlations between
digestibility and protein content were highly significant. Digestibility and protein
content were negatively correlated with stem elongation.
In the QTL detection in each population, a total of 90 QTLs for BER, LMS,
FT, LPB were identified and localized over all eight chromosomes, with a greater
concentration on chromosomes 1, 2, 7 and 8. The bottom of chromosome 7 carried
QTLs for FT, BER, LMS, and LPB on populations LR1, LR4 and LR5 with high R2
values. No QTL was found in this region in LR6. The strongest QTLs for FT in LR6
were located on chromosome 6 at the positions 15.4 –21.1 cM, explaining 15–19.4 %
of the variation. For quality traits, 19 QTLs were detected on chromosomes 1, 4, 5,
200 L. C. Lagunes Espinoza et al.

6, 7 and 8 for digestibility and 8 QTL were detected on chromosomes 1, 2, 3, 7 and

8 for protein content. In fact, all QTL for quality traits were located in regions where
QTLs for morphogenetic traits were also present. Either morphogenesis explained
most of the variation for quality, or genes involved in both types of traits co-located
in the genome.
The multipopulation QTL analysis produced four QTLs (herein called mcQTLs)
for LPB, two for BER, six for LMS, and two for FT, DIG and PC (Table 24.1), each
explaining between 3.0 and 22.9 % of the variation. For LPB and FT, the mcQTLs
on chromosome 7 had higher effects than the others mcQTLs, explaining 10.2 and
22.9 % of variation, respectively. For LPB, the mcQTL on chromosome 1 explained
almost 12 % of the variation. For BER, DIG and PC, the two mcQTLs had similar R2 .
This part of the variation explained by the mcQTLs was often lower than that of the
initial QTLs because not all populations × seasons produced these QTLs. QTLs that
were specific to one single cross such as the QTL for FT on chromosome 6 in LR6
were not revealed with this multi-population analysis, as if its effect was “diluted”
in the whole design. In most cases, the alleles of each parent had either positive or
negative effects, explaining the transgressions in the progenies. However Jemalong6
alleles except one induced long branches, high BER, low digestibility and early FT
and DZA315.16 alleles induced short branches, low BER, early FT and high PC.
All the mcQTLs were found in a position where at least one QTL was found in a
population. But some QTLs that seemed to co-localise in several populations (FT
on chromosome 1, BER on chromosome 1, LPB on chromosomes 4 and 8) were
not recovered in the multi-population analysis. The same positions for the mcQTLs
were observed for LPB and LMS on chromosome 1, for LPB, LMS and BER on
chromosome 2, for FT, LPB and LMS on chromosome 7, and for FT and LMS on
chromosome 8, suggesting partially common genetic regulation.
The in silico analysis of BACs included in the confidence interval of major mc-
QTLs for FT on chromosome 7 and LMS on chromosome 1 revealed three and ten
candidate genes related to these traits, respectively. In the confidence interval of
the mcQTL for FT on chromosome 7, genes related to floral induction were iden-
tified: zinc finger protein CONSTANS-like, FLOWERING LOCUS T and PEBP
genes. CONSTANS-like gene proved to contribute to variation for stem length and
flowering date in alfalfa (Herrmann et al. 2010). At the bottom of chromosome 1,
where a mcQTL for LMS was identified close to QTLs for LPB, eight genes were
detected (COP1, CLAVATA1, SBP, ARF/SAR, Auxin-binding protein, EMBRYO-
GENIC FLOWER 2, Aux/IAA protein and NAM) related to shoot and branching
development through hormone response and signalling. In the genomic region re-
vealed on chromosome 6 for FT, a gene FAR1 involved in the far-red responses
controlled by phytochrome A, and a member of SKP1 gene family, homologue of
ASK1 in Arabidopsis are annotated. In the QTL regions of quality traits, genes re-
lated to cell wall polysaccharides and protein biosynthesis, lignin pathway or plant
development metabolism were revealed on chromosomes 1, 3, 7 and 8. It is estab-
lished that stem growth contributes to decrease both stem digestibility because of an
increase in lignified tissues and plant protein content through reduction of the pro-
portion of leaves in the biomass. The presence of genes involved in quality pathways
Table 24.1 Location of QTLs for aerial morphogenesis and quality traits, proportion of explained variation (R2 ) and effects of parents in a multi-
population QTL analysis
Trait Chromosome Position Confidence R2 Effect of parents
interval
Jemalong6 DZA315.26 DZA45.5 A20 F83005.5
Length of primary branches 1 65.3 64–70 5.5 −1.72 −0.32 3.19 −1.67 0.51
(LPB), (cm)
2 52.3 49–58 6.0 2.40 −0.61 −1.31 2.16 −2.65
3 52.6 51–55 4.3 2.06 −1.27 −1.69 −1.36 2.27
7 56.5 54–60 10.2 3.29 −2.60 −4.42 2.17 1.56
Branch elongation rate (BER), 2 52.3 46–59 3.0 0.8 −1.9 2.6 0.2 −5.7
(10−3 cm.day−1 )
4 72.7 68–77 3.5 0.8 2.8 −1.2 −0.6 −1.8
Length of main stem* (LMS), 1 62.5 55–64 11.7 −3.18 −0.49 0.49 3.68 −0.50
(cm)
2 57.3 48–58 5.0 −0.59 −0.55 0.55 3.59 −3.00
7 22.9 17–26 8.5 −2.55 1.17 −1.17 4.25 −1.69
7 50 41–54 4.9 2.13 1.19 −1.19 −1.49 −0.64
8 5 1–9 5.2 1.16 1.09 −1.09 2.89 −4.06
8 31.5 28–37 7.8 2.37 −2.51 −1.50 −0.87 2.51
Flowering time (FT), (◦ C.D) 7 51.5 50–55 22.9 −76.6 57.7 149.7 −68.8 −62.0
8 5 0–9 3.9 −31.5 20.9 33.1 −36.2 13.7
Digestibility (DIG), (%) 7 56.5 53.9–59.3 10.9 −0.77 0.35 0.88 −0.93 0.47
8 5.0 3.2–20.2 7.6 −0.42 −0.23 0.41 −0.68 0.92
24 Multi-population QTL Detection for Flowering Time, Stem Elongation . . .

Protein content (PC), (%) 1 70.0 59.3–70.0 4.1 0.24 0.08 −0.27 −0.17 0.12
7 57.4 54.6–63.6 4.5 −0.32 0.21 0.40 −0.33 0.04
*Analysis conducted on three populations (DZA315.26 × DZA45.5, Jemalong6 × F83005.5 and Jemalong6 × A20)
201
202 L. C. Lagunes Espinoza et al.

in QTL regions of quality traits is consistent with the hypothesis that an intrinsic
genetic regulation of quality exists besides the physiological relationship between
morphogenesis and quality. The genomic regions or the genes detected in the model
species are candidates to explain phenotypic variation in cultivated legume species.
They could be the targets of new studies aiming at checking their role in a specific
crop and identifying the alleles conferring positive effects. These alleles could then
be used in marker-assisted selection.

Acknowledgments We thank Region Poitou-Charentes (France) for the post-doctoral grant for sup-
porting research stay of L.C. Lagunes Espinoza and Campus Tabasco of Colegio de Postgraduados
(Mexico) for permission to perform the research stay.

References

Ameline-Torregrosa C, Cazaux M, Danesh D, Chardon F, Cannon SB, Esquerre-Tugaye MT, Dumas

B, Young ND, Samac DA, Huguet T, Jacquet C (2008) Genetic dissection of resistance to
anthracnose and powdery mildew in Medicago truncatula. Mol Plant Microbe Interact 21:61–69
Arcade A, Labourdette A, Falque M, Mangin B, Chardon F, Charcosset A, Joets J (2004) BioMer-
cator: integrating genetic maps and QTL towards discovery of candidate genes. Bioinformatics
20:2324–2326
Basten CJ, Weir BS, Zeng ZB (1994) Zmap-a QTL cartographer. In: Proceedings of the 5th World
Congress on Genetics Applied to Livestock Production: computing strategies and software, Vol
22. (Eds C Smith, JS Gavora, B Benkel, J Chesnais, W Fairfull, JP Gibson, BW Kennedy, and
EB Burnside) pp 65–66. (Organizing Committee, 5th World Congress on Genetics Applied to
Livestock Production: Guelph, Ontario, Canada)
Basten CJ, Weir BS, Zeng ZB (2002) QTL cartographer, Version 1.16. Raleigh, NC, North Carolina
State University, Department of Statistics
Blanc G, Charcosset A, Mangin B, Gallais A, Moreau L (2006) Connected populations for detect-
ing quantitative trait loci and testing for epistasis: an application in maize. Theor Appl Genet
113:206–224
Herrmann D, Barre P, Santoni S, Julier B (2010) Association of a CONSTANS-LIKE gene to
flowering and height in autotetraploid alfalfa. Theor Appl Genet 121:865–876
Jourjon MF, Jasson S, Marcel J, Ngom B, Mangin B (2005) MCQTL: multi-allelic QTL mapping
in multi-cross design. Bioinformatics 21:128–130
Julier B, Huguet T, Chardon F, Ayadi R, Pierre JB, Prosperi JM, Barre P, Huyghe C (2007) Identifi-
cation of quantitative trait loci influencing aerial morphogenesis in the model legume Medicago
truncatula. Theor Appl Genet 114:1391–1406
Lander ES, Botstein D (1989) Mapping mendelian factors underlying quantitative traits using RFLP
linkage maps. Genetics 121:185–199
Pierre JB, Huguet T, Barre P, Huyghe C, Julier B (2008) Detection of QTLs for flowering date
in three mapping populations of the model legume species Medicago truncatula. Theor Appl
Genet 117:609–620
Chapter 25
Role of the RCT1 Gene in Anthracnose
Resistance in Alfalfa

B. Julier, I. Meusnier, L. Alaux, S. Flajoulot, P. Barre and J. Gouzy

Abstract Anthracnose, caused by Colletotrichum trifolii, is a severe disease of

alfalfa (Medicago sativa). The RCT1 gene, isolated from the model legume M. trun-
catula, is a candidate gene to explain genetic variation for anthracnose resistance
in alfalfa. A bulk segregant analysis was carried out to test this hypothesis: from
each of eight alfalfa varieties, 15 resistant (R) plants and 15 susceptible (S) plants
were selected and DNA was extracted. The whole gene including the upstream and
downstream regions (a total of 14 kb) was amplified by PCR for each individual and
the R and S plants were pooled for each variety. Sequencing was carried out using the
next generation sequencer 454 (Roche). The sequence reads, that averaged 295 bp,
were assembled to produce consensus sequences that can be considered as alleles.
Considering the five exons of the gene, five regions contained clear deletion/insertion
polymorphism but these polymorphisms were present in both the R and S pools. In-
dividual genotyping for these indels indicated that different alleles were present but
no specific allele was associated with the phenotype. These polymorphic regions in
RCT1 seemed not to explain the variation of anthracnose resistance in alfalfa. How-
ever, the presence of one rare allele inducing a lack of function was associated with
the resistance. A divergent selection for this allele would test its interest in breeding
programs.

Keywords Medicago sativa · Medicago truncatula · Colletotrichum trifolii ·

Candidate gene · Bulk segregant analysis · Next-generation sequencing

B. Julier () · I. Meusnier · L. Alaux · P. Barre

Unité de Recherche Pluridisciplinaire Prairies et Plantes Fourragères, INRA, UR 4,
Le Chêne RD 150, BP 80006, 86600 Lusignan, France
e-mail: [email protected]
S. Flajoulot
Jouffray-Drillaud, La Litière, 86600 Saint-Sauvant, France
J. Gouzy
Laboratoire des Interactions Plantes-Microorganismes, INRA-CNRS,
31326 Castanet-Tolosan, France
I. Meusnier
Centre de Biologie pour la Gestion des Populations, Campus international de Baillarguet,
INRA, UMR, CS 30016, 34988 Montferrier sur Lez Cedex, France

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 203

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_25, © Springer Science+Business Media Dordrecht 2013
204 B. Julier et al.

25.1 Introduction

Anthracnose, caused by Colletotrichum trifolii, determines severe yield losses in

alfalfa (Medicago sativa) forage production (Barnes et al. 1969; Raynal et al. 1989).
This fungus attacks stems and crowns and kills the plants, contributing to stand
decline. Three races have been described (Ariss and Vandemark 2007). Alfalfa resis-
tance to anthracnose is conferred by two independent dominant genes, An1 and An2
(Elgin and Ostazeski 1985). Resistance to race 1 and likely race 4 is conferred by
An1 (Mackie et al. 2003), and to race 2 by An2. Breeding has focused on the selection
of resistant varieties, using tests in controlled conditions based on artificial inocula-
tions (Gondran 1984; O’Neill 1991). Despite an oligogenic inheritance, quantitative
methods are used to select resistant plants, specifically because the autotetraploidy
of alfalfa makes genetic analyses complex. QTL studies were conducted to localise
the resistance genes on the genome, both in alfalfa (Mackie et al. 2007) and in the
model plant M. truncatula (Ameline-Torregrosa et al. 2008; Yang et al. 2007) in
which susceptible and resistant plants were identified. Partially consistent results
were obtained with QTLs found on chromosome 4 but also a QTL with strong effect
on chromosome 8 in alfalfa.
In M. truncatula, diploidy and the available genomics tools give the opportunity
to conduct fine mapping strategies aiming at discovering the gene responsible of a
QTL. Fine mapping analysis in a large population resulted in the identification of
the RCT1 gene, a TIR-NBS-LRR resistance gene (Yang et al. 2008). The function
of this gene was not validated in M. truncatula because of the difficulty to transform
some genotypes, but it was introduced in susceptible alfalfa genotypes, and conferred
resistance to anthracnose (Yang et al. 2008). Transgenic varieties are not generally
accepted by the European society, and transgene escapes risks are high in alfalfa.
Thus, the use of RCT1 to breed resistant transgenic cultivars is not possible for the
European market.
Given the results of Yang et al. (2008), we hypothesized that the alfalfa RCT1
ortholog could be responsible for anthracnose resistance in this species. The objective
of this study was to test the relationship of RCT1 sequence polymorphism with alfalfa
resistance to anthracnose. The polymorphisms related to resistance could be used to
develop markers useful to breeders to select for plants carrying high doses of resistant
alleles. The underlying hypothesis is that a gene identified to explain trait variation
in the model species would explain the variation for the same trait in alfalfa (Julier
and Meusnier 2010).

25.2 Material and Methods

The RCT1 gene of M. truncatula spans more than 3.5 kb. Upstream and downstream
regions as well as intronic sequences showed polymorphisms that could be related to
resistance expression (Yang et al. 2008). As a consequence, we sequenced the whole
gene, including non-coding regions. Because of the low sequence conservation in
25 Role of the RCT1 Gene in Anthracnose Resistance in Alfalfa 205

non-coding sequences but the high sequence conservation in coding regions between
alfalfa and M. truncatula, we designed primers in the genes flanking RCT1 in M.
truncatula.
The heterozygosity of alfalfa implied the need to sequence the four alleles of the
gene. Direct sequencing of small gene portions is feasible (Pierre et al. 2011) but
time-consuming and expensive. Next Generation Sequencing technologies offer the
possibility to sequence a large set of genomic sequences at a reasonable cost, and
have been adopted to get the complete sequence of RCT1, with its four alleles, in a
set of genotypes.
Eight varieties with different resistance level were chosen and anthracnose tests
were carried out on 100 seedlings of each variety with the strain C86-2 of C. trifolii.
For each variety, 15 resistant (R) plants and 15 susceptible (S) plants were selected,
and the genomic DNA was extracted. Two primer pairs were defined to amplify
the whole gene and the upstream and downstream regions (a total of about 14 kb).
A bulk segregant analysis (BSA) approach was used with the objective to identify
gene polymorphisms associated with resistance (Michelmore et al. 1991). The 454
(Roche) sequencing technology was used. The PCR products of the 15 R and 15
S plants were pooled for each variety. DNA libraries were tagged for each variety
and the libraries were then pooled. Consequently, up to 60 different alleles could be
present in each pool. The full RCT1 gene was cloned and Sanger-sequenced in one
alfalfa plant to produce a reference sequence.
The raw sequences were assembled based on the reference sequence to produce
consensus sequences that could be considered as alleles. As a very large sequence
polymorphism was present, the rest of the analysis was restricted to the five exons
of the gene. These consensus sequences were translated into proteins and R and
S sequences were compared. In the regions showing deletions, primer pairs were
designed to genotype the individuals of the R and S pools. The relationship between
genotype and phenotype was analysed.

25.3 Results and Discussion

Resistance is dominant over susceptibility, therefore susceptible genotypes should

only carry recessive alleles (r), their genotype is rrrr. By contrast, the resistant geno-
types carry at least one allele for resistance (R) and may be RRRR, RRRr, RRrr
or Rrrr. This notation is schematic because we do not know how many alleles of
susceptibility or resistance exist. If susceptibility corresponds to inactive allele, sev-
eral alleles of susceptibility may be observed due to accumulation of mutations. To
summarize, the susceptible (S) pools should only contain alleles for susceptibility,
except if a resistant individual has been misclassified as susceptible. The resistant (R)
pools should contain both alleles for resistance and susceptibility, their proportion
depending on the frequency of resistance alleles in the variety (Julier et al. 2004).
A last point could modify the analysis: the alleles of resistance or susceptibility
could be different in the varieties. These varieties were bred by different breeders
206 B. Julier et al.

Table 25.1 Evaluation of the

level of resistance of the eight Variety Resistant plants (%)
alfalfa cultivars to Galaxie 20.0
anthracnose, expressed as the Bar 2 27.0
percentage of resistant plants Prunelle 28.6
Kali 41.8
Canelle 42.0
Symphonie 44.9
Bar 1 47.4
Marshall 83.9

Fig. 25.1 Scheme of the RCT1 genomic region. Deletions indicated in red induce a reading frame
shift; those in green do not produce a reading frame shift. The arrows indicate the position of the
primers

who use partly different genetic backgrounds. However, the low level of genetic dif-
ferentiation among European varieties is favorable to the hypothesis that the genetic
determinant of resistance is the same in all varieties.
Resistance Test The eight varieties were evaluated for their resistance. As expected,
Marshall was the most resistant variety. Kali was chosen for its low level of resis-
tance, but had unexpected 42 % resistant plants. The other six varieties, known as of
intermediate resistance level, had among 20 and 47 % resistant plants. In all varieties,
15 resistant plants and 15 susceptible plants were chosen, except in Marshal in which
only seven susceptible plants were identified (Table 25.1).
RCT1 Sequencing The whole gene with the upstream and downstream regions was
cloned and sequenced for a single individual with Sanger method. It comprised a
14.5 kb sequence, with the same five exons structure as in M. truncatula. Within
this individual, polymorphism was identified, with some long insertions/deletions.
A reference sequence was built and exons were identified.
With the NGS method, the same genomic region was sequenced for 240 individu-
als, pooled as described above (potentially 960 alleles). A total of 5,70,819 sequences
was obtained with a mean size of 295 bp. Raw sequences were assembled based on
the reference sequence to produce consensus sequences that can be considered as
alleles. As a very large sequence polymorphism was present, the rest of the analysis
was restricted to the five exons of the gene. Five clear deletion/insertion polymor-
phisms were found: one in the ATG region of exon 1, two in exon 3 and two in exon
5 (Fig. 25.1).
25 Role of the RCT1 Gene in Anthracnose Resistance in Alfalfa 207

Table 25.2 Number of alleles

of 102, 241 and 244 bp in the Phenotype Number of Allele Allele Allele Missing
polymorphic region of exon 1 individuals 102 241 244 data
R 130 28 405 61 6
S 126 13 409 54 7

In all pools, both deleted and non-deleted alleles were found, the misclassification
of a single R individual in a S pool could induce this situation. However, this case
should be rare because only the plants showing clear phenotypes were used to build
the pools.
Polymorphism at the Individual Level For the deletions on exon 1, beginning
and end of exon 3, primers were defined to amplify a portion of less than 300 bp
(Fig. 25.1). As the two deletions of exon 5 were close to each other, a single primer
pair was used for both. Polymorphism was evidenced for each region with 2–5
different alleles. One, and the same, allele per region was highly frequent, whatever
the variety or the R or S pool. The difference in allele frequency between R and S
plants was never significant for the frequent alleles. For the deletion of exon 1 that
was the most promising because of ATG deletion, the deleted allele (102 bp) had a too
low frequency to be associated with resistance in this population (Table 25.2) even if
the χ 2 test reached P = 0.06. Surprisingly, the deleted allele that probably induced
a lack of function was associated with the resistance, a case already described in the
literature in other pathosystems. A divergent selection for the presence or absence
of this allele would test if the lack of function of RCT1 confers the anthracnose
resistance in alfalfa.

25.4 Conclusion

The polymorphic regions of RCT1 exons identified through the sequencing of the
whole gene with a bulk-segregant analysis did not explain the variation of anthracnose
resistance in the alfalfa varieties studied, even if a deleted rare allele was weakly
associated with the resistance. Non coding sequences of this gene or even other
genes may control the trait. In this specific case, the gene detected in M. truncatula
for anthracnose resistance may not be the one responsible for the same trait in alfalfa.
Indeed, a QTL study indicated that the position of a QTL for anthracnose resistance
did not correspond to the position of RCT1 (Cazaux 2008). A previous study on a
CONSTANS-like gene involved in flowering date and stem elongation (Herrmann
et al. 2010) showed that translational genetics between M. truncatula and alfalfa may
be efficient. However, this study shows the power of next-generation sequencing
technologies for genome analysis of a polyploid and heterozygous species.

Acknowledgments We thank the French Ministry of Agriculture for financial support (Contrat de
Branches 2008–11) and ACVF for scientific and technical contribution. We thank J. Lluch and
208 B. Julier et al.

O. Bouchez from “Plateforme Génomique” of Toulouse (France) for 454 sequencing and S.
Fouilloux from BioGeves at Surgères (France) for DNA quantification.

References

Ameline-Torregrosa C, Cazaux M, Danesh D, Chardon F, Cannon SB, Esquerre-Tugaye MT, Dumas

B, Young ND, Samac DA, Huguet T, Jacquet C (2008) Genetic dissection of resistance to
anthracnose and powdery mildew in Medicago truncatula. MPMI 21:61–69
Ariss JJ,Vandemark GJ (2007)Assessment of genetic diversity among nondormant and semidormant
alfalfa populations using sequence-related amplified polymorphisms. Crop Sci 47:2274–2284
Barnes DK, Ostazeski SA, Schillinger JA, Hanson CH (1969) Effect of anthracnose (Colletotrichum
trifolii) infection on yield, stand, and vigor of alfalfa. Crop Sci 9:344–346
Cazaux M (2008) Etude de la résistance de la légumineuse modèle Medicago truncatula à
Colletotrichum trifolii, agent de l’anthracnose. PhD Thesis Université de Toulouse
Elgin JH Jr, Ostazeski SA (1985) Inheritance of resistance to race 1 and race 2 anthracnose in Arc
and Saranac AR alfalfa. Crop Sci 25:861–865
Gondran J (1984) La verticilliose de la luzerne: détermination de l’agent causal, biologie du parasite,
répartition géographique, dégâts et méthodes de lutte. Thèse Université U.E.R. Sciences de
Poitiers
Herrmann D, Barre P, Santoni S, Julier B (2010) Association of a CONSTANS-LIKE gene to
flowering and height in autotetraploid alfalfa. Theor Appl Genet 121:865–876
Julier B, Meusnier I (2010) Alfalfa breeding benefits from genomics of Medicago truncatula. Ratar
Povrt/Field Veg Crop Res 47:395–402
Julier B, Bournoville R, Landré B, Ecalle C, Carré S (2004) Genetic analysis of lucerne (Medicago
sativa L.) seedling resistance to pea aphid (Acyrtosiphon pisum Harris). Euphytica 138:133–139
Mackie JM, Musial JM, O’Neill NR, Irwin JAG (2003) Pathogenic specialisation within Col-
letotrichum trifolii inAustralia, and lucerne cultivar reactions to all knownAustralian pathotypes.
Aust J Agric Res 54:829–836
Mackie JM, Musial JM, Armour DJ, Phan HTT, Ellwood SE, Aitken KS, Irwin JAG (2007) Iden-
tification of QTL for reaction to three races of Colletotrichum trifolii and further analysis of
inheritance of resistance in autotetraploid lucerne. Theor Appl Genet 114:1417–1426
Michelmore RW, Paran I, Kesseli RV (1991) Identification of markers linked to disease-resistance
genes by bulked segregant analysis: a rapid method to detect markers in specific genomic regions
by using segregating populations. Proc Natl Acad Sci 88:9828–9832
O’Neill N (1991) Anthracnose resistance. In: CC Fox et al. (ed) Standard tests to characterise alfalfa
cultivars, D1. North American alfalfa improvement conference
Pierre JB, Bogard M, Herrmann D, Huyghe C, Julier B (2011) A CONSTANS-like gene candidate
that could explain most of the genetic variation for flowering date in Medicago truncatula. Mol
Breed 28:25–35
Raynal G, Gondran J, Bournoville R, Courtillot M (1989) Ennemis et maladies des prairies. INRA,
Paris
Yang S, Gao M, Deshpande S, Lin S, Roe BA, Zhu H (2007) Genetic and physical localization of
an anthracnose resistance gene in Medicago truncatula. Theor Appl Genet 116:45–52
Yang S, Gao M, Xu C, Gao J, Deshpande S, Lin S, Roe BA, Zhu H (2008) Alfalfa benefits from
Medicago truncatula: the RCT1 gene from M. truncatula confers broad-spectrum resistance to
anthracnose in alfalfa. Proc Natl Acad Sci 105:12164–12169
Chapter 26
The EUCARPIA Multi-site Rust
Evaluation—Results 2010

F. X. Schubiger, J. Baert, T. Ball, B. Cagas, E. Czembor, U. Feuerstein, A. Gay,

S. Hartmann, H. Jakesova, M. Klima, B. Krautzer, H. Leenheer, C. Persson,
W. Pietraszek, L. Poinsard, U. K. Posselt, Y. Quitté, M. Romani, L. Russi,
S. Schulze, M. C. Tardin, M. Van Nes, E. Willner, L. Wolters and B. Boller

Abstract The EUCARPIA multi-site rust evaluation was repeated in 2010 for the
fourth time. The trials were sown at 24 sites in 11 countries in Europe. The 20 Ital-
ian/Hybrid (Lolium multiflorum and L. boucheanum, respectively) and 34 perennial
ryegrass (L. perenne) cultivars were sown in separate trials. The cultivars were grown
as rows in a completely randomized block design with four replicates. Rust incidence
was scored in the year of seeding by each of the participants. The method used to test

F. X. Schubiger () · B. Boller

Agroscope Reckenholz-Tänikon, Research Station ART, Reckenholzstrasse 191,
8046 Zürich, Switzerland
e-mail: [email protected]
J. Baert
ILVO-Eenheid Plant, 9090 Melle, Belgium
T. Ball
Euro Grass Breeding GmbH, Wardington, Banbury, OX17 1FE, Oxon, UK
B. Cagas
OSEVA PRO Ltd. Grassland Research Station Roznov Zubri, 75654 Zubri, Czech Republic
E. Czembor
IHAR Radzikow, 05-870 Blonie, Poland
U. Feuerstein
Euro Grass Breeding GmbH, 27330 Asendorf, Germany
A. Gay
Saatzucht Steinach, 94377 Steinach, Germany
S. Hartmann
Institut für Pflanzenbau u. Pflanzenzüchtung der Bayerischen Landesanstalt für Landwirtschaft,
85354 Freising, Germany
H. Jakesova
Hladke Zivotice 100, 74247 Hladke Zivotice, Czech Republic
M. Klima
Plant Breeding Station, 74247 Hladke Zivotice, Czech Republic
B. Krautzer
Bundesanstalt für alpenländische Landwirtschaft Gumpenstein, 8952 Irdning, Austria

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 209

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_26, © Springer Science+Business Media Dordrecht 2013
210 F. X. Schubiger et al.

cultivars for rust resistance was based on an estimation of the percentage of leaf area
affected. Crown rust (Puccinia coronata f. sp. lolii) was again the most frequently
observed rust on both ryegrass species. Stem rust (P. graminis f.sp. graminicola) was
reported at only one site for Italian and at four sites for perennial ryegrass. Variation
in resistance to crown rust among cultivars was significant at 17 sites for Italian and
at 16 sites for perennial ryegrass. The cultivars ‘Tarandus’, ‘Gosia’, ‘Caballo’ and
‘Domino’ showed the highest level of resistance of all the Italian ryegrass cultivars
tested. ‘Gwendal’ and ‘Bocage’ were the most crown rust resistant perennial ryegrass
cultivars. The ranking of the mean crown rust susceptibility of the cultivars was highly
correlated with the corresponding ranking of cultivars in the 2001, 2004 and 2007
trials, respectively. This was true for both perennial and Italian ryegrass. Coefficients
of rank order correlations of average cultivar disease scores between the years were
greater than rs = 0.97 (p < 0.05) for Italian and rs = 0.86 (p < 0.05) for perennial
ryegrass, respectively. Therefore, there is no evidence that crown rust resistance

H. Leenheer
DLF Trifolium Netherlands B.V., 4727 Moerstraten, Netherlands
C. Persson
Weibull AB, 26881 Svalöv, Sweden
W. Pietraszek
RAGT 2n Ferme expérimentale La Boucaudière, 35460 Montours, St. Brice en Cogles, France
L. Poinsard
S.A. Carneau, 59310 Orchies, France
U. K. Posselt
Universität Hohenheim, 70593 Stuttgart, Germany
Y. Quitté
Euro Grass Breeding GmbH, 49350 La planche Les Rosiers sur loire, France
M. Romani
CRA-FLC, 26900 Lodi, Italy
L. Russi
Universita’ degli Studi, 06121 Perugia, Italy
S. Schulze
Saatzucht Steinach Bornhof, 17219 Bocksee, Germany
M. C. Tardin
RAGT 2n Centre de recherche de Druelle, 12510 Druelle, France
M. Van Nes
Duitsekampweg 60, 6874 Wolfheze, Netherlands
E. Willner
IPK Genbank, Teilsammlung Nord, 23999 Malchow/Poel, Germany
L. Wolters
Euro Grass Breeding GmbH, Zelder 1, 6599 Ven-Zelderheide, Netherlands
26 The EUCARPIA Multi-site Rust Evaluation—Results 2010 211

of an individual cultivar was overcome by the rust pathogen over the 9 years of
experimentation.

26.1 Introduction

In 2000, the EUCARPIA Fodder crops and Amenity grasses section initiated a mul-
tisite rust evaluation trial. The aim of the trial was to determine the susceptibility
of different Italian (Lolium multiflorum), Hybrid (L. boucheanum) and perennial
(L. perenne) ryegrass cultivars to the most important rust species throughout Europe.
In repeating this trial every 3rd year, starting in 2001, we wanted to assess if rust
resistance of individual cultivars breaks down and is overcome by the rust pathogen
over the course of time. The results of the 2001, 2004 and 2007 trials were published
in the proceedings of the “Eucarpia Fodder crops and Amenity grasses” meetings
in Braunschweig in 2002, Perugia in 2006 and La Rochelle in 2009, respectively
(Boller et al. 2003; Schubiger et al. 2007, 2010a). In addition, a summary and a
comparison of the three trials were published in 2010 (Schubiger et al. 2010b). The
present paper reports on the results of the experiments undertaken in 2010.

26.2 Material and Methods

The trial was sown at 24 sites in 11 European countries. At 16 of these sites the
trial was carried out during each of the 4 years of evaluation (2001, 2004, 2007 and
2010). The same 33 perennial, 15 Italian and 3 Hybrid ryegrass cultivars were tested
as in the previous three trials. In addition, the Italian ryegrass cultivars ‘Crema’ and
‘Gosia’ and the perennial ryegrass cultivar ‘Maja’ were included as they were in
the trials of 2007. Nine Italian, two hybrid and 15 perennial ryegrass cultivars were
tetraploid, the other cultivars were diploid.
Twelve grams of seed of each cultivar were forwarded to each participant in the
form of encoded seed lots. At each site the seed was sown in spring in a completely
randomized block design with four replicates. The cultivars were grown as rows
(3 m long and 0.5 m apart). The perennial and Italian ryegrass cultivars were sown
in separate trials. The hybrid ryegrass cultivars were included in the Italian ryegrass
trial. The trials were cut and fertilized as was customary at each site.
The trials were scored for rust incidence between July and October one to three
times during different growth cycles and periods of abundant rust development.
Participants were asked to score the cultivars for each rust species occurring in the
field, separately. A scale from 1 to 9 was used: with 1 = no rust disease, 2 = trace
of rust, 3 = 5 %, 4 = 10 %, 5 = 25 %, 6 = 40 %, 7 = 60 %, 8 = 75 % and 9 =
more than 75 % of the foliage covered with rust. The rating values represented a
relative estimate of leaf area occupied by rust pustules, and not reaction type. For
any particular site, scoring data with an average score of at least two were included
212 F. X. Schubiger et al.

Table 26.1 Analysis of variance of mean rust disease scores. All F-values are significant at the
p < 0.001 level
Ryegrass species Lolium multiflorum Lolium perenne
Rust species P. coronata P. graminis P. coronata P. graminis
No. of cultivars 20 20 34 34
No. of sites 17 1 16 4
F-value for cultivars 137.8 23.7 66.6 28.5
F-value for sites 132.9 – 253.7 143.9
F-value for cv. x site interaction 4.4 – 4.7 3.1

in the analysis, provided that there were significant differences between the cultivars
at the site. If there were sites with more than one valid scoring per year, means of
the scorings (per row) were calculated and used in further analysis.

26.3 Results

Crown rust (Puccinia coronata) was the most serious and the most frequently ob-
served rust disease of Italian/Hybrid and perennial ryegrass. A relevant incidence
(average score of at least two and significant differences between cultivars) of crown
rust on Italian ryegrass was observed at 17 out of 23 sites and on perennial ryegrass
at 16 out of 24 sites, respectively. Stem rust (P.graminis) occurred on Italian ryegrass
at only one site (Les Rosiers, data not shown) and on perennial ryegrass at four sites,
respectively.
Over all sites, there was a highly significant difference (p < 0.001) in mean crown
rust scores among Italian/Hybrid ryegrass cultivars (Table 26.1). The cultivars ‘Taran-
dus’, ‘Gosia’, ‘Caballo’ and ‘Domino’ showed the highest level of resistance of all
the Italian/Hybrid ryegrass cultivars tested (Table 26.2). Despite the occurrence of
significant interactions of cultivars with sites, the Spearman rank order correlation
between the data of a particular site and the mean of all sites was always significant
(Table 26.2).
Mean crown and stem rust susceptibility scores of perennial ryegrass cultivars at
each site are presented in Tables 26.3 and 26.4. Analysis of variance revealed highly
significant differences in crown and stem rust susceptibility over all sites (Table 26.1).
‘Gwendal’ and ‘Bocage’ were the most crown rust resistant cultivars. ‘Gwendal’ and
‘Pastoral’ were the most stem rust resistant cultivars. The ranking of the cultivars
in respect to crown or stem rust susceptibility was very consistent. The Spearman
rank order correlations between the data of each site and the mean of all sites were
significant in all but four cases for crown rust (Druelle, Gumpenstein, Montours
and Radzikow). The rank order correlation between the mean disease scores of
the perennial ryegrass cultivars for the two rust pathogens was not significant at
p < 0.01(rs = 0.41).
Table 26.2 Crown rust (Puccinia coronata) disease scores of 17 Italian and 3 hybrid ryegrass cultivars at 17 sites. Data are means of one to three scorings
per site. Scores are made in a scale of 1–9; 1 = no rust. Cultivars are ranked according to the mean over sites. 4x = tetraploid; 2x = diploid
Cultivar Asen- Aston Bornhof Drue- Gumpen- H. Zivotice Lelystad Les Ros- Lodi Merel- Mont- Otter- Perugia Pulling Radzi- Swifter- Zurich Mean of
dorf D UK D lle F stein A CZ B. NL iers F I beke B ours F sum NL I D kow PL band NL CH all sites
Tarandus (4 x) 1.8 1.9 1.3 4.9 1.6 2.3 1.3 2.5 3.9 1.8 1.5 1.3 2.5 1.0 2.7 2.2 1.1 2.1
Gosiab (4 x) 2.8 1.5 1.5 3.3 2.1 4.2 1.0 2.0 3.3 1.5 1.1 1.0 1.8 2.0 3.8 2.0 1.1 2.1
Caballo (4 x) 2.3 2.0 2.0 4.6 1.6 2.6 1.3 3.0 3.6 1.3 1.8 1.8 2.7 1.0 2.8 1.5 1.3 2.2
Domino (4 x) 1.8 1.8 1.3 4.5 1.8 2.3 1.3 2.8 4.9 1.5 1.9 1.5 2.9 1.3 3.1 1.8 1.5 2.2
Zorro (4 x) 2.3 1.6 1.3 5.1 1.6 3.3 1.3 3.0 5.8 1.8 1.8 1.5 2.5 1.0 3.3 1.7 1.5 2.4
Bolero (4 x) 1.8 1.6 1.5 5.4 1.6 2.3 1.3 2.5 5.4 3.3 1.8 1.3 2.5 1.5 3.4 2.3 1.5 2.4
Tonyl (4 x) 2.0 2.9 1.0 4.9 1.8 2.8 1.8 2.5 4.6 2.0 2.5 2.0 2.0 1.5 3.5 2.0 1.4 2.4
Barprisma (2 x) 2.5 2.1 1.3 4.6 2.3 2.8 1.5 3.3 3.3 3.3 1.8 2.5 2.9 1.3 3.8 2.3 1.3 2.5
Aberexcb (4 x) 3.3 2.5 1.5 3.4 2.0 3.1 2.3 3.3 4.4 3.0 1.1 1.8 2.5 2.0 3.3 2.6 2.1 2.6
Fastyl (2 x) 2.8 2.5 2.0 5.0 2.3 2.3 1.5 2.3 3.6 3.5 3.1 1.8 2.8 1.5 3.6 2.3 1.4 2.6
Ellire (4 x) 2.3 2.1 1.5 5.5 1.5 2.8 1.5 3.0 6.3 2.5 3.5 1.8 3.7 1.8 3.5 2.1 2.0 2.8
Pirolb (2 x) 4.5 2.3 1.8 5.1 2.5 3.0 2.0 3.0 5.7 4.3 3.6 2.8 3.4 2.5 3.7 3.4 2.5 3.3
Meryl (2 x) 3.8 2.0 3.3 5.3 2.3 3.1 2.0 4.0 4.8 5.0 3.6 3.5 3.3 2.3 4.4 3.2 2.8 3.4
Crema (2 x) 5.8 5.5 2.3 5.0 3.4 3.0 3.0 2.5 5.6 4.8 7.4 3.5 3.6 2.0 4.1 3.8 2.5 4.0
Danergo (4 x) 2.8 2.0 4.0 6.4 2.0 5.2 3.3 5.8 5.9 7.0 4.3 5.0 3.8 2.5 4.2 2.3 3.3 4.1
Lolita (4 x) 3.5 1.9 5.0 6.9 2.1 5.8 3.0 5.3 8.0 4.0 4.0 4.8 4.5 2.5 4.7 2.2 3.5 4.2
Ligrande (2 x) 5.3 2.8 3.0 6.9 2.6 5.0 6.0 4.0 6.3 7.8 4.8 5.5 4.0 4.0 4.8 3.5 4.5 4.7
26 The EUCARPIA Multi-site Rust Evaluation—Results 2010

Gumpenb (2 x) 5.3 2.6 4.3 6.6 3.3 4.8 3.5 4.5 7.3 8.3 5.6 5.8 4.8 3.5 5.3 3.4 5.3 4.9
Lema (2 x) 5.0 2.1 4.3 6.8 2.8 4.8 4.5 4.8 7.0 7.8 6.9 5.8 3.8 4.8 5.2 3.8 4.5 5.0
Gordo (2 x) 4.8 2.4 4.0 7.0 2.6 5.9 4.3 6.0 6.9 7.8 5.1 6.5 5.2 4.8 5.3 3.8 5.8 5.2
Mean 3.3 2.3 2.4 5.4 2.2 3.6 2.4 3.5 5.3 4.1 3.4 3.1 3.2 2.2 3.9 2.6 2.5 3.2
LSD (P < 0.05) 1.0 0.5 1.8 0.8 0.4 0.6 1.1 1.6 1.6 1.3 0.7 1.3 1.2 0.9 0.7 0.7 1.0
Correlationa 0.82 0.54 0.82 0.82 0.73 0.72 0.94 0.75 0.78 0.94 0.89 0.94 0.88 0.89 0.87 0.78 0.94
a
Spearman rank order correlation with mean of all sites (all values are significant at P < 0.05)
b
Hybrid Ryegrass
213
 214

Table 26.3 Crown rust (P. coronata) disease scores of 34 perennial ryegrass cultivars at 16 sites. Data are means of one to three scorings per site. Scores are
made in a scale of 1–9; 1 = no rust. Cultivars are ranked according to the mean over sites. 4x = tetraploid; 2x = diploid
Cultivar Asend- Aston Bornhof Druelle Gumpen- H. Zivot- Lelystad Lodi Malc- Merelb- Monto- Ottersum Pulling Radzi- Swifter- Zurich Mean of
orf D UK D F stein A ice CZ B. NL I how D eke B urs F NL D kow PL band NL CH all sites
Gwendal (4 x) 2.8 1.8 2.5 2.8 2.3 2.3 2.7 1.3 2.8 2.7 2.5 1.0 2.0 2.5 3.2 1.3 2.3
Bocage (4 x) 2.8 2.3 4.3 2.5 2.0 2.5 2.7 2.0 2.3 3.3 3.1 1.3 1.0 2.5 2.9 1.8 2.4
Carrera (2 x) 3.5 2.0 3.3 3.3 2.6 2.5 3.3 1.4 2.3 3.3 2.6 1.8 1.5 3.5 3.7 1.3 2.6
Lacerta (4 x) 2.5 3.0 5.5 3.3 2.5 2.5 2.8 3.7 1.8 2.3 4.1 1.0 1.5 2.3 1.8 1.3 2.6
Pastoral (4 x) 3.8 2.3 5.0 3.1 2.4 1.8 3.3 2.6 2.3 3.3 3.1 1.0 1.3 2.3 3.7 2.5 2.7
Orval (4 x) 3.3 2.3 4.0 3.0 2.6 2.0 5.1 4.6 3.0 3.5 3.5 1.0 1.3 3.3 1.8 1.6 2.9
Aubisque (4 x) 3.8 2.8 6.0 3.5 2.0 2.3 3.5 1.4 3.3 2.8 2.8 1.0 1.5 2.5 4.3 2.8 2.9
Vincent (2 x) 4.0 2.5 3.8 4.0 3.3 2.3 4.1 1.8 2.8 4.3 3.3 2.0 1.3 3.5 2.4 1.3 2.9
Option (2 x) 4.0 2.8 5.3 4.4 3.3 2.3 3.8 2.4 2.5 4.0 4.5 1.5 1.5 3.0 3.3 1.3 3.1
Heraut (2 x) 4.5 2.8 4.8 4.1 3.0 2.3 4.1 2.1 2.8 4.5 3.6 2.5 1.3 4.0 2.5 2.1 3.2
Elgon (4 x) 5.0 3.0 7.3 2.6 2.5 2.3 5.4 2.2 4.0 3.0 2.5 1.5 1.8 3.0 3.8 3.1 3.3
Kells (2 x) 4.5 2.8 5.8 4.4 2.9 2.8 5.0 1.6 3.3 4.8 2.4 2.5 1.8 3.5 3.6 2.6 3.4
Roy (4 x) 3.3 3.0 6.5 2.5 2.1 3.0 5.0 3.1 3.3 4.5 3.6 2.0 1.8 2.5 4.8 3.1 3.4
Guru (2 x) 3.3 3.0 2.5 4.9 3.4 3.5 4.6 3.7 2.0 4.8 3.4 4.8 1.3 4.0 2.7 3.4 3.4
Barnhem (2 x) 5.3 3.0 5.3 3.4 3.6 2.3 5.8 2.2 3.0 4.8 2.5 3.0 1.3 3.8 4.3 1.8 3.4
Aberdart (2 x) 5.3 3.3 6.5 3.8 2.6 2.0 5.6 2.9 2.0 5.0 2.0 2.3 1.8 3.3 4.4 3.0 3.5
Corbet (2 x) 4.8 2.5 5.5 4.5 3.0 2.3 4.8 1.6 3.0 5.0 4.3 2.0 2.3 4.3 3.9 2.0 3.5
Fennema (2 x) 4.0 3.3 6.5 4.6 3.1 2.8 4.5 2.6 3.3 5.0 2.5 2.8 1.5 3.8 3.4 2.6 3.5
Kentaur (4 x) 5.0 3.3 7.5 2.5 2.1 4.3 4.9 4.4 5.0 3.5 2.4 1.8 2.3 2.3 5.1 3.1 3.7
Maja (4 x) 5.3 2.3 6.0 2.9 2.3 3.5 5.5 3.6 4.3 4.0 2.9 2.8 2.3 2.8 5.7 3.5 3.7
Weigra (2 x) 4.3 3.0 6.3 4.3 3.3 3.0 4.5 4.0 3.3 4.8 5.3 2.5 2.0 3.5 3.3 2.4 3.7
Sponsor (2 x) 4.8 3.5 7.5 4.5 3.0 2.3 5.5 1.3 2.8 5.5 2.8 2.8 1.8 3.5 4.8 3.4 3.7
Litempo (4 x) 4.5 3.3 7.3 3.4 2.3 3.5 5.2 4.1 3.8 3.3 3.0 1.8 1.8 3.5 5.5 3.9 3.7
Arabella (2 x) 4.8 3.3 6.8 4.6 3.3 2.8 5.1 3.4 2.8 5.3 4.5 2.3 1.8 4.0 4.0 2.0 3.8
F. X. Schubiger et al.
Table 26.3 (continued)
Cultivar Asend- Aston Bornhof Druelle Gumpen- H. Zivot- Lelystad Lodi Malc- Merelb- Monto- Ottersum Pulling Radzi- Swifter- Zurich Mean of
orf D UK D F stein A ice CZ B. NL I how D eke B urs F NL D kow PL band NL CH all sites
Gladio (2 x) 5.5 3.0 7.5 3.9 3.3 2.5 6.1 2.4 2.8 6.0 2.8 2.3 2.8 3.0 4.4 3.4 3.8
Terry (4 x) 5.5 4.0 8.3 2.9 2.4 3.0 5.8 4.9 4.3 4.0 2.9 2.0 2.8 3.0 5.0 3.1 4.0
Tivoli (4 x) 5.5 4.8 8.5 2.8 2.6 2.8 5.5 3.5 4.8 5.3 2.5 2.8 2.3 2.3 5.3 4.9 4.1
Foxtrot (2 x) 6.8 2.5 7.3 3.6 2.6 2.8 6.0 2.6 3.8 5.5 4.0 3.3 2.3 3.3 6.1 3.6 4.1
Aristo (2 x) 6.8 3.3 7.8 3.4 3.0 2.5 7.0 2.6 4.0 5.8 2.8 4.0 2.5 3.3 7.2 3.9 4.3
Sirocco (4 x) 6.0 3.5 8.0 3.4 2.5 3.8 6.2 5.8 5.3 5.0 5.3 1.8 2.5 3.3 6.1 4.0 4.5
Helmer (4 x) 7.0 4.0 8.5 3.0 2.3 3.5 6.7 6.6 4.0 4.5 3.3 3.3 2.0 3.0 6.3 4.6 4.5
Lipresso (2 x) 5.5 3.3 7.8 4.6 3.5 4.3 6.3 5.1 2.8 6.5 5.0 4.8 4.3 4.0 4.2 4.4 4.8
Condesa (4 x) 6.8 4.3 8.3 3.0 2.9 2.5 6.3 4.7 4.3 7.3 4.8 4.0 3.5 3.3 7.5 5.9 4.9
Aurora (2 x) 6.5 4.3 8.8 5.8 3.8 6.8 6.9 2.5 6.8 8.5 5.8 7.8 5.5 4.8 8.2 8.5 6.3
Mean 4.7 3.0 6.2 3.6 2.8 2.8 5.0 3.1 3.3 4.6 3.4 2.5 2.0 3.2 4.4 3.0 3.6
LSD (p = 0.05) 0.9 0.7 1.6 0.9 0.6 1.0 0.7 1.8 1.3 1.3 0.8 1.4 1.0 0.8 0.9 1.3
a
Correlation 0.88 0.79 0.88 ns ns 0.61 0.88 0.55 0.64 0.78 ns 0.73 0.80 ns 0.80 0.85
a
Spearman rank order correlation with mean of all sites (all values are significant at p < 0.05 except ns = not significant)
26 The EUCARPIA Multi-site Rust Evaluation—Results 2010
215
216 F. X. Schubiger et al.

Table 26.4 Stem rust (Puccinia graminis) disease scores of 34 perennial ryegrass cultivars at four
sites. Data are means of one to two scorings per site. Scores are made in a scale of 1–9; 1 = no rust.
Cultivars are ranked according to the mean over sites. 4x = tetraploid; 2x = diploid
Cultivar Les Rosiers F Montours F Radzikow PL Malchow D Mean of all sites
Gwendal (4 x) 1.0 2.5 2.6 2.3 2.1
Pastoral (4 x) 1.0 2.0 2.9 2.8 2.2
Roy (4 x) 1.0 2.3 3.0 2.8 2.3
Tivoli (4 x) 1.3 2.0 3.0 2.8 2.3
Bocage (4 x) 1.0 2.8 2.9 2.5 2.3
Maja (4 x) 1.0 2.3 3.3 2.8 2.3
Aubisque (4 x) 1.0 2.8 2.9 2.8 2.3
Aberdart (2 x) 1.3 2.3 3.5 2.5 2.4
Carrera (2 x) 1.5 2.5 3.4 2.8 2.5
Terry (4 x) 1.3 2.8 3.4 2.8 2.5
Elgon (4 x) 1.0 3.5 3.3 2.5 2.6
Orval (2 x) 1.0 3.5 3.3 2.5 2.6
Kentaur (4 x) 1.3 2.8 3.1 3.3 2.6
Lacerta (4 x) 2.0 3.0 3.0 2.8 2.7
Condesa (4 x) 1.5 2.3 3.6 3.5 2.7
Litempo (4 x) 1.5 2.5 3.5 3.5 2.8
Aristo (2 x) 2.0 3.0 3.5 3.0 2.9
Helmer (4 x) 2.3 3.0 3.3 3.0 2.9
Sirocco (4 x) 1.3 3.8 3.6 3.5 3.0
Foxtrot (2 x) 2.0 3.3 3.6 3.5 3.1
Gladio (2 x) 2.5 3.5 3.8 3.3 3.3
Guru (2 x) 4.0 3.0 3.5 3.8 3.6
Option (2 x) 3.5 3.3 3.5 4.3 3.6
Barnhem (2 x) 2.0 4.3 3.8 4.8 3.7
Vincent (2 x) 3.0 4.0 3.8 4.8 3.9
Heraut (2 x) 2.8 4.8 3.5 4.8 3.9
Weigra (2 x) 4.0 3.3 4.0 4.5 3.9
Kells (2 x) 3.0 5.3 4.0 4.3 4.1
Fennema (2 x) 2.8 5.5 3.9 4.5 4.2
Sponsor (2 x) 3.8 4.8 3.9 4.5 4.2
Arabella (2 x) 4.0 4.5 3.9 4.8 4.3
Lipresso (2 x) 4.0 4.0 4.5 4.8 4.3
Aurora (2 x) 4.3 3.8 4.4 6.0 4.6
Corbet (2 x) 4.8 5.8 4.3 5.3 5.0
Mean 2.2 3.4 3.5 3.6 3.2
LSD (p = 0.05) 0.9 0.9 0.5 1.0
Correlationa 0.92 0.87 0.91 0.92
a
Spearman rank order correlation with mean of all sites (all values are significant at p < 0.05)

The ranking of the Italian/Hybrid and perennial ryegrass cultivars in terms of

crown and stem rust resistance was consistent over the whole experimentation period
of nine years. The Spearman rank order correlations of average cultivar disease
scores between the years 2001, 2004, 2007 and 2010 are presented in Table 26.5.
For perennial ryegrass, the correlations involving 2007 were consistently lower than
those among 2001, 2004 and 2010.
26 The EUCARPIA Multi-site Rust Evaluation—Results 2010 217

Table 26.5 Coefficients of Spearman rank order correlations among mean crown (Puccinia coro-
nata) and stem rust (P. graminis) disease scores of Italian (Lolium multiflorum) and perennial
ryegrass (L. perenne) in 4 years of evaluation
Italian ryegrass Perennial ryegrass
Crown rust Crown rust Stem rust
2001 vs. 2004 0.99* 0.97* 0.95*
2001 vs. 2007 0.98* 0.92* 0.94*
2001 vs. 2010 0.99* 0.96* 0.94*
2004 vs. 2007 0.97* 0.90* 0.98*
2004 vs. 2010 0.99* 0.95* 0.92*
2007 vs. 2010 0.98* 0.86* 0.94*
*significant at p < 0.01

26.4 Discussion

In 2010, the results of the Eucarpia multisite rust evaluation trial revealed that several
Italian and perennial ryegrass cultivars are still resistant to crown and stem rust in a
wide range of sites across Europe.
As a consequence, the 2010 ranking of the ryegrass cultivars in terms of crown
and stem rust resistance was very consistent with the ranking of the trials in 2001,
2004 and 2007, respectively.
The trends for a breakdown of crown rust resistance of some cultivars in 2007
were not confirmed. In 2010, for example, ‘Orval’ ranked again in the 6th position
despite its 20th position in 2007. Therefore, there is no evidence that crown rust
resistance of an individual cultivar was overcome by the rust pathogen over the 9
years of experimentation.

References

Boller B, Schubiger FX, Streckeisen P (2003) The Eucarpia multisite rust evaluation—results 2001.
Vortr pflanzenzüchtung 59:198–207
Schubiger FX, Streckeisen P, Boller B (2007) The EUCARPIA multisite rust evaluation—results of
the trials 2004. In Rosellini D, Veronesi F (eds) Breeding and seed production for conventional
and organic agriculture. Proceedings of the Eucarpia meeting perugia, pp 154–158
Schubiger FX, Boller B et al (2010a) The EUCARPIA Multi-site rust evaluation—Results of the
trials 2007. In: Huyghe C (ed) Sustainable use of genetic diversity in forage and turf breeding.
Springer, pp 331–340
Schubiger FX et al (2010b) Susceptibility of European cultivars of Italian and perennial ryegrass
to crown and stem rust. Euphytica 176:167–181
Chapter 27
The Main Topics of Resistance Breeding
in Grasses in the Czech Republic

B. Cagaš and M. Svobodová

Abstract In the years 1995–2009 the incidence of important diseases in grasses

grown for seed in the Czech Republic was investigated. The goal of the study was
to find out whether there has been a significant increase in the level of resistance.
Each year a total of 500 samples of 19 cultivated grass species were taken on 2,000–
4,500 ha in various regions of the Czech Republic. Parasitic silver top (Fusarium
poae) was detected in 13.8–48.1 % of the locations under observation, with a statis-
tically insignificant downward trend of 0.19 % per year (P-value = 0.7518). Powdery
mildew (Blumeria graminis), occurring in 9.9–40 % of the locations showed a sig-
nificant downward trend of 1.2 % per year (P-value = 0.0373). Rusts (Puccinia
spp., Uromyces spp.) were recorded on 4.0–33.4 % of the sites; the trend remained
constant. Choke (Epichloë typhina) was found in 0.1–3.8 % of the locations, the
incidence of choke showed a significantly upward trend of 0.19 % per year (P-value
0.0030). Leaf spots (predominantly caused by Cladosporium spp., Drechslera spp.,
Mastigosporium spp. etc.) occurred in 7.5–35.4 % of the locations (an decreasing
incidence by 2 % per year, P-value = 0.0001).

Keywords Silver top · Leaf diseases · Rusts · Powdery mildew · Choke

27.1 Introduction

The land area of grasses grown for seed in the Czech Republic was subjected to strong
fluctuations in the last years, like in other European countries, and at present it is
estimated to be 13,000 ha. Grass seed production is still a very interesting branch of
plant production and grass seeds are mostly exported. However, their amounts and
quality are often significantly affected, especially in some years, by the action of
specific pests and diseases (Cagaš 2010). The spectrum of cultivated grass species
and varieties is unprecedentedly wide and has considerably increased since the year

B. Cagaš ()
OSEVA Development and Research Ltd., Zubří, Czech Republic
e-mail: [email protected]
M. Svobodová
Faculty of Agrobiology, Food and Natural Resources, Czech University of Life Sciences Prague,
Prague, Czech Republic

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 219

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_27, © Springer Science+Business Media Dordrecht 2013
220 B. Cagaš and M. Svobodová

60
50
40 silver top
% 30 powdery mildew
20 rusts
10 leaf diseases
0 choke
1994

1996

1998

2000

2002

2004

2006

2008

2010
Year

Fig. 27.1 Incidence of important diseases in grasses grown for seed in the years 1995–2009 in the
Czech Republic (%)

1995. In a number of species the spread of some diseases can be controlled by

traditional methods, in others by growing varieties with a higher resistance. The goal
of this study was to assess on the basis of long-term monitoring the prospects of
occurrence of some causal agents of diseases, to find out whether there has been
some progress in developing resistance to some diseases, to show the success rate
of resistance breeding in varieties grown for seed production in the Czech Republic
and to outline potential future trends of breeding for resistance.

27.2 Materials and Methods

Data on the incidence of the most serious diseases of grass seed stands (Fig. 27.1) were
obtained from the written reports of inspectors of Central Institute for Supervising
and Testing in Agriculture (ÚKZÚZ) whose task was to assess the occurrence of pests
and diseases during the growing season. In the period of study (1995–2009) each
year a total of 500 samples were collected to determine the incidence of a particular
disease. This study was carried out on 2,000–4,500 ha in 19 cultivated grass species
in different regions of the Czech Republic. Data on the number of varieties of both
forms of Italian ryegrass and meadow fescue grown in the Czech Republic in the
years 1995–2009 were obtained from the List of varieties of those particular years.
Data on the incidence of the most important diseases were processed by regression
analysis in the program Statgraphics version XV.

27.3 Results and Discussion

The incidence of important pests and diseases occurring on grasses grown for seed in
the Czech Republic in the years 1995–2009 is shown in Fig. 27.1. It reveals that para-
sitic silver top caused by the fungus Fusarium poae and transmitted predominantly by
the meadow plant bug (Leptopterna dolobrata) was detected in the observation period
27 The Main Topics of Resistance Breeding in Grasses in the Czech Republic 221

in 13.8–48.1 % of the locations under study, powdery mildew (Blumeria graminis)

occurred in 4.2–40 % of the locations, grass rusts (Puccinia spp., Uromyces spp.) in
4.0–33.4 % of the sites, leaf diseases (predominantly Cladosporium spp., Drechslera
spp., Mastigosporium spp., Pyrenophora spp.) in 7.5–35.4 % of the locations and
choke (Epichloë typhina) was detected in 0.1–3.8 % of the locations.
Parasitic silver top can be controlled by a combination of non-chemical and insec-
ticide treatments. Choke still remains a hard-to-solve problem; however, with respect
to the narrow range of the host species it is considered a minor problem (Cagaš 2009).
Out of the three remaining grass mycoses the most serious problem with respect to
high economic harmfulness is graminicolous rusts, mainly crown rust (Puccinia
coronata Corda var. coronata) and stem rust (Puccinia graminis subsp. graminicola
Urb.). Regression analysis of the incidence of the most serious diseases, where y = %
of localities with disease incidence, x = year 1–15, shows interesting facts. The inci-
dence of silver top (Fusarium poae) showed a slightly downward trend of 0.19 % per
year (y = −0.19x + 37.693), but under the same conditions in the future a marked
decline cannot be expected probably due to non-consistent chemical control of grass
seed stands. Moreover, the dependence is not significant (P-value = 0.7518). A sig-
nificant downward trend in incidence (P-value = 0.0373) was recorded in powdery
mildew (Blumeria graminis)—1.2 % per year (y = −1.2186x+29.735). In contrast,
a significantly upward trend of 0.19 % per year was seen in choke (Epichloë typhina)
(y = 0.1914x − 0.2314, P-value 0.0030) and especially in leaf diseases—almost
2 % per year (y = 1.9936x + 4.3181, P-value = 0.0001). The incidence curve of
graminicolous rusts suggests a steady trend (y = 0.0311x + 17.845) and it is ex-
pected that the incidence of rusts will be at the same level as it is now (Fig. 27.1).
The dependence, however, is not significant (P-value = 0.9436).
The findings suggest the necessity to aim breeding at increased resistance to stem
rust and crown rust in their major host species—perennial ryegrass, Italian ryegrass
and meadow fescue. To other important reasons belong the incidence of favorable
conditions supporting the overwintering of the spores and early rust infection. Breed-
ing for resistance to rust diseases has been a priority of research teams for a long
time and according to Heijden and Roulund (2010) good progress has been made
in this area and the resistance of present-day perennial ryegrass varieties to Puc-
cinia coronata has markedly increased. Has this trend become evident also in the
collection of varieties of the genus Lolium grown for seed in the last 15 years in the
Czech Republic? The number of varieties of both forms of Italian ryegrass, peren-
nial ryegrass and meadow fescue listed in the State Variety Book increased from 22
varieties (in the year 1995) to 114 (in the year 2011). The non-registered varieties
which are grown for seed also had a similar increase in tendency. The incidence of
rusts as given in Fig. 27.1 was studied in the years 1995–2009 in a large popula-
tion of varieties of several species. The study did not make any distinction between
P. coronata and P. graminis and included also other grass rusts. In this complex the
increase in resistance in ryegrasses and fescue was not very dramatic, as confirmed
by the regression line (Fig. 27.1). The multi-site rust evaluation trials organized on
different European localities can help to detect not only the widespread of both rusts
222 B. Cagaš and M. Svobodová

60
50
silver top
40 powdery mildew
% 30
rusts
20 leaf diseases
10 choke
0
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Year

Fig. 27.2 Regression lines of selected diseases incidence (%) (x = 1–15)

but possible sources of resistance (Schubiger et al. 2010). Breeding for resistance to
grass rusts, especially stem rust, which has been spreading very quickly in Central
Europe after the year 1999, is a vital task for Czech grass breeders (Fig. 27.2).

Acknowledgments The study was supported by the Grass and Legumes Seed Growers Association
of the Czech Republic (Research Project no. SPTJS 2009–2012).

References

Cagaš B (2009) Bude dusiváplísň ovitost ohrožovat v budoucnu naše semenářské porosty trav?
Pícninářské listy 2008 15:13–14
Cagaš B (2010) Choroby, škuº dci, abionózy a ochrana proti nim. In: Cagaš B (ed) A kol.:Trávy pě
stované na semeno. Vydavatelství Ing. Petr Baštan, Olomouc, pp 207–235
Heijden van der AG, Roulund N (2010) In: Huyghe C (ed) Sustainable use of genetic diversity in
forage and turf breeding pp 247–260, doi:10.1007/978-90-481-8706-5_36
Schubiger FX, Baert J, Bayle B, Bourdon P, Cagaš B, Černoch V, Czembor E, Eickmeyer F, Feuer-
stein U, Hartmann S, Jakesova H, Johnston D, Krautzer B, Leenheer H, Lellbach H, Persson C,
Pietraczek W, Posselt UK, Romani M, Russi L, Schulze S, Tardin MC, VanHee F, Van Kruijssen
L, Wilkins P, Willner E, Wolters L, Boller B (2010) Susceptibility of European cultivars of Italian
and perennial ryegrass to crown and stem rust. Euphytica 176:167–181
 Part V
Genetic Variation and Adaptation
Chapter 28
Origins of Diploid Dactylis from the Canary
Islands as Determined by DNA Sequencing

A. V. Stewart, N. W. Ellison and V. E. Martín Osorio

Abstract Two diploid forms of Dactylis are known to occur on the Canary Islands,
Dactylis smithii at low altitude and the newly discovered Dactyls metlesicsii at high
altitude. Nuclear ITS sequences of these two diploid forms were compared with a
wider set of diploids from Spain, Portugal and North Africa. The results suggest that
Dactylis has a dual origin in these islands. The newly discovered high altitude form
metlesicsii was found to have originated from the Iberian Peninsula unlike the low
altitude form smithii which originated from nearby North Africa. It is also likely that
hybrid forms occur where their ranges overlap.

28.1 Introduction

Dactylis is a genus represented by diploids, tetraploids and a hexaploid. It is dis-

tributed from China in the east to the Canary and Cape Verde Islands in the west,
over a wide range of climatic conditions, including temperate, sub-tropical, tropical,
Mediterranean and sub-alpine. Tetraploids cover almost the full range of the genus
while diploids are often little more than relic populations with very restricted ranges.
A hexaploid population occurs in Libya and Egypt. Many authors consider the genus
to be a single large and very diverse species, Dactylis glomerata L. (Stewart and Elli-
son 2011), while other authors divide the genus into many species and/or sub-species,
often in an inconsistent manner. Many authors divide the tetraploid forms into glom-
erata and hispanica types; the larger green productive glomerata types from high
rainfall regions and smaller bluish hispanica types from dry Mediterranean climates.
Unfortunately there is no consistent taxonomic treatment of the genus as a whole.

A. V. Stewart ()
PGG Wrightson Seeds, PO Box 175, Lincoln 7640, New Zealand
e-mail: [email protected]
N. W. Ellison
AgResearch Grasslands, Private Bag, 11008 Palmerston North, New Zealand
e-mail: [email protected]
V. E. Martín Osorio
Departamento de Biología Vegetal, Universidad de La Laguna, Islas Canarias, Spain
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 225

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_28, © Springer Science+Business Media Dordrecht 2013
226 A. V. Stewart et al.

In 1988, 15 diploid forms were reviewed by Lumaret (1988) but today

18 geographic diploids are known (Stewart and Ellison 2011). Although some forms
may not warrant their current taxonomic status the names still serve to provide a
valuable terminology for such geographically distinct and often isolated populations.
• altaica from the Alatau Mountains of Kazakhstan;
• aschersoniana in northern Europe and the Caucasus;
• castellata in Algeria;
• himalayensis in the Himalayan mountains;
• hyrcana on the Talish plain of North West Iran;
• ibizensis on the Balearic Islands near the East Coast of Spain;
• izcoi in North west Spain, formerly known as “Galician”;
• judaica in the mountains of Israel and Lebanon;
• juncinella at altitude in the Sierra Nevada mountains of Spain;
• lusitanica on the Sintra near Lisbon, Portugal;
• mairei in Algeria;
• metlesicsii at high altitude in the Canary Islands;
• parthiana at high altitude in the Elburz mountains of Iran;
• reichenbachii on the dolomite soils of the European alps and French Massif
Central;
• santai in Morocco;
• sinensis in the mountains of Xingjian Province of China;
• smithii at sub-tropical mid-altitudes of the Canary and Cape Verde Islands;
• woronowii on the steppes of Iran;
While the diploid forms cross very readily, the resulting hybrids often have a reduced
fertility with nuclear and cytoplasmic differences (Parker and Borrill 1968) but few
major structural chromosome differences (Wetschnig 1991). This partial sterility
is indicative of partial speciation, consistent with the large number of sub-species
known.
There have been a number of attempts to understand the historical relationships
among the diploid subspecies using morphological differences (Stebbins and Zohary
1959), isozymes (Lumaret 1988), phenolic compounds (Fiasson et al. 1987) and
chromosome karyotypes (Wetschnig 1991) and it is only recently that molecular
techniques have been employed.
Stewart and Ellison (2011) used nuclear ITS and chloroplast trnL intron sequences
to clarify the molecular phylogeny relationships. This showed that the genus is rela-
tively recent with the first diploids originating in central Asia over 150,000 year ago,
and during the subsequent interglacial period migrating west to Portugal and south
into Israel. As the northern climate cooled in the last glaciation, this continuous distri-
bution became disjunct, leaving many small remnant populations in milder southern
glacial refugia. At the same time the central Asian forms were able to migrate from
Israel across North Africa through the expanding grasslands as far as Morocco, and
on to the Canary and Cape Verde Islands. As the climate warmed again in the post
glacial period, the North African grassland forms became restricted to small disjunct
remnants, while the forests of Northern Europe were re-colonised by aschersoniana
28 Origins of Diploid Dactylis from the Canary Islands as Determined . . . 227

Table 28.1 ITS sequence differences between Dactylis forms

Accessions 55 92 425 571 601
izcoi Spain 5 C G G A A
lusitanica Portugal 6 C G G A A
metlesicsii Canary Islands 1 C G A/G A/G A/G
smithii Canary Islands 6 T C A G A/G
castellata Morocco 2 T C A G G
mairei Algeria 2 T C A G G
santai Algeria 6 T C A G G

from the Caucasus glacial refuge. Surprisingly many phylogenetic lineages consist
of both productive temperate forms as well as xeromorphic dryland forms indicating
Dactylis is quite plastic in its historical response to climate.
This paper presents molecular data on the recently reported Dactylis metlesicii
Schönfelder & Ludwig (Schönfelder and Ludwig 1996), a diploid form from high
altitudes in the Canary Islands, and compares it to the lower altitude Dactylis smithii
Link as well as to the phylogeny of the other diploids.

28.2 Materials and Methods

A single accession of Dactylis metlesicsii was collected on the Ruta de las Siete
Cañadas, Tenerife Island, (Herbario de la Universidad de la Laguna, TFC 44983).
DNA was extracted from approx. 10 mg of dried leaf material using the DNeasy
Plant Mini Kit (QIAGEN) according to the manufacturer’s instructions. The ITS
region was amplified using the primers EC-1 and EC-2 (Williams et al. 2001), and
the trnL (UAA) intron was amplified using the primers “c” and “d” (Taberlet et al.
1991). PCR products were purified and sequenced directly as described in Ellison
et al. (2006). DNA sequences were aligned, with manual adjustment to optimise
alignments where necessary, using MegAlign (DNASTAR). The tree was rooted
using the closely related species Lamarckia aurea (Catalan et al. 2004). Ploidy
levels were confirmed using flow cytometry of leaf tissue in comparison to a known
diploid standard. For each accession of each subspecies 3 or more genotypes were
sampled.

28.3 Results

The results of the ITS sequencing for a range of forms, as presented in Table 28.1,
show that the clear differentiation between Iberian forms and North African forms of
Dactylis as reported by Stewart and Ellison (2011). The smithii forms share similar
sequences with the North African forms except at position 601 where three out of four
accessions expressed a dual sequence, indicative of introgression from the Iberian
228 A. V. Stewart et al.

form. In contrast the single accession of metlesicsii expressed a greater similarity

with Iberian forms with three dual sequences (425, 571 and 601) indicative of some
introgression from the North African forms castellata and mairei. All forms in this
regions including metlesicsii share the same chloroplast trnL sequence (Stewart and
Ellison 2011).

28.4 Discussion

These results show that metlesicsii shares a greater molecular affinity with Spanish
forms of Dactylis in contrast to the lower altitude smithii which shares a greater
molecular affinity with adjacent North African forms of Dactylis.
This molecular result is consistent with the marked differentiation of phenolic
compounds found in high and low altitude tetraploid forms on Gran Canary Island
(Jay and Lumaret 1995). It is most probable that these tetraploids are based upon
their sympatric diploids as geneflow is usually from diploid to tetraploid in Dactylis
(Borrill and Lindner 1971; Lumaret and Barrientos 1990).
It is probable that the Canary Islands have been colonised by Dactylis forms from
both North Africa and the Iberian Peninsula, with the smithii African derivatives
predominating at low altitude and the metlesicsii Iberian derivatives dominating at
high altitudes. There are also indications that hybridisation has occurred between
these two forms, as could be expected, where they come into contact.
These results add the metlesicsii form to the overall phylogeny of diploid Dactylis
developed by Stewart and Ellison (2011) and provide a clear indication of historical
adaptive radiation within diploid Dactylis, from which the more common tetraploids
have developed. This knowledge will allow breeders to resynthesise many new and
unusual tetraploid forms not available in nature.

References

Borrill M, Lindner R (1971) Diploid-Tetraploid sympatry in Dactylis (Gramineae). New Phytol

70:1111–1124
Catalan P, Torrecilla P, Lopez Rodrıguez JA, Olmstead RG (2004) Phylogeny of the festucoid grasses
of subtribe Loliinae and allies (Poeae, Pooideae) inferred from ITS and trnL–F sequences. Mol
Phylogenet Evol 31:517–541
Ellison NW, Liston A, Steiner JJ, Williams WM (2006) Molecular phylogenetics of the clover genus
(Trifolium-Leguminosae). Mol Phylogenet Evol 39:688–705
Fiasson JL, Ardouin P, Jay M (1987) A phylogenetic groundplan of the specific complex Dactylis
glomerata. Biochem Syst Ecol 15:225–230
Jay M, Lumaret R (1995) Variation in the subtropical group of Dactylis glomerata L. 2. Evidence
from phenolic compound patterns. Biochem Syst Ecol 23:523–531
Lumaret R (1988) Cytology, genetics and evolution in the genus Dactylis. Crit Rev Plant Sci 7:55–91
Lumaret R, Barrientos E (1990) Phylogenetic relationships and gene flow between sympatric diploid
and tetraploid plants of Dactylis glomerata (Gramineae). Plant Syst Evol 169:1615–6110
28 Origins of Diploid Dactylis from the Canary Islands as Determined . . . 229

Parker PF, Borrill M (1968) Studies in Dactylis. 1 Fertility relationships in some diploid subspecies.
New Phytol 67:649–662
Schönfelder P, Ludwig D (1996) Dactylis metlesicsii (Poaceae), eine neue art der Gebirgsvegetation
von tenerife, kanariische inseln. Willdenowia 26:217–223
Stebbins GL, Zohary D (1959) Cytogenetic and evolutionary studies in the genus Dactylis. I mor-
phological, distribution and inter relationships of the diploid subspecies. Univ Calif publ bot
31:1–40 (California university press, Berkeley, Los Angeles p 40)
Stewart AV, Ellison NW (2011) The genus Dactylis. In: C. Kole (Ed) Wild crop relatives: genomic
and breeding resources, millets and grasses. Springer, New York, p 73–87
Taberlet P, Gielly L, Pautou G, Bouvet J (1991) Universal primers for amplification of three non-
coding regions of chloroplast DNA. Plant Mol Biol 17:1105–1109
Wetschnig W (1991) Karyotype morphology and some diploid subspecies of Dactylis glomerata L.
(Poaceae). Phyton Horn 31:35–55
Williams WM, Ansari HA, Ellison NW, Hussain SW (2001) Evidence of three subspecies in
Trifolium nigrescens Viv. Ann Bot London 87:683–691
Chapter 29
Introduction and Adaptation of Cynodon
L. C. Rich Species in Australia

M. C. Jewell, W. F. Anderson, D. S. Loch, I. D. Godwin and C. J. Lambrides

Abstract Cynodon L. C. Rich comprises warm season grass species that are valuable
as turf and forage in warm climates across the globe. Traditionally, Cynodon breed-
ing programs have aimed to produce varieties with improved turf and forage grass
quality characteristics. More recent goals, such as improved abiotic stress tolerance,
aim to address challenges associated with climate change. Germplasm collections
representing a fraction of the global distribution of Cynodon comprise the plant
resource for these breeding programs. We compared a core collection consisting of
116 Australian genotypes with an international core collection of 59 genotypes rep-
resenting Cynodon’s global distribution. Our aims were to determine whether unique
genetic diversity exists amongst Australian Cynodon that may enhance international
breeding resources. This research will facilitate the incorporation of Australian Cyn-
odon genetic resources, which have expanded and naturalized under harsh Australian
climates, into existing germplasm collections.

29.1 Introduction

Cynodon spp. comprise perennial, widely adapted, warm-season grasses that serve
multiple purposes in all continents of the world and are reported to have excellent heat
and drought tolerance (Taliaferro 2003; Zhou et al. 2009). Enormous morphological
variability exists within the genus (Harlan and de Wet 1969; Harlan et al. 1970).
Cynodon breeding programs aim to improve quality characteristics, such as yield,
vigour, and uniformity, as well as abiotic stress tolerance, which requires sampling
and characterization of germplasm collected from diverse environments around the
world.
There a currently nine recognised Cynodon species (Harlan et al. 1970). Cos-
mopolitan C. dactylon var. dactylon (2n = 4x = 36; common bermudagrass) is the

M. C. Jewell () · D. S. Loch · I. D. Godwin · C. J. Lambrides

School of Agriculture and Food Sciences, The University of Queensland,
Brisbane, QLD 4072, Australia
e-mail: [email protected]
W. F. Anderson
United States Department of Agriculture—Agriculture Research Station (USDA-ARS),
PO Box 748, Tifton 31793, GA, USA

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 231

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_29, © Springer Science+Business Media Dordrecht 2013
232 M. C. Jewell et al.

most widespread variety. Other cultivars include triploid hybrids produced by crosses
between C. dactylon var. dactylon and diploid species, such as C. transvaalensis
Burtt-Davy (Taliaferro 2003). Most Cynodon species apparently originated from
eastern and tropical areas across the southern half of Africa and are currently ac-
cepted as being introduced in the USA (Merril 1940), Canada (Macoun 1902), New
Zealand (Cheeseman 1925; Langdon 1954), and Australia. However, the question
of whether Cynodon spp. were present in Australia prior to European settlement has
not been resolved (Brown 1814 #56; Langdon 1954 #72).
Cynodon germplasm collection began in the early 20th Century (Taliaferro 2003)
and a number of collections have since been developed (Anderson and Wu 2007;
Wu 2011). Despite the interest in retention of Cynodon resources for breeding, the
extent of genetic diversity within these collections represents only a fraction of the
diversity that is available in the wild (Wu 2011). Further collection of the genus
across its current range is likely to sample genotypes that have evolved under stresses
associated with diverse environments.
The broad aim of this study was to assess whether Cynodon genetic diversity in
Australia will expand international breeding resources. Specific goals were to deter-
mine whether unique genetic diversity exists in Australia and to increase knowledge
of origins and introductions of Australian Cynodon.

29.1.1 Materials and Methods

29.1.1.1 Plant Materials, EST-SSR Genotyping and Core Collection

Development

The plant materials used in this study comprised a core collection of 116 genotypes
representing 1070 Cynodon genotypes from Australia (Australian core collection;
AC). Core collection selection was based on EST-SSR marker diversity, climate,
ploidy, morphology, environment type (managed or wild), and performance under
a range of abiotic stresses (Jewell et al. 2012). Compared to the source germplasm
(the germplasm present in the entire source population), the AC contains increased
gene diversity (H = 0.18 versus H = 0.14), > 95% of the total number of alleles,
and similar proportions of accessions representing specific climates, ploidy levels,
and environment type. DNA extraction procedures and EST-SSR analyses have been
reported previously (Jewell et al. 2010; Kearns et al. 2009; Zhou et al. 2009). DNA
from 59 international accessions representing global distribution was provided by the
USDA-ARS in Tifton, GA (international collection; IC; Anderson et al. 2009). These
accessions were a subset of a core collection of 600 accessions maintained at the Crop
Genetic and Breeding Research Unit (USDA-ARS Coastal Plain Experiment Station,
Tifton, GA, USA). This core collection was selected to maximize the phenotypic
and geographic diversity among these 600 accessions (Anderson 2005). The 59
accessions utilised in this analysis encompassed the geographic and putative species
diversity present in the international core collection.
29 Introduction and Adaptation of Cynodon L. C. Rich Species in Australia 233

Table 29.1 Sample size (n), Nei’s heterozygosity (H), number of alleles (A), number of geno-
types (G) and proportion of polymorphic bands (PPB) for the Australian core collection and the
international collection
Parameter Australian International
n 116 59
H 0.17 ± 0.03 0.18 ± 0.03
A 114 112
G 116 57
PPB 0.84 0.74

29.1.1.2 Genetic Variation and Population Structure

EST-SSR bands were scored as either present or absent and treated as dominant mark-
ers. Total number of alleles, rare alleles (alleles present in < 5% of individuals), and
private alleles (exclusive to either the AC or the IC) were calculated in Excel (2003).
Nei’s heterozygosity (H), number of genotypes (G) and proportion of polymorphic
bands (PPB) were calculated in R software version 2.12.1 (The R Development Core
Team 2005) using the aflpdat package (Ehrich 2006). Tess software, which defines
sub-populations (clusters; K) according to the likely number of ancestral genotypes
represented in the population, was used to infer population structure (Francois and
Durand 2010). Q values give proportional memberships of individuals to each clus-
ter. Ten runs of K = 2–20 were performed using the BYM admixture model with a
neighbour-joining tree as the starting configuration. Default values were used for all
parameters. The most likely K (Kmax ) was the K corresponding to the lowest mean
deviance information criterion (DIC). Clumpp (Jakobsson and Rosenberg 2009) was
used to permute Q matrices from ten runs of Kmax and bar plots were constructed
from permuted Q matrices using Distruct (Rosenberg 2004). Diva-gis software (Hi-
jmans et al. 2005) was used to generate maps showing the global distribution of the
accessions used in this study.

29.1.2 Results

29.1.2.1 Genetic and Allelic Variation

Genetic divergence between the IC and the AC was negligible (F ST = 0.039, sd =

0.009). Heterozygosity (H) and number of genotypes (G) were similar between the
AC and the IC (accounting for sample size differences; Table 29.1). However, PPB
was slightly higher in the AC than in the IC. The number of alleles detected was
similar between the AC (114) and the IC (112). Almost half the total number of
alleles across both collections were considered rare. Of these, eight were found to
be common in the AC but rare in the IC, while only five were found to be rare in
the AC but common in the IC. There were 16 alleles that were only present among
Australian genotypes and one of these was also common in Australia (Table 29.2).
234 M. C. Jewell et al.

Table 29.2 Total number of alleles, rare alleles, alleles that were rare in both the Australian and
international collections, alleles that were common in Australia but rare in the international collec-
tion, alleles that were rare in Australia but common in the international collection, alleles that were
private in Australia and alleles that were private in the international collection
Marker Alleles Rare Australian Australian Australian Rare/ International Australian
Alleles Rare/Interna- Common/Inter- International Private Private
tional Rare national Rare Common
Tri-63 19 12 3 0 2 5 2
Tri-69 6 3 0 1 1 1 0
Tri-72 7 2 0 0 1 2 1
Tri-58 6 2 1 0 0 1 0
Tri-76 7 4 0 0 0 1 3
Tri-10 8 4 3 0 0 0 1
Tri-87 4 2 1 0 0 0 1
Di-35B 2 1 1 0 0 0 1
Tri-79 12 7 6 1 0 1 0
Tetra-8 8 2 1 1 0 1 0
Tetra-3 7 2 1 0 0 1 0
Tri-91 12 4 1 3 1 0 2
Tri-17 8 6 4 1 0 0 1
Tri-56 12 6 4 0 0 1 1
Tri-74 2 0 0 0 0 0 0
Tri-88 8 4 1 1 0 0 3
Total 128 61 27 8 5 14 16

29.1.2.2 Population Structure

Five sub-populations were detected using Tess (Fig. 29.1a). The ratio of IC to AC
individuals was highest in clusters 1 and 2 and lowest in clusters 3 and 5 (Fig. 29.2).
Cluster 3 contained the highest proportion of admixture and was represented entirely
by IC individuals. However, one of these was an Australian accession of the IC.
Cluster 5 consisted almost exclusively of non-admixed IC individuals. The only
exception was one AC individual which shared < 50% (Q = 0.43) homology with
the cluster 5 ancestral genotype. All other AC individuals were found in clusters 1,
2 and 4. Sub-populations did not separate species into discrete clusters. The global
distribution of clusters is shown in Fig. 29.1b.

29.1.3 Discussion

This study showed that allelic diversity was similar in the Australian and international
collections. This does not indicate that a population bottleneck associated with intro-
duction and colonisation processes contributed to the current distribution of Cynodon
in Australia. Introduced populations arising from bottleneck events are usually char-
acterised by lower levels of genetic diversity compared with source populations. This
29 Introduction and Adaptation of Cynodon L. C. Rich Species in Australia 235

Fig. 29.1 Distribution of Cynodon in terms of (a) clusters identified using the spatially explicit
Bayesian population structure program TESS and (b) geographic proximity identified using DIVA-
GIS software with individuals colored according to TESS–inferred clusters

Cluster 5

Cluster 4

Cluster 3

Cluster 2

Cluster 1

Total

Australia International

Fig. 29.2 Ratio of Australian and international accessions in each of Five clusters

was the reported cause of reduced levels of genetic diversity in introduced popula-
tions of the invasive weed Rubus alceifolius Poir. (Rosaceae) compared with that
of populations representing the native range of this species (Amsellem et al. 2000).
Similar observations have reportedly been associated with genetic bottlenecks dur-
ing domestication of many crop species including sunflower, (Mandel et al. 2011),
maize (Vigouroux et al. 2005), soybean (Kuroda et al. 2010), and barley (Matus and
Hayes 2002). It is possible that multiple introductions have led to the distribution
of Cynodon in Australia and hybridisations between individuals arising from these
introductions have led to maintenance of genetic diversity within the genus.
The collection clustered into five sub-populations. However, these did not cor-
respond to distinct clusters of species. This may be explained by the extensive
multiplicity of Cynodon species and sub-species nomenclature, which has led to
enormous taxonomic confusion within the genus (The Plant List 2010). Previous
236 M. C. Jewell et al.

population structure analyses of the international collection have also found that
species did not form discrete clusters. This was reportedly due to mis-naming of ac-
cessions prior to Harlan et al.’s (1970) revision of the genus, contamination among
nursery plots since establishment of the collection in the 1940’s, and taxonomic
confusion during species classification (Anderson et al. 2009).
Despite these confusions, the population structure analysis, which shows that
Australian germplasm is largely restricted to only 3 of the 5 Tess clusters, suggests
that fewer ancestral populations contributed to the current distribution of Australian
than global germplasm. This implies that the origins of Australian Cynodon may
involve colonisation by species endemic to other countries. These findings will need
to be further investigated through analyses investigating the phylogenetic species
diversity of Cynodon in Australia.
The identification of rare and private alleles within Australian Cynodon shows that
there may be genetic diversity present in Australia that may not be represented in
current international breeding resources. Furthermore, the presence of alleles that are
common in Australia but rare in the international collection suggests that Cynodon
in Australia may be undergoing range expansion and adaptation (Barret and Schluter
2008). The presence of private, but not rare, alleles in Australia may also represent
adaptive responses involving mutations occurring since Cynodon’s introduction to
Australia. However, the more parsimonious explanation for these observations is
associated with the insufficient sample size of the international germplasm. There-
fore, these speculations will need to be confirmed through further analyses, such as
adaptation and association mapping studies.

Acknowledgments This work was funded by an Australian Research Council (ARC) linkage
project (EcoTurf ; ID: LP0775239). Financial support was also provided by the South East Queens-
land Council of Mayors, Jimboomba Turf, and the AW Howard Memorial Trust. We thank the
Central Glasshouse Services Unit, students and technical staff at the University of Queensland for
assistance in collection, establishment, and maintenance of plant collections.

References

Amsellem L, Noyer JL, Le Bourgeois T, Hossaert-Mckey M (2000) Comparison of genetic diversity

of the invasive weed Rubus alceifolius Poir. (Rosaceae) in its native range and in areas of
introduction, using amplified fragment length polymorphism (AFLP) markers. Mol Ecol 9:
443–455
Anderson MP, Wu YQ (2007) Fingerprinting of Bermudagrass DNA. USGA Turfgrass Environ Res
Online 6:1–8
Anderson WF (2005) Development of a forage bermudagrass (Cynodon sp.) core collection. Grassl
Sci 51:305–308
Anderson WF, Maas A, Ozias-Akins P (2009) Genetic variability of a forage bermudagrass core
collection. Crop Sci 49:1–12
Barret RDH, Schluter D (2008) Adaptation from standing genetic variation. Trends Ecol Evol
23:38–44
Brown R (1814) In: A voyage to terra australis. G. and W. Nichol, London, p 590
29 Introduction and Adaptation of Cynodon L. C. Rich Species in Australia 237

Cheeseman TF (1925) Manual of the New Zealand flora, 2 edn. N. Z. board of science and art,
Wellington
Ehrich D (2006) AFLPdat: a collection of R functions for convenient handling of AFLP data. Mol
Ecol Notes 6:603–604
Francois O, Durand E (2010) Spatially explicit Bayesian clustering models in population genetics.
Mol Eco Resour 10:773–784
Harlan JR, de Wet JMJ (1969) Sources of variation in Cynodon dactylon (L). pers. Crop Sci
9:774–778
Harlan JR, de Wet JMJ, Huffine WW, Deakin JR (1970) A guide to the species of Cynodon
(Gramineae), Bull B-673. Oklahoma Agricultural Experiment Station, Stillwater
Harlan JR, de Wet JMJ, Rawal KM, Felder MR, Richardson WL (1970) Cytogenetic studies in
Cynodon L. C. Rich. (Gramineae). Crop Sci 10:288–291
Hijmans RH, Guarino L, Cruz M, Rojas E (2005) DIVA-GIS, version 7.4.0.1. A geographic
information system for the management and analysis of genetic resource data. In
Jakobsson M, Rosenberg NA (2009) CLUMPP: a cluster matching and permutation program
for dealing with label switching and multimodality in analysis of population structure.
Bioinformatics 23:1801–1806
Jewell MC, Frere CH, Prentis PJ, Lambrides CL, Godwin ID (2010) Characterisation and mul-
tiplexing of EST-SSR primers in Cynodon (Poaceae) species. Am J Bot 97:e99-e101 (Primer
Notes and Protocols)
Jewell MC, Zhou Y, Loch DS, Godwin ID, Lambrides CJ (2012) Maximizing genetic, morpho-
logical, and geographic diversity in a core collection of Australian bermudagrass. Crop Sci
52:879–889
Kearns R, Zhou Y, Fukai S, Ye C, Loch D, Godwin ID, Holton T, Innes D, Stirling H, Cao N, Jewell
M, Lambrides CJ (2009) Eco-Turf: water use efficient turfgrasses from Australian biodiversity.
Acta Horticulturae 829:113–118
Kuroda Y, Kaga A, Tomooka N, Vaughan D (2010) The origin and fate of morphological inter-
mediates between wild and cultivated soybeans in their natural habitats in Japan. Mol Ecol
19:2346–2360
Langdon RFN (1954) The Origin and Distribution of Cynodon dactylon (L.) Pers. The University
of Queensland Papers, Department of Botany 4:41–44
Macoun J (1902) Catalogue of Canadian plants. In: Geological Survey of Canada Montreal, Quebec,
pp 1883–1892
Mandel J, Dechaine J, Marek L, Burke J (2011) Genetic diversity and population structure in
cultivated sunflower and a comparison to its wild progenitor, Helianthus annuus; L. Theor Appl
Genet 123:693–704
Matus IA, Hayes PM (2002) Genetic diversity in three groups of barley germplasm assessed by
simple sequence repeats. Genome 45:1095–1106
Merril ED (1940) Proceedings of the sixth Pacific Science Congress of the Pacific Science Associ-
ation In. University of California, Berkeley, Stanford University, and San Francisco, July 24th
to August 12th, 1939, pp 629–639
Rosenberg NA (2004) DISTRUCT: a program for the graphical display of population structure.
Mol Ecol Notes 4:137–138
Taliaferro CM (2003) Bermudagrass (Cynodon (L.) Rich). In: Casler MD, Duncan RR (eds)
Turfgrass Biology, Genetics, and Breeding. John Wiley and Sons, Inc., Hoboken, pp 235–256
The Plant List (2010) Available at: http://www.theplantlist.org/. In:
Vigouroux Y, Mitchll S, Matsuoka Y, Hamblin MT, Kresovich S, Smith JSC, Jaqueth J, Smith OS,
Doebley JF (2005) An analysis of the genetic diversity of New World maize races assessed by
DNA microsatellites. Am J Bot 95:1240–1253
Wu YQ (2011) Cynodon. In: Kole C (ed) Wild Crop Relatives: Genomic and Breeding Resources,
Millets and Grasses. Springer-Verlag, Berlin, pp 53–71
Zhou Y, Lambrides C, Kearns R, Ye CR, Cao N, Fukai S (2009) Selecting for drought tolerance
among Australian green couch grasses (Cynodon spp.). Crop Pasture Sci 60:1175–1183
Chapter 30
Variation in Traits Associated with Carbon
Sequestration for a Range of Common
Amenity Grass Species

S. J. Duller, J. MacDuff, D. Allen and R. Mathews

Abstract Carbon sequestration in the soil profile depends on a number of soil, plant,
climatic and management factors, and their interactions. Plants affect the quantity,
quality and placement of carbon in the soil profile. Ecological studies have demon-
strated considerable variation in soil organic matter (SOM) under different types
of grassland but the potential for exploiting genetic variation amongst grasses for
carbon deposition in the soil has not yet been explored. Neither has there been any
attempt to select and breed amenity or forage grass genotypes exhibiting enhanced
carbon sequestration.

30.1 Introduction

Carbon sequestration is the relatively long term storage of carbon in the stabilised
soil organic matter (SOM) fraction and depends on a number of soil, plant, climate
and management factors, and their interactions (Bardgett 2011). Besides constituting
one of the measures identified under article 3.4 of the Kyoto Protocol as necessary
for mitigation of global climate change, efforts to enhance carbon sequestration are
justified by the decline in SOM reported, irrespective of soil type and land-use, across
England and Wales (Bellamy et al. 2005).
Long term equilibrium levels of SOM are substantially higher in grassland soils,
especially permanent pasture, compared with arable cropped soils. Soussana et al.
(2004) calculated a figure of 25 t carbon ha−1 for the average difference in soil
organic carbon stock (0–30 cm) between temperate lowland cropland and pasture.
Consequently, options for enhancing carbon storage in grassland systems have so
far focused on conversion from temporary to permanent grassland and alterations
in N input (Soussana et al. 2004). In general, the kinetics of carbon accumulation
following a change in vegetation or management are non-linear: rapid change occurs

S. J. Duller () · J. MacDuff · D. Allen · R. Mathews

Institute of Biological Environmental and Rural Sciences, Aberystwyth University,
SY23 3EB, Aberystwyth, Ceredigion, Wales, UK
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 239

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_30, © Springer Science+Business Media Dordrecht 2013
240 S. J. Duller et al.

Table 30.1 List of varieties established, grouped by species

Lolium Festuca rubra Festuca rubra Festuca rubra Agrostis Deschampsia caespitosa
perenne ssp. littoralis ssp. commutata ssp. rubra capillaris Koeleria cristata
AberImp AberFlight Bl 3051 Adinda AberRoyal Abercampsia
Ace AberCharm Bargreen Boreal BarKing Barcampsia
Ba13516 AberGem Baroxi Cindy AberRegal
Bargold Barcrown Center Greenvie Heriot BarKoel
Bizet Barpearl Raisa Hollywood Lance –
Brightstar Cezanne – – Avalonc –
Cadix Count Pintora Aberfleeceb Vespac –
Greenfair Mocassin – Quatrob Highlandd –
a
F. longifolia
b
F. ovina
c
A. canina
d
A. castellana

during the early years followed by an asymptotic approach to a new equilibrium

value over many decades (Post and Kwon 2000).
A number of candidate traits associated with carbon sequestration can be identified
from experimental and modelling studies (Jones and Donnelly 2004). These can be
classified according to whether they affect:
a. Quantity, achieved through the balance between photosynthesis and respiration,
the net primary production.
b. Quality, through tissue composition, for example lignin content affecting the
rate of decay.
c. Placement of carbon in the soil profile, including leaf litter production, root
depth, turnover and decomposition.
The objective of this preliminary study is to assess variation in selected traits likely
to affect net carbon sequestration in amenity grasses, grown under field conditions
and receiving contrasting levels of fertiliser application.

30.2 Materials and Methods

The study utilised existing field-plots, established 18 months previously. In total

40 cultivars were grown representing five species groups, see Table 30.1. The trial
consisted of 1 m2 plots randomised in two split blocks replicated twice. From estab-
lishment onwards two blocks received zero fertiliser and two blocks 360 kg N/ha/year,
plus maintenance phosphate and potash in six applications. Prior to this trial begin-
ning, all plots were cut once a week throughout the growing season to a height of
30 mm, with the clippings removed.
Measurements were made over a ten week period. All plots were cut immediately
prior to the start and again after four and ten weeks, for dry matter yields. Following
the final cut, detailed measurements were confined to perennial ryegrass and strong
30 Variation in Traits Associated with Carbon Sequestration for a Range . . . 241

Table 30.2 Herbage dry

weights for five species at Species Cultivar Herbage Dry Weight (g/m2 )
final harvest. Means of Number
No Nitrogen Nitrogen
cultivars
Agrostis ssp 8 17.5 230.8
Festuca rubra ssp rubra 5 16.9 121.7
F. rubra ssp littoralis 8 17.3 175.2
F. rubra ssp commutata 5 14.0 112.9
Lolium perenne 8 3.0 23.0

creeping red fescue plots. This involved taking duplicate soil cores (0–30 cm depth)
from each plot, for separation into live shoot and root biomass. Leaf litter was also
collected from quadrats within the plots over the ten week period. Fractions were
oven dried and weighed.

30.3 Results

It is clear from this study that there are differences in above ground biomass between
species especially at the high nitrogen application level (Table 30.2). Detailed mea-
surements of leaf litter and root biomass were confined to the Lolium and Festuca

3.5

3.0

2.5

2.0
dry weight (g/quadrat)

1.5 -N
+N
1.0

0.5

0.0
e

t
p

r
d

x
ar
6

ze
Ac

ai
di
im

ol
51

tst
Bi

nf
Ca
rg
er

13

ee
igh
Ba
Ab

Ba

gr
Br

Fig. 30.1 Leaf litter production by cultivars of Lolium perenne over 10 week period
242 S. J. Duller et al.

7.0

6.0

5.0
dry weight (g/quadrat)

4.0

-N
3.0
+N
2.0

1.0

0.0
Adinda Boreal Cindy Greenvie Hollywood

Fig. 30.2 Leaf litter production by cultivars of Festuca rubra ssp rubra over 10 week period

rubra plots, and provide evidence of variation between cultivars (populations) of the
same species (Figs. 30.1–30.4). In particular the Lolium cultivars show variation with
and without nitrogen for leaf litter production, (Fig. 30.1). For the Festuca rubra cul-
tivars Adinda and Greenvie produced more leaf litter in the high nitrogen treatment,
but the cultivar ‘Cindy’ appeared to produce more leaf litter at low levels (Fig. 30.2).
The data for root mass was less conclusive and only Bizet and the stay green variety,
Ba 13516 stand out as having a greater root mass at higher nitrogen levels. There
were no significant differences between the varieties in the zero fertiliser treatment
for root mass for either species (Figs. 30.3 and 30.4).

30.4 Conclusion

The scope of this trial was to provide an initial comparison of potential carbon
sequestering traits across different species and genotypes. This variation has the
potential to be exploited to enhance the effectiveness of future grass varieties in the
storage of carbon in the soil.
The dynamics of carbon sequestration suggest that both the enhancement of short-
term carbon accumulation (0–5 years) and long-term equilibrium level of soil organic
carbon (> 50 years) are targets that warrant consideration with respect to genetic
improvement of amenity grasses. Enhancing the short-term rate of carbon accumu-
lation would benefit arable to grassland conversion schemes and the restoration of
brown-field sites.
30 Variation in Traits Associated with Carbon Sequestration for a Range . . . 243

4.5

3.5

2.5
-N
2
+N
1.5

0.5

0
e

t
p

ir
ar
6

ze
Ac
Im

di
ol
51

a
tst
Bi

nf
Ca
rg
er

13

ee
igh
Ba
Ab

Ba

Gr
Br

Fig. 30.3 Average Lolium perenne root material in soil cores

3 -N
+N
2

0
Adinda Boreal Cindy Greenvie Hollywood

Fig. 30.4 Average Festuca rubra ssp rubra root material in soil cores
244 S. J. Duller et al.

Candidate traits associated with carbon sequestration can be classified according

their effect and further detailed work could be targeted at
• Quantity, biomass produced above and below ground, including leaf/root turnover
• Quality, resistance to decomposition and chemical form, including C:N ratio
• Placement in the soil profile and the risk of loss through decomposition and release
of carbon
Future work will require measurement of SOM levels under field conditions over
many years to determine more thoroughly the variation within grass species and the
subsequent development of enhanced carbon sequestering varieties.

References

Bardgett RD (2011) Plant-soil interactions in a changing world. F1000 Biol Rep 3:16
Bellamy PH, Loveland PJ, Bradley RI, Lark RM, Kirk GJD (2005) Carbon losses from all soils
across England and Wales 1978–2003. Nature 473:245–248
Jones MB, Donnelly A (2004) Carbon sequestration in temperate grassland ecosystems and the
influence of management, climate and elevated CO2 . New Phytol 164(3):423–439
Post WM, Kwon KC (2000) Soil carbon sequestration and land-use change: Processes and potential.
Glob Change Biol 6:317–328
Soussana JF, Loiseau P, Vuichard N, Ceschia E, Balesdent J, Chavallier T, Arrouays D (2004) Carbon
cycling and sequestration opportunities in temperate grasslands. Soil Use Manag 20:219–230
Chapter 31
Suitability of Grass Species
for Phytoremediation of Soils Polluted
with Heavy-metals

G. Żurek, M. Pogrzeba, K. Rybka and K. Prokopiuk

Abstract Highly heavy metal (HM) polluted soil from non-ferrous mine and smelter
in Poland (further named as “Waryński”) were used in pot experiment for six grass
varieties. Growing media with three different amounts of “Waryński’ soil (0, 33 and
100 %) mixed with unpolluted field soil were used. The grass varieties showed sig-
nificant differences in ability to accumulate HM from soil. Performed measurements
proved proportional reduction of plant growth with increasing amount of polluted
soil in the growing medium: 100 % Waryński soil inhibited plant growth so strongly
that it was not useful for further analysis. Depending on the variety, it has been esti-
mated that using different grass varieties, it could be possible to extract: 2.6–10.2 g
of lead, 10.2–34.2 g cadmium and 250–2,562 g zinc, as calculated for total grass
yield from 1 ha.

Keywords Bioaccumulation · Lead · Cadmium · Zinc

31.1 Introduction

The term ‘phytoremediation’ refers to the environmental cleanup techniques, where

plants stabilize, extract or degrade water, air or soil pollutants. It offers a coast-
effective and environment-friendly alternative or complementary technology for
conventional remediation such as soil incineration or excavation and pump-and-treat
system (Pilon-Smith 2005).
Pollutants could be of organic (i.e. fuels, solvents, explosives, pesticides, herbi-
cides, chemicals etc.) or inorganic (salts of heavy metal ions) nature. Presence of
high levels of pollutants as heavy metals in soil is stress factor for plants, and may
produce deleterious effects on many physiological processes (Seregin and Ivanov
2001). Finally, plants reduce growth, yield and generative reproduction.

G. Żurek () · K. Rybka · K. Prokopiuk

Plant Breeding & Acclimatization Institute,
National Research Institute, 05-870 Radzików, Błonie, Poland,
e-mail: g.zurek@ihar. edu.pl
M. Pogrzeba
Institute for Ecology of Industrial Areas, 40-844 Katowice, Poland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 245

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_31, © Springer Science+Business Media Dordrecht 2013
246 G. Żurek et al.

Soils contaminated by inorganic pollutants are serious problem, and it is esti-

mated that ca. 1.5 mln places in Europe need to be remediated, from which 300,000
pose a potential threat to surroundings (Kucharski et al. 2005). After sufficient plant
growth and metal accumulation, plants are harvested and removed, resulting removal
of metals from polluted site. Despite HM uptake, plants cover the soil and erosion
and pollutants leaching will thus be reduced (Chhotu and Fulekar 2009). Biomass
removed from polluted areas should be used only for non-food purposes, i.e. conver-
sion to energy. In some cases metals can be extracted from the HM-rich ash and serve
as a source of revenue, thereby offsetting the expense of remediation (Ghosh and
Singh 2005). Variety of techniques and number of plants species need to be evaluated
for cleanup of polluted areas. From many species examined for phytoremediation of
HM, trees and grasses seems most interesting. Trees, such as Salix are high biomass
producers, and also have effective nutrient uptake, high evapotranspiration rate and a
pronounced clone specific capacity for HM uptake (Pulford and Watson 2003). But,
due to its high water consumption and specific harvest equipment, its practical use
is limited to areas, located in the vicinity of water on large and flat fields. But this
is not always in the case of polluted areas. Therefore, we tested grass species with
relatively high biomass yields and low site and harvest requirements. The aim of our
studies was to evaluate the suitability of grass species for phytoremediation of HM
polluted soils by evaluation of the effect of HM ions contamination on plant growth
and determination of Pb, Cd and Zn ions concentration in biomass.

31.2 Materials and Methods

Six grass varieties were selected for pot studies: tall fescue (Festuca arundi-
nacea Schreb.) ‘Terros’ and ‘Rahela’, reed canary grass (Phalaris arundinacea L.)
‘Keszthelyi’, tall oat grass (Arrhenatherum elatius J. et C. Presl.) ‘Wiwena’, tall
wheatgrass (Elytrigia elongata (Host) Nevski)‘Bamar’ and switchgrass (Panicum
virgatum L.) ‘Shelter’. Metal pots (18.8 dm3 ) were filled with soil mixture com-
posed of typical field soil and highly HM-polluted soil from non-ferrous mine and
smelter in Poland (further named as “Waryński” soil). Different amounts (0, 33 and
100 %) of “Waryński” soil were used. Mentioned soil was previously used in other
experiments, as source of man-made, HM contamination (Kucharski et al. 2005;
Japenga et al. 2007). Soil mixtures were further analyzed for pH, soil organic matter
(SOM) contents and concentration of lead, cadmium and zinc (Table 31.1).
Grass seed was sown in April of 2010 and aboveground plant biomass was har-
vested at the end of the season. Considering low quantity of biomass harvested from
100 % “Waryński” soil, biomass for analysis of concentrations of Pb, Cd and Zn
was harvested from pots with 0 and 33 % of “Waryński” soil. Bioaccumulation fac-
tor (BAF) was expressed as ratio of concentration of HM in plant biomass to the
concentration of mentioned chemicals in soil.
31 Suitability of Grass Species for Phytoremediation of Soils Polluted with Heavy-metals 247

Table 31.1 Selected chemical parameters of soil mixtures used in experiment

Amount of “Waryński” Soil mixtures characteristics
soil in mixture (%)
pH KCl HM concentration (ppm) SOM
(%)
Pb Cd Zn
0 6.1 7.7 0.2 17.7 1.00
33 6.3 659.0 36.5 1966.0 1.80
100 6.4 1412.0 90.2 4685.0 3.63

Table 31.2 Dry matter yields, concentration of HM and BAF of tested grass varieties
Species and “Waryński” DM yield HM BAF
variety name soil in (g/plant) in biomass (ppm)
mixture (%)
Pb Cd Zn Pb Cd Zn
Festuca 0 17 0.73 0.19 19.5 0.095 1.188 1.102
arundinacea 33 7.2 0.52 4.96 95.750.00078 0.136 0.049
RAHELA 100 2 – – – – – –
Festuca 0 9.1 1.65 0.31 30.8 0.216 1.906 1.74
arundinacea 33 7.9 0.99 2.64 85.85 0.00149 0.072 0.044
TERROS 100 1.5 – – – – – –
Elytrigia 0 10.5 0.25 0.14 18.45 0.033 0.875 1.042
elongata 33 7.8 1.2 1.77 42.6 0.00182 0.048 0.022
BAMAR 100 1.5 – – – – – –
Phalaris 0 6.1 0.38 0.1 34.5 0.049 0.625 1.949
arundinacea 33 7.7 0.59 1.49 259.0 0.0009 0.041 0.132
KESZTHELYI 100 2.4 – – – – – –
Arrhenatherum 0 5.4 0.31 0.1 20.45 0.041 0.594 1.155
elatius 33 7.4 0.68 2.47 71.25 0.00102 0.068 0.036
WIWENA 100 2.5 – – – – – –
Panicum 0 13.7 0.22 0.05 13.9 0.028 0.281 0.785
virgatum 33 12.7 0.54 0.75 44.45 0.00082 0.020 0.023
SHELTER 100 2.4 – – – – – –
Mean 0 10.3 0.59 0.15 22.93 0.077 0.911 1.296
33 8.45 0.75 2.34 99.82 0.00114 0.064 0.051
100 2.0 – – – – – –
LSDα = 0.05 0 3.42 0.649 0.081 6.233 0.085 0.507 0.352
33 3.61 0.423 0.414 17.45 0.0006 0.0113 0.0089
100 > 0.001 – – – – – –

31.3 Results and Discussion

Reduction of plant growth, finally expressed in decreasing values of dry matter

of plants was proportional to the increase of HM concentration in soil mixture
(Table 31.2).
Despite of high concentration of Pb in polluted soil mixture, its concentration in
plant biomass was low. It was only 28 % more than in plants grown in unpolluted
soil. It has been mentioned by Kucharski et al. (2007) that Pb bioavailability in
248 G. Żurek et al.

“Waryński” soil was relatively low as compared to Cd or Zn. It has been shown in
our results that Cd and Zn concentration in plant biomass grown on polluted soil
was 1,507 and 335 %, respectively, higher than concentration in plant biomass from
unpolluted soil. Bioaccumulation factors (BAF) for plants grown on polluted soil
were lower than grown on unpolluted soil. Such relation for Zn and Cd has also been
reported by Baran and Jasiewicz (2009). Values of BAF for zinc from polluted soil
was close related to BAF for unpolluted soil (r = 0.80, α < 0.001). It is therefore
possible that tested varieties exposed some genetic properties for accumulation of
zinc, which were not so clear in case of cadmium (r = 0.53, α < 0.05) and lead
(r = 0.24, ns). HM may also interact—for example, Zn2+ ions may counteract Cd2+
absorption (Seregin and Ivanov 2001).
The highest concentrations of HM as well as BAF-s for Pb, Cd and Zn were
noted for E.e. ‘Bamar’, F.a. ‘Rahela’ and P.a. ‘Keszthelyi’, respectively. Considering
reduction of biomass yields on HM polluted soils, it is possible to extract from 2.6 g
(F.a. ‘Rahela’) to 10 g (F.a. ‘Terros’) of lead, from 10 g (P.v. ‘Shelter’) to 34 g (A.e.
‘Wiwena’) of cadmium and from 320 g (E.e. ‘Bamar’) to 2,562 g (P.a. ‘Keszthelyi’)
of zinc in biomass harvested from 1 ha.
Grass species used in this study are not the highest effective HM accumulators. For
example in case of willow, cadmium accumulation more than 200 g per hectare per
year was reported (Pore˛bska and Ostrowska 1999). But in case of phytoremediation,
research should focus not only on the absolute amount of HM yearly extracted but
also on diversity of species used for mentioned process, due to relatively long time
scale and numerous, currently unpredictable factors involved in future success.

References

Baran A, Jasiewicz C (2009) Toksyczna zawartość cynku i kadmu w glebie dla różnych gatunków
ros̀lin. Ochrona Środowiska i Zasobów Naturalnych 40:157–164
Chhotu DJ, Fulekar HM (2009) Phytoremediation of heavy metals: recent techniques. African J
Biotechnol 8(6):921–928
Ghosh M, Singh S P (2005) A review on phytoremediation of heavy metals and utilization of its
byproducts. Appl Ecol Environ Res 3(1):1–18
Japenga J, Koopmans GF, Song J, Romkens PFAM (2007) A feasibility test to estimate the duration
of phytoextraction of heavy metals from polluted soils. Int J Phytorem 9:115–132
Kucharski R, Sas-Nowosielska A, Małkowski E, Japenga J, Kuperberg JM, Pogrzeba M, Krzyżak J
(2005) The use of indigenous plant species and calcium phosphate for the stabilization of highly
metal-polluted sites in southern Poland. Plant and Soil 273:291–305
Pilon-Smith E (2005) Phytoremediation. Annu Rev Plant Biol 56:15–39
Pulford ID, Watson C (2003) Phytoremediation of heavy metal-contaminated land by trees—a
review. Environ Int 29:529–540
Pore˛bska G, Ostrowska A (1999) Heavy metal accumulation in wild plants: implications for
phytoremediation. Polish J Environ Stud 8(6):433–442
Seregin IV, Ivanov VB (2001) Physiological aspects of cadmium and lead toxic effects on higher
plants. Russian J Plant Physiol 48:523–544
Chapter 32
Targeting Lucerne Cultivars to Saline-soil
Environments

L. Pecetti, P. Annicchiarico, L. De Rosa and S. Proietti

Abstract Lucerne (Medicago sativa L.) is rated as moderately susceptible to soil

salinity, a major stress whose incidence is increasing in drought-prone regions
because of irrigation with saline water. Three-year dry-matter yield (DMY) of 14 pop-
ulations (cultivars or landraces) of different origin was assessed across ten agricultural
environments of Algeria, Italy, Morocco and Tunisia whose electrical conductivity
(ECe ) in the 0–30 cm soil layer ranged between 0.2 and 6.0 dS m−1 . The adaptive re-
sponses of the populations were modeled by factorial regression as a function of ECe
and one or two additional significant covariates. Ameristand 801S (a benchmark
variety for salt tolerance) and the Moroccan landrace Erfoud 1 displayed positive
genotype × environment (GE) interaction in environments with high soil salinity.
Nine populations and one additional Algerian landrace (Tamantit) underwent a seed
germination test under saline conditions, to verify the ability of this test to predict the
field-based adaptive responses to soil salinity of the populations. The salt concen-
tration required to inhibit germination of 50 % of viable seeds (IC(50)) ranged from
1.13 % NaCl of Prosementi to 2.61 % NaCl of Ameristand 801S. IC(50) values of
the populations displayed moderately high correlation (r = 0.70; P < 0.03) with the
genotype slope as a function of ECe in the factorial regression (β ECe ). Salt-tolerant
genetic resources could be found in north African landraces which evolved in less
favourable oasis environments, such as Tamantit and Erfoud 1.

Keywords Factorial regression analysis · Genotype × environment interaction ·

Germination test · Medicago sativa · Salt tolerance

32.1 Introduction

Lucerne (Medicago sativa L.) is the main perennial forage legume grown in the
Mediterranean basin and other semi-arid areas. Soil salinity is a major stress affect-
ing crop production in drought-prone regions whose incidence is increasing because
of irrigation with saline water. Lucerne is rated as moderately susceptible to soil

L. Pecetti () · P. Annicchiarico · L. De Rosa · S. Proietti

Centro di Ricerca per le Produzioni Foraggere e Lattiero-Casearie (CRA-FLC), Lodi, Italy
e-mail: [email protected]

S. Barth, D.Milbourne (eds.), Breeding Strategies for Sustainable Forage 249

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_32, © Springer Science+Business Media Dordrecht 2013
250 L. Pecetti et al.

salinity, with 30 % yield reduction occurring under salinity that corresponds to a

soil electrical conductivity (ECe ) of 6.1 dS m−1 (Maas 1986). An increase of salt
tolerance through selection is feasible, as genetic variation exists for this trait among
populations (Smethurst et al. 2008). Narrow-sense heritability estimates for this trait
were reportedly between 0.23 and 0.61 depending on the germplasm (Al-Khatib
et al. 1994). Salt tolerance is a complex trait involving the integrated contribution
of several physiological and biochemical mechanisms at various levels of the plant
structure. Smethurst et al. (2008) identified two major mechanisms of tolerance in
lucerne, one based on sodium exclusion from leaves and the other implying effective
sodium compartmentation in cell vacuoles. Early screening for salt tolerance has
been proposed by means of seed germination testing under saline condition, but the
consistency of its results with those for the adult plant is still controversial (John-
son et al. 1992; Smethurst et al. 2008). The North American Alfalfa Improvement
Conference (NAAIC) has defined a standard test for assessing cultivar salt tolerance
from seed germination under saline conditions (Rumbaugh 1986).
The main objective of this study was to verify whether field-based adaptive re-
sponses to soil salinity of lucerne populations could partly be predicted just by
screening for seed germination under saline conditions. As a first step of this inves-
tigation, three-year dry-matter yield (DMY) of 14 lucerne populations was assessed
across ten agricultural environments which varied widely for soil salinity as expressed
by soil ECe values. Factorial regression modeling of the population responses, whose
results are reported in detail elsewhere (Annicchiarico et al. 2011), indicated soil ECe
as one of the main environmental variables affecting the site-specific adaptation of
the populations as expressed by their genotype × environment (GE) interaction for
DMY. Field-based salt tolerance of the populations was estimated by their regression
parameter for response to soil ECe .

32.2 Materials and Methods

Fourteen populations, including six farm landraces from north Africa and Italy and
eight commercial varieties from Italy, France, Australia and the USA (Table 32.1),
were evaluated for three-years in ten agricultural environments encompassing six
locations of Algeria, Italy, Morocco and Tunisia. Crop management in the ten envi-
ronments implied rainfed conditions, irrigation with 9-week summer suspension, or
continuous irrigation (Annicchiarico et al. 2011). Soil electrical conductivity in the
0–30 cm soil layer ranged between 0.2 and 6.0 dS m−1 across environments. Two
locations (Alger in Algeria, and Médenine in Tunisia) approached the threshold of 2
dS m−1 for salinity damage, while the Algerian location of Hmadna featured fairly
high soil salinity (5.2 and 6.0 dS m−1 in its rainfed and irrigated environments, re-
spectively). Each experiment was designed as a randomized complete block design
(RCBD) with four replications. DMY was recorded over three-years on a 2.4 m2
harvest area of the 5 m2 large plots. The total number of harvests varied depending
on the growth potential of the environment, ranging between nine (Mateur rainfed in
32 Targeting Lucerne Cultivars to Saline-soil Environments 251

Table 32.1 Name, code, germplasm type, origin, code, slope of genotype × environment interaction
for dry-matter yield as a function of soil electrical conductivity (ECe ), and NaCl concentration
required to inhibit germination of 50 % of viable seeds (IC(50)), for 14 lucerne cultivars evaluated
in 10 Mediterranean environments and 10 populations tested for seed germination under saline
conditions
Population Code Type Origin β ECe (dS m−1 ) IC(50) (% NaCl)a
ABT 805 Ab Com SE USA −0.005 –
Ameristand 801S Am Com SW USA 0.713* 2.306 a
Coussouls Co Com S France 0.172 –
Demnat 203 De Lan Morocco 0.114 –
Erfoud 1 Er Lan Morocco 0.575† 1.588 bc
Gabès 2355 Ga Lan Tunisia 0.329 1.367 d
Magali Mg Com W France −0.369 1.200 ef
Mamuntanas Ma Lan Sardinia −0.082 1.202 e
Melissa Me Com S France 0.110 –
Prosementi Pr Com N Italy −0.696* 1.130 f
Rich 2 Ri Lan Morocco −0.156 1.433 cd
SARDI 10 Sa Com S Australia −0.608* 1.303 de
Sicilian ecotype Sc Lan Sicily 0.257 1.355 d
Siriver Si Com S Australia −0.353 –
Tamantit Ta Lan Algeria – 1.785 b
Com commercial variety, Lan farm landrace
†
, *: regression slope different from zero at P < 0.06 and P < 0.05, respectively; data from
Annicchiarico et al. (2011)
a
Values followed by same letter do not differ at P < 0.05 according to Probit analysis

Tunisia) and 23 (Oued Tessaout under continuous irrigation in Morocco). GE interac-

tion effects for DMY were modeled by factorial regression (Denis 1988). The slope
of GE interaction for DMY as a function of ECe (β ECe ) in the factorial regression
model estimated the adaptation to saline-soil conditions of each population. Positive
β ECe values indicated positive GE interaction for DMY in saline-soils environments
and, hence, salt tolerance of the population.
A subset of nine populations, including four with contrasting β ECe value, and one
additional Algerian landrace (Tamantit), were chosen for the following assessment
of salt tolerance based on seed germination under saline conditions (Table 32.1). We
adopted the standard NAAIC test (Rumbaugh 1986), which contemplates seed ger-
mination in each of eight concentrations of NaCl: 0.00, 0.50, 0.75, 1.00, 1.25, 1.50,
1.75 and 2.00 % (wt/wt) in deionised water. The eight concentrations corresponded
to 0.0, 85.6, 128.3, 171.1, 213.9, 256.7, 299.4 and 342 mM NaCl solutions, or to 0.0,
8.3, 11.6, 15.5, 17.2, 23.1, 26.7 and 30.0 dS m−1 electrical conductivity. Differently
from the protocol by Rumbaugh (1986), 30 scarified seeds per Petri plate were ger-
minated instead of 25, and three replications in a RCBD were used instead of two.
The plates were placed in the dark in a germination cabinet maintained at 25 ◦ C,
counting germinated seeds after seven days. Germination data, corrected for possi-
ble hard seeds, were analysed by Probit analysis. Population mean values and the
respective 95 % confidence intervals were estimated for the NaCl concentration (%)
252 L. Pecetti et al.

required to inhibit germination of 50 % of viable seeds (IC(50)). The correlation

between β ECe and IC(50) values of the populations was assessed for data of the
nine populations in common between field evaluation and germination test.

32.3 Results and Discussion

Population and environment main effects and GE interaction for total DMY across
the ten agricultural environments were significant (P < 0.01). The highest mean
yield (48.12 t ha−1 ) was recorded in Oued Tessaout under continuous irrigation,
and the lowest (12.16 t ha−1 ) in Hmadna rainfed, where the crop was exposed to
severe drought and salinity stress. Environment mean yield was closely correlated
with annual and spring-summer water availability, and tended to positive association
with milder winter temperatures, lower soil salinity and higher number of harvests
(Annicchiarico et al. 2011).
The best factorial regression model included number of harvests, soil electrical
conductivity and spring-summer (April-September) water available for the crop as
significant (P < 0.05) environmental covariates. It explained 53 % of GE interaction
variation for DMY. A simpler model including only spring-summer water available
and soil electrical conductivity accounted for 42 % of the GE interaction sum of
squares (Annicchiarico et al. 2011).
The two factorial regression models provided similar estimates of population salt
tolerance as expressed by their β ECe parameter. The β ECe values for the three-
covariate model which are reported in Table 32.1 were used for correlation with
data of seed germination under saline water. Ameristand 801S was the population
featuring the highest salt tolerance on the basis of its high positive β ECe value
(implying large positive GE interaction in environments with high soil salinity).
Indeed, this cultivar has been considered as a benchmark variety for salt tolerance
in lucerne (Smethurst et al. 2008). Also the Moroccan landrace Erfoud 1 showed
positive β ECe value (P < 0.06), unlike the remaining cultivars and particularly
Prosementi and SARDI 10, which were highly susceptible on the basis of their
negative β ECe value (Table 32.1).
The two factorial regression models allowed to scale up to other environments the
indications on site-specific, top-yielding populations. In particular, they highlighted
the outstanding adaptation of Ameristand 801S to saline-soil environments, as well
as the specific adaptation of Erfoud 1 to saline-soil environments without severe
drought stress (Annicchiarico et al. 2011).
Four populations with contrasting β ECe value, namely, Ameristand 801S, Erfoud
1, Prosementi and SARDI 10, and five populations with intermediate field response to
saline-soil environments, underwent the seed germination test under saline conditions
along with the Algerian landrace Tamantit. The IC(50) values of the populations
ranged from 1.13 % of Prosementi to 2.61 % of Ameristand 801S (Table 32.1). The
latter cultivar confirmed also here its outstanding salt tolerance, which was obtained
through several cycles of phenotypic selection for this trait and is reportedly related
32 Targeting Lucerne Cultivars to Saline-soil Environments 253

to a sodium exclusion mechanism (Smethurst et al. 2008). According to indications

by Rumbaugh (1986), cultivars with IC(50) greater than about 1.6 % NaCl should
be considered as salt tolerant, and those with IC(50) lower than about 1.45 % NaCl
as salt susceptible. In the current investigation, Ameristand 801S and the Algerian
landrace Tamantit exceeded the threshold value for stress tolerance, and Erfoud 1
was just near it. All other populations had IC(50) values indicating susceptibility,
with the Moroccan landrace Rich 2 being close to the threshold value of 1.45 %
(Table 32.1).
The correlation between the factorial regression coefficient β ECe and IC(50) of
the nine common populations was moderately high, namely, r = 0.70 (P < 0.03),
supporting the adoption of the germination test as a rapid means to roughly predict
the adaptation to saline-soil conditions of lucerne populations. In contrast, Johnson
et al. (1992) found that selection based on seed germination in saline water was not
correlated with forage yield in saline soils of adult plants. However, as also inferred
from Smethurst et al. (2008), the concentration of the 80 mM NaCl solution applied
for the screenings by Johnson et al. (1992) was likely insufficient for an efficient
discrimination between tolerant and susceptible germplasm.
The adaptation to salt stress of Erfoud 1 and Tamantit may be related to the spe-
cific conditions of the oasis environments in which these landraces evolved. Our
results suggest that useful lucerne germplasm for salt-tolerance breeding might still
be substantially untapped. Novel sources of tolerance could provide a valuable ge-
netic base for the improvement of the crop, or could be used in target crosses between
germplasms characterized by different tolerance mechanisms, once their tolerance
mechanisms were revealed.

Acknowledgments The research was carried out within the EU-funded project PERMED (INCO-
CT-2004-509140). This work is dedicated to the lovely memory of Loredana De Rosa.

References

Al-Khatib M, McNeilly T, Collins JC (1994) The genetic basis of salt tolerance in lucerne (Medicago
sativa L.). J Genet Breed 48:169–174
Annicchiarico P, Pecetti L, Abdelguerfi A, Bouizgaren A, Carroni AM, Hayek T, M’Hammadi Bouz-
ina M, Mezni M (2011) Adaptation of landrace and variety germplasm and selection strategies
for lucerne in the Mediterranean basin. Field Crops Res 120:283–291
Denis JB (1988) Two-way analysis using covariates. Statistics 19:123–132
Johnson DW, Smith SE, Dobrenz AK (1992) Genetic and phenotypic relationships in response to
NaCl at different developmental stages in alfalfa. Theor Appl Genet 83:833–838
Maas EV (1986) Salt tolerance of plants. Appl Agric Res 1:12–26
Rumbaugh MD (1991) Salt tolerance of germinating alfalfa seeds. In: Fox CC, Berberet R, Gray
FA, Grau CR, Jessen DL, Peterson MA (eds) Standard tests to characterize alfalfa cultivars, 3rd
edn. NAAIC, Beltsville, MD, USA, pp A-3. Available online at http://www.naaic.org. Accessed
31 Aug 2011
Smethurst CF, Rix K, Garnett T, Auricht G, Bayart A, Lane P, Wilson SJ, Shabala S (2008) Multiple
traits associated with salt tolerance in lucerne: revealing the underlying cellular mechanisms.
Funct Plant Biol 35:640–650
Chapter 33
Comparison of Seed Mixtures for Technical
Revegetation at High Altitude

P. Spoleto, A. Tosca, G. Della Marianna, F. Gusmeroli, L. Pecetti

and M. Romani

Abstract With the aims of improving the success of sowings, enhancing the biodi-
versity, and reducing the impact of interventions (both in terms of energy input and
genetic pollution), the use of seed mixtures including ‘site-specific’species is increas-
ingly recommended for revegetation interventions at high altitude. This germplasm
is ecologically adapted to the prevailing pedoclimatic conditions and native to the
target region. In this three-year study performed in Alpe Palù (Chiesa in Valmalenco,
Sondrio, Italy) at 2,020 m a.s.l. elevation, ground cover and botanical composition
of native seed mixtures were compared versus those of commercially available mix-
tures. Both a simpler and a more complex version (with greater number of species)
were evaluated for both native and commercial mixtures. The commercial mixtures
colonised faster the bare soil in the year following the sowing, but in the subsequent
years the native mixtures reached over 80 % of ground cover being no longer differ-
ent from the former ones. The complexity of mixture had only slight effects on the
soil cover. A relative prevalence of grasses was evident in the commercial mixtures,
and of non-legume dicots in the native ones. The better botanical balance featured
by the native mixtures is likely to result in a more-lasting stability of the covers and
an increase of biodiversity in the long term.

Keywords Alps · Ecological restoration · ‘Site-specific’ species

33.1 Introduction

Mountain areas altered by infrastructures for winter tourism need revegetation to re-
store the landscape appearance as well as preventing environmental damages due to
water run-off and soil erosion (Florineth 1992; Krautzer et al. 2001; Peratoner 2003).

F. Gusmeroli () · G. Della Marianna

Fondazione Fojanini di Studi Superiori, via Valeriana 32, 23100 Sondrio, Italy
e-mail: [email protected]
P. Spoleto · A. Tosca
Fondazione Minoprio, viale Raimondi 54, 22070 Vertemate con Minoprio, CO, Italy
L. Pecetti · M. Romani
Centro di Ricerca per le Produzioni Foraggere e Lattiero-Casearie (CRA-FLC)
viale Piacenza 29, 26900 Lodi, Italy

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 255

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_33, © SpringerScience+Business Media Dordrecht 2013
256 P. Spoleto et al.

Table 33.1 Composition and species complexity of the five seed mixtures sown at high altitude
Seed mixture Grasses Legumes Other dicots
Local, simple (A)
Composition (%) 76 18 6
No. of species 4 3 2
Local, complex (B)
Composition (%) 76 18 6
No. of species 8 3 2
Commercial, simple (C)
Composition (%) 80 20 0
No. of species 4 3 0
Commercial, complex (D)
Composition (%) 88 11 1
No. of species 9 5 1
Commercial, site-specific (E)
Composition (%) 82.3 16.5 1.2
No. of species 7 5 2

Species and varieties selected for lowland turf or forage purposes are often used for
sowing at high altitude because of their large availability of seed at reasonable price.
However, this germplasm may prove unsuitable to the peculiar environmental condi-
tions, while requiring costly maintenance (as fertilisation and mowing) and affecting
the local biodiversity (Peratoner 2003). To improve the success of sowings, enhanc-
ing the biodiversity, and reducing the impact of interventions, the use of seed mixtures
including ‘site-specific’ species is increasingly recommended (Krautzer et al. 2004).
This germplasm is native to the same environmental context and, therefore, adapted to
the prevailing pedoclimatic conditions and compatible with the existing ecosystems.
In the current study, seed mixtures obtained from local populations of site-specific
species were evaluated on a high-altitude ski slope of the Italian Alps, and com-
pared with commercial mixtures including either site-specific or non-site-specific
germplasm. Aim of the work was to assess the behaviour of the mixtures in terms of
ground cover and contribution to the site biodiversity.

33.2 Materials and Methods

The experiment was established in Alpe Palù, Chiesa in Valmalenco (Rhaetian Alps,
Italy) at 2,020 m a.s.l. elevation. Seed of site-specific species was obtained from
native populations collected in three valleys of the same mountain district, viz.
Valchiavenna, Valmalenco and UpperValtellina, and two mixtures of different species
complexity were assembled with this germplasm (coded as A and B in Table 33.1).
The local seed mixtures were evaluated together with three commercial mixtures.
Two of them (coded as C and D) were also characterised by different complexity and
mostly included non-site-specific species and varieties. The third one was a mixture
33 Comparison of Seed Mixtures for Technical Revegetation at High Altitude 257

Table 33.2 Mixture (over 3 test years) and year mean values of ground cover and botanical family
composition of cover
Ground cover (%) Family contribution to cover (%)
Grasses Legumes Other dicots
Seed mixture
Local, simple (A) 75.3 34.0 38.6 16.7
Local, complex (B) 72.6 31.2 37.0 20.0
Commercial, simple (C) 84.1 47.3 48.5 0.3
Commercial, complex (D) 84.6 39.8 43.5 10.8
Commercial, site-specific (E) 83.4 47.6 37.8 6.4
Year
2007 75.6 b 35.7 c 32.7 b 7.1 b
2008 75.9 b 44.7 a 44.1 a 11.1 a
2009 88.5 a 39.5 b 46.3 a 14.2 a
ANOVA probability
Mixture (M) <0.05 <0.05 ns <0.001
Year (Y) <0.05 <0.01 <0.001 <0.01
Block (B) ns <0.05 ns ns
M ×Y <0.05 <0.05 <0.01 <0.001
M×B ns ns <0.01 <0.05
Means followed by different letters differ at P < 0.05 according to Duncan’s test
ns not significant (P > 0.05)

complex mostly including site-specific germplasm (code E). The seed rate adopted
for mixtures C and D (30 g m−2 ) was within the common range of 20–50 g m−2 for
revegetation of ski slopes (Peratoner 2003). The site-specific seed mixtures were
sown at lower rate (15 g m−2 ) as suggested by previous experiences with this kind of
germplasm (Krautzer 2005). Hand sowing took place on November 15, 2006 and was
facilitated by applying 6 g m−2 of a commercial polyvinyl acetate water suspension
as bonding material to the seed. Plots were 2 × 2 m and arranged in a randomised
complete block design with four replications. Data were recorded in August of each
of the three years following that of sowing, visually estimating the percentage of
ground cover per plot and the contribution to cover of botanical families as grasses,
legumes and other dicots. An analysis of variance (ANOVA) was carried out accord-
ing to a three-way mixed factorial design, with block and year as random factors and
mixture as fixed factor. The latter factor was tested roughly on a combination of vari-
ances, as suggested by Cochran (1951). Single year data were submitted to two-way
ANOVA. Means were compared by Duncan’s multiple range test at P ≤ 0.05.

33.3 Results and Discussion

A ground cover close to 90 % was reached across all mixtures by the third year
of evaluation (Table 33.2). The local site-specific seed mixtures had lower ground
cover percentage over the three test years compared with the commercial mixtures
258 P. Spoleto et al.

(Table 33.2). A significant mixture × year interaction for this character was ob-
served, though (Table 33.2). The commercial mixtures colonised faster the bare
soil, with a covering percentage in the first year exceeding 75 versus about 65 % of
the native mixtures. In the subsequent years, however, the commercial mixtures sub-
stantially maintained the initial cover, while the native ones steadily spread, reaching
85 % of soil cover by the third year, being no longer different from the former ones
(Fig. 33.1a). Although sown at a halved seed rate compared with the other commer-
cial mixtures, the mixture E based on site-specific species showed similar ground
cover, confirming the usefulness of this kind of germplasm for sowings at high al-
titude (Krautzer 2005). The local mixtures, also sown at 15 g m−2 rate, had a slow
start, which could be partly explained by their lower seed quality compared with
that of commercial material. Peratoner (2003) found that lower competitive ability
of site-specific species played an important role in the establishment phase, while
specific ecological features became relevant later on. In the current study, after just
three years from the sowing the local mixtures filled their gap resulting in no dif-
ference between site-specific-species mixtures sown at lower seed rate and mixtures
sown at higher seed rate.
The mixtures also differed over the years in the contribution of grasses and di-
cots other than legumes to the cover vegetation (Table 33.2). A variable percentage
between about 4 and 11 % of ground (depending on the mixture) was covered by
non-angiosperm vegetation such as mosses (mostly) and lichens. The three commer-
cial mixtures had higher percentage of grasses than the two local mixtures, whereas
the latter had, in turn, greater proportion of other dicot species (Table 33.2). Be-
cause of its richness in grasses, the site-specific mixture E resembled more the other
commercial mixtures than the two local ones. These latter mixtures featured better
botanical balance, which might result in a more-lasting stability of the covers and an
increase of biodiversity in the long term, as observed by Peratoner (2003).
Despite significant mixture × year interactions (Table 33.2) for the botanical
family contributions, some trends were consistent across years, with mixtures A
and B being always different from mixtures C and E for the relative contribution of
grasses and of other dicot species to the ground cover (Fig. 33.1b, c). The behaviour
of the complex commercial mixture D was peculiar in this respect, showing a drop of
cover provided by grass species in the third year of growth, that was compensated by
an increase of other dicots (Fig. 33.1b, c). Peratoner (2003) reported a similar trend
in the German Alps in plots sown with commercial seed mixtures, where lowland
forage species decreased suddenly by the third year.
The different botanical composition of mixtures C and D at the third year was
the only case of inconsistent behaviour between mixtures of same germplasm kind
and different species complexity. Apart from this, indeed, the complexity of seed
mixtures had only slight effects on their covering ability or botanical composition.
However, a proper assessment of possible effects of the mixture complexity could
be better made over a longer period of time than the current one.
In conclusion, the mixtures including native materials appeared, despite the lower
seeding rate and the possibly lower seed quality, to be suitable for a profitable use
100

90
a

80
a
ab
70
b
b
60

50
2007 2008 2009
Local, simple Local, complex
Commercial, simple Commercial, complex
a Commercial, site-specific

60 30
a
a
50 a 25
a ab a a
bc a
40
a 20
a bc b a
30
c b 15 b
b a
b
b ab
20 10 b
bc b b
10 5
cd
33 Comparison of Seed Mixtures for Technical Revegetation at High Altitude

c
0 0
d b
2007 2008 2009 2007 2008 2009
Local, simple Local, complex Local, simple Local, complex
Commercial, simple Commercial, complex c Commercial, simple Commercial, complex
b Commercial, site-specific Commercial, site-specific
259

Fig. 33.1 Ground cover percentage (a), and contribution percentage of grasses (b) and other dicot species (c) to the cover of five seed mixtures of different nature
and complexity, in 3 years following that of sowing at 2,020 m a.s.l. In each graph and for each year, values with different letters differ at P < 0.05 according to
Duncan’s test. No lettering in a year meant no significant variation (P > 0.05) among mixtures according to F test in the two-way analysis of variance
260 P. Spoleto et al.

in the revegetation at high altitude, especially for areas of high natural value, where
the absence of genetic pollution is a preliminary requirement to any intervention.

Acknowledgments We are grateful to Dr B. Krautzer, BAL Gumpenstein, Irdning, Austria, for

kindly providing seeds of the site-specific species Anhtyllis vulneraria and of the commercial,
site-specific mixture. The research was carried out in the framework of the projects SemenSci and
SemTek funded by the General Directions ‘Agricoltura’ and ‘Sistemi Verdi’ of Regione Lombardia,
Italy.

References

Cochran WG (1951) Testing a linear relation among variances. Biometrics 7:17–32

Krautzer B (2005) Exploitation of native plant material for green covers in environmental recla-
mation. In: Piano E (ed) Atti del Convegno ‘Inerbimenti e Tappeti Erbosi per l’Agricoltura,
l’Ambiente e la Società’, vol 2. ISCF, Lodi, pp. 109–117
Krautzer B, Bohner A, Partl C, Venerus S, Parente G (2001) New approaches to restoration of alpine
ski slopes. Organic grassland farming. Int Occas Symp Eur Grassland Fed 6:193–196
Krautzer B, Peratoner G, Bozzo F (2004) Site specific grasses and herbs, seed production and use
for restoration of mountain environments. FAO plant production and protection series 32. FAO,
Rome
Florineth F (1992) Establishment of greens in high altitudes in Southern Tyrol. Rasen-Turf-Gazon
3:74–80
Peratoner G (2003) Organic seed propagation of indigenous species and their use in ecological
restoration of ski runs in mountain regions. Ph.D. dissertation, Kassel University Press, Kassel,
Germany
Chapter 34
Genetic Diversity for Cell Wall Digestibility
in a Diverse Lolium perenne Collection

H. Muylle, C. Van Waes, F. Van Parijs, G. Obianugba, J. Baert

and I. Roldán-Ruiz

Abstract The digestibility of the cell wall has a significant impact on the fodder
quality in Lolium. This is especially the case at reproductive maturity, when the
lignification of the cell wall and the proportion of stems increase. The relevance of
this factor becomes clear if we take into consideration that a 1 % increase of the
digestibility of the fodder improves the daily intake per cow by 0.2 kg and results in
an increase of the daily milk production of 0.4 kg per cow (Gilliland, Quality Counts
on Northern Irish Recommended List, 2007). Detailed information on the variability
available for cell wall digestibility in individual ryegrass genotypes and how it differs
for different organs is currently not available. This information could, however, be
exploited to breed ryegrasses with a higher feeding value. In a preliminary study we
identified significant differences in the cell wall digestibility of the leaf fraction of
ten Lolium perenne genotypes. The glucose release varied between 2.58 and 3.89 g
glucose/20 g dry biomass. Significant differences for glucose release were identified
between leaf and stem samples. To get a better estimation of the variability available
for cell wall digestibility in L. perenne, a subset of 30 genotypes selected within a
broader collection of 300 genotypes from current varieties were analyzed in detail.
Preliminary results indicated that improving total digestibility might be achieved
through enhanced cell wall digestibility in stems. However more genotypes should
be analysed before firm conclusions can be drawn.

34.1 Introduction

Total digestibility is a major selection criterion in Lolium breeding. The relevance

of this factor becomes clear if we take into consideration that a 1 % increase of the
digestibility of the fodder improves the daily intake per cow by 0.2 kg and results in

H. Muylle () · C. Van Waes · F. Van Parijs · J. Baert · I. Roldán-Ruiz

Institute for Agricultural and Fisheries Research (ILVO), Caritasstraat 21,
9090 Melle, Belgium
e-mail: [email protected]
G. Obianugba
National Center for Genetic Resources and Biotechnology (NACGRAB),
Federal Ministry of Science and Technology,
Moor Plantation, P.M.B 5382, Ibadan, Nigeria

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 261

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_34, © Springer Science+Business Media Dordrecht 2013
262 H. Muylle et al.

an increase of the daily milk production of 0.4 kg per cow (Gilliland 2007). Recently,
major advances have been achieved in increasing water soluble sugar content (WSC)
as a means to enhance total digestibility (Humphreys et al. 2010). Another approach
to improve total digestibility is increasing the cell wall digestibility. The cell wall is a
major component of the feedstock and its digestibility has a significant impact on the
fodder quality in swards, especially at reproductive maturity, when the lignification
of the cell wall increases (Groot et al. 2003). The contribution of cell wall digestibility
to total digestibility has been studied in several forage crops as maize, orchardgrass,
bromegrass and canarygrass (Barriere et al. 2009; Casler et al. 2008). In Lolium,
little is known on cell wall digestibility, especially at the individual genotype and
organ level. This knowledge is however essential to facilitate breeding for this trait.
Therefore, the objectives of this study were to estimate the variability in cell
wall digestibility present in a limited set of genotypes and to estimate differences
in cell wall digestibility in relation to plant organ. For this we used a small-scale
saccharification assay requiring little amounts of plant material. In a second step we
analyzed the correlation between cell wall digestibility and quality parameters in a
larger collection of 30 genotypes varying in WSC and total digestibility, using NIRS
(Near InfraRed Spectroscopy). The relationship between heading date and cell wall
digestibility was also estimated in these 30 genotypes.

34.2 Material and Methods

Plant Material Ten diploid perennial ryegrass genotypes (Set 1) were used for a
preliminary small-scale experiment. Four of these genotypes (10370-22, 10370-45,
Ba12938-40 and Ba12990-6) were derived from three wild accessions of the ECPGR
core collection of Lolium (10370, Ba12938 and Ba12990). Four genotypes were
selected among advanced ILVO breeding material (1554-1, 5297, 5311 and KC98R-
138-1) and two genotypes (ABE and OPT) are the parents of an ILVO mapping
population (Studer et al. 2010). These two parental genotypes, ABE and OPT, were
selected from the varieties Aberdart and Option, respectively.
Thirty genotypes (Set 2) were selected out of a collection of 300 individual geno-
types from 12 varieties (25 plants/variety), displaying a wide range in WSC and total
digestibility (Table 34.1). Selection of Set 2 was based on the quality parameters
measured on two consecutive cuts in spring 2008 on non vernalised plants without
clonal replicates of the genotypes. Set 2 was used to estimate correlations between
NDFD (NDF digestibility) and total digestibility, ADL (lignin fraction) and heading
date.
Growth Conditions Plants of Sets 1 and 2 were cloned in three replicates in autumn
2009 and grown in the glasshouse during winter 2009–2010 in 12 L containers that
were transferred to open air by the beginning of March 2010.
Biomass Harvested The plants were harvested on April 26th and June 9th 2010. The
June 2010 cut was used for quality analysis. Stem and leaf material was separated
manually.
34 Genetic Diversity for Cell Wall Digestibility in a Diverse Lolium perenne Collection 263

Table 34.1 WSC (percentage Genotype nr. Variety WSC2008 Tot. Dig.2008
of dry matter), and total
digestibility (percentage of 301 Merks 36.5 82.8
dry matter) of Set 2. Values 302 Merks 20.5 70.7
are averages of two cuts 303 Merks 27.9 79.3
(spring 2008) of 304 Melways 23.3 77.0
non-vernalized clones 305 Melways 32.1 79.8
without clonal replications of 306 Barnhem 32.8 80.1
genotypes 307 Barnhem 35.4 82.0
308 Carillon 36.6 82.0
309 Carillon 28.8 78.7
310 Asturion 30.8 78.1
311 Asturion 27.8 77.9
312 Asturion 21.7 72.0
313 Tomaso 29.4 74.8
314 Tomaso 26.7 71.0
315 Tomaso 27.0 77.0
316 Meloni 33.0 80.9
317 Meloni 34.8 81.1
318 Barata 31.1 79.3
319 Barata 16.8 67.6
320 Barata 19.0 68.3
321 Gandalf 25.0 73.1
322 Gandalf 25.8 73.9
323 Sibasa 25.0 76.2
324 Sibasa 29.8 76.2
325 Sibasa 21.8 73.1
326 Orantas 20.5 70.1
327 Orantas 23.7 75.2
328 Aberzest 25.7 74.2
329 Aberzest 39.6 83.6
330 Aberzest 40.8 83.1
Mean 28.3 76.6
Min. 16.8 67.6
Max. 40.8 83.6

Saccharification Assay Cell wall digestibility of Set 1 was determined using a small
scale saccharification assay derived from Gomez et al. (2008). 100 mg of dried plant
materials were ground in a Retch Tissuelyser (Qiagen) for 6 minutes at 30 Hz. Twenty
mg per genotype were taken in four replicates and suspended in 1 ml 70 % ethanol.
Samples were incubated in a thermoshaker at 700 rpm for 4 hours at 50 ◦ C. After a
short spin, the supernatant was discarded and the washing step was repeated 3 more
times. Finally, samples were suspended in acetone after which the supernatant was
removed. The samples were dried overnight and subsequently the weight of dry
biomass remaining (cell wall) was recorded. Enzymatic hydrolysis was carried out
using the enzyme mix AccelleraseTM 1,000 (Genencor) in 0.1 M NaOAc buffer pH
4.8. The samples were incubated at 50 ◦ C and 700 rpm during 72 hours. Glucose
release was measured at specific time intervals (0 and 72 hours) using the GOD POD
assay (Bergmeyer 1974).
264 H. Muylle et al.

Fig. 34.1 Glucose release

(mg glucose/20 mg dry
material) from leaf material
harvested in June 2010 of ten
genotypes using a small-scale
saccharification assay

Determination of Quality Parameters For Set 2 NDF, ADF, ADL, NDFD and total
digestibility were determined using NIRS calibration lines available at ILVO (Plant
Sciences Unit—Crop Husbandry). Reference methods for NDF, ADF, ADL and
NDFD were carried out according to the methods described by Goering and Van
Soest (1970), and Van Soest et al. (1991). Total digestibility was analysed by the
cellulase method of De Boever et al. (1988).
Statistical Analysis Phenotypic data were subjected to analysis of variance to iden-
tify significant differences among genotypes. Pearson correlation coefficients were
calculated to determine the relationships between the different parameters. The
software STATISTICA v9 (Statsoft, USA) was used for these calculations.

34.3 Results and Discussion

In a preliminary experiment, we used a small scale saccharification assay to determine

cell wall digestibility on small amounts of tissue. Significant differences in cell wall
digestibility were identified among the ten genotypes of Set 1. Glucose release of
the leaf material (June 2010 cut) varied between 2.58 and 3.89 g glucose/20 g dry
biomass (Fig. 34.1).
The June 2010 cut of genotype ABE contained stem material and therefore, for
this genotype, the harvested biomass was divided in three fractions: leaf, sheath and
stem. The cell wall digestibility of these three organs was determined separately.
Glucose release was significantly different for the three fractions (3.51, 2.95 and
2.57 g glucose/20 g dry biomass for leaf, sheath and stem respectively). This small
scale test proved to be reproducible (results not shown) and able to identify significant
differences among the ten genotypes of Set 1 and between different organs in one
34 Genetic Diversity for Cell Wall Digestibility in a Diverse Lolium perenne Collection 265

Fig. 34.2 Correlation

between NDFD values
(percentage of total NDF) of
stem material of the June
2010 cut of 30 genotypes. a
ADL content (percentage of
absolute dry weight); b total
digestibility (percentage of
absolute dry weight);
c heading date (DOY: day of
the year recorded in 2011).
Dotted horizontal lines
represent the division into
early, intermediate and late
heading genotypes
266 H. Muylle et al.

genotype (ABE). This method allows thus to study spatio-temporal differences in

cell wall digestibility within one plant and among genotypes.
Given the identified (and expected) lower cell wall digestibility in stem material in
the ABE genotype, 30 genotypes (Set 2) covering a broad a range of total digestibility
and WSC values was studied more closely in order to identify the range of cell wall
digestibility in the stem fraction. Our aim was to identify prototypes with extreme
values for quality parameters as WSC, total digestibility and NDFD. The cut made in
June 2010 on Set 2 was separated in leaf and stem material. NIRS was used to deter-
mine NDFD (NDF digestibility), total digestibility, NDF (cell wall fraction), ADF
and ADL (lignin fraction) of stem material. NDFD values varied between 66 and
87 % and were significantly correlated (Fig. 34.2) with other quality parameters such
as lignin content (r = 0.905) and total digestibility (r = 0.927). A lower, but signif-
icant correlation was identified also between NDFD and heading date (r = 0.721).
Despite the correlation found between NDFD and heading date, variation useful for
breeding purposes could be observed within each maturity class. For example, the
NDFD of intermediate heading genotypes (between DOY [Day of the Year] 147
and 157) varied between 65.7 and 82.0 %, indicating available genetic variation for
NDFD within this maturity class.

34.4 Conclusion

The small-scale saccharification assay used here was able to identify significant
differences in cell wall digestibility among genotypes and among different plant parts.
This will enable the study of the spatio-temporal evolution of cell wall digestibility
in single genotypes, and to compare genotypes.
Significant variation for NDFD was observed in stem material of a June cut in a
set of 30 genotypes. A high correlation was identified between NDFD and heading
date, indicating that caution is necessary when comparing results for cuts of plants of
different maturity classes. On the other side, our data suggest that sufficient variability
for NDFD might be available within maturity classes, and that selection for higher
NDFD is possible. In addition, Set 2 has been harvested in 2011, taking into account
maturity (results not shown). If differences among genotypes are confirmed in 2011,
we will be able to conclude that genetic variation for NDFD exists in Lolium that
can be explored for breeding purposes.

References

Bergmeyer HU (1974) Methods of enzymatic analysis. Academic Press, New York.

Barriere Y, Mechin V, Riboulet C, Guillaumie S, Thomas J, Bosio M, Fabre F, Goffner D, Pichon
M, Lapierre C, Martinat JP (2009) Genetic and genomic approaches for improving biofuel
production form maize. Euphytica 170:183–202
Casler MD, Jung HG, Coblentz WK. (2008) Clonal selection for lignin and etherified ferulates in
three perennial grasses. Crop Sci 48:424–433
34 Genetic Diversity for Cell Wall Digestibility in a Diverse Lolium perenne Collection 267

De Boever J, Cottyn B, Andries J, Buysse F, Vanacker J (1988) The use of a cellulase technique to
predict digestibility, metabolizable and net energy of forages. Anim Feed Sci Technol 19:247–
260
Gilliland T (2007) Quality counts on Northern Irish recommended list. Agri-Food Biosciences
Institute Plant Testing Station, Crossnacreevy. http://www.sinclairmegill.com/images/T%
20gilliland% 20NI%20Recommended%20List%202007.pdf. Accessed 30 Aug 2011
Goering HK, Van Soest PJ (1970) Forage fiber analyses (apparatus, reagents, procedures, and some
applications). Agricultural handbook no.◦ 379. U.S. Department of Agriculture, Washington
Gomez LD, Bristow JK, Statham ER, McQueen-Mason SJ (2008) Analysis of saccharification
in Brachypodium distachyon stems under mild conditions of hydrolysis. Biotechnol Biofuels
1(1):15
Groot JCJ, Lantinga EA, Neuteboom JH, Deinum B (2003) Analysis of the temperature effect on
the components of plant digestibility in two population of perennial ryegrass. J Sci Food Agric
83:320–329
Humphreys M, Feurestein U, Vandewalle M, Baert J (2010) Ryegrasses. In: Boller B, Posselt U,
Veronesi F (eds) Fodder crops and amenity grasses: handbook of plant breeding. Springer, New
York, pp. 211–260
Studer B, Kölliker R, Muylle H, Asp T, Frei U, Roldán-Ruiz I, Barre P, Tomaszewski C, Meally H,
Barth S, Skot L, Armstead IP, Lübberstedt T (2010) EST-derived SSR markers used as anchor
loci for the construction of a consensus linkage map in reygrass (Lolium spp.). BMC Plant Biol
10:177 (http://www. biomedcentral.com/1471-2229/10/177). Accessed 30 Aug 2011
Van Soest PJ, Robertson JB, Lewis BA (1991) Methods for dietary fiber, neutral detergent fiber,
and nonstarch polysaccharides in relation to animal nutrition. J Dairy Sci 74:3583–3597
Chapter 35
Variability Among Accessions of Forage Vetch
for Basic Agronomic and Morphological Traits
under Agro-Ecological Conditions of Serbia

Z. Lugić, J. Radović, G. Sokolović, G. Jevtić, T. Vasić and S. Andjelković

Abstract The vetches are important annual forage crops in Serbia. Forage and
seed production, as well as morphological traits of 33 accessions of common vetch
(V. sativa) and eight accessions of hairy vetch (V. villosa), were estimated in the agro-
ecological conditions of Serbia. The accessions were obtained from South Australia
where they had been bred or cultivated in the past 20 years. Field nurseries were es-
tablished at the Institute of forage crops in Krusevac in the autumn of 2008 and 2009,
with 40 spaced plants per accession. In the spring of 2009 and 2010, morphological
traits, and forage and seed yield components were measured. The results for both
years for all traits were analyzed by ANOVA. A cluster analysis was performed on
the basis of all investigated traits, using the complete linkage method with Euclidean
distances. The results indicated a high level of variability of the accessions and a
high potential for genetic improvement of all investigated traits.

Keywords Variability · Agronomic traits · Morphology · Vetch

35.1 Introduction

Among annual legumes, peas and vetches are a very significant source of qual-
ity forage in Serbia, ranking directly after soybean. They are grown on about
35,000 hectares (SYB 2009). Their importance is based on a high adaptability to
different growing conditions, the possibility of sowing in different periods of the
year, and the production of relatively stable yields. Vetch forage is especially impor-
tant for small and medium farms, where grown in mixtures with cereals, it is used
to produce hay, haylage or silage (Djordjević and Dinić 2003). Breeding of forage
vetches in Serbia began in the seventies and has resulted in a significant number of
cultivars of winter and spring forms of common vetch (Vicia sativa L.) and win-
ter hairy vetch (Vicia vilosa Roth.). The main aims in vetch breeding have been to
increase forage yield, improve biomass quality and increase tolerance to different

Z. Lugić () · J. Radović · G. Sokolović · G. Jevtić · T. Vasić · S. Andjelković

Institute for Forage Crops, Globoder, 37251 Kruševac, Serbia
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 269

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_35, © Springer Science+Business Media Dordrecht 2013
270 Z. Lugić et al.

biotic and abiotic stresses. In breeding programs, local wild populations and domes-
tic landraces were most used, while the usual methods include mass selection, and
pedigree and bulk methods from hybrid populations (Mihailović et al. 2010). The
aim of this paper is to examine the potential variability of the main agronomic traits
of 33 accessions of common vetch and eight accessions of hairy vetch of different
origins, but which were grown and had been involved in breeding programs in South
Australia for the last 20 years.

35.2 Materials and Methods

The study included a total of 41 vetch accessions. 33 were common vetch (1–33) and
eight were hairy vetch (34–41) obtained from SARDI (South Australian Research
and Development Institute), Adelaide. Sowing was carried out in the autumn of 2008
and 2009 in space plants nurseries with plant to plant distance 60 × 60 cm and 30
plants per accession. In order to assess winter survival, the number of surviving
plants (SP) was determined in the following spring. In the phase of full flowering
in the next year, the following measurements on individual plants were made: plant
height (PH), green matter yield per plant (GMY), dry matter yield per plant (DMY)
and number of lateral branches (NLB). Seed yield components, number of pods per
plant (NPP), number of grains per pod (NGP) and number of grains per plant (NGPL)
were determined in ten randomly chosen individual plants. For each accession, the
weight of 1,000 seeds (1000SW) was determined. Average results for both years for
all traits were analyzed by ANOVA. Cluster analysis was performed on the basis of
all investigated traits, using the complete linkage method with Euclidean distances
(Statistica 5.0, Stat Soft Inc.).

35.3 Results and Discussion

The results presented in Table 35.1 indicate a very high variability of the accessions
for all traits, as indicated by the high variation coefficients. As pointed out by Matić
et al. (2010), the existence of accessions with desirable traits offers the possibility
of crossing and recombination to create new cultivars. Percentage of wintered plants
was largely dependent on the accession and ranged from 35 % (34) to 96 % (16),
suggesting that a large number of accessions has sufficient resistance to low tem-
peratures and winter conditions. The average height of plants of all accessions was
56.61 cm, which is significantly less compared to the results for the cultivars grown
in dense swards under agro-ecological conditions of Serbia (Erić et al. 2007). The
average 2-year results for GMY and DMY (67.25 and 20.45 g/plant) indicate high
variability and high potential of the accessions for this trait. Most common vetch
accessions showed better results than hairy vetch, and accessions no. 2, 3, 6, 12
and 14 had extremely high forage yields which were significantly higher compared
35 Variability Among Accessions of Forage Vetch for Basic Agronomic . . . 271

Table 35.1 Average 2-year values (2009–2010) of studied accessions

Accession SP PH GMY DMY NLB NPP NGP NGPL 1000
No. (%) (cm) (g) (g) SW (g)
Common vetch
1 75 64.10 110.06 34.87 10.25 9.25 4.50 41.63 75.00
2 81 85.56 218.13 65.40 11.90 49.00 6.87 336.63 83.90
3 81 76.80 193.16 62.43 8.00 27.67 6.22 172.11 79.90
4 78 75.20 131.66 36.33 9.00 34.30 6.87 235.64 69.30
5 77 78.53 186.8 56.43 8.08 37.08 5.45 202.10 70.70
6 97 76.86 187.56 57.23 10.22 35.67 7.22 257.54 82.30
7 75 64.43 80.93 20.16 9.86 32.00 5.24 167.70 79.60
8 83 61.46 104.73 24.06 10.69 55.54 6.77 376.01 80.10
9 86 55.13 80.40 25.20 11.92 39.58 5.50 217.70 77.20
10 92 47.56 77.60 19.53 8.00 26.39 5.62 148.31 92.00
11 71 45.80 46.96 11.86 8.50 18.00 5.78 104.04 59.90
12 88 64.30 124.73 35.36 11.71 25.57 6.19 158.28 69.90
13 74 63.66 78.06 15.70 9.76 25.71 6.90 177.40 77.00
14 91 82.30 124.23 35.50 8.23 12.62 4.50 56.80 90.90
15 92 65.36 111.50 29.00 7.20 18.40 4.53 83.35 79.60
16 96 52.60 53.20 13.46 6.86 13.00 5.10 66.30 81.10
17 82 53.83 74.83 20.90 8.67 19.78 6.89 136.28 85.70
18 87 55.80 77.80 18.80 9.80 10.60 5.00 53.00 72.50
19 72 53.46 46.86 12.66 6.17 9.50 4.33 41.14 75.50
20 81 50.80 48.66 15.76 9.83 19.67 6.28 123.53 52.40
21 75 41.63 39.96 13.13 7.00 14.43 5.86 84.56 79.90
22 77 42.50 22.6 8.60 6.57 7.43 4.57 34.00 74.20
23 51 46.00 21.93 7.713 7.40 9.60 4.13 39.65 75.70
24 76 45.20 22.26 9.13 8.00 10.80 4.93 53.24 77.10
25 85 52.23 46.33 13.36 8.88 14.25 5.33 75.95 84.00
26 57 56.76 46.50 18.26 8.71 16.57 4.26 70.59 80.10
27 63 50.70 43.73 17.73 7.71 13.57 5.48 74.36 73.60
28 51 44.03 19.86 9.43 9.75 12.00 4.88 58.56 69.90
29 48 50.63 30.63 15.10 7.50 10.00 4.17 41.70 56.90
30 63 53.90 30.13 11.63 8.60 17.80 5.40 96.12 43.50
31 80 54.86 42.33 15.30 12.18 23.55 5.50 129.53 68.00
32 72 48.50 46.20 18.20 7.30 16.80 5.92 99.46 88.00
33 66 51.10 70.20 23.86 7.50 16.83 4.78 80.45 76.50
Hairy vetch
34 45 43.80 18.36 6.433 11.13 132.13 3.31 437.45 44.80
35 71 70.06 36.70 13.46 14.44 68.11 3.18 216.59 36.60
36 41 42.23 8.20 3.166 12.29 7901 349 161.00 42.35
37 80 50.66 6.90 3.533 1122 8118 351 285.60 41.12
38 90 61.93 28.73 10.06 9.14 90.71 3.05 276.67 44.50
39 47 78.43 44.96 16.46 15.67 65.33 2.59 169.20 40.60
40 63 57.80 15.84 5.70 12 58.67 2.85 167.21 42.90
41 46 61.23 24.36 8.16 10.27 58.00 2.61 151.38 35.40
X 56.61 67.25 20.45 9.23 31.81 4.89 141.86
CVa 21.36 80.28 76.71 24.77 86.49 30.67 69.09
LSD0.05 2.37 3.84 2.02 1.65 1.96 1.59 23.72
LSD0.01 3.13 5.09 2.67 2.18 2.59 2.12 31.39
SP surviving plants, PH plant height, GMY green matter yield per plant, DMY dry matteryield per
plant, NLB number of lateral branches, NPP number of pods per plant, NGP number of grains per
pod, NGPL number of grains per plant, SW weight of 1,000 seedsa
a
CV refers to variation among plants (genotypes) within accessions
272 Z. Lugić et al.

Fig. 35.1 Cluster diagrams for common vetch (above) and hairy vetch (below) accessions based
on the studied traits

to the results for spring vetch by Mihailović et al. (2004). In contrast, hairy vetch
accessions showed higher values for parameters of seed yield per plant. Although all
accessions of hairy vetch had fewer grains per pod (NGP), due to the large number
of pods per plant (NPP) they produced larger number of seeds per plant (NGP). Also,
large differences among the accessions for 1,000 seed weight were determined, and
35 Variability Among Accessions of Forage Vetch for Basic Agronomic . . . 273

all common vetch accessions had significantly larger seeds than the hairy vetch. The
cluster diagram based on the studied traits showed four groups of populations of
common vetch. Within breeding germplasm accessions of vetch, populations two
and eight were phenotypically the most distant from the rest of the collection and
will be considered during further studies. In the dendrogram for hairy vetch, popu-
lation 34 (SA Haymaker) was clearly distinct from other populations which showed
a series of linkage distance (Fig. 35.1).

Acknowledgments This research was financed by the Ministry of Education and Science of the
Republic of Serbia, project TR31057.

References

Djordjević N, Dinić B (2003) Siliranje leguminoza Agricultural Research Institute Serbia Vizartis,
Belgrade, Serbia, pp 1–201
Eric P, Mihailovic V, Cupina B, Mikic A (2007) Jednogodišnje krmne biljke Naučni Institut za
ratarstvo i povrtarstvo, Novi Sad, Serbia, pp 111–143
Matic R, Nagel S, Kirby G, Robertson B, Mikić A, Mihailović V (2010) Vetches (Vicia spp) adoption
and utilization in Australia. Biotechnol Anim Hus, 2, vol 26:17–27
Mihailović V, Prentovik T, Vasiljević S, Katić S, Mikić A (2004) Forage yield and forage yield
components of spring vetch genotypes (Vicia sativa L). Acta Agri Serbica, vol IX(17):407–411
Mihailović V, Mikić A (2010) Novel directions of breeding annual feed legumes in Serbia.
Biotechnol Anim Hus, 2, vol 26:81–89
Statistical yearbook of Serbia (2009) Statistical Office of the Republic of Serbia, Belgrade, Serbia
Chapter 36
Genetic Variation of Root Characteristics
and Deep Root Production in Perennial Ryegrass
Cultivars Contrasting in Field Persistency

D. Sokolovic, S. Babic, J. Radovic, J. Milenkovic, Z. Lugic,

S. Andjelkovic and T. Vasic

Abstract The identification and breeding of genotypes with improved drought tol-
erance play an important role in developing grasses with better performance and
persistence. Plants should maintain some root contact with ground water to maintain
cell turgor and to survive during drought periods. Plants with longer roots can better
exploit the available, deeper, soil water. Breeding idea is to increase soil water uptake
and drought tolerance of perennial ryegrass by improving root architecture with pre-
served above ground biomass yield. The objectives of this study were to determine
variability of root dept and distribution, shoot dry matter yield (DMY) and root/shoot
ratio of three Lolium perenne cultivars contrasting in filed persistency. The trial was
conducted in 0.9 m sand plastic tubes with 30 plants per cultivar in three replications.
Plants were irrigated daily with complete nutrient solution and trimmed once after 80
days of growth. The roots were washed out three months after planting, cut in 10 cm
increments, dried and weighed. Data were analyzed by standard ANOVA. There were
significant differences between cultivars and those with better persistency showed
higher proportions of deep roots and 8 % heavier roots in total. Similar root/shoot
ratios for all cultivars have showed that plants which have invested more dry matter
into root have not automatically had less shoot yield. Genotypes with deeper root
systems can be used for drought resistance improvement in Lolium perenne.

Keywords Drought resistance · Perennial ryegrass · Root depth · Variability

36.1 Introduction

In recent years long periods of drought have been experienced regularly during
summer mounts in Serbia. Therefore soil water deficit tolerance has become an
extremely important breeding criterion, especially in all rain fed forage plants such
as perennial grasses. Improvement of this complex trait of perennial forage grasses
can be achieved in many ways, since plants have many strategies to cope with drought.

D. Sokolovic () · S. Babic · J. Radovic · J. Milenkovic · Z. Lugic · S. Andjelkovic · T. Vasic

Institute for forage crops, Globoder, 37251 Krusevac, Serbia
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 275

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_36, © Springer Science+Business Media Dordrecht 2013
276 D. Sokolovic et al.

They can “escape” the dry season (with storage organs, rhizomes, dormant buds),
adjust growth and development patterns (reduce leaf area, regulate stomatal control,
enhance water use efficiency and extract more water) and successfully recover after
drought (minimal growth during drought then regrow rapidly when stress had been
removed with maintained tiller and plant density) (Ludlow 1980). Very few of these
options have been explored within perennial forage grasses. Most attempts to develop
more drought-tolerant plants have concentrated on determination of survivors after
drought or evaluating genotypes in dry conditions, both in the field, which is time
consuming and expensive process. Thought this procedure is initially an efficient
process, the later steps in drought breeding should involve methods to accelerate
further selection cycles and make them more efficient.
Perennial ryegrass, which is one of the most important forage grasses with many
superior characteristics in fodder production, suffers from drought susceptibility
which makes it less suitable as a component of Serbian grasslands. However, there are
a lot of genotypes found across the Serbian landscape which show better performance
in production (Sokolovic et al. 2003, 2007), and which have been incorporated in
local cultivars with better persistency (Sokolovic et al. 2010).
Resistance to drought in Serbia in forage perennial grasses has generally been
evaluated through observation of prolonged field persistency, determined via the
number of plants surviving after several warm and dry summers. This mechanism
and strategy is of most benefit to select plants which survive the normal dry periods
with high regrowth rates when it rains. Grass plants need to keep some root contact
with ground water to survive because they need water constantly to maintain cell
turgor. Therefore, our goal was to increase the efficient use of available water through
improved root architecture and soil water uptake by looking for perennial ryegrass
plants with root systems which can better exploit the available, deeper, soil water.
Drought tolerance is closely related to the distribution and penetration of root
systems in the deeper portions of the soil for a number of grass species (Carrow 1996;
Huang et al. 1997). Due to relatively high narrow sense heritability for deep root
growth (Lehman and Engelke 1991) it is likely that selection for deeper root portions
is possible. One important factor what should be taken into account in grass breeding
is the high positive correlation between root and shoot characteristics, especially
dry matter yield, in a number of species (Palazzo and Brar 1997). Nevertheless, it
is possible to obtain, after two cycles of selection, gains both in root production
(367 %) and in shoot weight (79 %) in forage type perennial ryegrass (Bonos et al.
2004). It is possible that incorporation of genetic markers associated with rooting
characteristics would accelerate such a selection process (Guthridge et al. 2001).
An argument against selecting for a massive root system is that it might consume
carbon at the expense of harvestable herbage, either for root growth or maintenance.
However, Parsons (1988) has shown that only a small proportion (10–15 %) of
assimilate is diverted to roots. Furthermore, during drought there is no shortage
of carbohydrate reserves, which can account for up to 30 % dry matter in Lolium
perenne (Thomas 1991) and 50 % in Dactylis glomerata (Volaire 1995). It is known
that there is useful variability for root mass and distribution, or root:shoot ratios in
perennial ryegrass (Crush et al. 2005, 2007) and other grasses (Bullitta 1996), and
36 Genetic Variation of Root Characteristics and Deep Root Production . . . 277

that any genotype which invests more dry matter into root is more drought resistant.
Carrow (1996) found a negative correlation between roots in the top 10 cm of soil
and drought resistance. Roots at depths of up to 30 cm may mainly play a role in
absorption of nutrients, which are concentrated in the topsoil. Deep roots tend to be
more important for absorbing water and those plants growth may be constrained by
nutrient deficiency, but at least the shoot tissues will receive enough water to keep
them alive. Even a few deep roots may be all that is necessary for survival.
Even though results are not totally consistent across all experiments about this
topic, the central premise of our breeding scheme for perennial ryegrass in Serbia is
to explore root characteristics of persistent cultivars and consequently increase soil
water uptake and drought tolerance by improving root architecture, while maintaining
above ground biomass yield. The objectives of this study were to determine variability
of root dept, root distribution and root/shoot ratio within and between three cultivars
of Lolium perenne found to contrast in field persistency and performance in previous
research (Sokolovic et al. 2010). Given the Serbian climate, the final long-term goal
is to select plants that can survive the normal dry periods characteristic of the Serbian
climate, and which exhibit with high regrowth rates subsequently when it rains.

36.2 Material and Methods

36.2.1 Plant Material

Three diploid perennial ryegrass cultivars contrasting in field persistency were objects
of this research. Serbian cultivar K11 and East European cultivar Mara showed in
previous experiments good field persistency and performance, while Cashel showed
higher drought susceptibility and plant density reduction (Sokolovic et al. 2010).
Seed was sown in containers in fertile potting mix Klasmann TS1 in glasshouse in
March 2010. Seedling were transferred in pots after 4 weeks when plants were of an
adequate size and grown till late summer 2010. Individual plants of all cultivars were
clonally divided on 15th of August into a minimum of four small parts—ramets, each
with three tillers. Three ramets were leaf and root trimmed to 2 cm and transferred
to the plastic root-screening tubes. Remaining parts of plants were kept in a stock
nursery.

36.2.2 Plastic Root Tube Experimental System

The method of root length and deep density evaluation was developed by Bonos
et al. (2004) and Crush et al. (2005). The trial was conducted in 0.9 m long polyvinyl
chloride (PVC) tubes, 75 mm in diameter (2 mm wall thickness) under greenhouse
conditions with 30 randomly chosen plants per cultivar in three cloned replications.
All tubes were cut in half lengthwise and rejoined with adhesive tape. Tubes were
278 D. Sokolovic et al.

inserted in three tables with wire net, each holding 90 tubes. Tubes were put on 10 mm
polythene foam on the floor, at an approximate angle of 25◦ from vertical, filled 5 cm
from bottom with gravel for drainage and with washed mortar sand progressively to
the top using a spray of water for uniform packing. Mortar sand particles were from
0.5–2 mm in size. Three tiller ramets were planted in the center of tubes and supplied
with drip nozzles for irrigation. Sand in every tube was covered with plastic foil
to decrease evaporation. Each tube received 100 ml day−1 of low strength complete
nutrient solution (Blamey et al. 1991, modified for Serbian conditions) in two separate
doses (morning and evening) through a pipe system. The shoots were trimmed once
after 80 days of growth, just before root analyses, dried and weighed. The roots were
extracted after three months of growth by laying the tubes down on plastic 5 × 5 mm
mesh and splitting them in two halves. Sand-root columns were rolled out, washed
under a water shower and cut into pieces 10 cm in length. Plants with 1 cm of roots
were weighed and planted in stock. Root parts were paper dried and weighted, then
dried in fan oven at 65 ◦ C over night and weighed.

36.2.3 Statistical Analyses

Measurements and calculations covered shoot dry matter (DM) weight, root DM
weight in 10 cm increments and root/ shoot ratio. The data were analyzed by standard
ANOVA. Root/shoot ratio has been calculated by ratio of total root DM weight and
shoot DM weight. Deep root/shoot ratio has been defined as the weight (g) of roots
that are deeper than 30 cm g−1 shoots.

36.3 Results and Discussion

There were significant differences in root extension and growth rates between
cultivars and those with better persistency showed higher deep root portions
(Fig. 36.1).
Evaluated within cultivar variability was very high for all investigated root traits
and the cv for root DMY ranged between 22 % in cv. Mara and 28 % in cv. Cashel.
There were no statistically significant differences between cultivars for root DM and
root/shoot ratio. Significant differences between K11 and Cashel, were scored for
proportion of root DM in top10 cm, deep root/shoot ratio and deep roots DM, while
significant difference for shoot DM weight between Mara and Cashel was obtained.
Statistically high significant differences among K11 and other two cultivars were
detected for proportion of plants with roots below 80 cm.
Within cultivars with improved persistency more plants had root portions below
80 cm than Cashel, especially in cv. K11 (40 %; Table 36.1). This indicates that those
genotypes should maintain root contact with deep soil water longer during drought
period and those few deep roots may play important role in drought survival. Also they
36 Genetic Variation of Root Characteristics and Deep Root Production . . . 279

Proportion of total root weight

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7

0,1

0,3
Root depth (m )

0,5

0,7

0,9
Perennial ryegrass cv Mara

Proportion of total root weight

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7

0,1

0,3
Root depth (m)

0,5

0,7

0,9
Perennial ryegrass cv K11

Proportion of total root weight

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7

0,1

0,3
Root depth (m)

0,5

0,7

0,9
Perennial ryegrass cv Cashel

Fig. 36.1 The proportion of total root DM obtained from 10 cm increments from 0.9 m deep plastic
sand tubes
280 D. Sokolovic et al.

Table 36.1 Average shoots and roots weights, proportions and ratios of perennial ryegrass cultivars
Traits Shoot DM Root DM Root/ Proportion Deep root/ Deep roots Plants with
cultivars weight weight shoot of root DM shoot ratio DM (mg) root below
(g) (g) ratio top 10 cm (below 60 cm) 80 cm (%)
Mara 0.54 0.64 1.21 45.4 0.18 16.22 43.33
K11 0.51 0.65 1.23 41.6 0.23 22.06 76.67
Cashel 0.48 0.6 1.27 49.6 0.14 9.44 36.67
LSD0.05 0.06 0.07 0.11 7.06 0.06 9.6 19.1
LSD0.01 0.11 0.11 0.18 11.7 0.1 15.9 31.8

showed 8 % heavier roots in total, but also higher shoot DM yield in similar proportion
(both differences are not statistically significant). Consequently, root/shoot ratio was
similar for all cultivars, showing that plants which invest more dry matter into root
do not have automatically decreased above ground DM yield.
Proportions of roots up to 10 cm deep showed that almost 50 % of roots in cul-
tivar Cashel laid in the shallow soil, which may be important in better nutrients
consumption, but it is negatively correlated with drought resistance (Carrow 1996).
More important is that cultivars Mara and especially K11 had about 100 % in-
creased deep roots DM (Table 36.1) which resulted in higher deep root/shoot ratio.
These differences obviously resulted in better field persistency, drought tolerance
and subsequently in higher DMY of cultivars K11 and Mara. Those two cultivars
also had high percentage of plants with root reaching below 80 cm, respectively. All
such plants are fitted in selection model and can be used for Lolium perenne drought
resistance breeding in Serbia.

Acknowledgments This research has been funded by Ministry of education and science of Republic
of Serbia (Project TR31057).

References

Blamey FPC, Edmeades DC, Asher CJ, Edwards DG, Wheeler DM (1991) Evaluation of solution
culture techniques for studying aluminium toxicity in plants, 905–912. In: Wright RJ et al (eds)
Plant and soil interaction at low pH. Kluwer Academic Press, Dordrecht, Netherlands
Bonos SA, Rush D, Hignight K, Meyer WA (2004) Selection for deep root production in tall fescue
and perennial ryegrass. Crop Sci 44:1770–1776
Bullitta S (1996) Response for selection for root dimension in a Mediterranean natural population
of Lolium rigidum Gaud. Plant Breed 115:63–66
Carrow RN (1996) Drought-avoidance characteristics of diverse tall-fescue cultivars. Crop Sci
2:371–377
Crush JR, Waller JE, Care DA (2005) Root distribution and nitrate interception in eleven temperate
forage grasses. Grass Forage Sci 60:385–392
Crush JR, Easton HS, Waller JE, Hume DE, Faville MJ (2007) Genotypic variation in patterns of
root distribution, nitrate interception and response to moisture stress of a perennial ryegrass
(Lolium perenne L.) mapping population. Grass Forage Sci 62:265–273
36 Genetic Variation of Root Characteristics and Deep Root Production . . . 281

Guthridge KM, Smith KF, McFarlane NM, Kirkwood BD, Jones ES, Forster JW (2001) Developing
molecular markers for traits associated with drought tolerance in perennial ryegrass (Lolium
perenne L.). In: Rowe B et al (eds) Proceedings of the 10th Australian Agronomy Conference,
“Science and Technology: Delivering Results for Agriculture”, Hobart, Tasmania, Australia,
http://www.regional.org.au/au/asa/2001/p/9/smith.htm#TopOfPage
Huang B, Duncan RR, Carrow RN (1997) Drought-resistance mechanisms of seven warm-season
turfgrasses under surface soil drying. II. Root aspects. Crop Sci 37:1863–1869
Lehman VG, Engelke MC (1991) Heritability estimates of creeping bentgrass root systems grown
in flexible tubes. Crop Sci 31:1680–1684
Ludlow MM (1980) Stress physiology of tropical plants. Trop Grasslands 14:136–145
PalazzoAJ, Brar GS (1997) Screening of 12 Festuca cultivars for rapid root development. J Turfgrass
Manag 2:15–25
Parsons AJ (1988) The effects of season and management on the growth of grass swards. In: Jones
MB, Lazenby A (eds) The grass crop, Chapman and Hall, London, UK
Sokolović D, Tomić Z, Lugić Z (2003) Dry matter yield components of perennial ryegrass (Lolium
perenne L.) populations. Grasslands Sci Euro 8:126–130
Sokolović D, Radović J, Lugić Z, Tomić Z, Babić S (2007) Genetic variability of perennial ryegrass
(Lolium perenne L.) autochthonous populations collected in Serbia for seed yield and seed yield
components, 253–256. In: Rosellini D, Veronesi F (eds) Proceedings of 26th EUCARPIA fodder
crops and amenity grasses section, Perugia, Italy
Sokolović D, Babić S, Marković J, Radović J, Živković B, Simić A (2010) Dry matter production
and nutritive value of perennial ryegrass cultivars collection, 341–346. In: C. Huyghe (ed)
Sustainable use of genetic diversity in forage and turf breeding, Springer, Heidelberg, Germany
Thomas H (1991) Accumulation and consumption of solutes in swards of Lolium perenne during
drought and after rewatering. New Phytol 118:35–48
Volaire F (1995) Growth, carbohydrate reserves and drought survival strategies of contrasting
Dactylis glomerata populations in a Mediterranean environment. J Appl Ecol 32:56–66
Chapter 37
The Study of Similarities Among Medicago
sativa L. Accessions

D. Knotová, J. Pelikán, T. Vymyslický and S. Raab

Abstract Medicago sativa is grown on 65,000 ha of agricultural land in the Czech

Republic. To produce high quality forage with high protein content Medicago sativa
is a crucial fodder species and thus takes priority by Czech breeders. Descriptions of
forage species of the family Fabaceae have been carried out in the Research Institute
for Fodder Crops, Ltd. under the project “National Programme on Conservation
and Utilization of Plant Genetic Resources and Agro-biodiversity”. 32 accessions
(varieties) of Medicago sativa from four states (Czech Republic, France, Yemen,
Hungary), were evaluated as spaced plants. 30 plants of each variety were planted on
the field in the year 2009. In the year 2010 51 morphological, yield and qualitative
characters were evaluated on ten plants of each origin. The average values of the
observations per character and accession created the input matrix for evaluation by
the method of cluster analysis. Similarities among accessions were determined on
the basis of these results and correspond more or less to the origin of accessions and
the geographical distance.

37.1 Introduction

Nowadays alfalfa covers 65,000 ha of agricultural land in the Czech Republic (Czech
Statistical Office 2011). Despite the present trend towards a reduction of the alfalfa
growing areas, alfalfa along with corn belongs to the high-yielding crops in the
Czech Republic. Its advantages are predominantly relatively stable and high yield,
drought resistance and persistence. From the nutritional point of view alfalfa is ap-
preciated for its high digestible protein yield per ha and its low cost per kg produced
digestible (Hrabě et al. 2004). In alfalfa breeding programs the knowledge of similar-
ities between cultivars is very valuable. Genetic similarities of alfalfa were studied,
for example by Wang et al. (2011), Julier (2010), Tucak (2008) and morphological
similarities were examined by Benabderrahim et al. (2009) and others.

D. Knotová () · J. Pelikán · T. Vymyslický

Research Institute for Fodder Crops, Ltd., Zahradni 1, 66441 Troubsko, Czech Republic
e-mail: [email protected]
S. Raab
Agricultural Research, Ltd., Zahradni 1, 66441 Troubsko, Czech Republic

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 283

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_37, © Springer Science+Business Media Dordrecht 2013
284 D. Knotová et al.

37.2 Materials and Methods

In the year 2010 a population of 32 alfalfa varieties from four countries was assessed.
The population included 14 varieties from the Czech Republic, five varieties from
France, five varieties from Yemen and eight varieties from Hungary. The evaluations
were made both in individual plantings and in the stand. In individual plantings
10 plants of each accession were evaluated. Stands were evaluated in the parcels of
10 m2 , established by the method of randomised blocks. The varieties from Yemen
were evaluated only in individual plantings because they are winterkilled in these cli-
matic conditions. The evaluation was based on 51 characters (Table 37.1). Weighed,
calculated and measured traits were converted using a classifier into numbers on a
nine point scale and the other traits were assessed directly using a classifier (Vacek
et al. 1985). Cluster analysis was performed in the software Statistica for Windows
(Statsoft, Inc. 2003) for all the accessions together. The complete linkage method
was used for clustering and Euclidean distance as the measure of distance.

37.3 Results and Discussion

As you can see in Fig. 37.1 the greatest similarity was found between the varieties
Reimance and Saatry from Yemen, followed by two pairs of Czech varieties. These
were the pairs of Holyna and Denisa; and Litava and Niva. In contrast, significantly
different from the entire population of varieties was the Czech variety Palava. In
fact, five clusters were formed. The first of them contained only the varieties from
Yemen. Another cluster was composed of two Czech, two Hungarian and two French
varieties. The varieties of this cluster excelled in leaf area, number of pods per fruit
and dry weight of bunch. The largest cluster included ten Czech varieties and the
French variety Julia. In another cluster there were Hungarian varieties and the Czech
variety Jitka and the French variety Harpe. The varieties in this cluster exceeded the
other clusters in terms of terminal leaflet length, number of pods per 100 flowers
and also height and width of pod spiral. The last cluster was composed of the Hun-
garian variety Alexandra, the French variety Exquise and the Czech variety Palava.
Table 37.2 shows that the lowest variation in three traits was achieved by the Czech
varieties Holyna (green biomass weight of bunch, dry weight of bunch and number
of pods per fruit) and Morava (number of internodes, length of central internodes
and seed weight per plant). The French variety Timbale and the Hungarian varieties
KM-Maraton, Alexandra and KM-Gyongy showed the lowest variation in two traits.
On the other hand, the greatest non-uniformity was found in the variety Wasany
from Yemen (leaf area size, stem thickness, number of internodes and number of
lateral branches), the Czech variety Denisa (length of central internodes, inflores-
cence length and seed weight per plant) and the Czech variety Jarka (terminal leaflet
length, stem length and pod spiral width).
37 The Study of Similarities Among Medicago sativa L. Accessions 285

Table 37.1 Weighed, calculated, measured and scored characters

Morphological, biological Weighed, calculated Scored characters
and economic characters and measured characters
Shape of leaf rosette in autumn X
Density of leaf rosette X
Number of stems in a bunch X
Stem length X
Stem thickness X
Shape of stem in cross section X
Stem hollowness X
Stem colouration X
Number of internodes on stem X
Length of the middle internode X
Number of lateral branches on stem X
Shape of the terminal leaflet X
Margin of the terminal leaflet X
Terminal leaflet tip X
Terminal leaflet length X
Terminal leaflet width X
Terminal leaflet area X
Leaf colour X
Leaf pubescence X
Leaf area X
Occurrence of compound leaves X
Inflorescence shape X
Inflorescence length X
Number of inflorescences per stem X
Number of florets per inflorescence X
Inflorescence colour X
Number of pods on the stem X
Number of pods per inflorescence X
Number of pods per 100 flowers X
Pod shape X
Pod colour X
Height of pod spiral X
Width of pod spiral X
Number of seeds per pod X
Seed shape X
Seed colour X
1,000-seed weight X
Resistance to winterkill X
Stand height at the start of blooming X
Stand height in full bloom X
Stand height 20 days after the first cut X
Stand re-growth X
Number of cut per year X
Total green biomass weight X
Green matter yield per plant X
Total hay yield of stand X
A ratio of hay yield at the first cut to total
hay yield per year X
Seed yield per stand X
Seed weight per plant X
Content of nitrous substances X
Fibre content X
286 D. Knotová et al.

Medicago sativa
Complete linkage
Euclidean distances
PalavaCZE
ExquiseFRA
AlexandraHUN
HarpeFRA
JitkaCZE
KM-GyongyHUN
KM-MaratonHUN
SzapkoHUN
KlaudiaHUN
JozsóHUN
JarkaCZE
ZuzanaCZE
JuliaFRA
LitavaCZE
NivaCZE
TerezaCZE
HolynaCZE
DenisaCZE
KamilaCZE
MoravaCZE
OslavaCZE
TimbaleFRA
KM-BossyHUN
Hunor 40HUN
MagdaCZE
GalaxieFRA
VlastaCZE
BahoudyYMD
ReimanceYMD
SaatryYMD
WasanyYMD
ColiYMD

4 6 8 10 12 14 16 18 20
Distance of connection

Fig. 37.1 Dendrogram of accessions of Medicago sativa L.

Table 37.2 Survey of maximum and minimum values of coefficients of variation as level of
uniformity in measured, weighed and calculated traits
Characters Variety Vx (%) min Variety Vx (%) max
Terminal leaflet length (mm) KM-Maraton 5.28 Jarka 19.98
Terminal leaflet width (mm) KM-Gyongy 8.33 Tereza 27.19
Leaf area (cm2 ) Alexandra 13.40 Wasany 35.64
Terminal leaflet area (cm2 ) Alexandra 12.57 Jozsó 31.98
Stem length (cm) Zuzana 6.09 Jarka 31.61
Stem thickness (mm) Timbale 7.38 Wasany 34.48
Number of internodes on stem Morava 9.13 Wasany 30.02
Length of the middle internode (cm) Morava 12.30 Denisa 37.00
Number of lateral branches on stem Jozsó 11.28 Wasany 43.70
Green matter yield per plant (kg) Holyna 17.07 Saatry 72.8
Number of stems in a bunch KM-Maraton 19.73 Bahoudy 75.96
Dry matter yield per plant (kg) Holyna 16.54 Saatry 84.35
Inflorescence length (cm) Timbale 13.41 Denisa 50.03
Number of inflorescence on stem Kamila 20.24 Vlasta 74.36
Number of florets per inflorescence Galaxie 20.60 Reimance 53.91
Number of pods on the stem Litava 24.78 Palava 101.91
Number of pods per inflorescence Holyna 22.05 Vlasta 59.83
Number of pods per 100 flowers Jitka 6.69 Palava 22.21
Height of pod spiral (mm) KM-Gyongy 11.13 Morava 39.95
Width of pod spiral (mm) Szapko 70.94 Jarka 20.62
Number of seeds per pod Denisa 21.88 KM-Bossy 57.68
Seed weight per plant (g) Morava 38.73 Denisa 118.43
37 The Study of Similarities Among Medicago sativa L. Accessions 287

37.4 Conclusions

On the basis of morphological and yield characteristics the studied population may
be split into clusters, which correspond more or less with the countries of origin.
These results can be further used in crop breeding programmes.

Acknowledgments The work was supported by the Ministry of Agriculture of the Czech Re-
public (National Programme on Conservation and Utilization of Plant Genetic Resources and
Agro-biodiversity) and Ministry of education, youth and sports (Institutional research project no.
MSM 2629608001).

References

Benabderrahim MA, Mansour H, Ali F (2009) Diversity of lucerne (Medicago sativa L.) populations
in south Tunisia. Pakistan J Bot 41:2851–2861
Czech Statistical Office (2011) Agriculture areas. http://www.czso.cz. Accessed June 2011
Hrabě F at al (2004) Grasses and forage-grasses in agricultural practice. Czech Republic
Julier B, Semiani Y, Laouar M (2010) Genetic diversity in a collection of lucerne populations from
the mediterranean basin evaluated by SSR markers. Sustainable use of genetic diversity in forage
and turf breeding, pp 107–112. doi:10.1007/978-90-481-8706-5_14
Statsoft Inc (2003) Statistica Cz (Software system for data analysis), version 6. Www.StatSoft.cz
Tucak M, Popovic S, Cupic T, Grljusic S, Bolaric S, Kozumplik, V (2008) Genetic diversity of
alfalfa (Medicago spp.) estimated by molecular markers and morphological characters. Period
Biol 110:243–249
Vacek V, Mrázková V, Sestrienka A, Sehnalová J, Bareš I, Hájek D (1985) Descriptor list of genus
Medicago L. Czech Republic
Wang X, Yang X, Chen L, Feng G, Zhang J, Jin L (2011) Genetic diversity among alfalfa
(Medicago sativa L.) cultivars in Northwest China. ACTA Agr Scand B-S P 61, pp 60–66.
doi:10.1080/09064710903496519
Chapter 38
Genetic Structure and Agronomic Value
of Italian Lucerne Landraces: A Synopsis

P. Annicchiarico

Abstract Because of their long history of cultivation and good characterization of

collecting sites, lucerne landraces from northern Italy are particularly useful for stud-
ies of genetic structure, evolutionary adaptation to specific growing conditions and
comparison with modern germplasm, which can drive the efficient management and
exploitation of landrace germplasm. This work aims to summarize the information
from such studies. Each study included ten to 13 farm landraces belonging to the
seven former commercial ecotypes of northern Italy, and four to seven well-adapted
varieties. Each landrace was represented by over 250 single plants grown in dense
stand under field conditions or in four artificial environments created by the combi-
nation of almost nil or high summer drought with sandy-loam or silty-clay soil. On
average, landrace germplasm was at least as performing as variety material for forage
yield and persistence over two or three years. Landraces exhibited specific adapta-
tion towards drought-stress levels similar to those of their environment of origin.
No mean difference between landrace and variety germplasm emerged for forage
quality (as estimated by the leaf:stem ratio) or seed yield. The estimated genetic
variance within population was always much larger than that among populations.
These results support the exploitation of landrace germplasm through breeding pro-
cedures able to exploit its remarkable within-population diversity and its adaptation
to specific environments.

Keywords Genetic resources · Genetic variation · Genotype × environment

interaction · Forage quality · Forage yield · Medicago sativa · Plant adaptation ·
Seed yield

38.1 Introduction

Lucerne (Medicago sativa L. subsp. sativa, 2n = 4x = 32) was introduced to Italy

from the Middle East around the second century BC and then, after its extensive culti-
vation during the Imperial Age and its disappearance during the Barbarian invasions,
was introduced again from Spain in the sixteenth century (Prosperi et al. 1995).

P. Annicchiarico ()
Centro di Ricerca per le Produzioni Foraggere e Lattiero-Casearie (CRA-FLC), Lodi, Italy
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 289

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_38, © Springer Science+Business Media Dordrecht 2013
290 P. Annicchiarico

Friulana di Udine
Premariacco
18
Leonicena Vicenza
19
Milan Venice
Lodi
10
Turin Cremonese Tipica Basso Friuli
17
Po river 16 Mantova

12 8 11 Polesana
Vogherese Reggio Bologna
4 25
3
Romagnola

Fig. 38.1 Site of origin of eleven farm landraces and production area of seven commercial ecotypes
of lucerne

Lucerne landraces were extensively grown in Italy until recently. Landrace seed,
grouped into 14 commercial ecotypes depending on its area of production, accounted
for about 62 % of the Italian seed market in 1997 (Annicchiarico et al. 2007). The ban
to commercialization of landrace seed from 2003, justified by the lack of effective
control on its origin, was accompanied by the collection of landraces with a long
track of multiplication in the same farm. The Italian history of landrace cultivation
contrasts with that of most European countries, whose earlier ban to marketing of
landrace seed (e.g. from 1970 in France) has led to much greater loss of indigenous
genetic resources (Julier 1996).
Because of their long history of cultivation and the large availability of accessions
with well-known environment of origin, lucerne landraces from northern Italy are
particularly useful for studies of genetic structure, evolutionary adaptation to spe-
cific growing conditions, and comparison with modern germplasm. The objective
of this work was to summarize the available information provided by such studies,
whose results can drive the efficient management and exploitation of landrace genetic
resources.

38.2 Materials and Methods

Each relevant study included at least one landrace for each of the seven former com-
mercial ecotypes of northern Italy, along with various well-adapted varieties. The
landraces were collected from farms that had long been registered as seed donors for
the relevant ecotype and had been multiplying their own seed for at least 20 years.
The collecting site of eleven landraces and the production area of the seven commer-
cial ecotypes are reported in Fig. 38.1. Dense planting was adopted in all studies,
given the only moderate reliability of spaced planting for assessing production traits
(Annicchiarico 2006a).
38 Genetic Structure and Agronomic Value of Italian Lucerne Landraces: A Synopsis 291

One study reported in Annicchiarico (2006b) included 4480 unreplicated geno-

types that derived from seed of eleven landraces and seven varieties, as well as
128 clones of a single control genotype used to estimate the environmental variance.
Each landrace was represented by 256 or 512 plants. Seedling and clonal material
was subdivided into 128 different units (grids), each including one clone and 35
plants with a fixed proportion for each population. The 128 units were arranged in
four parallel columns of 32 adjacent units, readjusting the data for the effects of
column of units, row of units and column of plants as described in Annicchiarico
(2004). The adopted plant density was 100 plants/m2 . This study allowed compar-
ing landrace and variety germplasm for mean value and for among-population and
within-population genetic variance relative to dry matter (DM) yield and final sur-
vival over 2 years and other agronomic traits. A second study based on a subset of
this material, i.e. 10 landraces and 6 varieties, investigated the genetic variation for
forage quality as estimated by the leaf:stem ratio (Annicchiarico 2007a).
Another study including 13 landraces and 4 varieties compared landrace and vari-
ety germplasm for three-year DM yield, survival and adaptation pattern across 4 large
artificial environments (24.0 × 1.6 × 0.8 m deep) created by the factorial combina-
tion of almost nil or high summer drought with sandy-loam or silty-clay soil. These
environments were able to reproduce the adaptive responses of lucerne cultivars
across agricultural environments of northern Italy (Annicchiarico and Piano 2005;
Annicchiarico 2007b). Each environment included four replications, and each plot
comprised 70 plants grown at the density of 178 plants/m2 . The genetic variation
for seed yield of this material was investigated in one environment similar to those
frequently used for seed production (Annicchiarico et al. 2007).
The genetic structure for adaptation pattern of landrace germplasm was estimated
here from DM yield data reported in Annicchiarico (2007c). These data relate to
18 parent genotypes, 6 for each of 3 landraces with contrasting adaptation pattern,
evaluated on the basis of their half-sib progenies grown in the 4 artificial environ-
ments with contrasting soil type and drought stress level. Among-population and
within-population components of variance relative to main effects and genotype ×
environment (GE) interaction of the 18 parents were estimated by a REML proce-
dure, considering the GE interaction variation for yield response as an indicator of
the variation for adaptation pattern.

38.3 Results and Discussion

On average, landrace germplasm was at least as performing as variety germplasm

for forage yield and final persistence over 2 or 3 years (Table 38.1). The assessment
of forage yield components in one experiment indicated the trend of landraces to-
wards less stems per plant and shorter stems which was compensated, however, by
greater plant survival (Table 38.1). The study in artificial environments revealed large
variation in adaptation pattern among landrace populations which was mainly due
to cultivar × drought stress interaction (Annicchiarico and Piano 2005). Better yield
292 P. Annicchiarico

2
Table 38.1 Mean value, and ratio of within-population (sW ) to among-population (sA2 ) genetic
variance, for dry matter (DM) yield and other traits of landrace (LAN) and variety (VAR) germplasm
groups
2 2
Trait No. of entries Mean valuea sW /sA ratio
LAN VAR LAN VAR LANb VARb Referencec
DM yield (2 years; g/plant) 11 7 9.51 8.75** 25.6** 11.1** A
Main stem length (cm) 11 7 57.5 58.8* 18.8** 15.1** A
No. stems/plant 11 7 6.01 6.46** 10.3** 6.6* A
Plant survival (2 years; %) 11 7 79.8 64.2** – – A
No. florets/inflorescence 11 7 17.2 17.1 ns 70.5** 4.5* A
Leaf-to-stem ratio 10 6 1.04 1.00 ns 48.0** 7.3** B
DM yield (3 years; t/ha) 13 4 36.5 36.9 ns – – C
Plant survival (3 years; %) 13 4 40.1 41.8 ns – – D
Seed yield (t/ha) 13 4 1.11 1.06 ns – – E
Adaptation patternd 3 – – – 21.8** – F
a
Mean comparison: ns not significant *p < 0.05; **p < 0.01
2
b
Within-group comparison of sW vs. sA2 *p < 0.05; **p < 0.01
c
A Annicchiarico (2006b), B Annicchiarico (2007a), C Annicchiarico and Piano (2005), D Annic-
chiarico (2007b), E Annicchiarico et al. (2007), F estimated from data in Annicchiarico (2007c)
d
As genotype × environment interaction for DM yield in 3 harvests spanning over 2 years

response under stress conditions was closely related to the level of summer drought
at collecting sites (Annicchiarico and Piano 2005). Greater plant survival tended to
parallel the landrace adaptation to specific environments (Annicchiarico 2007b), but
was also related to more frequent mowing at collecting sites (Annicchiarico 2006b).
Landrace and variety germplasm performed comparably also for forage quality
(as expressed by the leaf:stem ratio), seed yield, and one seed yield component,
i.e. the number of florets per inflorescence (Table 38.1). Landrace seed yield was
positively associated with frequent mowing and earlier year of seed production on
its farm of origin (Annicchiarico et al. 2007).
Within-population variation was distinctly larger than among-population variation
in both landrace and variety germplasm but its relative extent was larger in landraces
(Table 38.1), as expected from their wider genetic base. Despite the large variation
among landraces for adaptation pattern, within-population variation was much larger
also for this trait on the grounds of half-sib responses of the 18 genotypes sorted out
from 3 landraces with contrasting adaptive response (Table 38.1).
The good agronomic value of Italian landrace germplasm in specific agricultural
environments has been confirmed by studies including also landraces and varieties
from central Italy (Russi and Falcinelli 1997; Annicchiarico et al. 2012), or landraces
from Sardinia and Sicily and varieties from Europe, Australia and USA tested in
north-African and European locations (Annicchiarico et al. 2011). On the whole,
these findings suggest that the widespread historical adoption of landraces by Italian
farmers was also due to their good forage yield and persistence besides the lower cost
of their seed. The difficulty of modern varieties to distinctly outperform landraces
38 Genetic Structure and Agronomic Value of Italian Lucerne Landraces: A Synopsis 293

and old cultivars emerged not only in Italy but also elsewhere, for example in USA,
where lucerne breeding programmes improved mainly the tolerance to diseases while
failing to achieve substantial progress for intrinsic forage yielding ability or forage
quality (Lamb et al. 2006). Various factors may account for the low rate of yield
progress in lucerne relative to other crops, including its out-crossing mating sys-
tem, autotetraploidy, perennial growth cycle, and high rate of non-additive genetic
variance arising from gene interaction (Bingham et al. 1994; Brummer 1999).
Another factor hindering lucerne breeding progress is the high GE interaction,
which is widespread also in varieties and whose magnitude has been revealed also by
the site-specific nature of quantitative trait loci for forage yield (Robins et al. 2007).
The adaptation of Italian landraces and varieties to moisture-favourable or drought-
prone environments was related to different and partly incompatible adaptive traits,
complicating the selection for wide adaptation (Annicchiarico 2007b).
On the whole, the information summarized here supports the exploitation of
landrace germplasm through breeding procedures able to exploit its remarkable
within-population diversity and its adaptation to specific environments, especially
moisture-favourable or drought-prone ones.

Acknowledgments This work was carried out within the project ‘Plant Genetic Resources—FAO
Treaty’ funded by the Italian Ministry for Agriculture, Food and Forestry Policies.

References

Annicchiarico P (2004) A low-cost procedure for multi-purpose, large-scale field evaluation of

forage crop genetic resources. Euphytica 14:223–229
Annicchiarico P (2006a) Prediction of indirect selection for lucerne seed and forage yield from
space-planted evaluation. Plant Breed 125:641–643
Annicchiarico P (2006b) Diversity, genetic structure, distinctness and agronomic value of Italian
lucerne (Medicago sativa L.) landraces. Euphytica 148:269–282
Annicchiarico P (2007a) Inter- and intra-population genetic variation for leaf:stem ratio in landraces
and varieties of lucerne. Grass Forage Sci 62:100–103
Annicchiarico P (2007b) Lucerne shoot and root traits associated with adaptation to favourable or
drought-stress environments and to contrasting soil types. Field Crops Res 102:51–59
Annicchiarico P (2007c) Wide- versus specific-adaptation strategy for lucerne breeding in northern
Italy. Theor Appl Genet 114:647–657
Annicchiarico P, Piano E (2005) Use of artificial environments to reproduce and exploit genotype
x location interaction for lucerne in northern Italy. Theor Appl Genet 110:219–227
Annicchiarico P, Pecetti L, Romani M (2007) Seed yielding ability of landraces of lucerne in Italy.
Grass Forage Sci 62:507–510
Annicchiarico P, Pecetti L, Abdelguerfi A, Bouizgaren A, Carroni AM, Hayek T, M’Hammadi Bouz-
ina M, Mezni M (2011) Adaptation of landrace and variety germplasm and selection strategies
for lucerne in the Mediterranean basin. Field Crops Res 120:283–291
Annicchiarico P, Pecetti L, Torricelli R (2012) Impact of landrace germplasm, non-conventional
habit and regional cultivar selection on forage and seed yield of organically-grown lucerne in
Italy. J Agric Sci 150:345–355
Bingham ET, Groose RW, Woodfield DR, Kidwell KK (1994) Complementary gene interactions in
alfalfa are greater in autotetraploids than diploids. Crop Sci 34:823–829
294 P. Annicchiarico

Brummer EC (1999) Capturing heterosis in forage crop cultivar development. Crop Sci 39:943–954
Julier B (1996) Traditional seed maintainance and origins of the French lucerne landraces. Euphytica
92:353–357
Lamb JFS, Undersander DJ, Brummer EC, Sulc RM, Sheaffer CC, Rhodes LH (2006) Five decades
of alfalfa improvement: impact on forage yield, persistence, and nutritive value. Crop Sci
46:902–909
Prosperi JM, Guy P, Genier G, Angevain M (1995) Les luzernes ou le genre Medicago. In: Prosperi
JM, Guy P, Balfourier F (eds) Ressources génétiques des plantes fourragères et à gazon, pp
131–168. BRG-INRA, Paris
Robins JG, Luth D, Campbell TA, Bauchan GR, He C, Viands DR, Hansen JL, Brummer EC (2007)
Genetic mapping of biomass production in tetraploid alfalfa. Crop Sci 47:1–10
Russi L, Falcinelli M (1997) Characterization and agronomic value of Italian landraces of lucerne
(Medicago sativa). J Agric Sci 129:267–277
Chapter 39
The Use of Genebank Accessions in the Breeding
Programme of Lolium perenne

A. Ghesquiere and J. Baert

Abstract Using genetic resources could add some valuable genes to the genepool
used in the breeding programme. In 2005 we carried out paircrosses between material
from the breeding pool and 30 accessions from the ECP/GR Lolium core collection
with the aim of enhancing the heterosis and introducing valuable genes. The acces-
sions were selected based on geographical spreading of the collection sites, on the
genetic diversity analysed by AFLP markers and on the data generated on the Lolium
core collection experiment. The F1 seeds from the paircrosses were multiplied to
synthetic progeny families in 2007. From 23 multiplications we obtained enough
seeds to sow a yield trial in May 2008. The trial was cut five times in 2009 and
2010. The dry matter yield was measured and the rust resistance and sod density
were evaluated. In 2009 samples were taken at all cuts to determine the contents of
water soluble carbohydrates, digestible organic matter and crude protein. None of
the progenies performed better than the standard varieties (Aberavon, Barata, Can-
can and Premium). Only one of the progenies had a total dry matter yield that was
similar to the total dry matter yield of the standard varieties. The progenies of the
crossings between the genebank accessions and the breeding material showed no
heterosis effect on dry matter yield. Within the five best performing progenies we
will carry out further selections.

39.1 Introduction

There are thousands of accessions of forage crops stored in genebanks all over the
world. In most cases the description is limited to collection data and no evaluation
data are available. Due to the huge amount of accessions and the limited information
of these genetic resources choosing the appropriate accessions for breeding purposes
is difficult. Nevertheless, using genetic resources could add some valuable genes to
the genepool used in the breeding programme. In this study we crossed accessions
from the ECP/GR Lolium core collection with perennial ryegrass material from our
breeding programme. The progenies were tested in a yield trial.

A. Ghesquiere () · J. Baert

Institute for Agricultural and Fisheries Research (ILVO), Caritasstraat 21, 9090 Melle, Belgium
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 295

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_39, © Springer Science+Business Media Dordrecht 2013
296 A. Ghesquiere and J. Baert

39.2 Material and Methods

Under the aegis of the Forages Network of the ECP/GR (European Cooperative Pro-
gramme for Crop Genetic Resources Networks), a core collection of 162 populations
of perennial ryegrass has been formed and evaluated at 19 sites, including at ILVO
(Sackville Hamilton et al. 1998). None of the plants of this evaluation trial were used
in crossings in the breeding programme at ILVO. In 2005 we decided to carry out
paircrosses between material from the breeding pool and 30 accessions from the core
collection with the aim of enhancing the heterosis and introducing valuable genes.
The accessions were selected based on geographical spreading of the collection sites,
on the genetic diversity analysed by AFLP markers (Calsyn et al. 2005; Ghesquiere
et al. 2003) and on the data generated during the Lolium core collection experiment
(Sackville Hamilton et al. 1998; Edmondson 2005 and Humphreys, personal com-
munication). Since there was no seed available from the Polish accessions, we had
to exclude these from the experiment. All the early and intermediate heading acces-
sions were crossed with plants of the intermediate heading ILVO variety Plenty, the
late accessions were crossed with late material from our breeding programme. From
each accession we performed three paircrosses in crossing cells. The seeds were
harvested in the summer of 2005. Crosses from 5 accessions were discarded because
of a too large difference in heading date between the two plants in the crossing cell,
a too high disease susceptibility or a poor seed set. The three paircrosses of each of
the 25 accessions of the Lolium core collection with material from our breeding pool
were multiplied together. Therefore we have grown 72 F1-seedlings per cross, half
of them harvested on the three plants of the genebank accession, the other half on
the three plants of the breeding pool material. These 72 plants were allowed to freely
interpollinate in collective isolations, yielding 3 F1 based synthetic seed, which was
harvested in bulk in the summer of 2007. Seed yield, thousand seed weight and ger-
mination capacity of the seeds were measured. From 23 multiplications we obtained
enough seeds to sow a yield trial. Table 39.1 shows the country of origin of the
accessions used in the paircrosses, in the 3 F1 based synthetics and in the yield trial.
In May 2008 we have sown a yield trial to evaluate the performance of the 23
progenies. The trial was cut 5 times in 2009 and 2010. The dry matter yield (DMY)
was measured at all cuts and the rust resistance and sod density were evaluated.
In 2009 we took samples at all cuts to determine the contents of water soluble
carbohydrates, digestible organic matter and crude protein. These analyses were
performed by Near Infrared Spectroscopy (NIRS).

39.3 Results and Discussion

The mean seed yield per plant harvested on the 3 paircrosses varied from 16 to 894
seeds (a crossing between a late French population and a late breeding population
and between an intermediate French population and Plenty respectively). In the 3 F1
based synthetics there was also a large variation in the number of germinating seeds
39 The Use of Genebank Accessions in the Breeding Programme of Lolium perenne 297

Table 39.1 Number of Country of origin Number of accessions

accessions per country of
origin used in paircrosses In paircrosses In synthetics In yield trial
with material from the Belgium (BE) 1 1 1
breeding programme, in the 3 Bulgaria (BG) 1 1 1
F1 based synthetics and in the Czech Republic (CZ) 1 1 1
yield trial France (FR) 5 5 5
Germany (DE) 4 4 3
Greece (GR) 1 0 0
Hungary (HU) 3 3 2
Ireland (IE) 1 0 0
Italy (IT) 1 1 1
Norway (NO) 1 1 1
Romania (RO) 3 3 3
Spain (ES) 2 2 2
Switzerland (CH) 1 0 0
The Netherlands (NL) 1 1 1
United Kingdom (GB) 4 2 2

harvested per plant, varying from 185 (Hungarian population) to 2482 (Norwegian
population).
On average the total DMY was significantly lower for the progenies of the eco-
types than for the standard varieties (Aberavon, Barata, Cancan and Premium) in
2009 (9.08 and 10.29 ton/ha respectively). In 2010 the mean DMY did not differ
significantly: 8.22 ton/ha for the progeny families compared to 8.48 ton/ha for the
standard varieties. Figure 39.1 shows the total DMY in both testing years. The pro-
genies of a crossing between an intermediate Spanish accession and Plenty (ES2)
was yielding as much as the standard varieties in both years.
The sod density in early spring 2010 was on average similar for the varieties
and progenies of the ecotypes. The Hungarian, Norwegian and Romanian progenies
tended to have a better sod density after the winter. Since the sod density and the
DMY are similar for the standard varieties and the progenies of the ecotypes in
the second harvesting year, we can assume that most of the ecotypes have a good
persistency.
On average there was no difference between the rust resistance of the progenies of
the ecotypes and the standard varieties. One German, two French and the Norwegian
progenies were more susceptible than the standard varieties.
The digestible organic matter (DOM) was similar for the varieties and the proge-
nies of the accessions (80.9 and 80.3 respectively). The crude protein content (CP)
tended to be higher for the progenies versus the standard varieties (13.3 and 12.5
respectively), but the water soluble carbohydrates (WSC) tended to be higher for
the varieties (20.0 versus 18.5). There was a large variation in WSC ranging from
18.6 (Barata) to 21.4 (Aberavon) for the standard varieties and from 15.1 (Romanian
progeny) to 20.7 (Spanish progeny) for the progenies of the ecotypes. Table 39.2
gives an overview of the results of the rust resistance and the sod density evaluation
and the mean content of WSC, DOM and CP.
298 A. Ghesquiere and J. Baert

Fig. 39.1 Total dry matter yield (relative to the mean of the standard varieties Aberavon, Barata,
Cancan and Premium) of the standard varieties and the progenies of the crossings between European
accessions and material from our breeding programme in 2009 and 2010

Table 39.2 Results of the rust resistance and the sod density evaluation (scoring from 1 to 5;
5 = good) and the content of water soluble carbohydrates, digestible organic matter and crude
protein of the standard varieties (Aberavon, Barata, Cancan and Premium) and the progenies of the
crossings between European accessions and material from our breeding programme
Standard varieties Accessions
Mean Range Mean Range
Rust resistance 3.3 3.0–3.7 3.0 1.0–4.3
Sod density 3.2 2.0–4.0 3.2 2.3–4.3
Digestible organic matter 80.9 80.0–82.9 80.3 78.0–82.1
Water soluble carbohydrates 20.0 18.6–21.4 18.5 15.1–20.7
Crude protein 12.5 12.0–12.9 13.3 12.5–14.5

None of the progenies performed better than the standard varieties and we will
not create directly new candidate varieties. Within the five best performing progenies
(CZ1, ES2, GB2, HU2, RO3) we will carry out further selections.

39.4 Conclusion

Using genetic resources in the breeding programme can add some valuable genes
to the genepool. Due to the limited information of these genetic resources it is hard
to choose the appropriate accessions. No heterosis effect on dry matter yield after
39 The Use of Genebank Accessions in the Breeding Programme of Lolium perenne 299

crossing diverse accessions with material from our breeding pool was observed. In
this study the progenies of some crossings had a good persistency and will be used
for further selections.

References

Calsyn E, Ghesquiere A, Baert J, De Riek J (2005) Study of genetic diversity between and within
ryegrass populations of the ECP/GR collection by means of AFLP markers. Report of a working
group on forages of ECP/GR: 8th meeting. Linz, Austria, 10-12 April 2003: 122–131.
Edmondson RN (2005) Appendix II. Statistical analysis of the IGER European Lolium core collec-
tion trial. Report of a working group on forages of ECP/GR: 8th meeting. Linz, Austria, 10-12
April 2003: 188–189
Ghesquiere A, Calsyn E, Baert J, De Riek J (2003) Genetic diversity between and within ryegrass
populations of the ECP/GR collection by means of AFLP markers. Czech J Genet Plant Breed
39:333–336
Sackville Hamilton NR, Marum P, Thomas ID, Balfourier F, Boller B, Connolly V, Dologa G,
Fritsen H, Ghesquiere A, Horvath L, Negri V, Oliveira JA, Ostrem L, Posselt UK, Reheul D,
Schmidt J, Sevcikova M, Willner E, Chapurin VF, Shamov D, Vaitsis T, Gass T (1998) The
european ryegrass core collection: some Preliminary Results. Proceedings of the 21th meeting
of the fodder crops and amenity grasses secion of eucarpia, pp 145–148
Chapter 40
Characterization and Evaluation of Genebank
Accessions as a Pre-selection Instrument for
Plant Breeding Objectives and Strategies

S. Nehrlich, E. Willner and K. J. Dehmer

Abstract The federal German ex-situ collection of agricultural and horticultural

plants at the Leibniz Institute of Plant Genetics and Crop Plant Research (IPK)—
the IPK Genebank—with its headquarters in Gatersleben and a satellite collection
for fodder crops at Malchow/Poel, is one of the most comprehensive plant genetic
resources (PGR) collections worldwide. In addition to collection, conservation, doc-
umentation and distribution activities, the characterization and evaluation comprises
a major part of the work at the IPK Genebank, particularly in field trials of collected
material with common geographical origin or habitat. Evaluated traits of grasses
include plant development before and after winter, growth type, length and width of
flag leaf and plant height or resistance to diseases. We assessed Irish Lolium perenne
accessions for relevant breeding traits to give an example of IPK Genebank activi-
ties. On the basis of the monitored traits we are able to predict the behavior of plants
against abiotic stress (e.g. drought, late frost and winter hardiness), serving as a
pre-selection instrument for different plant breeding objectives and strategies.

Keywords Lolium perenne L. · Diseases · Abiotic stress

40.1 Introduction

The Malchow site of the IPK Genebank harbours the oilseed and fodder crop col-
lections of the Leibniz Institute of Plant Genetics and Crop Plant Research. The
Malchow collection has more than 14,000 accessions with 17 genera and 155 species
of grasses, crucifers and legumes. In addition to maintenance and supply, a very im-
portant aim of the genebank is the scientific characterization and evaluation of the
plant genetic resources over decades, in order to provide breeders and scientists
with high-quality and well described seed/plant material. The L. perenne collec-
tion contains some of the best characterized and evaluated accessions of the IPK
Genebank.

S. Nehrlich () · E. Willner · K. J. Dehmer

Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Genebank,
Satellite Collections North, Inselstr. 9, 23999 Malchow/Poel, Germany
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 301

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_40, © Springer Science+Business Media Dordrecht 2013
302 S. Nehrlich et al.

In the last two decades, an initiative between the IPK Genebank and German grass
breeders was started to cooperate in collecting trips and plant material multiplication,
as well as in characterization and evaluation trials. As result of this initiative, the IPK
Genebank obtained new seed samples (ecotypes) that increased the range of diversity
within the genebank collection. The grass breeders, on the other hand, benefitted from
this cooperation by receiving new potential (well-adapted) starting material (Boller
et al. 2010).

40.2 Material and Methods

153 Irish L. perenne accessions from two different collection trips by Dr. V. Conolly/
Teagasc (1981/1982) and by Dr. U. Feuerstein/Euro Grass Breeding and E. Will-
ner/IPK Genebank in 2002 were evaluated and characterized in the Malchow Satellite
Collection of the IPK. The material was split into three parts: 98 accessions from
the 1981/1982 trip, 55 genebank accessions from the 2002 collection trip and 8 cul-
tivars as standards (Aberelf, Barcley, Cancan, Fennema, Gladio, Juwel, Respect and
Sambin). The plants were cultivated in the field in 2008–2010. For each genotype,
10 plants were grown and repeated four times in randomised blocks.
Quality traits were evaluated each year, including disease susceptibility, growth
before and after winter, spring development, heading and flowering date, homogene-
ity and green matter yield. For evaluating the collected material the most important
forage breeding aspects like rust susceptibility, green matter development, before
winter and spring development were screened using a 1–9 scale (1 = minimum,
9 = maximum).

40.3 Results and Discussion

153 Irish L. perenne accessions were categorized into five maturity groups (MG)
according to the Federal Plant Variety Office in Germany (BSA). The five categorized
maturity groups, depending on heading date, were MG 1 (n = 84) for very early, MG
2 (n = 31) for very early to early, MG 3 (n = 9) for early, MG 4 (n = 27) for early
to middle, MG 5 (n = 2) for middle to late. Due to the large number of accession
numbers in the maturity groups only the most important (according to minimum,
maximum) and interesting accessions of MG 1, 2 and 4 were selected for the results
presented here. Because of the lower number of accessions, the maturity group 5
was not compared with MG 1–4.
For the comparisons between the individual MGs, the results were shown using
box plots.
For rust susceptibility (Fig. 40.1a) the only difference was detected between MG
1 and 4 showing that early to middle maturity types are less susceptible to rust. There
were no differences detectable between the very early and MG 2 and 3. Genotypes
in maturity group 3 and 4 revealed a low susceptibility to rust. For green matter
40 Characterization and Evaluation of Genebank Accessions as a Pre-selection . . . 303

9 9
8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1
0 0
MG 1 MG 2 MG 3 MG 4 MG 5 MG 1 MG 2 MG 3 MG 4 MG 5
a b
9 9

8 8

7 7

6 6

5 5

4 4

3 3

2 2

1 1
0 0
MG 1 MG 2 MG 3 MG 4 MG 5 MG 1 MG 2 MG 3 MG 4 MG 5
c d
Fig. 40.1 Box plots presenting the variability of evaluated traits between the five maturity groups.
a Rust susceptibility, 2009, b Green matter estimation in 2009, c Plant development before winter,
2009, d Plant development in spring, 2010

(Fig. 40.1b), no differences between the groups (MG 1–3) were observed. However,
accessions of group 4 displayed a higher variation compared to the other groups.
For the trait plant development before winter, there was considerable variability
between groups 1–4, with evaluation scores from low (3) to high (8) plant growth
(Fig. 40.1c). For the plant development in spring, a huge amount of variability was
found (Fig. 40.1d) in the second winter season with a long cold period in 2010. Very
early to early maturity groups (MG1, 2 and 4) were particularly variable and differed
in their evaluation values from 1 to 8.
Thus, the analysis of rust susceptibility, green matter development, growth before
winter and in spring within the MGs showed a huge variability according to the
evaluated traits and in comparison to the standards.
Within the first maturity group (Fig. 40.2a) some accessions (1, 4 and 5) indicated
better performance with respect to all of the analysed traits, while accessions with a
high susceptibility to rust showed reduced biomass production and a delayed or min-
imum spring development (No. 26 and 28) in comparison to the standards Fennema,
Juwel and Sambin. In the second maturity group (Fig. 40.2b), compared to Respect
as a standard variety, several entries of Irish material showed better results for some
traits. For example, accession number 5 for plant development after winter and in
spring. In the third maturity group (Fig. 40.2c) only a small amount of variation was
observed between the accessions. The rust susceptibility ranked between 4 and 6.
The fourth maturity group (Fig. 40.2d) contained some better performing accessions
304 S. Nehrlich et al.

Fig. 40.2 Selected Irish L. perenne accessions of the five maturity groups compared to cultivars
as standards, which are marked with S compared to rust susceptibility, green matter development,
growth before winter and in spring for maturity group 1 (a), 2 (b), 3 (c), 4 (d), 5 (e)
40 Characterization and Evaluation of Genebank Accessions as a Pre-selection . . . 305

compared to the standard (Gladio). Accession numbers 2–4 produced high biomass,
had a good plant development before winter and spring and a moderate rust suscepti-
bility (between 4 and 5). In the fifth maturity group (Fig. 40.2e) only two accessions
were observed. In comparison to the standards (Barcley and Cancan) there were no
differences in this group.
The results demonstrate that some of the collected material had superior perfor-
mance in forage breeding traits compared to the standard varieties. The Irish ecotypes,
collected in meadows and pastures, are presumably well-adapted to the Irish climate.
However, for German cultivation conditions, harder and longer winter periods have
to be considered as factores influencing plant development and crop growth (Figs.
40.1a, 40.1c). Regarding plant development before winter and in spring, the results
varied enormously and it can be expected that this material will provide some useful
accessions suitable for the German climate. In contrast, rust susceptibility between
the maturity groups hardly varied.
Similar results were observed with Romanian collected plant material in the four
regions of Crisana, Transylvania, Carpathians and Subcarpathians. Willner et al.
(2010) described 455 Romanian ecotypes for rust susceptibility and only detected
slight differences between and within the four regions. However, in comparison to
standard varieties and traits, collected plant material had partly better economical
qualities and climate adapted traits than standard varieties.

40.4 Conclusions

Promising candidates for breeding objectives could be identified with a wide range
of variability between accessions that were partly better than standard varieties,
especially in the very early to middle maturity group. High variability within maturity
groups concerning green matter, rust susceptibility, and growth before winter and
in spring was observed. An extensive variability for breeding and research purposes
exists within and between maturity groups. Based on these results, plants with desired
performance against abiotic stress (e.g. winter hardiness) can be pre-selected by IPK.

References

Willner E, Hünmörder S, Dehmer KJ (2010) Sustainable use of genetic diversity in forage and turf
breeding. Springer, Netherlands
Boller B, Posselt UK, Veronesi F (2010) Fodder crops and amenity grasses. Handbook of Plant
Breeding. Springer, New York
Chapter 41
Exploitation of ‘Site-Specific’ Alpine Grass
Germplasm for Revegetation at High Altitude

L. Pecetti, M. Romani, P. Spoleto, A. Tosca, G. Della Marianna

and F. Gusmeroli

Abstract The use of ‘site-specific’ germplasm, that is, species adapted to the
prevailing pedo-climatic conditions and native to the same geographic context, is in-
creasingly recommended for revegetation interventions at high altitude. Germplasm
of two Alpine grass species, viz Phleum rhaeticum and Poa alpina, collected at 12
and 15 sites, respectively, across three valleys of the Rhaetian Alps, Italy, was evalu-
ated for a set of morpho-physiological characters in a mountain (1,300 m a. s. l.) and
a lowland location (81 m a. s. l.), with the main goal of identifying superior popula-
tions for further activities of selection and the secondary aim of assessing the effect
exerted by an environment markedly different from those of origin on growth and
seed production of site-specific species. One valley featured a slight trend of better
seed yield and more suitable plant morphology. However, individual collection sites
were by far the most important source of variation in both species. The collection of
natural populations across a given mountain district seems an appropriate strategy
to gather useful genetic variation in site-specific species. The environment of evalu-
ation interacted with the germplasm, affecting the morphology, seed yield, disease
susceptibility and survival of site-specific germplasm. Under the Italian conditions,
growth and seed multiplication in mountain sites is advised.

Keywords Alps · Ecological restoration · Phleum rhaeticum · Poa alpina ·

‘Site-specific’ species

41.1 Introduction

Revegetation of mountain landscapes altered by infrastructures for winter tourism

is necessary to restore their aesthetical appeal and prevent possible environmen-
tal damages due to increased water run-off and soil erosion. Suitable germplasm

L. Pecetti () · M. Romani

Centro di Ricerca per le Produzioni Foraggere e
Lattiero-Casearie (CRA-FLC), viale Piacenza 29, 26900 Lodi, Italy
e-mail: [email protected]
P. Spoleto · A. Tosca
Fondazione Minoprio, viale Raimondi 54, 22070 Vertemate con Minoprio, CO, Italy
G. Della Marianna · F. Gusmeroli
Fondazione Fojanini di Studi Superiori, via Valeriana 32, 23100 Sondrio, Italy

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 307

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_41, © Springer Science+Business Media Dordrecht 2013
308 L. Pecetti et al.

for revegetation at high altitude must be able to combine successfully the techni-
cal requirement of ground cover with good adaptation to the peculiar pedoclimatic
conditions. To reduce the risk of failure of revegetation at high altitude, the use of
‘site-specific’ germplasm, that is, species adapted to those conditions and native to
the same geographic context, is increasingly recommended (Krautzer 2005). A set
of key site-specific species belonging to the Poaceae, Fabaceae and a few other fam-
ilies have been identified for the formation of suitable seed mixtures (Krautzer et al.
2004).
Alpine cat’s tail [Phleum rhaeticum (Humph.) Rauschert] and Alpine bluegrass
(Poa alpina L.) are included among site-specific species. Better knowledge is needed
on the existing level of variation within species for morpho-physiological traits,
including seed production, which may affect their adoption, to drive possible further
activities of germplasm collection and selection. In this study, germplasm of these
two species was evaluated for a set of morpho-physiological characters in a mountain
and a lowland location, with the main goal of identifying superior populations and the
secondary aim of assessing the effect exerted by an environment markedly different
from those of origin on growth and seed production of site-specific species.

41.2 Materials and Methods

Twelve populations of Alpine cat’s tail and 15 of Alpine bluegrass collected above
1,800 m a. s. l. in three valleys of the Rhaetian Alps, Lombardy, Italy (namely,
Valchiavenna, Valmalenco and Upper Valtellina) were evaluated in a mountain lo-
cation of the same Alp district (Bormio, 1,300 m a. s. l.) and in a lowland location
(Lodi, 81 m a. s. l.). The trials were laid according to randomised complete block
designs with four replications. Originally collected seeds were sown in plug trays
and seedlings were subsequently field transplanted. Each single-row plot per block
included eight spaced plants in Bormio (35 cm apart) and six in Lodi (50 cm apart)
for the Alpine cat’s tail, and ten plants in both locations (35 cm apart) for the Alpine
bluegrass. Rows were in all cases 70 cm apart. On all plants of both species and
in both locations heading time (as days from 1 April to the complete emergence of
three inflorescences) and length of the main culm including its inflorescence were
recorded in the second year of growth. The number of seeds per plant was counted
at maturity in Ph. rhaeticum while it is still being processed in P. alpina. In the latter
species, the seed yield potential was visually estimated on a 9-level scale from 1
(minimum) to 9 (maximum). The susceptibility to rust (mostly Puccinia coronata
Corda, but some populations also showed susceptibility to P. graminis Pers.) and the
spring plant diameter were further recorded in P. alpina in the first and second year,
respectively. The disease reaction was recorded as individual plant score from 1 (no
symptoms) to 9 (> 75 % of foliage covered with rust) in Lodi, and as percentage
of plants per plot with score ≥ 5 (up to 25 % of affected foliage) in Bormio. An
analysis of variance (ANOVA) was carried out for all traits in both species (except
susceptibility to rust) including the fixed factors ‘location of evaluation’ and ‘valley
41 Exploitation of ‘Site-Specific’ Alpine Grass Germplasm for Revegetation . . . 309

of origin’, and the random factors ‘population within valley’ and ‘block within loca-
tion’. Susceptibility to rust in P. alpina was analysed by ANOVA separately for the
two locations, testing the variation of valleys and populations, and a Spearman rank
correlation coefficient was computed between the population mean values of the two
disease assessments.

41.3 Results and Discussion

As expected, the location of evaluation generally was a significant source of variation

(Table 41.1). Plant development was slower in the colder location of Bormio relative
to Lodi, as indicated by the delay of 33 days in mean heading time for the Alpine cat’s
tail, and 25 days for the Alpine bluegrass, in the former compared with the latter site.
Seed yield, either actual (in Ph. rhaeticum) or potential (in P. alpina), was greater
in the mountain than in the lowland location (Table 41.2). It is a common finding
that natural populations grown outside their environments of origin undergo sizeable
changes in morphology and physiology, including reproductive traits (Merrell 1981).
Successful seed production of site-specific species in lowland sites of other European
countries has been reported, though (Krautzer 2005), and the low seed production in
Lodi was likely to be accounted for by the level of summer heat and drought of this
location rather than by just its altitude.
Differences among valleys of origin were moderate at most (Table 41.1). The
Ph. rhaeticum germplasm from Valmalenco was slightly later heading than those
from the other valleys. A trend of Valchiavenna to produce more seeds per plant in
Bormio (+ 6 % than Upper Valtellina and + 33 % than Valmalenco) did not reach the
P < 0.10 significance level. This trend was consistent with the significantly (P < 0.05)
greater seed yield potential also observed in P. alpina germplasm from Valchiavenna
at Bormio (32 % greater mean score than Valmalenco and 59 % greater mean score
than Upper Valtellina). Germplasm from Valchiavenna tended to larger plant diame-
ter in both locations, with differences among valleys significant at P < 0.05 in Bormio
and P < 0.15 in Lodi (data not reported). Greater differences among valleys for
P. alpina in Bormio than in Lodi accounted for the sizeable location × valley
interaction for most characters in this species (Table 41.1).
Individual populations were by far the most important source of variation in both
species (Tables 41.1 and 41.2). Selective pressures in the individual sites of origin
were likely to cause the diversity among populations observed for most traits, in
accordance with the genecological assumption that genetic differences among natu-
ral populations are related with the environments in which they evolved (Sackville
Hamilton et al. 2002). The reported apomictic reproductive system of P. alpina (Huff
2010) might also account for the great separation among populations. The collection
of natural populations across a given mountain district seems an appropriate strategy
to gather useful genetic variation for introduction and/or breeding purposes, being
aware that possible drawbacks may occur when using non-local germplasm sources
in revegetation interventions (e.g., misadaptation; ‘genetic pollution’ of resident
310 L. Pecetti et al.

Table 41.1 Probability levels of the main sources of variation in the analysis of variance for a set
of morphological and seed production traits in two grass species from the Rhaetian Alps, Italy,
evaluated in two altitude-contrasting locations
Source of Heading time Length of No. seeds Plant diameter Seed yield
variation main culm per plant potential
Phleum rhaeticum:
Location of < 0.001 ns < 0.001 – –
evaluation (L)
Valley of origin (V) < 0.05 ns ns – –
Population within ns < 0.001 < 0.05 – –
valley [P(V)]
L ×V ns ns ns – –
L × P(V) ns ns ns – –
Poa alpina:
Location of < 0.001 < 0.001 – < 0.001 < 0.10
evaluation (L)
Valley of origin (V) ns ns – < 0.05 < 0.05
Population within < 0.001 < 0.001 – < 0.001 < 0.001
valley [P(V)]
L ×V < 0.10 ns – < 0.01 < 0.001
L × P(V) < 0.001 < 0.05 – < 0.001 < 0.001
ns not significant (P > 0.10)

Table 41.2 Mean and minimum and maximum individual values among 12 populations of Phleum
rhaeticum and 15 populations of Poa alpina evaluated in two altitude-contrasting environments.
Only characters with significant (P < 0.05) variation among populations are reported
Bormio (1,300 m a. s. l.) Lodi (81 m a. s. l.)
Mean Min Max Mean Min Max
Phleum rhaeticum:
Length of main culm (cm) 41.0 26.4 47.5 38.5 30.2 51.7
No. seeds/plant 7,200 1,680 12,165 1,256 56 3,072
Poa alpina:
Heading time (dd > Apr 1) 41.8 40.1 46.9 16.3 13.6 21.4
Length of main culm (cm) 29.2 22.6 36.8 20.8 17.4 26.7
Plant diameter (cm) 13.4 9.7 18.8 21.4 19.3 24.1
Seed yield potential (1–9) 5.1 2.8 7.7 4.5 3.0 6.2
Susceptibility to rusta 64.9 30.0 90.0 4.2 1.8 6.4
a
Recorded: (i) on a scale from 1 (no symptoms) to 9 ( > 75 % foliage covered with rust) in Lodi;
(ii) as the plot percentage of plants with score ≥ 5 (up to 25 % of affected foliage) in Bormio

germplasm). The risk of such drawbacks is limited when there is close ecological
correspondence between the target area for revegetation and the sites of origin of the
germplasm used for sowing (McKay et al. 2005).
The environment of evaluation largely interacted with the population responses
in Alpine bluegrass, affecting the phenology, morphology and seed yield potential
(Table 41.1). The population response to rust was also prone to interaction effects
41 Exploitation of ‘Site-Specific’ Alpine Grass Germplasm for Revegetation . . . 311

with the location of evaluation, as indicated by the nil rank correlation coefficient
(r = −0.11; P = 0.70) between the population values of the indexes of disease
susceptibility in the two locations. Remarkable inversions of ranking were observed
in some cases between locations, such as for the populations ‘Campolungo’, that was
among the most susceptible populations in Bormio and among the least susceptible
ones in Lodi, and ‘Val Loga 2’, that was among the most susceptible populations
in Lodi and among the least susceptible ones in Bormio. The race-specificity of the
resistance determinants in P. alpina is unknown. Studer et al. (2007) found differences
in position and magnitude of QTLs among individual evaluation locations in Italian
ryegrass (Lolium multiflorum Lam.), suggesting a differential quantitative response
to local pathogen races. In addition to race specialisation, environmental effects
might also influence rust resistance (Roderick et al. 2000).
Promising ecotypes were identified in both species, which will be used in the
ultimate selection of adapted high-seed-yielding germplasm for sowing at high alti-
tude. One of the aims of the funding institution is, in fact, the implementation of a
seed production chain of site-specific species involving smallholders of the Alpine
area. Information has been added by this study on suitable locations for the seed
multiplication of site-specific species. Under the Italian conditions, growth and seed
multiplication of this germplasm in mountain sites is advised.

Acknowledgments The research was carried out in the framework of the projects SemenSci and
SemTek funded by the General Directions ‘Agricoltura’ and ‘Sistemi Verdi’ of Regione Lombardia,
Italy.

References

Huff DR (2010) Bluegrasses. In: Boller B, Posselt U, Veronesi F (eds) Fodder crops and amenity
grasses. Handbook of plant breeding, vol 5. Springer, New York, pp 345–379
Krautzer B, Peratoner G, Bozzo F (2004) Site specific grasses and herbs, seed production and use
for restoration of mountain environments. FAO plant production and protection series, vol 32.
FAO, Italy
Krautzer B (2005) Exploitation of native plant material for green covers in environmental recla-
mation. In: Piano E (ed) Atti del Convegno ‘Inerbimenti e Tappeti Erbosi per l’Agricoltura,
l’Ambiente e la Società’, vol 2. ISCF, Italy, pp 109–117
McKay JK, Christian CE, Harrison S, Rice KJ (2005) How local is local?–A review of practical
and conceptual issues in the genetics of restoration. Rest Ecol 13:432–440
Merrell DJ (1981) Ecological genetics. Longman, London
Roderick HW, Thorogood D, Adomako B (2000) Temperature-dependent resistance to crown rust
infection in perennial ryegrass, Lolium perenne. Plant Breed 119:93–95
Sackville Hamilton NR, Skøt L, Chorlton KH, Thomas ID, Mizen S (2002) Molecular genecology
of temperature response in Lolium perenne: 1. preliminary analysis to reduce false positives.
Mol Ecol 11:1855–1863
Studer B, Boller B, Bauer E, Posselt UK, Widmer F, Kolliker R (2007) Consistent detection of QTL
for crown rust resistance in Italian ryegrass (Lolium multiflorum Lam.) across envronments and
phenotyping methods. Theor Appl Genet 115:9–17
 Part VI
Agronomy/Performance
and Compositional Analysis
Chapter 42
The Impact of Perennial Ryegrass Variety
Throughout the Growing Season on in vitro
Rumen Methane Output

P. J. Purcell, M. O’ Brien, T. M. Boland, M. McEvoy and P. O’Kiely

Abstract The selection and feeding of perennial ryegrass varieties may affect enteric
methane (CH4 ) output due to changes in rumen fermentation dynamics as a result of
differences in herbage chemical composition. Thus, the objective of this study was
to determine the effects of perennial ryegrass variety (PRV) harvested throughout the
growing season on herbage chemical composition and on in vitro rumen fermentation
variables and CH4 output. Seven PRV (Alto, Arrow, Bealey, Dunluce, Greengold,
Malone, Tyrella), managed under a simulated grazing regime, were incubated in a
batch culture for 24 h with rumen fluid and buffer. PRV had no effect (P > 0.05; SEM
0.41) on CH4 output per gram of DM incubated (CH4 i; mean values for Alto, Arrow,
Bealey, Dunluce, Greengold, Malone, Tyrella were 23.9, 24.0, 24.7, 25.3, 25.2, 24.2
and 24.7 (SEM 0.41) ml CH4 g−1 DM incubated, respectively). Although PRV had
an effect (P < 0.001; SEM 1.3) on total gas production per gram of DM incubated,
the scale of the effect was small (range of mean values among PRV was 148–160 ml),
and PRV had no effect (P > 0.05) on apparent DM disappearance during the in vitro
rumen incubation. Thus, the lack of an effect of PRV on CH4 i reflected the small
scale or lack of effects on herbage composition and in vitro rumen fermentation
variables. Hence, these results provide no encouragement that choices among the
PRV examined, produced within the management regimes operated, would reduce
enteric methane production.

42.1 Introduction

Grassland is the dominant (ca. 0.9) crop on agricultural land in Ireland, and any
enteric methane (CH4 ) mitigation strategies for ruminants must be effective within the
predominantly grass-based production systems used. Purcell et al. (2011a) showed

P. J. Purcell () · P. O’Kiely · M. O’ Brien

AGRIC, Teagasc, Grange, Dunsany, Co. Meath, Ireland
e-mail: [email protected]
P. J. Purcell · T. M. Boland
School of Agriculture, Food Science and Veterinary Medicine, University College Dublin,
Belfield, Dublin 4, Ireland
M. McEvoy
AGRIC, Teagasc, Moorepark, Fermoy, Co. Cork, Ireland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 315

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_42, © Springer Science+Business Media Dordrecht 2013
316 P. J. Purcell et al.

that altering grazing management strategy by making specific changes to herbage

mass and sward allowance had little effect on in vitro rumen CH4 output due to
limited effects on herbage chemical composition. However, Davies et al. (1991)
found differences in water soluble carbohydrate (WSC) concentration and organic
matter digestibility (OMD) between perennial ryegrass varieties (PRV). Changes in
the composition of herbage can affect the amount of CH4 produced in the rumen
(Janssen 2010). Thus, it is feasible that including PRV of higher nutritive value
within a predominantly grass-based animal production system may decrease in vitro
rumen CH4 output. The objective of this study was to determine the effects of PRV,
using varieties of differing country of origin (New Zealand or Ireland bred), ploidy
(diploid or tetraploid) and heading date (15 May–5 June), managed under a simulated
grazing regime, and thus harvested at successive stages of the growing season, on
herbage chemical composition, rumen fermentation variables and CH4 output using
an in vitro rumen gas production technique. This in vitro batch culture technique was
used because it is relatively low cost, rapid, and permits the simultaneous assessment
of a large number of treatments compared to in vivo methodologies.

42.2 Materials and Methods

Seven perennial ryegrass (Lolium perenne L.) varieties (Alto, Arrow, Bealey, Dun-
luce, Greengold, Malone, and Tyrella) were sown (2006) in a randomised complete
block (n = 3) design with repeated sampling (n = 10) of each of the plots (n = 7) per
block at Teagasc Moorepark (Fermoy, Co. Cork, Ireland). The plots (each 1.5 × 5 m)
were managed under a simulated grazing regime, being harvested on 10 sequential
occasions during the 2009 growing season (19 March, 8 April, 30 April, 22 May,
9 June, 1 July, 5 August, 4 September, 5 October and 10 November—defined as
Cuts 1–10, respectively). The plots received a total of 375 kg N ha−1 (60 kg N ha−1
9 weeks prior to Cut 1, and 35 kg N ha−1 after all cuts except Cut 10). The fertiliser
applied was calcium ammonium nitrate (CAN; 275 g N kg−1 ). Plots were harvested
to a 4 cm stubble height and, after thorough mixing, were stored at −18 ◦ C prior to
herbage chemical composition analysis.
When required, herbage samples were thawed at 4 ◦ C for 24 h and bowl-chopped.
Samples were dried (40 ◦ C for 48 h) and milled through a 1 mm sieve prior to chemical
analysis. Determination of OMD was carried out using the Tilley and Terry (1963)
technique, where the final residue was isolated by filtration rather than centrifugation.
Both ADF (acid detergent fibre) and NDF (neutral detergent fibre) were analysed
using the filter bag techniques (Ankom 2006a, b). The crude protein (CP; N × 6.25)
concentration was determined using a Leco FP 528 N analyser (AOAC 1990) and the
concentration of WSC was determined using the anthrone method (Thomas 1977).
For the collection of the rumen microbial inoculum, solid and liquid phase rumen
fluid (RF) samples were taken 1 h prior to feeding from four rumen fistulated steers
(682 kg mean steer liveweight) individually offered a restricted allowance (0.9 of ad
libitum intake) of a 60:40 (DM basis) grass silage and concentrate diet.
42 The Impact of Perennial Ryegrass Variety Throughout the Growing Season on . . . 317

For the in vitro rumen fermentation incubation, approximately 0.5 g of each dried,
milled sample was weighed into 160 ml fermentation bottles. Buffered mineral so-
lution (artificial saliva) was prepared according to McDougall (1948). The RF and
buffered mineral solution were maintained at 39 ◦ C and under a constant stream of
carbon dioxide (CO2 ) at all times. The RF was added to the buffered mineral solution
at a ratio of 1:4 (RF: buffered mineral solution) after which 50 ml of this buffered RF
was dispensed into each bottle. The gas headspace pressure inside each of the bottles
was recorded at the end of the 24 h incubation period using a detachable pressure
transducer, and the total gas produced (TGP) in each bottle was estimated using the
equation:
 
bottle headspace volume [mL]
TGP(mL) = × bottle headspace pressure (hPa)
atmospheric pressure [hP a]
A 0.8 ml sample of gas was then transferred to a pre-evacuated 2 ml screw-top glass
vial for determination of CH4 concentration. A 0.8 ml sample of liquid medium
was obtained from each bottle and placed in 2 ml Eppendorf tubes with 20 μl of a
9 M H2 SO4 solution. The volatile fatty acid (VFA) and CH4 concentrations were
measured by gas chromatography, using iso-caproic acid (0.04 M) as an internal
standard for the VFA as described by Ranfft (1973). After the 24 h incubation the
amount of sample disappearance, expressed as in vitro apparent dry matter (DM)
disappearance (aDMD), was estimated as the difference between the DM weight
incubated and the DM weight of the filtered (by sintered Pyrex glass crucibles;
porosity number 1) residue following oven drying at 98 ◦ C for 48 h. Both TGP and
aDMD for each sample were corrected for gas yield and particulate contamination
by inclusion of blank fermentation bottles containing buffered RF.
Data were statistically analysed as a randomised complete block design with
repeat sampling (n = 10) of each plot (n = 7) per block (n = 3) using the Proc MIXED
procedure in SAS. Pre-specified hypotheses were tested using a multiple comparison
procedure (Tukey).

42.3 Results and Discussion

Although some PRV by cut interactions were found for both herbage chemical com-
position and rumen fermentation variables in this study, the effects of PRV within
individual cuts were either in agreement with or did not markedly contradict the over-
all main effect of PRV averaged across all cuts throughout the grazing season for any
variable. Therefore, the overall effects of PRV and of Cut are discussed separately.

42.3.1 Herbage Composition

For the PRV, the NDF concentration was higher (P < 0.05) for the diploid varieties
Alto and Arrow than for the tetraploid varieties Bealey, Dunluce, Greengold and
318 P. J. Purcell et al.

Malone, possibly due to diploid varieties generally having a higher cell wall to cell
content ratio (Wilkins and Sabanci 1990). The WSC and ADF (mean annual value
across PRV of 234 g kg−1 DM; pooled SEM of 2.0) concentrations were also affected
by PRV (P < 0.05), but PRV had no effect (P > 0.05) on CP concentration (218 g kg−1
DM; SEM 3.4). Overall, the range in composition between PRV in this study was
much smaller than the range among cuts (e.g. the mean NDF values across PRV
and cuts ranged from 454–483 g kg−1 DM and 406–505 g kg−1 DM, respectively,
with corresponding WSC ranges of 94–122 and 59–190 g kg−1 DM). Thus, the small
scale of the effects of PRV on the concentrations of NDF, ADF and WSC were not
sufficient to create an effect on OMD (P > 0.05; 773 g kg−1 ; SEM 8.8). Thus, the
overall effect of PRV on herbage chemical composition was relatively small. The
generally high CP concentrations found in this study reflect the relatively high input
of N fertiliser applied.
The NDF, ADF, CP and WSC concentrations, and OMD, were all affected by
cut (P < 0.001). General increases in NDF and ADF concentrations were found from
Cuts 1–8, followed by general decreases for Cuts 9 and 10. These findings may reflect
changes in the overall ratio of leaf to stem within the sward and most likely caused
the concomitant general decrease in OMD found for Cuts 1–7 and general increase
for the remaining cuts, which is in accord with Dent and Aldrich (1968). The CP
concentration varied considerably through the season, being highest (P < 0.05) for
Cuts 2 and 10, and lowest (P < 0.05) for Cut 7. The concentration of WSC was higher
(P < 0.05) for Cut 1 than for all other cuts, which was most likely due to the longer time
interval between N fertiliser application and cutting time for Cut 1 (approximately
9 weeks) compared to all other cuts (no more than 5 weeks). The application of
N fertiliser can greatly reduce the WSC concentration of grass, especially fructans,
due to accelerated growth rate following application (Nowakowski 1962; Jones et al.
1965). No clear trend in WSC concentration was found throughout the remainder of
the growing season.

42.3.2 In vitro Rumen Fermentation Variables

The lack of an effect of PRV on the OMD of the herbage resulted in no effect
(P > 0.05) on aDMD (0.72 g g−1 ; SEM 0.012) after the 24 h fermentation. Although
ml of TGP per gram of DM incubated (TGPi) and the total VFA concentration (tVFA)
were both affected by PRV (P < 0.05), the overall scale of the effect on both variables
was small (ranges of mean values among PRV of 148–160 ml for TGPi and 42.5–
46.7 mmol l−1 for tVFA). Thus, the lack of an effect on aDMD and the small scale
effects on TGPi and aDMD indicate that the effect of PRV on the extent of the in vitro
rumen fermentation was small. This outcome explains the lack of an effect on CH4 i
(P > 0.05; Table 42.1), as the extent of in vitro rumen fermentation has been shown to
greatly influence CH4 output expressed relative to substrate incubated (Lovett et al.
2004; Purcell et al. 2011b). PRV had no effect (P > 0.05) on ml of CH4 output per
gram of DM disappeared during the in vitro rumen incubation (CH4 d; Table 42.1).
This finding reflects the lack of an effect of PRV on the direction of the in vitro rumen
42 The Impact of Perennial Ryegrass Variety Throughout the Growing Season on . . . 319

Table 42.1 Effects of perennial ryegrass variety (PRV) and stage of the growing season (Cut) on
ml of methane output per gram of dry matter incubated (CH4 i) and disappeared (CH4 d) after 24 h
of in vitro rumen incubation
Perennial ryegrass variety
Cut Alto Arrow Bealey Dunluce Greengold Malone Tyrella Mean
CH4 i
1 28.5 29.4 30.7 30.0 27.4 29.2 29.3 29.2a
2 23.6 26.6 27.3 27.7 29.1 26.9 27.8 27.0ab
3 25.8 24.2 23.4 24.0 24.9 25.9 24.5 24.7cd
4 24.0 22.7 20.6 23.6 24.7 24.1 22.9 23.2cde
5 21.7 22.0 21.6 21.5 23.0 19.2 22.5 21.6e
6 23.4 22.0 22.7 23.1 23.1 23.1 22.2 22.8de
7 23.2 22.2 23.9 25.0 23.5 24.1 23.2 23.6cde
8 21.6 23.3 24.4 23.2 25.4 22.9 24.0 23.5cde
9 23.0 24.5 26.3 28.0 24.3 23.7 24.9 25.0bc
10 24.3 22.8 26.1 27.0 26.3 23.1 25.5 25.0bcd
Mean 23.9 24.0 24.7 25.3 25.2 24.2 24.7 –
CH4 d
1 35.9 38.1 37.0 35.5 35.0 37.7 36.8 36.6a
2 31.1 31.8 32.1 33.2 33.3 31.9 32.6 32.3d
3 33.8 31.5 35.1 31.0 31.6 33.9 32.9 32.8bcd
4 34.8 33.1 31.4 34.2 35.8 37.1 33.1 34.2abcd
5 30.9 32.7 31.4 29.5 33.6 28.4 32.1 31.2cd
6 36.1 36.0 33.1 36.3 34.9 35.0 33.2 34.9abc
7 36.8 36.4 36.2 37.6 35.2 35.9 35.8 36.3a
8 33.1 35.9 37.3 37.9 37.1 33.8 36.5 36.0a
9 33.6 38.4 33.3 37.9 32.2 30.2 36.4 34.6ab
10 33.4 29.8 35.6 37.4 35.8 36.2 34.4 34.6abcd
Mean 33.9 34.4 34.3 35.1 34.4 34.0 34.4 –
Significance
CH4 i CH4 d
SEM Sig SEM Sig
PRV 0.41 NS 0.70 NS
Cut 0.43 *** 0.65 ***
Cut × PRV 1.15 * 1.71 ***
NS not significant (P > 0.05)
SEM standard error of the mean
Sig significance
***P < 0.001; *P < 0.05
a−e
Means within a column within a variable with common superscripts do not differ (P < 0.05).
Cuts 1–10, 29 March, 8 April, 30 April, 22 May, 10 June, 1 July, 6 August, 8 September, 5 October
and 10 November, respectively

fermentation after 24 h, as evidenced by the absence of an effect on the acetic acid to

propionic acid (A:P) ratio (2.59; SEM 0.020) and the non-glucogenic to glucogenic
VFA ratio (NGGR; (acetic acid + 2 butyric acid)/propionic acid; 3.35; SEM 0.032),
which influences the amount of hydrogen and, thus, CH4 produced per unit of feed
DM digested (Janssen 2010).
320 P. J. Purcell et al.

Cut had an effect on TGPi, aDMD and tVFA (P < 0.001). The CH4 i was also
affected (P < 0.001; Table 42.1) by cut, and there was a general decrease in CH4 i
from Cut 1 to Cut 5 after which no clear trend was observed. This outcome reflects the
apparent decrease in the extent of the fermentation for these cuts, as evidenced by the
general decrease in TGPi and tVFA observed from Cuts 1–5 and the general decrease
in aDMD between Cuts 2–4. Although cut had a significant effect (P < 0.001) on the
A:P ratio, NGGR and CH4 d (Table 42.1), no clear trend was found during the growing
season for any of these variables. Thus, it appears that the lack of a trend in CH4 d
reflected the similar finding for the direction of the in vitro rumen fermentation.
The finding that the direction of the in vitro rumen fermentation appeared to have
a greater influence on CH4 d throughout the growing season than the extent of the
fermentation, which appeared to have a greater influence on CH4 i, is in accord with
other similar in vitro rumen studies (Purcell et al. 2011b; Navarro-Villa et al. 2011).

42.4 Conclusion

No differences in in vitro rumen CH4 output were found between the PRV examined,
reflecting their small scale effects on herbage composition and in vitro rumen fer-
mentation variables. Hence, these results provide no encouragement that choices
among the PRV examined, produced within the management regimes operated,
would reduce enteric methane production. However, the technique utilised did not
take account of animal by PRV interactions that may occur under farm conditions.

Acknowledgments Funding for this study was provided by the Department of Agriculture, Fish-
eries and Food (RSF No. 07 517). The authors thank Jim Grant, Séan Lynch, and Grange
Laboratories staff.

References

Ankom (2006a) Acid detergent fiber in feeds filter bag technique. Ankom Technology, Macedon, p 2
Ankom (2006b) Neutral detergent fiber in feeds filter bag technique. Ankom Technology, Mace-
don, p 2
AOAC (1990) Official methods of analysis: 1st supplement. J Assoc Offic Anal Chem 15:3–4
Davies DA, Fothergill M, Jones D (1991) Assessment of contrasting perennial ryegrasses, with and
without white clover, under continuous sheep stocking in the uplands. 3. Herbage production,
quality and intake. Grass Forage Sci 46:39–49
Dent JW, Aldrich DTA (1968) Systematic testing of quality in grass varieties. Grass Forage Sci
23:13–19
Janssen PH (2010) Influence of hydrogen on rumen methane formation and fermentation balances
through microbial growth kinetics and fermentation thermodynamics. Anim Feed Sci Technol
160:1–22
Jones DIH, ap Griffith G, Walkers RJK (1965) The effect of nitrogen fertilizers on the water-soluble
carbohydrate content of grasses. J Agric Sci 64:323–328
42 The Impact of Perennial Ryegrass Variety Throughout the Growing Season on . . . 321

Lovett DK, Bortolozzo A, Conaghan P, O’Kiely P, O’Mara FP (2004) In vitro total and methane
gas production as influenced by rate of nitrogen application, season of harvest and perennial
ryegrass cultivar. Grass Forage Sci 59:227–232
McDougall EI (1948) Studies on ruminant saliva I. The composition and output of sheep’s saliva.
Biochem J 43:99–109
Navarro-Villa A, O’Brien M, López S, Boland TM, O’Kiely P (2011) In vitro rumen methane output
of red clover and perennial ryegrass assayed using the gas production technique (GPT). Anim
Feed Sci Technol 168:152–164
Nowakowski TZ (1962) Effects of nitrogen fertilizers on total nitrogen, soluble nitrogen and soluble
carbohydrate contents of grass. J Agric Sci 59:387
Purcell PJ, O’Brien M, Boland TM, O’Donovan M, O’Kiely P (2011a) Impacts of herbage mass
and sward allowance of perennial ryegrass sampled throughout the growing season on in vitro
rumen methane production. Anim Feed Sci Technol 166–167:405–411
Purcell PJ, O’Brien M, Boland TM, O’Kiely P (2011b) In vitro rumen methane output of perennial
ryegrass samples prepared by freeze drying or thermal drying (40 ◦ C). Anim Feed Sci Technol
166–167:175–182
Ranfft K (1973) Gas-chromatography determination of short-chain volatile fatty acids in the ruminal
fluid. Arch Tierernahr 23:343–352
Thomas TA (1977) An automated procedure for the determination of soluble carbohydrates in
herbage. J Sci Food Agric 28:639–642
Tilley JMA, Terry RA (1963) A two-stage technique for the in vitro digestion of forage crops. Grass
Forage Sci 18:104–111
Wilkins PW, Sabanci CO (1990) Genetic variation in leaf epidermal cell size and shape in lolium
perenne. Euphytica 47:233–239
Chapter 43
Origin and Yield of European Perennial
Ryegrass (Lolium perenne L.)
Varieties in Ireland

D. Grogan

Abstract The Department of Agriculture, Fisheries and Food have sown 63 In-
termediate and Late heading perennial ryegrass varieties in combined National
List/Recommended List evaluation trials at five locations in Ireland each year from
2004 to 2008. Breeder’s seed of candidate and commercial varieties originating
from breeding companies in eight European countries were included. Each sowing
year/experiment was harvested for two successive years under a six cut ‘general pur-
pose’ protocol. Annual total yields of dry matter per hectare were sorted by origin
of germplasm and averaged for each country. Intermediate varieties had an average
yield of 14.8 T DM/ha, 0.3–0.4 T DM/ha greater than Late varieties (14.5 T DM/ha)
from all sources except Belgium. Intermediate varieties from Germany, Denmark,
Ireland, Northern Ireland and the UK were 0.4 T DM/ha higher than varieties from
France and the Netherlands, while in the Late category varieties from Germany, Ire-
land and the UK had an average yield of at least 0.5 T DM/ha greater than varieties
from Denmark, France and the Netherlands.

Keywords Perennial ryegrass · Variety evaluation · Ireland

43.1 Introduction

The climatic conditions of Ireland supports a predominately grassland dependent

agri-business. Grassland covers approximately 85 % of the arable area of the island
and is by far the most important agricultural land use. Given the wide climatic
range of growing conditions across the European market, selection criteria in grass
breeding and evaluation programmes need to focus on specific ecozones if breeding
advances are to be achieved and passed on to farmers. The use of dormancy zones,

D. Grogan ()
Crop Evaluation & Certification, Department of Agriculture, Fisheries and Food,
Backweston Farm, Leixlip, Co. Kildare, Ireland
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 323

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_43, © Springer Science+Business Media Dordrecht 2013
324 D. Grogan

Table 43.1 Site details of NL/RL Variety Evaluation Trials for perennial ryegrass in Ireland
2004–2010
Site Geographic Altitude (m) Soil Type (loam) Org. Matter (g/kg) pH
Coordinates
Athenry 53◦ 18 N 8◦ 45W 35 Peaty 79 7.1
Backweston 53◦ 22 N 6◦ 30W 50 Clay 63 7.2
Fermoy 52◦ 08 N 8◦ 17W 53 Med 59 5.8
Piltown 52◦ 21 N 7◦ 20W 15 Clay 49 5.5
Raphoe 54◦ 52 N 7◦ 36W 65 Med 62 5.6

based on Lucerne adaptation, has been found to be a useful means of classifying such
regions (Long and Gilliland 2010). Ireland is within Zone 6, which is a maritime
region that includes Britain and the coastal regions of north west France and Spain.
There is ample evidence that ryegrass breeding programmes focused on this agri-
environmental zone have achieved notable success in the past (Van Wijk and Reheul
1991; Wilkins and Humphreys 2003). Recent work has indicated that strong genotype
by environment interactions are present in identifying superior genotypes under Irish
conditions (Conaghan et al. 2008; Grogan and Gilliand 2010). The objective of this
study is to establish the extent of variation in total annual yield in recent Irish national
perennial ryegrass evaluation trials when varieties are examined by country of origin.

43.2 Materials and Methods

Each year from 2004 to 2008, thirty Intermediate (or mid-season heading), and
thirty-three Late heading perennial ryegrass varieties from eight European countries
were evaluated by the Department of Agriculture, Fisheries and Food in combined
National List and Recommended List trials. These groups were sown separately
in a randomised complete block design with four replications at five locations
(Table 43.1.)
Plots of 11.4 m2 were broadcast sown in the autumn at a seed rate of 30 kg/ha for
diploids and 40 kg/ha for tetraploids. Fertiliser N was applied annually at 350 kg/ha,
and P, K and S were applied as indicated by annual soil analysis to meet growth
requirements. These fertilizer rates were intended to simulate intensive grassland
use. Each sowing was harvested by Haldrup plot harvester (Logstor, Denmark) for
two successive years under an annual six cut ‘general purpose’ cutting protocol. Total
plot yields in kilograms were recorded and a subsample (approx 300 g) oven dried at
800 ◦ C for 16 h to determine Dry Matter yield. Results of each cut from each location
were subjected to ANOVA using Genstat and/or AgroBase. Annual average total DM
yield for each variety was sorted by origin of seed as declared on application forms
used for entry to NL/RL trials.
43 Origin and Yield of European Perennial Ryegrass (Lolium perenne L.) . . . 325

Table 43.2 Number of perennial ryegrass varieties in Irish NL/RL evaluation trials 2004–2010
Year sown/experiment
04 04 05 05 06 06 07 07 08 08 Mean Mean
Origin Inter Late Inter Late Inter Late Inter Late Inter Late Inter Late
Belgium 1 2 1 3 0 1 0 0 0 0 0.4 1.2
Germany 1 3 1 2 1 4 2 5 6 8 2.2 4.4
Denmark 1 0 1 1 0 3 1 4 2 2 1.0 2.0
France 0 2 1 4 2 6 2 5 2 5 1.4 4.4
Ireland 9 6 7 6 8 6 6 2 7 4 7.4 4.8
N. Ireland 4 6 4 6 6 6 8 6 5 4 5.4 5.6
Netherlands 12 10 13 7 11 3 10 6 6 6 10.4 6.4
UK 2 3 2 4 2 4 1 4 0 4 1.4 3.8

Table 43.3 Yield of inter perennial ryegrass varieties in Irish NL/RL evaluation trials 2004–2010
Sowing Year 2004 2005 2006 2007 2008
Harvest Year 2005 2006 2006 2007 2007 2008 2008 2009 2009 2010 Mean
Origin Intermediate varieties, average annual yield T DM/ha
Belgium 14.9 13.3 13.6 13.9 13.9
Germany 15.8 14.0 15.6 15.5 17.7 15.0 15.8 15.4 14.6 13.2 15.3
Denmark 16.0 13.3 15.5 14.6 16.6 16.2 15.1 13.2 15.1
France 15.1 15.1 16.3 14.6 14.8 14.2 13.9 12.6 14.6
Ireland 15.8 13.7 14.9 15.0 17.2 15.4 15.7 15.4 14.6 13.2 15.1
N. Ireland 15.7 13.5 14.8 15.0 17.1 15.4 15.7 15.5 14.8 13.5 15.1
Netherlands 15.4 13.6 14.5 14.6 16.6 14.9 15.2 14.7 14.4 13.1 14.7
UK 15.4 13.5 14.8 14.7 17.0 15.2 15.2 14.7 15.1
Mean 15.6 13.5 14.9 14.8 17.0 15.1 15.6 15.2 14.6 13.1 14.9

43.3 Results and Discussion

43.3.1 Number of Varieties

The number of varieties tested from each country varied from a small number of
Belgian varieties (on average 0.5–1 variety per annum) to between six and ten each
year from the Netherlands (Table 43.2). The number of Dutch varieties tested has
been declining in recent years.

43.3.2 Yield of Varieties

Annual yields from all sites for each harvest year are presented for Intermediate
varieties in Table 43.3, and for Late varieties in Table 43.4.
326 D. Grogan

Table 43.4 Yield of late perennial ryegrass varieties in Irish NL/RL evaluation trials 2004–2010
Sowing Year 2004 2005 2006 2007 2008
Harvest Year 2005 2006 2006 2007 2007 2008 2008 2009 2009 2010 Mean
Origin Late varieties, average annual yield T DM/ha
Belgium 16.3 13.3 13.9 13.8 15.2 13.7 14.3
Germany 16.1 13.5 14.6 14.8 16.9 15.1 14.0 14.1 14.5 13.2 14.7
Denmark 13.9 14.0 16.3 14.7 14.0 14.1 14.1 12.9 14.2
France 16.0 13.5 13.8 14.0 15.7 14.3 13.5 13.6 14.1 12.6 14.1
Ireland 16.0 13.4 14.3 14.7 16.4 15.2 14.1 14.5 14.8 13.4 14.7
N. Ireland 15.8 13.3 13.9 14.5 16.4 15.0 14.2 14.3 14.4 13.2 14.5
Netherlands 15.7 13.1 13.9 14.0 16.1 14.5 13.9 14.0 14.0 12.9 14.2
UK 16.3 13.4 14.3 14.7 16.9 15.0 14.6 14.6 15.3 14.1 14.9
Mean 16.0 13.4 14.1 14.3 16.2 14.7 14.1 14.2 14.4 13.2 14.5

Overall, Intermediate varieties out yielded Late varieties by an average of 0.4 T

DM/ha. The lowest yields occurred in 2006 and 2010 from 2nd year harvests (13.1–
13.5 T DM/ha), while the highest yields were recorded from 1st year harvests in
2007 (16.2–17.0 T DM/ha).
Varieties from countries climatically closest to the evaluation sites (i.e. Ireland,
Northern Ireland, and UK Wales) on average yielded 0.4 T DM/ha more than north
western European material (Germany, Denmark and Netherlands), and 0.6 T DM/ha
more than French material.
There are apparent differences in potential total annual DM yield based on origin of
seed. However, use of other traits such as seasonality of yield, ruminant digestibility,
and persistency are needed to give a full assessment of true forage quality.

References

Conaghan P, Casler MD, McGilloway DA, O’Kiely P, Dowley LJ (2008) Genotype x environment
interactions for herbage yield of perennial ryegrass swards in Ireland. Grass Forage Sci 63:
107–120
Grogan D, Gilliland TJ (2010) A review of perennial ryegrass variety evaluation in Ireland. In:
O’Donovan M, Hennessey D (eds) Grasses for the Future. Perennial ryegrasses: current and
future genetic potential. Teagasc, Cork, pp 99–116
Long D, Gilliland TJ (2010) A review of the procedures and priorities for marketing of improved
ryegrass varieties. In: O’Donovan M, Hennessey D (eds) Grasses for the future. Perennial
ryegrasses: current and future genetic potential. Teagasc, Cork, pp 171–183
Van Wijk AJP, Reheul D (1991) Achievements in fodder crops breeding in maritime Europe. In:
Den Nijs APM, Elgersma A (eds) In: Fodder crops breeding: achievements, novel strategies
and biotechnology. Proceedings of the 16th meeting of the fodder crops section of eucarpia,
Wageningen, Netherlands, pp 13–18
Wilkins PW Humphreys MO (2003) Progress in breeding perennial forage grasses for temperate
agriculture. J Agric Sci 140:129–150
Chapter 44
Yield Dynamics and Quality in White Clover
and Perennial Ryegrass in the First cut
of the Establishment Year

B. Ćupina, A. Mikić, Ð. Krstić, S. Antanasović, P. D’Ottavio and P. Erić

Abstract A field trial was carried out in the Vojvodina province, Serbia, during
2009 and 2010 in rainfed conditions. The study assessed the dynamics of yield for-
mation, changes in morphological characteristics, as well as the nutritive values of
white clover and perennial ryegrass grown as monocultures. Five varieties of white
clover (Chieftain, Susi, Aran, Avoca and Rivendale) and eight of perennial ryegrasses
(Cashel, Shandon, Magician, Greengold, Glenstal, Millenium, Sarsfield and Glen-
car), developed at and provided by Teagasc, Ireland, have been used for this research.
After four measurements a significant differences in examined parameters with white
clover and perennial ryegrass varieties were registered. In all studied parameters the
highest values were recorded in the fourth measurement. The stolon length ranged
from 1.4 cm (Avoca) to 4.1 cm (Chieftain). Susi had the highest number of stolons
(2.1), and Rivendale had the lowest (0.9) but also the highest number of leaves per
stolon (3.7). The leaf length ranged from 6.1 cm (Avoca) to 19.3 cm (Aran). Concern-
ing ryegrass, the number of lateral shoots ranged from 2.8 (Millenium) to 4.8 (Magi-
cian). The highest number of leaves per shoot was in Cashel (4.8), and the lowest was
reported for Millenium and Greengold (3.3). The average shoot height to the first leaf
varied between 3.2 cm (Glenstal) and 4.4 cm (Cashel). The highest yield of white
clover was achieved by Rivendale (1.19 t ha−1 ), while the highest yield of perennial
ryegrass were recorded by Millennium (8.26 t ha−1 ) and Magician (8.03 t ha−1 ). The
average crude protein content was higher in white clover (22.8 %) than in perennial
ryegrass (12.8 %), while the crude fibre content of white clover and perennial ryegrass
were 18.1 and 34.7 %, respectively. These parameters led to a lower digestibility in
perennial ryegrass, reflected through the monitored parameters of NDF and ADF.

B. Ćupina () · Ð. Krstić · S. Antanasović · P. Erić

Faculty of Agriculture, Department of Field and Vegetable Crops, University of Novi Sad,
Trg Dositeja Obradovića 8, 21000 Novi Sad, Serbia
e-mail: [email protected]
A. Mikić
Institute of Field and Vegetable Crops, Maksima Gorkog 30, 21000 Novi Sad, Serbia
P. D’Ottavio
Department of Agricultural, Food and Environmental Sciences, Polytechnic University of Marche,
Via Brecce Bianche, 60131 Ancona, Italy

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 327

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_44, © Springer Science+Business Media Dordrecht 2013
328 B. Ćupina et al.

Keywords Perennial ryegrass · Quality · Varieties · White clover · Yield dynamics ·

First cut

44.1 Introduction

Two legumes, alfalfa (Medicago sativa L.) and red clover (Trifolium pratense L.),
and permanent grasslands have the greatest importance for the forage production in
the agroecological conditions of Serbia. However, significant characteristics such
as yield and particularly quality in white clover (Trifolium repens L.) and perennial
ryegrass (Lolium perenne L.) have not been studied to a sufficient extent.
White clover is a perennial type of legume, widely prevalent. It is a very resource-
ful plant, although with somewhat lower yields than grass, but it is an important
component in forage mixtures for improving quality (Turkington 1989; Collins and
Rhodes 1989). It is a weak competitor for light and nutrients in comparison to most
grass companions in binary (Soussana et al. 1995) or complex legume-grass mix-
tures (Nyfeler et al. 2009; Annicchiarico and Proietti 2010). It is usually grown with
one or more grass companions, the most common being perennial ryegrass (Hill
and Michaelson-Yates 1987). When grown in mixtures, white clover and ryegrass
provide pastures of high productivity and quality (Lucero et al. 1999). One of the
frequent problems occurring with grass-legume mixtures is the timely determination
of the physiological maturity for the first cut in the establishment year.
The objective of this study is to determine the potential of white clover and peren-
nial ryegrass in the conditions of the Serbian province of Vojvodina. Additionally,
it aimed at defining the growth dynamics and the yields formation in perennial rye-
grass and white clover grown separately in the first cut of the establishment year
by monitoring their forage yields, forage yield components and forage dry matter
quality.

44.2 Materials and Methods

A field experiment was conducted under the rain-fed conditions during 2009 and
2010 at the Experimental Field of the Institute of Field and Vegetable Crops at
Rimski Šančevi, Serbia (45◦ 20 N and 19◦ 51 E and 84 m a.s.l). This research
included five varieties of white clover (Chieftain, Susi, Aran, Avoca and Rivendale)
and eight varieties of perennial ryegrass (Cashel, Shandon, Magician, Greengold,
Glenstal, Millenium, Sarsfield and Glencar). All these varieties have been developed
at Teagasc, Ireland, except the white clover Rivendale, bred by DLF Trifolium,
Denmark.
Two monofactorial trials were set up as random block systems in three replications.
In both years and in each trial, varieties were sown inApril with a plot size of 5 m2 . The
seeding rate for perennial ryegrass was 25 kg ha−1 and for white clover 10 kg ha−1 .
44 Yield Dynamics and Quality in White Clover and Perennial Ryegrass . . . 329

Both species were sown separately as pure stands. During the growing period, all the
usual agronomic measures for both crops were applied, specifically manual weed
control. Samples of 20 plants were taken from each plot every 20 days during the
development of the first cut in both years, namely on May 15th, June 5th, June
25th and July 15th. In white clover, stolon length (cm), number of stolons, petiole
length (cm), number of leaves per stolon, total number of leaves and dynamics
of formation of dry matter yield (t ha−1 ) were measured. In ryegrass, number of
lateral shoots, number of leaves per shoot and shoot height (cm) and dry matter yield
(t ha−1 ) were measured. In the first cut, that is, in its fourth measurement stages,
quality parameters were analysed according to Van Soest (1991) such as the content
of crude protein (CP), crude fibre (CFb), as well as neutral detergent fibres (NDF)
and acid detergent fibres (ADF).
The results were statistically processed by using ANOVA (analysis of variance)
and a Fisher’s LSD test to detect significant differences between varieties for all mean
values at P = 0.05. The results were presented and discussed as a 2 year average.

44.3 Results and Discussion

After four measurements, there were significant differences in the examined pa-
rameters in white clover and perennial ryegrass varieties. Generally, in all studied
parameters the highest values were recorded in the last, namely fourth, measurement,
being the objective of the following discussion.

Table 44.1 Dynamics and changes in morphological characteristics and dry matter (DM) yield of
white clover varieties
Variety Stolon Stolon Leaf Leaf DM Stolon Stolon Leaf Leaf DM
length number length number yield length number length number yield
Sample Date May 15th June 5th
Chieftain 0.8a 0.9a 4.1a 1.3a 0.44a 1.7a 1.0a 5.8a 1.3b 0.63a
Susi 0.6b 0.7b 2.8b 1.5a 0.10d 1.3b 0.9a 6.4a 1.8a 0.26c
Aran 0.4c 1.0a 1.8c 1.3a 0.35b 0.6d 1.0a 3.7b 1.7a 0.51b
Avoca 0.1e 0.3c 1.1d 0.6c 0.09d 0.3e 0.6b 1.5c 1.1c 0.15d
Rivendale 0.2d 0.3c 1.0d 1.0b 0.23c 0.8c 0.5c 2.0bc 1.3b 0.31c
2009 0.5a 0.6a 2.0b 1.2a 0.26a 1.0a 0.8a 2.8b 1.5a 0.38a
2010 0.4a 0.6a 2.3a 1.0b 0.23b 0.9b 0.7a 4.9a 1.4b 0.36a
Average 0.4 0.6 2.2 1.1 0.24 0.9 0.8 3.9 1.4 0.37
Sample Date June 25th July 15th
Chieftain 2.2a 1.0b 5.5a 1.8bc 0.91a 4.1a 1.8ab 10.8b 2.8bc 1.11a
Susi 1.5c 1.6a 5.8a 2.6a 0.47c 3.6b 2.1a 10.0b 2.9b 0.43d
Aran 1.4d 0.9bc 5.3a 1.9b 0.74b 3.5b 1.5b 19.3a 2.5c 0.97b
Avoca 0.6e 1.0b 1.9b 1.5c 0.32d 1.4c 1.3bc 6.1c 3.4a 0.54c
Rivendale 1.7b 0.8c 2.1b 1.7bc 0.52c 2.1d 0.9c 10.1b 3.7a 1.19a
2009 1.5a 1.1a 3.6b 2.0a 0.57a 3.1a 1.6a 8.8b 3.3a 0.92a
2010 1.4b 1.0b 4.6a 1.8b 0.61a 2.8b 1.4a 13.7a 2.9b 0.78b
Average 1.5 1.0 4.1 1.9 0.59 2.9 1.5 11.3 3.1 0.85
*LSD followed by different letters are significantly different (P ≤ 0.05)
330 B. Ćupina et al.

Table 44.2 Dynamics and changes in morphological characteristics and DM yield of perennial
ryegrass varieties
Variety Number Number Shoot height DM Number Number Shoot height DM
of lateral of leaves to the 1st leaf yield of lateral of leaves to the 1st leaf yield
shoots per shoot shoots per shoot
Sample Date May 15th June 5th
Cashel 2.4a 3.4a 2.5a 4.24b 2.8a 3.9a 3.4a 4.77b
Shandon 2.1b 2.9bc 1.7c 3.95b 2.5b 3.9a 2.9b 5.18b
Magician 2.1b 2.8bc 2.0b 4.61ab 1.9c 3.4bc 2.8bc 5.09b
Greengold 1.8c 1.9e 2.0b 4.17b 2.0c 3.1cd 2.3d 4.81b
Glenstal 1.9bc 3.0b 1.8bc 4.41b 2.0c 3.6b 2.3d 4.94b
Millennium 2.5a 2.4d 2.0b 6.83a 2.6ab 3.0d 2.4d 6.92a
Sarsfield 2.6a 2.6c 1.5d 4.00b 2.6ab 3.3c 2.7c 4.47b
Glencar 1.7c 2.1de 1.4d 4.17b 2.1c 3.4bc 2.0e 4.35b
2009 2.3a 3.0a 2.0a 4.84a 2.5a 3.6a 2.8a 5.41a
2010 2.0b 2.6b 1.7b 4.26b 2.2b 3.3b 2.4b 4.72b
Average 2.2 2.8 1.9 4.55 2.3 3.4 2.6 5.07

Sample Date June 25th July 15th

Cashel 3.6a 4.4a 4.2a 5.91bc 3.9c 4.8a 4.4a 6.32bc
Shandon 3.7a 4.0b 4.1a 7.12ab 4.1c 4.2b 4.3a 7.08bc
Magician 3.5a 3.6c 3.2c 6.77b 4.8a 3.8c 3.7b 8.03a
Greengold 2.6c 3.2d 3.2c 5.34bc 3.0e 3.3d 3.5bc 6.19c
Glenstal 2.8bc 3.8bc 3.7b 7.45a 3.2de 4.2b 3.2c 7.88ab
Millennium 2.7bc 3.2d 3.5b 7.85a 2.8f 3.3d 3.3c 8.26a
Sarsfield 3.0b 3.4cd 2.8d 5.31c 4.4b 3.6d 3.5bc 6.73bc
Glencar 2.6c 3.7bc 3.2c 6.13bc 3.4d 3.9c 3.6b 7.15ab
2009 3.2a 3.9a 3.7a 6.87a 3.9a 4.1a 3.9a 7.57a
2010 2.9b 3.4b 3.3b 6.10b 3.5b 3.6b 3.5b 6.84b
Average 3.0 3.7 3.5 6.48 3.7 3.9 3.7 7.21
*LSD followed by different letters are significantly different (P ≤ 0.05)

The stolon length ranged from 1.4 cm in Avoca to 4.1 cm in Chieftain. Susi had
the highest number of stolons 2.1, while the lowest was in Rivendale with 0.9. The
values of leaf length ranged from 6.1 cm (Avoca) to 19.3 cm (Aran). The highest
number of leaves per stolon (3.7) and dry matter yield (1.19 t ha−1 ) were produced
by Rivendale (Table 44.1).
As for ryegrass (Table 44.2), the highest number of lateral shoots was recorded in
Magician (4.8), while the lowest one was in Millennium (2.8). The number of leaves
per shoot ranged from 3.3 in Greengold and Millenium to 4.8 in Cashel. Glenstal had
the lowest values of shoot height to the first leaf (3.2 cm), while the highest values
were in Cashel (4.4 cm). Despite the fact that it did not have high values of some of
the monitored parameters, Millenium (Table 44.2) had the highest dry matter yield
(8.26 t ha−1 ), while the lowest yield was measured in Greengold (6.19 t ha−1 ).
The quality parameters differed within varieties. The average crude protein content
was higher in white clover (22.8 %) than in perennial ryegrass (12.8 %), while the
crude fibre content of white clover and perennial ryegrass were 18.11 and 34.66 %,
respectively. The average values of NDF andADF in white clover varieties were 38.20
and 32.24 %. This is supported by Ayres et al. (1998). In perennial ryegrass varieties,
they were 60.67 and 39.28 %. These parameters indicate a lower digestibility in
perennial ryegrass (Table 44.3).
44 Yield Dynamics and Quality in White Clover and Perennial Ryegrass . . . 331

Table 44.3 Quality parameters of white clover and perennial ryegrass varieties
Variety Crude protein Crude fibre NDF ADF
White clover
Chieftain 16.00c 15.49c 47.63a 42.23a
Susi 23.16b 16.94b 36.12b 31.08b
Aran 22.10b 25.70a 37.10b 29.57bc
Avoca 24.88a 16.52bc 34.67b 28.45c
Rivendale 24.04ab 15.90bc 35.51b 29.86bc
2009 22.23b 17.24b 36.32b 31.02b
2010 23.37a 18.98a 40.09a 33.46a
Average 22.80 18.11 38.20 32.24
Perennial ryegrass
Cashel 11.03bc 33.09b 63.30ab 40.95ab
Shandon 10.25cd 37.92a 65.11a 42.07a
Magician 22.76a 37.39a 57.86bc 38.22b
Greengold 11.79b 36.44a 58.90bc 38.47b
Glenstal 10.20cd 33.49b 61.21abc 41.47ab
Millennium 9.75d 27.74c 60.34abc 40.26ab
Sarsfield 10.67cd 34.17b 56.40c 34.21c
Glencar 11.02bc 37.04a 62.23abc 38.58b
2009 11.53b 33.13b 57.82b 37.33b
2010 12.84a 36.19a 63.51a 41.22a
Average 12.80 34.66 60.67 39.28
*LSD followed by different letters are significantly different (P ≤ 0.05)

44.4 Conclusion

In all studied parameters, the highest values were recorded in the last, fourth, mea-
surement. There are significant differences in morphological properties, dry matter
yield and quality between tested varieties of white clover and perennial ryegrass.
The white clover variety Rivendale and the perennial ryegrass variety Millennium
produced the highest forage yields and had the best forage quality.

Acknowledgments This research was supported by the Ministry of Education and Science of the
Republic of Serbia (Project No. 31016). Patrick Conaghan of Teagasc is kindly acknowledged for
providing all the tested varieties.

References

Annicchiarico P, Proietti S (2010) White clover selected for enhanced competitive ability widens
the compatibility with grasses and favours the optimization of legume content and forage yield
in mown clover-grass mixtures. Grass Forage Sci 65:318–324
Ayres JF, Nandra KS, Turner AD (1998) A study of the nutritive value of white clover (Trifolium
repens L.) in relation with different stages of phonological maturity in the primary growth phase
in spring. Grass Forage Sci 53:250–259
Collins RP, Rhodes I (1989)Yield of white clover populations in mixtures with contrasting perennial
ryegrass. Grass Forage Sci 44:111–115
332 B. Ćupina et al.

Hill J, Michaelson-Yeates TPT (1987) Effects of competition upon the productivity of white clover-
perennial ryegrass mixtures: analysis of and interrelations between characters. Plant Breed
98(2):161–170
Lucero DW, Grieu P, GuckertA (1999) Effects of water deficit and plant interaction on morphological
growth parameters and yield of white clover (Trifolium repens L.) and ryegrass (Lolium perenne
L.) mixtures. Eur J Agron 11:167–177
Nyfeler D, Huguenin-Elie O, Suter M, Frossard E, Connolly J, Lusher A (2009) Strong mixture
effects among four species in fertilized agricultural grassland led to persistent and consistent
transgressive overyielding. J Appl Ecol 46:683–691
Soussana JF, Vertès F, Arregui MC (1995) The regulation of clover shoot growing point density and
morphology during short-term clover decline in mixed swards. Eur J Agron 4:205–215
Turkington R (1989) The growth, distribution and neighbor relationships of Trifolium repens in a
permanent pasture. VI. Conditioning effects of neighbours. J Ecol 77:734–746
Van Soest PJ, Robertson JB, Lewis BA (1991) Methods for dietary fiber, neutral detergent fiber and
non-starch polysaccharides in relation to animal nutrition. J Dairy Sci 74:3583–3597
Chapter 45
Influence of Plant Growth Promoting
Rhizobacteria on Alfalfa, Medicago sativa L.
yield by Inoculation of a Preceding Italian
Ryegrass, Lolium multiflorum Lam

D. Delić, O. Stajković-Srbinović, D. Kuzmanović, N. Rasulić, S. Maksimović,

J. Radović and A. Simić

Abstract This study was conducted to test the hypothesis that plant growth pro-
moting rhizobacteria (PGPR) can promote the growth of Italian ryegrass, Lolium
multiflorum Lam. as well as the growth and nodulation of subsequent alfalfa, Med-
icago sativa L. In a pot experiment, the influence of PGPR on yield and nitrogen
content of Italian ryegrass and alfalfa was studied with the aim to improve their
growth and provide effective alfalfa nitrogen fixation under unfavourable soil condi-
tions. Plants were inoculated with seven strains belonging to Sinorhizobium meliloti,
Azotobacter spp. and Enterobacter sp. A beneficial effect on yield and N-assimilation
in Italian ryegrass was obtained due to the inoculation of the plants with Azotobacter
vinelandii and some Sinorhizobium meliloti strains. In addition, Italian ryegrass seed
inoculation with particular rhizobial strains the year before alfalfa growing provided
abundant nodulation and better growth of alfalfa.

Keywords Italian ryegrass · Alfalfa · Inoculation · Yield · PGPR · Sinorhizobium

45.1 Introduction

In recent years, there has been a growing interest in using bacterial inoculants (biofer-
tilizers) because of the ability of particular rhizobacteria (PGPR) to promote growth
and yield in great number of plants including legumes and cereals through different
mechanisms, i.e. increasing nitrogen (N) uptake (biological N fixation-BNF), synthe-
sis of phytohormones (auxin, cytokinin), minerals solubilization and iron chelation

D. Delić () · O. Stajković-Srbinović · D. Kuzmanović · N. Rasulić · S. Maksimović

Institute of Soil Science, Teodora Drajzera 7, 11000 Belgrade, Serbia
e-mail: [email protected]
J. Radović
Institute for Forage Crops, Globoder, 37251 Kruševac, Serbia
A. Simić
Faculty of Agriculture, University of Belgrade, Nemanjina 6, 11000 Belgrade, Serbia

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 333

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_45, © Springer Science+Business Media Dordrecht 2013
334 D. Delić et al.

(Antoun and Prévost 2005; Mehboob et al. 2009; Goos et al. 2001; Egamberdiyeva
et al. 2004). These biofertilizers represent an alternative to plant growth enhancement
by chemical fertilizers. BNF is promoted by both free living soil bacteria (Azotobac-
ter, Enterobacter, Bacillus) and by symbiotic association of mostly nodule bacteria
(rhizobia) and legumes. Azotobacter species are aerobic heterotrophic soil bacteria.
Yields of rice, cotton and wheat in many investigations were increased significantly
with applications of Azotobacter due to its ability to fix N2 (Kennedy et al. 2004).
Enterobacter is a free-living nitrogen fixer that is distributed to the soil matrix via
animal feces. These N2 -fixing microbes grow in humid environments on leaf surfaces
or in leaf sheaths (phyllosphere), the soil, and root surfaces (Paul and Clark 1996).
Alfalfa (Medicago sativa L.) is one of the most important leguminous forage
crops with the ability of BNF with specific rhizobia (Sinorhizobium meliloti). This
association belongs to the most significant agricultural systems for BNF (on average
250 kg N ha−1 ). The presence, density and effectiveness of natural populations of
S. meliloti vary in the soil and depend on the soil type and land use regime (Delić-
Vukmir et al. 1994). Low number of S. meliloti (≤ 102 g−1 of soil) may be caused
by bad soil characteristics or the absence of a host plant for a long period of time.
Artificial inoculation of legumes with N fixing rhizobia is always applied on soils
with no recent cultivation of alfalfa as well as on soils with low fertility with the aim
to increase the number and effectiveness of rhizobia in the soil. The possibility for
increasing the number of B. japonicum before growing soybean for the first time by
inoculation of the previous crops was shown (Goos et al. 2001).
Italian ryegrass, Lolium multiflorum Lam. is one of the best forage grasses in
Serbia, producing high-quality forage from early spring to late summer Simić et al.
2009). It is valuable in crop rotations with row crops (e.g. soybean, maize, sunflower,
sugar beet) for maintaining soil structure and health. Italian ryegrass is well-adapted
to high rainfall but can be grown where a minimum of about 500 mm rainfall occurs
during the growing season. It also can be used as emergency forage after winterkill
of alfalfa. It establishes quickly and produces a lot of forage in a short period of time.
The object of this study was to evaluate the possible effects of a preceding ryegrass
inoculation with S. meliloti strains and free-living rhizobacteria on the number of
applied rhizobia in the soil as well as on the yield and N content of ryegrass as the
preceding crop and of alfalfa as the subsequent crop.

45.2 Material and Methods

Azotobacter vinelandi strain Av, Azotobacter chroococum strain AoNDD, Enter-

obacter sp.strain E1 and four effective S. meliloti strains 218, L3Si, 4148ss and 207,
from the Collection of the Institute of Soil Science were used for the inoculation of
Italian ryegrass and alfalfa. The effects of these strains on the yield and N-content of
Italian ryegrass cultivar K-29 t and alfalfa cultivar K-28 (Institute for Forage Crops,
Kruševac) were examined in a pot experiment under greenhouse conditions, using
45 Influence of Plant Growth Promoting Rhizobacteria on Alfalfa, Medicago . . . 335

non-sterile soil with no history of cultivation of these crops. The soil used had the
following characteristics: pH (in H2 O) 7.4, 900 mg N kg−1 , 21 mg kg−1 available P.
The first part of the experiment was set up in 2009 with Italian ryegrass. The exper-
iment consisted of seven inoculated treatments and two control treatments without in-
oculation and without (Ø) or with (NØ) mineral nitrogen fertilization (27 mg N kg−1 )
in five completely randomized replicates. Enterobacter, Sinorhizobium and Azo-
tobacter strains were cultivated on appropriate medium for 24, 48 and 72 h,
respectively. Italian ryegrass seeds were surface-sterilized with a 0.1 % HgCl2 solu-
tion. Five seeds per pot were planted and inoculated with 2 ml plant−1 of the liquid
culture of the single strains containing > 109 cells per ml. The pots were kept in a
greenhouse conditions for six weeks.
The second part of the experiment consisted of the determination of the number of
S. meliloti strains in the pots after cutting the Italian ryegrass by the Plant infection
count method (Vincent 1970). The most probably number (MPN) of S. meliloti
strains applied in the rhizobial inoculated treatments was compared with the soil
before sowing.
The third part of the experiment was conducted in 2010 with alfalfa growing
in the pots with the soil where the Italian ryegrass had grown after removal of the
roots of the Italian ryegrass. Alfalfa was inoculated with the same strains applied for
the inoculation of Italian ryegrass. In addition, the yield and the N-content of the
subsequent crop were compared with alfalfa grown without the preceding crop and
inoculated with the same rhizobial strains. The pots were kept in semi controlled
conditions for six weeks. Roots of alfalfa were carefully removed from the pots,
washed free of soil and the number of nodules was recorded.
In the first and third part of the experiment shoot height of both species was mea-
sured. Plant shoots were dried in an oven at 70 ◦ C to constant weight and the average
dry weight per plant was calculated. The percentage of shoot N was determined from
dried and ground plant samples using the CNS analyser (Vario model EL III (ELE-
MENTAR Analysensysteme GmbH, Hanau, Germany) and it was used to calculate
total N content in mg per pot. The data were statistically processed by the LSD and
Duncan test using the statistical program SPSS 10.0. Correlation coefficients were
calculated to study the associative relations among the measured traits.

45.3 Results and Discussion

In our investigation Italian ryegrass as crop preceding alfalfa was inoculated with four
single effective rhizobial strains, one Enterobacter strain as well as two single strains
of Azotobacter spp. The height of the ryegrass plants was 26.57–29.82 cm without
significant differences among treatments. The highest average values of ryegrass
shoot dry weight (SDW) were obtained in the inoculated treatments with A. vinelandi
strain Av, S. meliloti strain 207 and A. chroococcum strain Ao NDD (Table 45.1).
These results indicated that shoot dry weight (SDW) was significantly influenced
by inoculation with these strains in respect to the other inoculated treatments and
336 D. Delić et al.

Table 45.1 Effect of bacterial inoculants on plant parameters of Italian ryegrass as preceding crop
and alfalfa as subsequent crop
Treatments Bacterial strains Italian ryegrass Alfalfa
2009 2010
SDW Total N SDW Total N Number
mg content mg content of nodules
per mg per per pot mg per per plant
pot pot pot
Inoculation Free-living Av 752.90a 31.80a 3,729.33b,c,e 120.84b,c 10.44b,c
of preceding nitrogen E1 622.95bc 27.40b 3,466.33b,c,e 107.45b,c 18.20b
and of fixers Ao NDD 658.15a,b 24.50bcd 3,655.33b,c,e 102.72c 18.64b
subsequent S. meliloti L3Si 651.35b,c 27.40b 4,939.33a 172.39a 11.33b,c
crop 4148 637.45b,c 23.35cd 3,087.30c 100.31c 26.74a
207 701.20a 26.35bc 3,540.33b,c 107.63bc 27.84a
218 568.85c 21.30d 4,216.30a,b 121.84b,c 17.08b,c
Controlsf Ø1 584.65c 22.75cd 2,180. 00d 47.45d 11.73b,c
NØ 785.40a 34.85a 5,397.21a 178.10a 7.16c
Alfalfa S. meliloti L3Si / / 4,042.62b 137.04b 8.00c
common 4148 / / 2,900.00c 94.25c 15.00b,c
inoculation 207 / / 3,321.67b,c 96.33c 16.00b,c
218 / / 3,530.00b,c 97.66c 7.00c
Controlf Ø2 / / 2,550.00c,d 49.00d 12b,c
LSD 0.05 87.55 3.6 896.71 22.26 7.02
a−d
Means in a column followed by the same letter are not significantly different according to
Duncan’s multiple range test at the 5 % level (p ≤ 0.05)
e
S. meliloti strain 207 was applied for inoculation of subsequent crop
f
Uninoculated controls: NØ-with N and Ø1,2 -without N

control Ø. There were no significant differences between these treatments and the
control treatment with N fertilization (NØ) which pointed out the N fixating and
PGPR ability of these strains.
Rhizobia are known to produce growth promoting substances in the presence
of a rhizosphere which is not the one of a host plant (Kavimandan 1985). Thus,
besides the usual N fixing ability of rhizobia that is expressed only with specific
legumes, it can be assumed that S. meliloti strain 207 has some of PGPR properties
which could affect the SDW increase of Italian ryegrass. Similar results have been
noted by Kavimandan (1985, 1986) and Avis et al. (2008). They found that wheat
seed inoculation with several rhizobial species increased wheat growth and N uptake
under N-limiting conditions. The phosphorus content was significantly increased in
cotton plants inoculated with S. meliloti (Egamberdiyeva et al. 2004).
Among the free living strains applied, the treatment with a A. vinelandi strain
resulted in the highest value of total N content (31.80 mg per pot−1 ) in the shoot
dry matter which was well correlated with SDW (r = 0.912). Yields of rice, cotton
and wheat in many investigations were increased significantly after applications of
Azotobacter due to its ability to fix N2 (Kennedy et al. 2004). Treatments inoculated
with the other bacterial strains realized lower SDW and total N content and there
were no significant differences among them and untreated control-Ø.
45 Influence of Plant Growth Promoting Rhizobacteria on Alfalfa, Medicago . . . 337

Table 45.2 Effect of previous

S. meliloti strains Index of increase
crop seed inoculation on
a −1
Sinorhizobium meliloti Inoculants MPN g soil
number in soil the following L3Si 95 × 102 6.8
year 218 30 × 102 2.1
207 140 × 102 10
Øb3 14 × 102 1
a b
MPN most probable number, Ø3 (Control)
unseeded soil

The number of rhizobial bacteria was determined before sowing and after cutting
the ryegrass inoculated with the investigated S meliloti strains. In the soil before sow-
ing the number of autochthonous S. meliloti was low (14 × 102 g−1 soil) (Table 45.2).
The rhizobial number determined after cutting the ryegrass increased by two to ten
times depending on the rhizobial strain applied which is in agreement with results of
some authors obtained in experiments with other crops (Diatloff 1969; Domit et al.
1990; Goos et al. 2001). This inoculation of Italian ryegrass as crop preceding alfalfa
had an influence on the increase of the number of S. meliloti and together with the
inoculation of the subsequent alfalfa lead to a double inoculation of alfalfa.
In 2010 alfalfa was grown in the pots after the removal of annual ryegrass. All
alfalfa plants as a subsequent crop were inoculated with the same rhizobial strains
applied for the inoculation of the ryegrass. The rhizobial strain 207 was applied in
the pots where ryegrass was inoculated with the free-living bacteria. This is a double
inoculation in the same pots with the same strains: seed inoculation of ryegrass as pre-
ceding crop and seed inoculation of alfalfa as subsequent crop. For comparison with
this double inoculation new treatments were set up in the new pots with alfalfa com-
monly inoculated without preceding crop. The height of alfalfa was not significantly
affected neither by common nor double inoculation. The height of the plants varied
between 44.73 and 54.09 cm without significant differences among treatments (data
not shown). The double inoculation of alfalfa by the strains L3Si and 218 gave the
highest SDW, 4,939.33 and 4,216.30 mg per pot, respectively (Table 45.1). Among
alfalfa treatments without preceding crop, strain L3Si had also the highest effective-
ness expressed by SDW (4,042 mg per pot) and total N content (137.04 mg per pot),
however significantly lower than in the double inoculated treatment.
There were no significant differences between the SDW and N content of alfalfa
obtained in treatments inoculated with strain 207 after the preceding crop inoculated
with free living bacteria and commonly inoculated alfalfa with strain 207. According
to our results it can be assumed that SDW of alfalfa was not influenced by the inoc-
ulation with strains of free living rhizobacteria. The benefit of free-living N fixing
bacteria concerns cereals as ryegrass, while their benefit for legumes is negligible
(Rai and Gaur 1988). Double rhizobial inoculation increased significantly the SDW
of alfalfa significantly in comparison with treatments without preceding crop. These
results indicated that, in a soil with low rhizobial density where alfalfa was not grown
for a long time and with low available N, the rhizobial inoculation of ryegrass with
338 D. Delić et al.

effective rhizobial strains can increase yield and N content of alfalfa as a subsequent
crop. The total N content in the shoots of alfalfa was highly correlated with SDW
(r = 0.966) while there was no correlation between nodule number on one side and
SDW and total N content (r = −0.277) on the other side. Both double and com-
mon inoculations with L3Si, 218, 207 and 4148ss significantly increased the total
N content of alfalfa with respect to the untreated controls. However, only plants
double inoculated with L3Si had the highest total N content (172.39 mg per pot) in
comparison with plants commonly inoculated (Table 45.1). It can be assumed that
the rhizosphere of the preceding crop had influence on the rizobial number increase.
There was effect of the increased number of S. meliloti on nodule number after in-
oculation of Italian ryegrass in some alfalfa treatments (strains 4148 and 218). The
increase of SDW and total N content of alfalfa double inoculated by highly effective
strains L3Si, 218 was likely caused by the increased S. meliloti number after the
inoculation of the ryegrass as a preceding crop.

45.4 Conclusion

Our results indicate that Italian ryegrass yield and N assimilation was increased
by inoculation with A. vinelandi strain Av and S. meliloti strain 207. Nevertheless,
Italian ryegrass seed inoculation with S. meliloti strains led to an increased number
of S. meliloti in a soil with a low density of this species. The Italian ryegrass seed
inoculation with S. meliloti strains L3Si and 218 provided an abundant rhizobial
number in the soil the year before alfalfa growing. This leads together with a common
inoculation of alfalfa to a better yield of alfalfa. This double inoculation of alfalfa
should be applied in soils without or poor in S. meliloti.

Acknowledgments This research was supported by the Ministry of Science of the Republic of
Serbia, Project TR 37006.

References

Antoun H, Prévost D (2005) Ecology of plant growth promoting rhizobacteria. In: Siddiqui ZA (ed)
PGPR: Biocontrol and biofertilization. Springer, Dordrecht
Avis TJ, Gravel V, Antoun H, Tweddell RJ (2008) Multifaceted beneficial effects of rhizosphere
microorganisms on plant health and productivity. Soil Biol Biochem 40:1733–1740
Egamberdiyeva D, Juraeva D, Poberejskaya S, Myachina O, Teryuhova P, Seydalieva L, Aliev A
(2004) Improvement of wheat and cotton growth and nutrient uptake by phosphate solubilizing
bacteria. In: Jordan DL, Caldwell DF (eds) Proceedings of the 26th southern conservation tillage
conference for sustainable agriculture. North Carolina, June 8–9
Goos RJ, Johnson BE, Carr PM (2001) Establishment of Bradyrhizobium japonicum for soybean
by inoculation of a preceding wheat crop. Plant Soil 235:127–133
Delić-Vukmir D, Lugić Z, Radin D, Knežević-Vukčević J, Simić D (1994) Presence and density
of root nodulation rhizobium meliloti bacteria in different soil types of the krusevac region.
Mikrobiologija 31:117–122
45 Influence of Plant Growth Promoting Rhizobacteria on Alfalfa, Medicago . . . 339

Diatloff A (1969) The introduction of Rhizobium japonicum to soil by seed inoculation of non-host
legumes and cereals. Aust J Exp Agric Anim Husb 9:357–360
Domit LA, Costa JA, Vidor C, Pereira JS (1990) Inoculation of cereal seeds with Bradyrhizobium
japonicum and its effect on soybeans grown in succession. Rev Brasileira Ci Solo 14:313–319
Kavimandan SK (1985) Root nodule bacteria to improve yield of wheat. Plant Soil 86:141–144
Kavimandan SK (1986) Influence of Rhizobial inoculation on yield of wheat (Triticum aestivum
L.). Plant Soil 95:297–300
Kennedy IR, Choudhury ATMA, Kecskes ML (2004) Non-symbiotic bacterial diazotrophs in crop-
farming systems: can their potential for plant growth promotion be better exploited? Soil Biol
Biochem 36:1229–1244
Mehboob I, Muhammad N,Ahmad ZZ (2009) Rhizobial association with non-legumes: mechanisms
and applications. Crit Rev Plant Sci 28:432–456
Paul EA, Clark FE (1996) Soil microbiology and biochemistry, Academic Press, New York
Rai SN, Gaur AC (1988) Characterization of azotobacter spp. and effect of Azotobacter and
Azospirillum as inoculant on the yield and N-uptake of wheat crop. Plant Soil 109:131–134
SimićA,Vučković S, Maletić R, Sokolović D, Ðordević N (2009) Impact of seeding rate and interrow
spacing on italian ryegrass for seed in the first harvest year. Turk J Agr Forest 33:425–433
Vincent MJ (1970) Manual for the practical study of root-nodule bacteria. In: IBP handbook, vol
15. Blackwell, Oxford
Chapter 46
Optimal Plant Type of Pea for Mixed Cropping
with Cereals

P. Annicchiarico, P. Ruda, C. Sulas, M. Pitzalis, M. Salis, M. Romani

and A. M. Carroni

Abstract Pea (Pisum sativum L.) in mixed stand (MS) with one small-grain cereal
or in pure stand (PS) is gaining interest for silage production. Breeding programmes,
however, target essentially the grain crop, selecting semi-dwarf germplasm. The aim
of this study was to assess the impact of pea plant stature on biomass production and
competitive ability against cereals. Three semi-leafless pea genotypes with a com-
parable phenology but contrasting plant stature, i.e. (i) one breeding line (‘1/15b’)
lacking dwarfing genes, (ii) one semi-dwarf line with moderate stature (cv. ‘Attika’)
and (iii) one semi-dwarf line with short stature (cv. ‘Spirale’), were grown in PS
and in binary mixtures with barley (cv. ‘Cometa’) or triticale (cv. ‘Amarillo’) in
a Mediterranean environment. Harvest and dry matter yield assessment occurred at
waxy stage of the pea grain. Seed densities for pure stands (100 seeds/m2 for pea; 320
seeds/m2 for cereals) were halved for mixtures. The pea genotypes ‘1/15b’, ‘Attika’
and ‘Spirale’ ranked in this order for plant height and pea yield in PS or MS, total
(pea + cereal) yield in MS, and pea proportion in MS, indicating the strict association
of pea stature with biomass yield and competitive ability. Pea yield and proportion
were lower and with smaller genotypic differences in MS with triticale than in MS
with barley. The only drawback of the pea ‘1/15b’ was its greater susceptibility to
lodging, which emerged as a problem only in PS.

Keywords Dwarfing gene · Competitive ability · Forage · Intercropping ·

Mixtures · Plant stature

P. Annicchiarico () · M. Romani

Centro di Ricerca per le Produzioni Foraggere e Lattiero-Casearie (CRA-FLC), Lodi, Italy
e-mail: [email protected]
P. Ruda · C. Sulas · M. Pitzalis · M. Salis · A. M. Carroni
Unità di Ricerca per i Sistemi Agropastorali in Ambiente Mediterraneo (CRA-AAM), Sanluri, Italy

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 341

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_46, © Springer Science+Business Media Dordrecht 2013
342 P. Annicchiarico et al.

46.1 Introduction

Field pea (Pisum sativum L.) is largely grown in European crop-livestock systems
as a high-protein feed crop, owing to its high grain yield potential and the value of
its straw as a forage (Carrouée et al. 2003). In addition, pea is gaining interest as a
crop for silage production in mixed stand (MS) with one small-grain cereal (such as
triticale, barley or wheat) or in pure stand (PS). Silage produced from pea PS has
higher protein content but tends to display lower yield per unit area and poorer silage
fermentation and quality than that from pea-cereal MS. However, pea silage with sat-
isfactory quality can be obtained by adding an inoculant (Lactobacillus plantarum)
and/or by ensiling material whose dry matter content has overcome 30 % through de-
layed harvest and a wilting period (Borreani et al. 2009). Mixed cropping of pea with
cereals tends to maximize the crop forage yield and silage quality but usually im-
plies a marked competitive disadvantage for the pea component (Hauggaard-Nielsen
and Jensen 2001; Lithourgidis et al. 2011), especially in fertile soils and/or under
moderate nitrogen fertilization (Corre-Hellou et al. 2006). In northern Italy the pea
proportion in MS may drop below 20 % when adopting a semi-dwarf pea cultivar
and component seed rates that are halved relative to seed rates used in pure stands
(Tomasoni et al. 2006). The pea competitive disadvantage may be limited either by
increasing largely the pea proportion in the seed mixture (Lithourgidis et al. 2011),
or by selecting pea varieties with better ability to compete for light and nutrients
(Hauggaard-Nielsen and Jensen 2001).
Pea breeding programmes target essentially the grain crop, selecting varieties with
dwarfing genes to increase the crop harvest index and standing ability (Ranalli 1995).
Additional selected traits for improving the standing ability are semi-leaflessness
(i.e. the modification of leaflets in tendrils) and stem stiffness (Stelling 1989; Ranalli
1995). The semi-dwarf plant type may be less suited to silage production in PS or MS
than a tall type, as it may imply lower pea biomass production and competitiveness
against cereals. Research on wheat has highlighted that tall germplasm tends to dis-
play greater biomass production (Annicchiarico et al. 2005) and better competitive
ability against weeds (Vandeleur and Gill 2004) than semi-dwarf material. In gen-
eral, extensive shoot elongation and early growth are the main traits associated with
inter-specific competitive ability of crop species (Lemerle et al. 2001). In a study
comparing six leafed pea cultivars in a binary mixture with barley, a tall indetermi-
nate cultivar and a very short cultivar were the best- and worst-competing genotypes,
respectively (Hauggaard-Nielsen and Jensen 2001). Pea competitive ability against
weeds was associated with taller stature in most (Mc Donald 2003; Annicchiarico
and Filippi 2007) but not all studies (Townley-Smith and Wright 1994). Leaf type
showed negligible influence on pea competitive ability (Mc Donald 2003; Townley-
Smith and Wright 1994), supporting the general interest of leaflessness because of
its positive effect on the standing ability.
The impact of pea stature on the suitability for silage production should also
consider the fact that semi-dwarf cultivars may differ for plant height, owing to the
several different dwarfing genes and their possibly different effect (Huyghe 1998).
46 Optimal Plant Type of Pea for Mixed Cropping with Cereals 343

This study aimed to produce information on the optimal pea ideotype for silage
production, by comparing three semi-leafless genotypes with a comparable phe-
nology but contrasting plant stature for biomass production in PS and MS and pea
competitive ability in MS with barley or triticale. One genotype lacked dwarfing
genes, whereas two were semi-dwarf with moderate or short stature.

46.2 Materials and Methods

The tall, semi-leafless plant type was represented by the advanced line ‘1/15b’, se-
lected by CRA-FLC from a cross between the cultivars ‘Santana’ (semi-dwarf, early
flowering and with good standing ability) and ‘Forrimax’ (tall and late flowering).
The progenies from this cross underwent selection for early flowering, standing abil-
ity and winter survival under field conditions in Lodi (Po Valley) for three seasons
starting from the F3 generation. The cultivars ‘Attika’ and ‘Spirale’, both featuring
early flowering, good standing ability and wide adaptation to Italian environments
(Annicchiarico and Iannucci 2008), represented the semi-dwarf germplasm with
moderate and short stature, respectively. ‘Attika’ was the tallest of 46 recent semi-
dwarf European and Australian varieties tested by Annicchiarico et al. (2003) in
climatically-contrasting Italian sites, whereas ‘Spirale’ displayed high grain yield
and harvest index associated with short stature in the same evaluation. The cereal
companions in pea-cereal binary mixtures were the varieties ‘Cometa’ of barley
(Hordeum vulgare L.) and ‘Amarillo’ of triticale (× Triticosecale Wittmack), which
ranked among the best-yielding cultivars of the respective species in the Italian
network of autumn-sown variety trials for 2010.
Two trials were performed in different fields of the Mediterranean site of Sanluri
(Sardinia) in 2011. One included the PS of barley and pea genotypes, and pea-barley
binary mixtures; the other included the PS and MS of triticale and pea genotypes.
Both trials were designed as a split-plot with four replications, holding pea cropping
condition (PS or MS) on main plots and pea genotypes on subplots. The cereal PS
was added to MS subplots. Seed densities for PS were those recommended, i.e., 100
seeds/m2 for pea and 320 seeds/m2 for cereals, halving them and mixing the seed for
MS. Plot size was 4.0 × 1.5 m, with rows spaced 0.18 m apart. The autumn sowing
was delayed until January 10, owing to unfavourable climatic conditions. Harvests
took place between May 18 and May 23 at waxy stage of the pea grain, assessing
forage dry matter (DM) yield of pea and cereal genotypes in PS and MS. Forage DM
content of pea and cereals were assessed on a fresh herbage sample of at least 3 kg
per plot. This sample was also used for assessing the pea proportion on total DM in
MS, which estimated the pea competitive ability. Onset of flowering, plant height at
crop harvest, and pea susceptibility to lodging at crop harvest (expressed on a visual
5-level scale ranging from 1 = erect to 5 = extensively lodged) were also recorded.
The rainfall amount from January 1 to May 15 was 224 mm, slightly higher than the
long-term value for the site (195 mm).
344 P. Annicchiarico et al.

46.3 Results and Discussion

The delayed autumn sowing retarded the reproductive development of both cereals
and reduced their DM yield in PS and their competitive ability in MS relative to an
ordinary cropping season.
The differences in plant stature among pea genotypes were larger between tall and
semi-dwarf material than between semi-dwarf genotypes, but ‘Attika’ was distinctly
taller than ‘Spirale’ in all PS or MS conditions (Table 46.1). Only the pea genotype
lacking dwarfing genes displayed taller stature than any cereal companion in MS.
The pea genotypes ‘1/15b’, ‘Attika’ and ‘Spirale’ ranked in this order not only for
plant height but also for pea DM yield in PS or MS, total (pea + cereal) DM yield in
MS, and pea proportion in MS (Table 46.1). The respective values of these genotypes
averaged across the two experiments were 8.25, 6.86 and 5.52 t ha for pea DM yield
in PS; 5.01, 3.04 and 2.00 t ha for pea DM yield in MS; 7.91, 7.07 and 6.91 t ha for
total DM yield of mixtures; and 0.61, 0.43 and 0.29 for pea proportion in MS. These
results highlighted the positive association of pea stature with pea forage yield in
PS or MS and with pea competitive ability in MS. They also suggested that better
competitive ability of the less competitive species, i.e. pea, may have a positive
effect also on total yield of the mixture, in agreement with prior observations on
legume-grass binary mixtures including genotypes of a poorly-competing species
such as white clover in association with highly-competing grass species or varieties
(Annicchiarico and Piano 1994).
The two cereals showed similar DM yield in PS, but triticale displayed greater
competitive ability against pea than barley associated with its taller stature (Ta-
ble 46.1). The tall pea line ‘1/15b’ was at competitive disadvantage only with triticale
on the basis of pea proportion data (Table 46.1). In general, pea yield and pea pro-
portion were lower and with smaller genotypic differences in MS with triticale than
in MS with barley (Table 46.1).
PS or MS including the pea genotype ‘1/15b’ also tended to outperform the cereal
PS (Table 46.1). The only drawback of the tall genotype was its greater susceptibility
to lodging relative to semi-dwarf material, which emerged as a problem only in PS
(where its lodging susceptibility score averaged nearly 4.5 compared with 2.9 for
‘Attika’ and 1.9 for ‘Spirale’). Thus, the semi-leafless trait and the selection for
standing ability could not provide this genotype with sufficient standing ability in
PS in this test environment.
On the whole, our results indicate that the tall semi-leafless pea plant type is
superior to the semi-dwarf semi-leafless plant type for silage production in MS with
cereals. This indication is likely to be reinforced under timely autumn sowing, which
is expected to increase the cereal competitive ability. The seed production of the tall
pea type in PS, however, may suffer of losses due to insufficient tolerance to lodging
(whereas its seed production in MS is hindered by current regulations for producing
certified seed). The semi-dwarf pea type with maximized plant stature is preferable
to the tall plant type for silage production in PS, owing to its better standing ability.
46 Optimal Plant Type of Pea for Mixed Cropping with Cereals 345

Table 46.1 Pea plant height at crop harvest, forage dry matter (DM) yield of the crop and its
components, and pea proportion on total DM, for three pea genotypes with contrasting plant stature
grown in pure stand (PS) and mixed stand (MS) with barley or triticale in Sardinia
Experimenta Pea height Total DM Pea DM Cereal DM Pea
(cm)bc yield (t/ha)b yield (t/ha)b yield (t/ha)b proportionb
With barley
PS, barley – 5.93 de – 5.93 a –
PS, pea ‘1/15b’ 118.1 a 8.83 ab 8.83 a – –
PS, pea ‘Attika’ 72.3 c 7.22 cd 7.22 b – –
PS, pea ‘Spirale’ 54.4 d 5.62 e 5.62 c – –
MS, barley + pea ‘1/15b’ 103.2 b 9.23 a 7.05 b 2.18 c 0.76 a
MS, barley + pea ‘Attika’ 66.5 c 7.79 bc 3.63 d 4.16 b 0.46 b
MS, barley + pea ‘Spirale’ 49.7 d 7.25 cd 2.23 d 5.02 ab 0.31 c
With triticale
PS, triticale – 5.94 bc – 5.94 a –
PS, pea ‘1/15b’ 118.5 a 7.67 a 7.67 a – –
PS, pea ‘Attika’ 69.4 b 6.50 b 6.50 b – –
PS, pea ‘Spirale’ 48.9 c 5.42 c 5.42 c – –
MS, triticale + pea ‘1/15b’ 108.5 b 6.59 b 2.96 d 3.63 c 0.45 a
MS, triticale + pea ‘Attika’ 61.4 b 6.35 bc 2.46 d 3.89 c 0.39 a
MS, triticale + pea ‘Spirale’ 45.5 d 6.57 b 1.78 e 4.79 b 0.27 b
a
Pea genotypes are: line ‘1/15b’, tall; ‘Attika’, tall within semi-dwarf type; ‘Spirale’, short within
semi-dwarf type (all semi-leafless). Cereal genotypes are ‘Cometa’ for barley and ‘Amarillo’ for
triticale
b
Column means within experiment with different letters differ at P < 0.05 according to Duncan’s
test
c
Mean cereal plant height at mixed stand harvest: 83.2 cm for barley, 97.1 cm for triticale

Acknowledgments The pea line ‘1/15b’was selected within the project ‘Plant Genetic Resources—
FAO Treaty’ funded by the Italian Ministry for Agriculture, Food and Forestry Policies.

References

Annicchiarico P, Piano E (1994) Interference effects in white clover genotypes grown as pure stands
and binary mixtures with different grass species and varieties. Theor Appl Genet 88:153–158
Annicchiarico P, Filippi L (2007) A field pea ideotype for organic systems of northern Italy. J Crop
Improv 20:193–203
Annicchiarico P, Iannucci A (2008) Adaptation strategy, germplasm type and adaptive traits for
field pea improvement in Italy based on variety responses across climatically contrasting
environments. Field Crops Res 108:133–142
Annicchiarico P, Iannucci A, Filippi L (2003) Cultivar di colture proteiche a confronto in areali
contrastanti. L’Inform Agr 59(42):73–76
Annicchiarico P,Abdellaoui Z, Kelkouli M, Zerargui H (2005) Grain yield, straw yield and economic
value of tall and semi-dwarf durum wheat cultivars in Algeria. J Agric Sci 143:57–64
Borreani G, Revello Chion A, Colombini S, Odoardi M, Paoletti R, Tabacco E (2009) Fermentative
profiles of field pea (Pisum sativum), faba bean (Vicia faba) and white lupin (Lupinus albus)
silages as affected by wilting and inoculation. Anim Feed Sci Technol 151:316–323
Carrouée B, Crépon K, Peyronnet C (2003) Les protéagineux: intérêt dans les systèmes de
production fourragers français et européens. Fourrages 174:163–182
346 P. Annicchiarico et al.

Corre-Hellou G, Fustec J, CrozatY (2006) Interspecific competition for soil N and its interaction with
N2 fixation, leaf expansion and crop growth in pea-barley intercrops. Plant Soil 282:195–208
Hauggaard-Nielsen H, Jensen ES (2001) Evaluating pea and barley cultivars for complementarity
in intercropping at different levels of soil N availability. Field Crops Res 72:185–196
Huyghe C (1998) Genetics and genetic modifications of plant architecture in grain legumes: a
review. Agronomie 18:383–411
Lemerle D, Gill GS, Murphy CE, Walker SR, Cousens RD, Mokhtari S, Peltzer SJ, Coleman R,
Luckett DJ (2001) Genetic improvement and agronomy for enhanced wheat competitiveness
with weeds. Aust J Agric Res 52:527–548
Lithourgidis AS, Vlachostergios DN, Dordas CA, Damalas CA (2011) Dry matter yield, nitrogen
content, and competition in pea-cereal intercropping systems. Eur J Agron 34:287–294
McDonald GK (2003) Competitiveness against grass weeds in field pea genotypes. Weed Res
43:48–58
Ranalli P (1995) Improvement of pulse crops in Europe. Eur J Agron 4:151–166
Stelling D (1989) Problems of breeding for improved standing ability in dried peas, Pisum sativum
L. J Agron Crop Sci 163:21–32
Tomasoni C, Annicchiarico P, Borrelli L, Filippi L (2006) Insilati di leguminose da granella
consociate a cereali. L’Inform Agrario 62(14):57–61
Townley-Smith L, Wright AT (1994) Field pea cultivar and weed response to crop seed rate in
western Canada. Can J Plant Sci 74:387–393
Vandeleur RK, Gill GS (2004) The impact of plant breeding on the grain yield and competitive
ability of wheat in Australia. Aust J Agric Res 55:855–861
Chapter 47
Dry Matter Recovery and Aerobic Stability
of Maize Whole-Crop, Cob and Stover
Silages—Harvest Date and Cultivar Effects

J. P. Lynch, P. O’Kiely and E. M. Doyle

Abstract Forage maize (Zea mays L.) has the potential to produce high yields of
excellent quality feed for ruminants. However, for regions with a cool overcast
climate, improvements in maize silage production systems are required to reduce
variation in yield and quality. Ensiling can reduce yield through dry matter (DM)
losses and may alter feed quality. This study investigated the effects of harvest date on
the DM recovery and aerobic stability of whole-crop, cob and stover silages produced
from contrasting cultivars of maize. Six cultivars of forage maize, four of which were
categorised as conventional (Tassilo, FAO 190; Beethoven, FAO 200; Andante FAO
200 and Nescio, FAO 230) and two categorised as high biomass (Atletico, FAO 280
and KXA 7211, FAO 260) were sown in 72 m2 plots under plastic mulch on 7 May
2008. Within each of three replicate blocks, harvest date (16 September, 7 October
and 28 October) constituted the main plots and cultivar the sub plots within a split-
plot design. Samples of whole crop, stover and cob from each plot were precision
chopped and 6 kg of each were ensiled in laboratory silos for 130 days at 15 ◦ C. After
opening, sub-samples were subjected to chemical and microbial analyses, while
aerobic stability and deterioration were estimated by measuring silage temperature
during 8 days exposure to air. The rate of DM recovery of ensiled whole-crop, cob
and stover was unaffected (P > 0.05) by harvest date or cultivar. No overall effects of
harvest date or cultivar were observed on the aerobic stability or aerobic deterioration
of whole-crop or stover silages. Cob silages harvested on 16 September underwent
more (P < 0.05) aerobic deterioration than cob silages harvested at later dates. Cob
silages produced from Nescio underwent less (P < 0.05) aerobic deterioration than
for Tassilo, Beethoven and Andante.

J. P. Lynch () · P. O’Kiely

Animal & Grassland Research and Innovation Centre, Teagasc, Grange,
Dunsany, Co. Meath, Ireland
e-mail: [email protected]
J. P. Lynch · E. M. Doyle
School of Biology and Environmental Science, UCD, Belfield, Dublin, 4, Ireland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 347

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_47, © Springer Science+Business Media Dordrecht 2013
348 J. P. Lynch et al.

47.1 Introduction

Forage maize (Zea mays L.) silage has the potential to support high animal perfor-
mance in ruminant production systems compared to more conventional conserved
forages in Ireland and Britain (Fitzgerald and Murphy 1999; Phipps et al. 2000;
Keady et al. 2007). The limitations often imposed by relatively low solar radiation
and temperatures in Ireland, which often result in sub-optimal yields and feed qual-
ity, need to be overcome for maize to be considered a sustainable and economically
viable alternative forage. Previous studies in Ireland have focused on the effects
which crop management factors, such as cultivar selection, the use of plastic mulch,
seeding rate and sowing date (Keane 2002; Keane et al. 2003; Farrell and Gilliland
2011) have on the yield and nutritive value of whole-crop maize previous at time of
harvest. However, the subsequent ensilage of forages can alter their nutritional value
and may incur qualitative and quantitative losses through the production of effluent
or gas during the fermentation process (McDonald et al. 1991). In addition, sub-
stantial losses due to poor aerobic stability and aerobic deterioration can occur when
silage chemical components are respired by aerobic organisms. These can decrease
the amount of dry matter (DM) available, produce undesirable by-products which
place an animals health at risk, such as mycotoxins (Muck 2010).
The objectives of this study were to determine the effects of cultivar and harvest
date on the DM recovery and on the aerobic stability of whole-crop, cob and stover
silages.

47.2 Materials and Methods

The experiment had a split-plot design, with three main plots (date of harvest) and
six sub plots (maize cultivar) in each of three replicate blocks. Of the six cultivars of
maize, four conventional cultivars were representative of what is sown by commercial
livestock farmers for silage production in Ireland (Tassilo FAO 190; Beethoven, FAO
200; Andante, FAO 200; Nescio, FAO 230) and two were categorised as high biomass
cultivars (Atletico, FAO 280; KXA 7211, FAO 260). The three harvest dates of
16 September, 7 October and 28 October represented an early, normal and late har-
vest, respectively, for a midland site in Ireland. On each harvest date, whole-crop, cob
and stover samples were manually harvested and subsequently precision-chopped.
A 6 kg sub-sample of each chopped sample was ensiled in a laboratory silo (O’Kiely
and Wilson 1991) for 130 days at approximately 15 ◦ C. Following the ensilage period,
effluent (if any) was collected and weighed. Silage was weighed, aseptically mixed
and sub-sampled for chemical analyses. Aerobic stability was determined by placing
3.6 kg of each silage into a polystyrene box (2.5 cm thick; 59 cm × 39 cm × 22 cm)
lined with polythene and loosely covered with a polystyrene lid. A thermocouple
was placed in the centre of each silage sample. The temperature of the silage was
recorded on an hourly basis over 8 day period by a data logger (SQ ELTEK 80T;
47 Dry Matter Recovery and Aerobic Stability of Maize Whole-Crop, Cob . . . 349

Eurolec Instrumentation Ltd, Dundalk, Ireland). Reference temperatures were ob-

tained from containers of water stored beside the boxes of silage. The indices of
aerobic stability and deterioration were expressed as (i) hours elapsed until the tem-
perature rose more than 2 ◦ C above the reference temperature and (ii) accumulated
temperature rise during 120 h exposure to air. Chemical analyses were as described
by McEniry et al. (2006). Data were analysed as a split-plot design using a model
that accounted for harvest date as the main plot, cultivar as the sub plot, and their
interactions, using the PROC GLM procedure of the SAS statistical program (SAS
2002). Treatment contrasts were made using the Fischer least significant differences
test.

47.3 Results and Discussion

No effect of harvest date or cultivar (P > 0.05) was observed on the DM recovery of
ensiled whole-crop, cob or stover (Table 47.1). However, the mean (SD) whole-crop
DM recovery value of 908 g (45.4 g) silage DM/ kg DM ensiled was lower than
reported by Johnson et al. (2002; 940–1000 g silage DM/kg DM ensiled). The addi-
tional losses in the present study likely resulted from the relatively low herbage DM
content leading to effluent production and a more extensive fermentation resulting
in increased CO2 production.
The values for the indices of aerobic stability (135–192 h for silage temperature
to increase more than 2 ◦ C above ambient temperature) and aerobic deterioration
(1–9 ◦ C accumulated temperature rise during 120 h exposure to air) in the present
study indicate that these whole-crop silages were quite aerobically stable. This lim-
ited aerobic activity reflected a relatively high content of acetic acid in the present
study. Acetic acid is inhibitory to yeast, the primary organisms initiating aerobic de-
terioration (Muck 2010), and yeast numbers were generally low in these whole-crop
silages. No effect of harvest date or cultivar (P > 0.05) were observed on the aero-
bic stability or deterioration of whole-crop or stover silages as all these silages had
high acetic acid concentrations and low yeast numbers, and were thus aerobically
stable. Cob silages from crops harvested on the 16 September had a more extensive
deterioration after exposure to air than silages from crops harvested at later dates.
In conclusion, later harvesting resulted in improved DM recovery and aerobic
deterioration of cob silages, while whole-crop and stover silages were unaffected by
harvest date. The high biomass crops did not confer a disadvantage in DM recovery
or aerobic stability of whole-crop, cob or stover silage when compared to the conven-
tional cultivars. The DM recovery and aerobic deterioration values for whole-crop
maize silages were generally intermittent between the values for the two individually
ensiled components of the crop.

Acknowledgments Funding for this study was provided under the National Development Plan
through the Research Stimulus fund administered by the Department of Agriculture, Fisheries
and Food (RSF 07 501). The provision of maize seed by Seed Technology Ltd., Ballymountain,
 350

Table 47.1 Silage fermentation characteristics, DM recovery and aerobic stability and deterioration of whole-crop, cob and stover silages
Harvest 16 September 7 October 28 October SEMf Significance
a
Cultivar T B An N At K T B An N At K T B An N At K H C H×C H C H×C
Whole- TFPb 185 181 220 171 182 201 112 95 134 128 202 176 76 65 76 76 148 102 6.2 8.5 14.8 *** *** *
crop AAb 33 41 36 30 44 38 39 31 37 28 48 45 35 28 43 27 60 34 3.1 4.5 7.8 *
pH 3.6 3.7 3.6 3.5 3.7 3.6 3.9 3.6 3.7 3.8 3.6 3.6 4.3 4.2 4.1 4.4 4.2 4 0.09 0.08 0.13 *
DMRc 858 863 859 866 805 923 979 921 924 949 962 938 944 895 948 927 874 919 21.8 19.5 33.8
ITR > 2 ◦ Cd 190 181 172 192 105 175 134 158 144 167 192 135 162 138 192 192 136 192 11.2 18.1 31.3
ACT 120 he 3 2 4 2 4 4 9 2 2 1 1 1 3 5 1 2 1 1 1.5 1.4 2.4
Cob TFPb 61 67 114 80 156 160 66 61 75 55 216 99 47 37 66 43 73 88 11.4 12.6 21.9 ***
AAb 11 12 31 17 55 46 15 9 25 11 57 34 16 14 18 10 33 23 3.4 4.9 8.4 ***
pH 3.7 3.7 3.8 3.9 4.1 4 3.7 3.7 3.7 3.7 3.7 3.8 4 4.1 4.2 4.2 3.8 3.9 0.04 0.05 0.8 ** **
DMRc 921 908 858 872 837 844 988 925 900 967 983 934 934 983 975 974 948 976 20.7 21.7 37.6
ITR > 2 ◦ Cd 48 41 98 173 101 113 83 117 77 126 192 167 192 192 126 192 133 128 21.3 16.4 24.3 *
ACT 120 he 44 46 26 1 31 20 18 9 15 7 1 1 2 1 1 2 1 3 3.3 3.7 6.4 * * *
Stover TFPb 173 173 155 137 167 155 156 107 165 120 135 144 89 88 121 66 145 83 3.6 7.0 12.2 *** ** *
AAb 68 72 62 66 72 62 72 61 75 63 57 70 49 36 59 41 64 47 3.7 2.1 3.6 * * ***
pH 4.2 4.1 4.5 4.2 4.4 4.4 4.2 4.4 4.5 4.3 3.8 4 4.4 4.3 4.3 4.6 4.4 4.4 0.06 0.06 0.1 **
DMRc 855 847 875 873 812 872 943 956 953 920 903 930 903 937 883 866 889 892 15.1 22.5 38.9
ITR > 2 ◦ Cd 162 147 192 192 170 192 192 192 192 192 192 192 167 192 143 173 143 192 10.5 12.7 22.0
ACT 120 he 6 9 2 2 3 1 1 3 2 1 1 2 2 0 2 2 1 1 1.0 1.2 2.1
*P < 0.05; **P < 0.01; ***P < 0.001
H harvest date, T Tassilo, B Beethoven, An Andante, N Nescio, At Atletico, K KXA 7211, TFP total fermentation products, AA acetic acid
a
Cultivar C
b
Gram per kilogram DM
c
Gram silage DM/kg DM ensiled
d
Interval (h) until temperature rises more than 2 ◦ C above ambient temperature (ITR > 2 ◦ C; index of aerobic stability)
e
Accumulated temperature rise (◦ C) during 120 h exposure to air (ACT, index of aerobic deterioration)
f
Standard error of the mean
J. P. Lynch et al.
47 Dry Matter Recovery and Aerobic Stability of Maize Whole-Crop, Cob . . . 351

Ferrybank, Waterford, Ireland, the input into crop production and ensilage by B. Weldon and Grange
farm staff and the chemical analyses by Grange laboratory staff are acknowledged.

References

Farrell AD and Gilliland TJ (2011)Yield and quality of forage maize grown under marginal climatic
conditions in Northern Ireland. Grass Forage Sci 66(2) 3–10
Fitzgerald JJ and Murphy JJ (1999) A comparison of low starch maize silage and grass silage and
the effect of concentrate supplementation of the forages or inclusion of maize grain with the
maize silage on milk production by dairy cows. Livest Prod Sci 57:95–111
Johnson LM, Harrison JH, Davidson D, Mahanna WC, Shinners K and Linder D (2002) Corn silage
management: effects of maturity, inoculation, and mechanical processing on pack density and
aerobic stability. J Dairy Sci 85:434–444
Keady TWJ, Lively FO, Kilpatrick DJ and Moss BW (2007) Effects of replacing grass silage with
either maize or whole-crop wheat silages on the performance and meat quality of beef cattle
offered two levels of concentrates. Animal 1:613–623
Keane GP (2002) Agronomic factors affecting the yield and quality of forage maize in Ireland:
effect of sowing date and plastic film treatment. Grass Forage Sci 57:3–10
Keane GP, Kelly J, Lordan S and Kelly K (2003) Agronomic factors affecting the yield and quality
of forage maize in Ireland: effect of plastic film system and seeding rate. Grass Forage Sci
58:362–371
McDonald P, Henderson AR and Heron SJR (1991) The biochemistry of silage. Chalcombe
Publications, Bucks, UK
McEniry J, O’Kiely P, Clipson NJW, Forristal PD and Doyle EM (2006) The microbiological and
chemical composition of baled and precision-chop silages on a sample of farms in County Meath.
Irish J Agr Food Res 45:73–83
Muck RE (2010) Silage microbiology and its control through additives. Revista Brasileira de
Zootecnia 39:183–191
O’Kiely P and Wilson RK (1991) Comparison of three silo types used to study in-silo processes.
Irish J Agr Res, 30:53–60
Phipps RH, Sutton JD, Beever DE and Jones AK (2000) The effect of crop maturity on the nutritional
value of maize silage for lactating dairy cows. 3. Food intake and milk production. Brit Soc
Anim Sci 71:401–409
SAS (2002) Statistical analysis institute. SAS Institute Inc., Cary, N.C., USA
Chapter 48
Performance of Forage Soya Bean (Glycine max)
Cultivars in the Northern Balkans

V. Mihailović, A. Mikić, V. Ðordević, B. Ćupina, V. Perić, Ð. Krstić, M. Srebrić,

S. Antanasović and T. E. Devine

Abstract Soya bean (Glycine max (L.) Merr.) is the most important grain legume
crop in many West Balkan Countries. A programme on the alternative uses of soya
bean such as forage, biomass or green manure has been recently launched in the
Institute of Field and Vegetable Crops, the Faculty of Agriculture in Novi Sad and
the Maize Research Institute Zemun Polje in Belgrade. A small-plot trial has been
carried out in Novi Sad and Zemun Polje during 2009 and 2010 with four US forage
soya bean cultivars. In both years and at both locations, all four cultivars were sown
in late April, with a target sowing density of 75 viable seeds m−2 , and cut in the
stages of full flowering or first pods development, mostly in the second half of July.
In a 2-year average, the cultivar Donegal had the highest yields of both green forage
(63.9 t ha−1 ) and forage dry matter (15.1 t ha−1 ). In single years, the highest yields
were recorded in the cultivar Donegal, in Novi Sad in 2010, with 82.4 t ha−1 of green
forage and 18.4 t ha−1 of forage dry matter.

Keywords Forage DM yield · Green forage yield · Serbia · Soya bean

48.1 Introduction

Pea (Pisum sativum L.) and common vetch (Vicia sativa L.) are the most traditional
and widely cultivated annual forage legumes in Serbia and other Balkan countries.
Both have autumn and spring sown forms and are often grown in mixtures with
small grains such as oat (Avena sativa L.) and triticale (× Triticosecale spp.). Annual

A. Mikić () · V. Mihailović · V. Ðordević

Institute of Field and Vegetable Crops, 21000 Novi Sad, Serbia
e-mail: [email protected]
e-mail: [email protected]
B. Ćupina · Ð. Krstić · S. Antanasović
Faculty of Agriculture of the University of Novi Sad, 11185 Belgrade, Serbia
V. Perić · M. Srebrić
Maize Research Institute Zemun Polje, 21000 Novi Sad, Serbia
T. E. Devine
USDA, ARS, Sustainable Agricultural Systems Laboratory, Beltsville, USA

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 353

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_48, © Springer Science+Business Media Dordrecht 2013
354 V. Mihailović et al.

forage legumes may be used as green forage, forage dry matter, forage meal, silage
or haylage (Ćupina et al. 2011). They are also considered valuable green manure
crops in organic farming and sustainable agriculture (Ćupina et al. 2004).
Soya bean (Glycine max (L.) Merr.) is the most important annual legume crop
in Serbia, with an average harvested area of 150,000 ha (Mikić et al. 2009). It is
cultivated mostly for the production of grain rich in protein and oil. Soya bean crop
may also be a source of quality forage during the summer, when other annual forage
brassicas or legumes are already cut (Bilgili et al. 2003; Devine et al. 2010).
The first forage soya bean cultivars in USA introduced from Asia were not adapted
to the new conditions. At the same time, soya bean breeding programmes for grain
produced highly adapted cultivars with enhanced disease, insect, and nematode re-
sistances. Both these and traditional hay-type cultivars were used in the USDA-ARS
forage soya bean breeding programme by conventional methods and remarkably tall
and lodging resistant lines were developed (Devine et al. 2007).
This research aims at assessing the potential of soya bean for forage production in
climatic regions of Europe similar to that of Serbia and northern Balkans, by testing
the performance of some of the commercial US forage soya bean cultivars developed
at USDA-ARS and widely used throughout USA.

48.2 Materials and Methods

A small-plot trial was carried in 2009 and 2010 at Rimski Šančevi near Novi Sad and
Zemun Polje near Belgrade, including four US forage soya bean cultivars, namely
Derry (Devine et al. 1998a), Donegal (Devine and Hatley 1998), Tara (Devine and
McMurtrey 2004) and Tyrone (Devine et al. 1998b).
In both years and at both locations, the trial was established in the last week of
April. All cultivars were sown at a target density of 75 viable seeds m−2 (Acikgoz
et al. 2007) and at a row spacing of 20 cm (Seiter et al. 2004). The dominant weather
conditions during the trial period at Rimski Šančevi and Zemun Polje are given in
Table 48.1.
Each cultivar was cut in full bloom, as an optimal stage and a balance between yield
and quality in most annual forage legumes (Mihailović et al. 2009). In both years and
at both locations, this was in mid- to late August. Plant samples taken immediately
before cutting were used to record the main forage yield components, which were
plant height (cm), number of internodes (per plant) and number of photosynthetic
active leaves (per plant). Green forage yield (t ha−1 ) was based on the green forage
yield per plot, measured immediately after cutting. Forage dry matter yield (t ha−1 )
was calculated on the basis of the ratio between the green mass of forage samples of
1 kg before and after the drying at the controlled room temperature until the confirmed
constant mass.
The results were processed by analysis of variance (ANOVA) using the LSD test,
as well as the significance of genotypes, environments and genotype × environment
48 Performance of Forage Soya Bean (Glycine max) Cultivars in the Northern Balkans 355

Table 48.1 Monthly average temperatures and precipitation at Rimski Šančevi and Zemun Polje
during the trial
Location Year April May June July Average
Average monthly temperature (◦ C)
Rimski Šančevi 2009 15 18 20 23 19
2010 13 17 20 23 18
Long-term 11 17 20 21 17
Zemun Polje 2009 16 20 21 24 20
2010 13 18 20 24 19
Long-term 13 18 22 23 19
Monthly precipitation sum (mm)
Rimski Šančevi 2009 2 47 123 57 229
2010 71 95 174 98 438
Long-term 47 59 85 70 261
Zemun Polje 2009 7 27 72 31 137
2010 65 89 155 103 412
Long-term 43 41 76 56 216

(G × E) interactions (Steel and Torrie 1960) for forage dry matter yield, as the most
important agronomic characteristic.

48.3 Results and Discussion

As for the average monthly temperature, both soya bean growing seasons were mostly
warmer in comparison to the long-term averages, especially at Rimski Šančevi. The
season of 2009 was generally warmer than 2010, especially during April and July.
Regarding the monthly precipitation sums, at both locations, the season of 2009 was
drier than the long-term averages, while the season of 2010 was extremely wetter,
with a specific emphasis in June, with doubled long-term values at both locations.
Since the extremity of both years, a specific analysis of the interaction between
cultivars and locations was purposely left out for the next seasons and more data.
There were significant differences between the average values of all three forage
yield components among the four tested soya bean cultivars (Table 48.2).
The average plant height varied from 141 cm in Tyrone and 142 cm Derry to
161 cm Donegal (Table 48.2). The smallest plant height (138 cm) was both Derry
and Tyrone at Zemun Polje. The greatest plant height (166 cm) was in the cultivar
Donegal at Rimski Šančevi. This was lower than that recorded in performance trials
at Orange, Virginia, USA (Darmosarkoro et al. 2001), where Tyrone reached a height
of 180 cm.
Donegal had the highest average values of both number of internodes (29 per
plant) and number of active leaves (25 per plant). Tyrone had the smallest average
number of both internodes (24 per plant) and active leaves (21 per plant).
356 V. Mihailović et al.

Table 48.2 Two-year average of forage yield components of four soya bean cultivars at Rimski
Šančevi and Zemun Polje
Cultivar Location Plant height Number of internodes Number of leaves
(cm) (per plant) (per plant)
Derry Rimski Šančevi 146 26 22
Zemun Polje 138 29 24
Average 142 28 23
Donegal Rimski Šančevi 166 27 23
Zemun Polje 155 30 26
Average 161 29 25
Tara Rimski Šančevi 152 29 25
Zemun Polje 159 25 22
Average 156 27 24
Tyrone Rimski Šančevi 144 24 22
Zemun Polje 138 23 20
Average 141 24 21
LSD0.05 6 3 2
LSD0.01 8 4 3

Donegal had the highest average values of fresh yield (63.9 t ha−1 ) and forage dry
matter (15.1 t ha−1 ), while the cultivar had the lowest average forage yields, with
44.9 t ha−1 of green forage and 10.9 t ha−1 of forage dry matter (Table 48.3).
The highest green forage yields at individual locations and years were measured
in the cultivar Donegal at Rimski Šančevi in 2010 (82.4 t ha−1 ) and at Zemun Polje in
the same year (75.5 t ha−1 ), due to a particularly rainy and moderately warm growing
season at both locations. The same cultivar also had the highest forage dry matter
yields, also in 2010, with 18.4 t ha−1 at Rimski Šančevi and 17.9 t ha−1 Zemun Polje.
It should be noted that the cultivar Donegal had as excellent 2-year performance
in the conditions of Serbia as it had in the trial carried out at the Royal Agricultural
College at Cirencester, UK (Koivisto et al. 2003), where it produced an average yield
of 10.0 t ha−1 of forage dry matter.
If compared the traditional Balkan annual forage legumes, the four tested forage
soya bean cultivars proved at least as equally suitable for forage production as forage
pea or common vetch. On average, the autumn sown forage pea cultivars of Serbian
origin may produce nearly 48 t ha−1 of green forage and more than 10 t ha−1 of
forage dry matter (Mihailović et al. 2004), while the average yields of the Serbian
common vetch cultivars may surpass 39 t ha−1 of green forage and 10 t ha−1 of
forage dry matter (Mihailović et al. 2005).
The stands of all four cultivars, despite their great height, remained erect until
cutting and without a single plant lodged. Also, by some still non-clarified reason,
some cultivars such as Tara suffered from a reduced number of plants during their
growing period, resulting in smaller number of plants per plot and an excessive
growth of the above ground parts. Such cultivars, despite the fact that both forage
soya bean growing seasons were extreme in comparison to the average, may be
48 Performance of Forage Soya Bean (Glycine max) Cultivars in the Northern Balkans 357

Table 48.3 Forage yields of four soya bean cultivars at Rimski Šančevi and Zemun Polje in 2009
and 2010
Cultivar Location Year Green forage yield Forage dry matter yield
(t ha−1 ) (t ha−1 )
Derry Rimski Šančevi 2009 28.4 7.1
2010 73.4 17.6
Average 50.9 12.4
Zemun Polje 2009 31.0 7.4
2010 64.4 15.6
Average 47.7 11.5
Mean 49.3 11.9
Donegal Rimski Šančevi 2009 49.6 12.4
2010 82.4 18.4
Average 66.0 15.4
Zemun Polje 2009 48.2 11.6
2010 75.5 17.9
Average 61.9 14.8
Mean 63.9 15.1
Tara Rimski Šančevi 2009 39.1 9.8
2010 63.2 14.0
Average 51.2 11.9
Zemun Polje 2009 52.0 12.5
2010 60.2 13.7
Average 56.1 13.1
Mean 53.6 12.5
Tyrone Rimski Šančevi 2009 51.5 12.9
2010 40.2 9.5
Average 45.9 11.2
Zemun Polje 2009 44.9 10.8
2010 43.0 10.2
Average 44.0 10.5
Mean 44.9 10.9
LSD0.05 13.45 2.91
LSD0.01 18.96 4.03

considered not suitable for both cultivation in the conditions of Serbia and use in
developing novel genetic variability targeting similar regions.
There were significant differences at the level of 0.05 (*) between genotypes
(F = 20.16∗ ), environments (F = 51.84∗ ) and genotype × environments (F =
2.51∗ ). As could be seen from the difference in the ranking of the cultivars in two
individual locations, a significant genotype × environment interaction indicated the
linear function of the additive environment effects (Mather and Jinks 1982).

48.4 Conclusions

The preliminary performance of forage soya bean cultivars of US origin in the con-
ditions of Serbia opens the possibility of cultivating this crop as a valuable source
of forage during the summer. A possible ideotype of the forage soya bean cultivar
358 V. Mihailović et al.

for this region could be one with stable forage yields, instead of broad variation in
the overall performance between individual growing seasons. The research will be
continued with emphasis upon forage quality, especially the content of crude protein
and crude fibre fractions, as well as on the reliable seed production in order to secure
the potential commercialisation of the newly developed forage soya bean cultivars
and their introduction to the Serbian market and fields.

Acknowledgments The Ministry of Education and Science of the Republic of Serbia.

References

Acikgoz E, Sincik M, Oz M, Albayrak S, Wietgrefe G, Turan ZM, Goksoy AT, Bilgili U, Karasu
A, Tongel O, Canbolat O (2007) Forage soybean performance in Mediterranean environments.
Field Crop Res 103:239–247
Bilgili U, Sincik M, Goksoy AT, Turan ZM, Acikgoz E (2003) Forage and grain yield performances
of soybean lines. J Cent Eur Agric 6:397–402
Ćupina B, Erić P, Krstić Ð, Vučković S (2004) Forage catch crops in sustainable agriculture and
organing farming. Acta Agric Serb IX(17 special issue):451–459
Ćupina B, Mikić A, Stoddard FL, Krstić Ð, Justes E, Bedoussac L, Fustec J, Pejić B (2011) Mu-
tual legume intercropping for forage production in temperate regions. In: Lichtfouse E (ed)
Sustainable agriculture reviewes 7: Genetics, biofuels and local farming systems, Springer,
Dordrecht
Darmosarkoro W, Harbur MM, Buxton DR., Moore KJ, Devine TE, Anderson IC (2001) Growth,
development, and yield of soybean lines developed for forage. Agron J 93:1028–1034
Devine TE, Hatley EO (1998) Registration of ‘Donegal’ forage soybean. Crop Sci 38:1719–1720
Devine TE, Mcmurtrey JE (2004) Registration of ‘Tara’ soybean. Crop Sci 44:1019–1020
Devine TE, Hatley EO, Starner DE (1998a) Registration of ‘Derry’ forage soybean. Crop Sci
38:1719
Devine TE, Hatley EO, Starner DE (1998b) Registration of ‘Tyrone’ forage soybean. Crop Sci
38:1720
Devine T, Starner D, Darmosarkoro W, Moore K, Lucey R, Hatley E (2007) Breeding soybeans for
forage production. Ratar Povrt/ Field Veg Crop Res 44(II):49–54
Devine T, Mikić A, Ðordević V, Perić V, Mihailović V, Srebrić M, Ćupina B, Krstić Ð (2010) First
attempts of growing forage soybean in Serbia. Biotechnol Anim Husb 26(special issue 2):73–79
Koivisto JM, Devine TE, Lane GPF, Sawyer CA, Brown HJ (2003) Forage soybeans (Glycine max
(L.) Merr.) in the UK: test of new cultivars. Agron 23:287–291
Mather K, Jinks JL (1982) Biometrical genetics. Chapman Hall, London
Mihailović V, Erić P, Mikić A (2004) Growing peas and vetches for forage in Serbia and Montenegro.
Grassl Sci Eur 9:457–459
Mihailović V, Mikić A, Karagić Ð, Pataki I, Krstić Ð (2005) Genetic variability of yield and its
components in spring vetch cultivars. Grassl Sci Eur 10:303–306
Mihailović V, Mikić A, Ćupina B, Krstić Ð, Erić P, Hauptvogel P, Karagić Ð (2009) Forage yields
in urban populations of Hungarian vetch (Vicia pannonica Crantz) from Serbia. Grassl Sci Eur
14:417–420
Mikić A, Perić V, Ðordević V, Srebrić M, Mihailović V (2009) Anti-nutritional factors in some grain
legumes. Biotechnol Anim Husb 25(5–6): 1181–1188
Seiter S, Altemose CE, Davis MH (2004) Forage soybean yield and quality responses to plant
density and row distance. Agron J 96:966–970
Steel RGD, Torrie JS (1960) Principles and procedures of statistics. MacGraw Hill, New York
Chapter 49
Effects of Trinexapac-Ethyl (Moddus) on Seed
Yields and Its Quality of Eleven Temperate
Grass Species

R. Macháč

Abstract In small plot field trials conducted for 2 years the effects of plant growth
regulator (PGR) trinexapac-ethyl in selected grass species were studied. Eleven tem-
perate grass species were used in trials: Perennial ryegrass, annual ryegrass, meadow
fescue, red fescue, Kentucky bluegrass, timothy, cocksfoot, loloid and festucoid type
of festulolium, yellow oat grass and tall oat grass. Evaluations were made total seed
yield, thousand seed weight, germination, germination energy and number of seeds
per unit. Grasses treated by PGR achieved of higher seed yield especially in conse-
quence of higher number of seeds. The qualities of seeds (TSW, germination) were
comparable to untreated control.

Keywords Grasses · PGR · Trinexapac-ethyl · Seed yield

49.1 Introduction

The grass seed production has a long history in the Czech Republic, however, the
production and seed yields fall under average of EU. Main cause of low yields is non-
keeping of basic agronomical practices and low inputs. To achieve high seed yields
it is necessary to supply sufficient amounts of nitrogen which has significantly effect
on the photosynthesis and thereby total productivity of plant. However, increasing
of nitrogen supply increase also growth and prolongation of stems that are more
inclinable to lodging. Lodging has been identified as one of the most important
factors reducing grass seed yield. Losses due to lodging have been estimated to
be as great as 60 % (Rolston et al. 1997). Lodged stems are exposed to higher
competitive for light and nutrients. Developing seeds may abort or fall due to less
effective photosynthesis and decreasing of assimilate supply. Lodged stands are as
well more predisposed to diseases. Last but not least is complication of harvest and
increasing of seed losses due to no-cutting of lodged stems, increasing moisture of
thrashed material etc. In the Czech Republic only chlormequat-chloride (CCC) was

R. Macháč ()
OSEVA development and research Ltd., Grassland Research Station at Zubří,
Hamerska 698, 756 54 Zubří, Czech Republic
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 359

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_49, © Springer Science+Business Media Dordrecht 2013
360 R. Macháč

registered for stem length shortening and reduction of lodging in last years. CCC
works on the beginning of gibberellin biosynthesis by inhibiting of kaurene synthesis
(precursor of gibberellins). Plant growth regulator trinexapac-ethyl is widely used on
grass seed crops abroad (Chastain et al. 2003; Haldrup 2007; Rijckaert 2007, etc.).
Rademacher (2000) claims that trinexapac-ethyl is inhibiting the activity of enzyme
3-β hydroxylase that the transforms inactive gibberellins form GA20 on highly active
forms GA1 and GA4 . In consequence this inactivation the stem growth is reduced and
stalk wall is stronger. Finally, the plant height is lower and the crop is less susceptible
to lodging. Main topic of our research was to verify the influence of trinexapac-ethyl
on seed yield of eleven temperate grass species and its quality in Central Europe
conditions.

49.2 Materials and Methods

Field small-plot trials were conducted at Grassland Research Station at Zubri (North-
east Moravia, 360 m a.s.l., cambisol, average air temperature 7.5 ◦ C, precipitation
864 mm) for 2 years (2007–2008). The trials were conducted with eleven grass
species: perennial ryegrass (Lolium perenne L.) cv. Olaf, annual ryegrass (Lolium
multiflorum Lam. ssp. multiflorum) cv. Jivet, meadow fescue (Festuca praten-
sis Huds.) cv. Roznovska, red fescue (F. rubra L.) cv. Ta-gera, timothy (Phleum
pratense L.) cv. Sobol, loloid type of Festulolium cv. Lofa, festucoid type of Festu-
lolium cv. Hykor, cocksfoot (Dactylis glomerata L.) cv. Dana, Kentucky bluegrass
(Poa pratensis L.) cv. Slezanka, tall oat grass (Arrhenatherum elatius (L.) Beauv. ex
J. S. et K. B. Presl) cv. Roznovsky and yellow oat grass (Trisetum flavescens (L.)
P. Beauv.) cv. Roznovsky. The plot size was 10 m2 . Each trial was arranged in a
randomized complete block design (with another seven pesticide treatments) with
four replications. Two tested treatments with trinexapac-ethyl (TE) were applied:
single application of dose 0.2 l TE ha−1 at GS 31–32 and split application two times
0.1 l TE ha−1 , first application at GS 29 and second application at GS 32.
All treatments were performed with wheelbarrow sprayer driven by compressed
air (Lurmark 01F80 nozzles, a pressure 0.25 MPa, spraying volume 300 l ha−1 ). The
standard treatment (MCPA + clopyralid + fluroxypyr) was used for weed control.
Fertilizers application: autumn 45 kg N, 20 kg P2 O5 and 60 kg K2 O per ha, spring—
only nitrogen at dose depending on grass species 80–110 kg ha−1 . The trials plots
were combined directly with plot combine Wintersteiger Elite. Harvested seed was
dried and subsequently cleaned by laboratory cleaner Westrup-Kamas for seed yield
determination. Seed quality (TSW, germination) and number of seed per inflores-
cence were analyzed in the lab of GRS Zubri. The results were analyzed by ANOVA
and Tukey’s post hoc test on significance level 95 % (Statistica 8.0). Due to insertion
to the randomized blocks with pesticide treatments the number of degree of freedom
(27) was satisfactory for statistical analyses.
49 Effects of Trinexapac-Ethyl (Moddus) on Seed Yields and Its Quality of Eleven . . . 361

Table 49.1 The effect of trinexapac-ethyl on seed yield selected grass species
Species Treatment Seed yield
2007 2008 Average
Kilogram Tukey Kilogram Tukey Kilogram Relative
per hectare per hectare per hectare
Perennial ryegrass Untreated 934 d 555 e 744 100
TE 200 1230 ab 708 a 969 130
TE 2×100 1238 ab 696 abc 967 130
Annual ryegrass Untreated 1748 a 1655 c 1702 100
TE 200 1793 a 1843 ab 1818 107
TE 2×100 1860 a 1758 a 1809 106
Meadow fescue Untreated 719 de 356 ab 538 100
TE 200 765 cd 378 ab 572 106
TE 2×100 801 bcd 521 a 661 123
Red fescue Untreated 902 ab 1195 bc 1048 100
TE 200 995 a 1349 a 1172 112
TE 2×100 983 a 1227 ab 1105 105
Kentucky blue grass Untreated 346 bcd 517 b 432 100
TE 200 471 a 676 a 574 133
TE 2×100 414 ab 629 ab 521 121
Timothy Untreated 835 a 640 ab 737 100
TE 200 859 a 707 a 783 106
TE 2×100 903 a 729 a 816 111
Cocksfoot Untreated 838 abcd 626 ab 732 100
TE 200 800 abcd 763 a 781 107
TE 2×100 862 abc 6945 ab 778 106
Festulolium Lofa Untreated 894 bcdef 735 bc 814 100
TE 200 1055 abc 767 ab 911 111
TE 2×100 998 abcde 877 a 938 115
Festulolium Hykor Untreated 938 a 952 abc 945 100
TE 200 937 a 1099 a 1018 107
TE 2×100 951 a 1148 a 1049 111
Tall oat grass Untreated 517 bc 367 c 442 100
TE 200 672 a 506 ab 589 133
TE 2×100 549 b 514 a 532 120
Yellow oat grass Untreated 267 c 208 a 237 100
TE 200 285 a 231 a 258 109
TE 2×100 303 b 239 ab 271 114
TE trinexapac-ethyl

49.3 Results and Discussion

The positive effect of trinexapac-ethyl on seed yield was recorded in larger or smaller
rate on all of selected grass species. However, there was insignificant decreasing of
seed yield of some species in 2007. With regard to dry weather in growing seasons
(especially in 2007) the grasses only very few lodged. It is possible to suppose, that
if grasses would more lodged the difference in seed yield between untreated plots
and plots treated by TE were to be higher. The effect of TE application on seed yield
of selected grass species is shown in the Table 49.1. Thousand seed weight (TSW)
362 R. Macháč

80

60
seeds per inflorescence

40

20

0
untreated TE 200 TE untreated TE 200 TE untreated TE 200 TE untreated TE 200 TE
2x100 2x100 2x100 2x100
Perennial ryegrass Annual ryegrass Meadow fescue Red fescue
2007 2008

Fig. 49.1 Effect of trinexapac-ethyl on number of seed in inflorescence of perennial ryegrass,

annual ryegrass, meadow fescue and red fescue including confidence intervals (p = 0.05)

was at most grass species insignificant lower on treatments when the TE was applied
in comparison with untreated plots. Only in timothy in year 2007 and in yellow oat-
grass in both trial years were observed significantly higher TSW on plots treated with
TE (data not shown). Minimal and also insignificant differences between treatments
were recorded in germination and germination energy (data not shown).
Significant differences were recorded in the number of seed in inflorescence (see
Fig. 49.1). At the most grass species the number of the seed was higher at treated
grasses; however, at the some species then number of the seed was lower in conse-
quence with higher number of fertile stems. Positive influence of trinexapac-ethyl
application on increase of number of perennial ryegrass seed was observed by Sil-
berstein et al. (2002), who found out that TE increases number of created ripe seed,
while the potential number of seed was not affected. Application TE so has positively
effect on Floret site utilization (FSU). According to Young et al. (2007) the nitrogen
rate increasing has significant effect on increasing of both, actual and potential seed
number, but insignificant effect on FSU in perennial ryegrass. Reduction in plant
height and lodging was recorded on all the treated plots.
Application of trinexapac-ethyl on cool season grass seed crops can significant
increases seed yields mainly due to number of seed increasing. There is slight vari-
ation in the increasing of seed yield between grass species, but average increasing
of seed yield is about 12–15 %. Based on trials the PGR Moddus (trinexapac-ethyl
250 g l−1 ) in rate 0.8 l ha−1 preferably applied at growth stage GS 31–32 or 0.4 l ha−1
applied two times (GS 29 and GS 32) has been allowed for minor use in seed crops
of each grass species under study in the Czech Republic.
49 Effects of Trinexapac-Ethyl (Moddus) on Seed Yields and Its Quality of Eleven . . . 363

References

Chastain TG, Young WC III, Garbacik CJ, Silberstein TB (2003) Seed partitioning and yield re-
sponses to trinexapac ethyl in perennial ryegrass. In: Proceedings of the 5th international herbage
seed conference, Gatton, Australia, pp 104–108
Haldrup C (2007) Growth regulation, fungicides and nitrogen interaction in seed crop production.
In: Aamlid TS, Havstad LT, Boelt B (eds) Seed production in the northern light. Proceedings of
the 6th international herbage seed conference, Gjennestad, Norway, pp 211–213
Rademacher W (2000) Growth retardants: effects on gibberellin biosynthesis and other metabolic
pathways. Ann Rev Plant Physiol Plant Mol Biol 51:501–531
Rijckaert G (2007) Effects of trinexapac-ethyl (Moddus) in seed crops of Italian ryegrass and
timothy. In: Aamlid TS, Havstad LT, Boelt B (eds) Seed production in the northern light.
Proceedings of the 6th international herbage seed conference, Gjennestad, Norway, pp 231–235
Rolston MP, Rowarth JS,Young WC III, Mueller-Warrant GW (1997) Grass seed crop management.
In: Fairey DT, Hampton JG (eds) Forage seed production, vol 1. Temperate species. CAB
International, Wallingford, UK
Silberstein TB, Young WC III, Chastain TG, Garbacik CJ (2001) Response of cool season grasses
to foliar applications of Palisade (trinexapac-ethyl) plant growth regulator. In: Young WC III
(ed) Seed production research at Oregon State University USDA-ARS cooperating. Department
of Crop and Soil Science Ext/Crs 121, 4/02
Silberstein TB, Young WC III, Chastain TG, Garbacik CJ (2002) Response of perennial ryegrass
to spring nitrogen fertility and plant growth regulator applications. In: Young WC III (ed.) Seed
production research at Oregon State University USDA-ARS cooperating. Department of Crop
and Soil Science Ext/Crs 122 5/03 pp 15–18
Young WC III, Silberstein TB, Chastain TG, Garbacik CJ (2007) Response of creeping fescue
(Festuca rubra L.) and perennial ryegrass (Lolium perenne L.) to spring nitrogen fertility and
plant growth regulator applications in Oregon. In: Aamlid TS, Havstad LT, Boelt B (eds) Seed
production in the northern light. Proceedings of the 6th international herbage seed conference,
Gjennestad, Norway, pp 201–205
Chapter 50
The Chemical Composition of a Range of Forage
Grasses Grown Under Two Nitrogen Fertiliser
Inputs and Harvested at Different Stages
of Maturity

C. King, J. McEniry, M. Richardson and P. O’Kiely

Abstract Grass species, rate of N fertiliser application and plant maturity at harvest
represent three of the most important grassland management factors affecting plant
chemical composition. This study investigates the effects of two N fertiliser inputs
and five harvesting dates in the primary growth on the yield and chemical compo-
sition of five common grass species. Perennial ryegrass, Italian ryegrass, cocksfoot,
timothy and tall fescue, were grown in triplicate field plots under two inorganic ni-
trogen fertiliser inputs (low = 0 kg N ha−1 , high = 125 kg N ha−1 ) and harvested at
five sequential dates (fortnightly from 12 May–7 July; Harvests 1–5) in the primary
growth in both 2009 and 2010. At each harvest date, herbage was weighed to estimate
DM yield and representative samples were used to determine herbage chemical com-
position. In general for the five grass species investigated, herbage dry matter (DM),
neutral detergent fibre (NDF), acid detergent fibre (ADF) and acid detergent lignin
(ADL) concentrations increased (P < 0.001), while dry matter digestibility (DMD),
buffering capacity (BC) and water soluble carbohydrate (WSC) concentration de-
creased (P < 0.001) with advancing harvest date. An exception to this trend was for
the cocksfoot where a decrease (P < 0.001) in DM concentration was observed from
Harvest 4–5. Cocksfoot had the lowest DM yield while no difference (P > 0.05) was
observed between the other grass species. Although timothy and PRG had a similarly
high DMD, of the five grass species timothy had the highest (P < 0.001) NDF and
ADF concentration. The IRG had the highest (P < 0.001) WSC concentration and
lowest (P < 0.001) BC making it the most suitable species for ensiling.

C. King () · J. McEniry · P. O’Kiely

Animal & Grassland Research and Innovation Centre,
Teagasc, Grange, Dunsany, Co. Meath, Ireland
e-mail: [email protected]
C. King · M. Richardson
School of Civil, Structural and Environmental Engineering, University College Dublin,
Dublin 4, Belfield, Ireland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 365

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_50, © Springer Science+Business Media Dordrecht 2013
366 C. King et al.

50.1 Introduction

Approximately 35 % of the utilised agricultural area in Europe is under permanent

grassland. These grasslands contribute substantially to agricultural production sys-
tems, while also providing an important resource in terms of biodiversity, carbon
sequestration in soils and, in recent years, bioenergy and biorefining (Hopkins and
Wilkins 2006). These grasslands are usually managed to enhance the production of
herbage and its quality. Three of the main management factors affecting herbage
yield and chemical composition are grass species, rate of nitrogen fertiliser applica-
tion and stage of maturity at harvest (Buxton 1996). For example, grass harvested
at an early vegetative growth stage will generally have a higher digestibility, crude
protein (CP) and ash content than herbage harvested at a later stage of growth (Bux-
ton and O’Kiely 2003), with the optimal time for harvesting being determined by
the particular grassland use. Inorganic nitrogen fertiliser is widely used to improve
herbage yields, but can also effect the chemical composition of the herbage (e.g.
increased CP with increasing N fertiliser application rate). Although the majority
of reseeded temperate grassland in Europe is dominated by perennial ryegrass, dif-
ferent grass species may offer a potential alternative for non-agricultural uses. The
objective of this study was to investigate the effects of two N fertiliser inputs and
five harvesting dates in the primary growth on the yield and chemical composition
of five common grass species

50.2 Materials and Methods

Five common grass species, perennial ryegrass (PRG; Lolium perenne L. var. Gan-
dalf), Italian ryegrass (IRG; Lolium multiflorum Lam. var. Prospect), cocksfoot
(Dactylis glomerata L. var. Pizza), timothy (Phleum pratense L. var. Erecta) and tall
fescue (Festuca arundinacea Schreb. var. Fuego) were grown in field plots (20 m2 )at
Teagasc, Grange, under two inorganic fertiliser nitrogen inputs (low = 0 kg ha−1 ,
high = 125 kg ha−1 ; applied as urea (460 g N kg−1 ) in mid-March when soil tem-
perature > 6 ◦ C) and harvested at five sequential dates (fortnightly from 12 May–7
July; Harvests 1–5; n = 150 plots) in the primary growth during both 2009 and 2010
(n = 300 samples). At each harvest date, herbage was harvested and weighed using
a Haldrup forage plot harvester (J. Haldrup, Løgstor, Denmark) cutting to a 6 cm
stubble height. A representative 2 kg sample of each herbage was taken and stored at
−18 ◦ C prior to chemical analyses. After harvesting in 2009, the grass plots received
250 kg ha−1 of a compound fertiliser (240 g N, 25 g P and 100 g K kg−1 ). Dry matter
(DM) concentration was estimated following drying in a ventilated oven with forced
air circulation at 98 ◦ C for 16 h. Replicate samples (200 g) were also dried at 40 ◦ C
for 48 h before being milled (Wiley mill; 1 mm screen). Dried, milled samples were
used for the determination of DMD, NDF, ADF, ADL, ash, BC, CP and WSC as
previously described by Purcell et al. (2011). Data were analysed as a split-split plot
design using the Proc MIXED procedure of SAS, Version 9.1.2 (SAS 2004) with
50 The Chemical Composition of a Range of Forage Grasses Grown . . . 367

harvest date as the main-plot, nitrogen fertiliser as the sub-plot and grass species as
the sub-sub plot, and with year and replicate blocking being accounted for.

50.3 Results and Discussion

On average, with advancing harvest date there was an increase (P < 0.001) in DM
yield (except from Harvest 4 to Harvest 5 where values did not differ (P > 0.05)),
together with an increase (P < 0.001) in herbage DM concentration and an increase
(P < 0.001) in the concentrations of NDF, ADF and ADL (Table 50.1. An exception
to this trend was for the cocksfoot where a decrease (P < 0.001) in DM was observed
from Harvest 4–5. As a plant matures the proportion of cell wall components (e.g.
cellulose, hemicellulose and lignin) increases, reflecting the general decrease in
plant leaf to stem ratio and the increasing cell wall content of the stems in particular
(Hatfield 1993). This was accompanied by increased lignification within the cell wall
fraction, which resulted in a decrease (P < 0.001) in herbage DMD. Furthermore,
herbage CP concentration is higher in grass leaves than in stems and decreased
(P < 0.01) with advancing harvest date in accord with Ballard et al. (1990). The
decrease (P < 0.001) in BC and WSC concentration with advancing plant maturity
can also be attributed to the decrease in the plant leaf to stem ratio and is in Sagreement
with Keating and O’Kiely (2000).
On average, of the two nitrogen fertiliser treatments employed in this study, the
high N treatment gave a higher (P < 0.05) DM yield (Table 50.2). However, of the
five grass species only tall fescue had a significantly higher (P < 0.05) DM yield
with the high N fertiliser treatment. This is in contrast to Reid (1985) who reported
that ryegrass and cocksfoot produced the best responses at high rates of N fertiliser
application. The increase (P < 0.001) in herbage BC and CP concentration following
addition of N fertiliser is in agreement with Keady and O’Kiely (1996) who reported
that herbage BC is positively correlated to CP concentration. In contrast, herbage
DM and WSC concentrations were higher (P < 0.001) for the low N treatment. An
exception to this trend was for Harvest 4 and 5 where herbage DM concentration
did not differ (P > 0.05) between the two N fertiliser treatments. Nowakowski (1962)
reported that WSC concentration decreased in grasses with increasing levels of N
fertilisation due to an accelerated growth rate following application. The increase in
BC and decrease in WSC concentration with N fertiliser addition may have negative
implications for grass ensilability.
Of the five grass species investigated, on average cocksfoot had the lowest
(P < 0.001) DM yield (with no difference (P > 0.05) observed between the other grass
species) and DMD (along with IRG and tall fescue), and the highest BC (P < 0.001),
ash (P < 0.001) and CP concentrations (P < 0.001; along with timothy; Table 50.1).
Green et al. (1971) reported that all varieties of ryegrass out-yielded other common
grass species, which is in contrast to this study where, with the exception of cocksfoot
which had the lowest DM yield, no difference (P > 0.05) was observed between the
grass species. The PRG and timothy grasses had similarly higher (P < 0.001) DMD
368 C. King et al.

Table 50.1 Yield (t DM ha−1 ) and chemical composition (g kg−1 DM, unless indicated otherwise
in footnotes) of five grass species harvested at five sequential dates in the primary growth (averaged
across year and nitrogen fertiliser treatment)
S H Variables
Yield DM DMD NDF ADF ADL Ash CP WSC BC
PRG 1 5.04 195 827 454 239 7.8 83.9 141 245 522
PRG 2 6.95 189 776 518 287 11.3 81.8 122 196 464
PRG 3 8.20 207 681 588 344 23.7 72.2 93 159 391
PRG 4 12.01 286 666 606 350 25.1 68.1 80 160 305
PRG 5 9.81 296 613 616 372 34.5 67.9 72 148 272
IRG 1 5.46 211 780 453 242 7.3 81.6 126 262 426
IRG 2 7.01 220 746 470 261 11.1 76.2 117 268 371
IRG 3 9.39 252 698 526 313 23.6 67.4 89 224 331
IRG 4 9.79 291 644 556 339 27.2 74.9 86 185 282
IRG 5 8.72 295 623 583 361 32.3 79.2 81 147 277
TF 1 4.93 198 770 496 261 14.4 79.8 146 196 485
TF 2 6.33 190 726 541 301 15.5 82.7 126 157 443
TF 3 9.25 223 681 602 361 25.5 78.7 103 126 407
TF 4 9.69 280 657 617 365 30.3 75.3 89 129 340
TF 5 9.42 280 608 613 366 32.6 79.9 80 127 336
CF 1 4.42 184 771 493 254 14.2 87.4 154 178 511
CF 2 5.50 182 720 558 309 24.0 86.3 136 146 477
CF 3 6.77 224 664 599 347 26.4 84.9 108 113 425
CF 4 9.04 309 579 650 376 33.7 77.4 100 87 340
CF 5 6.95 252 596 635 366 37.1 94.8 95 74 363
TIM 1 4.47 191 821 518 261 11.2 79.8 153 163 493
TIM 2 6.23 178 758 571 321 22.3 82.6 138 101 469
TIM 3 9.37 202 710 651 382 28.4 74.1 107 80 396
TIM 4 10.84 270 662 656 380 36.6 71.3 107 81 329
TIM 5 9.75 309 616 643 389 44.5 65.6 87 101 304
SEM
Species (S) 0.289 2.6 6.2 4.4 2.0 0.77 0.91 1.5 3.4 3.8
Harvest (H) 0.335 5.6 9.3 4.4 3.6 1.27 0.89 2.1 5.2 4.9
S×H 0.558 8.0 13.8 7.5 5.9 2.11 2.06 3.7 8.0 9.0
Levels of significance
Species *** *** *** *** *** *** *** *** *** ***
Harvest *** *** *** *** *** *** ** *** *** ***
S×H * *** * *** *** * *** NS *** ***
PRG perennial ryegrass, IRG Italian ryegrass, TF Tall fescue, CF Cocksfoot, TIM timothy,
Harvest1 12 May, Harvest2 26 May, Harvest3 9 June, Harvest4 23 June, Harvest5 7 July, DM
dry matter (g kg−1 ), DMD dry matter digestibility (g kg−1 ), NDF neutral detergent fibre, ADF acid
detergent fibre, ADL acid detergent lignin, CP crude protein, WSC water soluble carbohydrates,
BC buffering capacity (mEq kg−1 DM), SEM Standard error of the mean, NS not significant
*P < 0.05; **P < 0.01; ***P < 0.001

values compared to the other grasses, despite timothy having the highest (P < 0.001)
NDF and ADF concentrations and a similarly high (P > 0.05) ADL concentration to
cocksfoot The IRG had the highest DM (P < 0.001) and WSC (P < 0.001) concentra-
tions, and the lowest (P < 0.001) NDF and ADF concentrations and BC. In general,
50 The Chemical Composition of a Range of Forage Grasses Grown . . . 369

Table 50.2 Yield (t DM ha−1 ) and chemical composition (g kg−1 DM, unless indicated otherwise
in footnotes) of five grass species grown under two nitrogen fertiliser (N) inputs (averaged across
year and harvest date)
S N Variables
Yield DM DMD NDF ADF ADL Ash CP WSC BC
PRG Low 7.89 242 723 554 311 18.1 72.1 84 207 366
IRG Low 8.32 264 698 516 303 18.6 73.9 83 239 313
TF Low 7.14 242 703 568 327 23.1 77.7 97 164 379
CF Low 5.95 231 676 579 328 25.6 85.9 104 127 420
TIM Low 7.60 243 719 612 349 27.2 71.1 102 122 358
PRG High 8.91 227 703 559 325 22.9 77.4 119 156 422
IRG High 7.83 243 698 519 304 22.1 77.9 117 195 362
TF High 8.71 227 674 580 334 24.2 80.9 121 130 423
CF High 7.12 230 655 595 333 28.6 86.4 133 112 444
TIM High 8.66 217 709 604 344 30.0 78.2 135 89 418
SEM
Species (S) 0.289 2.6 6.2 4.4 2.0 0.77 0.91 1.5 3.4 3.8
Nitrogen (N) 0.235 1.2 5.1 3.8 1.1 0.37 0.54 1.0 2.7 2.7
S×N 0.368 3.7 8.2 5.4 2.9 1.13 1.30 2.2 4.6 5.5
Levels of significance
Species *** *** *** *** *** *** *** *** *** ***
Nitrogen * *** NS NS * ** *** *** *** ***
S×N * * NS * * NS NS NS *** *
PRG perennial ryegrass, IRG Italian ryegrass, TF Tall fescue, CF Cocksfoot, TIM timothy, DM
dry matter (g kg−1 ), DMD dry matter digestibility (g kg−1 ), NDF neutral detergent fibre, ADF acid
detergent fibre, ADL acid detergent lignin, CP crude protein, WSC water soluble carbohydrates,
BC buffering capacity (mEq kg−1 DM), Low N 0 kg N ha−1 , high N 125 kg N ha−1 , SEM Standard
error of the mean, NS not significant
*P < 0.05; **P < 0.01; ***P < 0.001

the higher WSC concentration and the lower BC observed for the two ryegrasses
may make these grasses more suitable for silage production (Buxton and O’Kiely
2003).
There were no significant interactions (P > 0.05) between harvest date, nitrogen
fertiliser and grass species for any of the variables measured.

50.4 Conclusions

On average, herbage grown under the high N treatment and harvested at a later date
had a higher DM yield. With the exception of cocksfoot which had the lowest DM
yield, there was no difference between the other grass species. The later harvested
herbage had a higher concentration of NDF and had a lower digestibility. Although
timothy and PRG had a similarly high DMD, on average timothy had the highest
NDF and ADF concentrations. The IRG had the highest WSC concentration and
lowest BC making it the most suitable species for ensiling.
370 C. King et al.

Acknowledgments Funding for this research was provided under the National Development Plan,
through the Research Stimulus Fund (#RSF 07 557), administered by the Department of Agriculture,
Food & the Marine, Ireland.

References

Ballard RA, Simpson RJ, Pearce GR (1990) Losses of the digestible components of annual ryegrass
(lolium rigidum gaudin) during senescence. Aust J Agr Res 41:719–731
Buxton DR (1996) Quality-related characteristics of forages as influenced by plant environment
and agronomic factors. Anim Feed Sci Tech 59:37–49
Buxton DR, O’Kiely P (2003) Preharvest plant factors affecting ensiling. In: Buxton DR, Muck RE,
Harrison JH (ed) Silage science and technology. American society of agronomy, crop science
society of America, soil science society of Americ. USA
Green JO, Corrall AJ, Terry RA (1971) Relationships between stage of growth, yield and forage.
Grass species and varieties: Grassland Research Institute
Hatfield RD (1993) Cell wall polysaccharide interactions and degradability. In: Jung HG, Buxton
DR, Hatfield RD and Ralph J (ed) Forage cell wall and digestibility. Society of Agronomy,
Madison, pp 685–714
Hopkins A, Wilkins RJ (2006) Temperate grassland: key developments in the last century and future
perspectives. J Agr Sci 144:503–523
Keady TWJ, O’Kiely P (1996) An evaluation of the effects of rate of nitrogen fertilization of
grassland on silage fermentation, in-silo losses, effluent production and aerobic stability. Grass
Forage Sci 51:350–362
Keating T, O’Kiely P (2000) Comparison of old permanent grassland, Lolium perenne and Lolium
multiflorum swards grown for silage: 4. Effects of varying harvesting date. Irish J Agr Food Res
39:55–71
Nowakowski TZ (1962) Effects of nitrogen fertilizers on total nitrogen, soluble nitrogen and soluble
carbohydrate contents of grass. J Agr Sci 59:387
Purcell PJ, O’Brien M, Boland TM, O’Kiely P (2011) In vitro rumen methane output of perennial
ryegrass samples prepared by freeze drying or thermal drying (40◦ C). Anim Feed Sci Tech
166:175–182
Reid D (1985) A comparison of the yield responses of four grasses to a wide range of nitrogen
application rates. J Agr Sci 105:381–387
SAS (2004) SAS for Windows (v.9.1.2). Cary, NC: Statistical Analysis System Institute
Chapter 51
NIRS Calibration Strategies for the Botanical
Composition of Grass-Clover Mixtures

M. Cougnon, C. Van Waes, J. Baert and D. Reheul

Abstract In literature, different calibrations to predict the species composition of

grass legumes mixtures or mixtures of different grass species are described. Mostly,
these calibrations were developed using so called “artificial samples”. These artificial
samples are obtained by mixing pure (ground) material of the species for which the
calibration is developed in known proportions. The plant material used for these arti-
ficial samples may have been grown in mixtures or in pure stands. Calibrations based
on artificial samples mostly have very good calibration statistics but fail to predict
real validation samples. “Real samples” are obtained by hand separation of species
mixtures into the different species followed by recomposition. The advantage of the
use of artificial samples relative to real samples is that a lot of calibration samples
with a different composition can be obtained with a relative small labour input. We
built calibrations to predict the white clover content in grass clover mixtures, based
on real and artificial samples with the same composition, and validated them with
the same independent samples. Calibrations based on real samples performed far
better than calibrations based on artificial samples. The failure of the latter can be
explained by the lack of environmental variation in their spectra. We recommend a
calibration strategy based on fewer but more diverse hand sorted samples, rather than
making a lot of artificial samples that contain relatively little spectral information.

51.1 Introduction

A visual estimation of the composition of multi-species grass swards is a fast but sub-
jective method. Hand sorting of a sample is very precise but time consuming as one
needs samples of a representative magnitude. Hand sorting is too much labour de-
manding in large field experiments. NIRS (Near Infrared Reflectance Spectroscopy)
can work both precisely and quickly, but a calibration database is needed.

M. Cougnon () · D. Reheul

Department of Plant Production, Faculty of Bioscience Engineering, Ghent University,
Coupure Links 653, 9000 Ghent, Belgium
e-mail: [email protected]
C. Van Waes · J. Baert
ILVO Plant Sciences, Burg. Van Gansberghelaan 109, 9820 Merelbeke, Belgium

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 371

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_51, © Springer Science+Business Media Dordrecht 2013
372 M. Cougnon et al.

Several calibrations to predict the botanical composition of multi-species swards

were developed in the past. They differ according to the species involved and by the
strategy used to build the calibration.
Coleman et al. (1985) and Petersen et al. (1987) built calibrations for the botanical
composition based on so called artificial samples: species were grown in single
species plots, harvested material was mixed in known proportions, varying between
0–100 % with 5 % increments. These calibrations had good calibration statistics,
and were validated, to a certain extent, with independent samples.
Another way to build calibration samples is by determining the botanical compo-
sition of samples harvested in multi-species pastures/trials by separating them and
recomposing them afterwards (Shaffer et al. 1990; Wachendorf et al. 1999). Sam-
ples obtained in this way are called real samples. The advantage of artificial samples
compared to real samples is that more samples can be made with less labour.
Pitmann et al. (1991) compared NIRS calibrations based on these two different
strategies and concluded that despite the excellent statistics of calibrations based on
artificial samples, they were not acceptable for estimating the composition of pasture
samples. Our own preliminary research confirmed this finding. In order to understand
the failure of calibrations based on artificial samples, we compared similar sets of
artificial samples and real samples.

51.2 Material and Methods

Calibrations were built to predict the botanical composition of swards in a trial

comparing the performance of white clover (TR) with either perennial ryegrass (LP)
or tall fescue (FA) or a mixture of both grass species with white clover. The mixtures
differ in the initial proportion of the ryegrass component in the seed mixture (0 %LP,
12.5 %LP, 25 %LP or 100 %LP on a seed number base) and the ploidy level of the
ryegrass (diploid or tetraploid). The trial was established in 2009, and the yield and
the botanical composition were recorded since 2010. At each harvest, a representative
sample of the harvested material was sorted into the different species.
The sorted plant material of the first cut in 2011 was used to build calibrations in
two different ways (Fig. 51.1). “Real samples” samples were obtained by sorting a
sample of the harvested material of each plot. After weighing the different species,
they were put together again, dried (16 h at 75 ◦ C) and finally ground (Brabender
shear mill, 1 mm sieve). We obtained 25 calibration samples in this way. “Artificial
samples” were based on plant material obtained from plots that contained a single
grass species (FA or LP) and white clover. Pure FA, LP and TR were obtained
by hand sorting grass and clover. This plant material was dried and ground. The
powder was mixed in proportions that matched the composition that was found in
the real samples. Hence, 2 sets of 25 calibrations samples with the same botanical
composition were created.
NIRS spectra were collected with a Foss NIRSystems 5000 and WinISI II 1.50
software. The inverse reflectance (log(1/R)) was measured from 1100–2500 nm in
51 NIRS Calibration Strategies for the Botanical Composition of Grass-Clover Mixtures 373

Mixture 1 H, S Proportion A = x1 R, D, G Cal Sample 1

Proportion B = y1 Composition known

Species A+B
Real samples

Mixture 2 Proportion A = x2 Cal Sample 2

H, S R, D, G
Proportion B = y2 Composition known
Species A+B

Mixture 3 H, S Proportion A = x3 R, D, G Cal Sample 3

Species A+B Proportion B = y3 Composition known

Mixing in known
Species A H, D, G 100 % A Cal Sample 1’
Artificial samples

x,y proportions

Pure stand Powder Cal Sample 2’

Cal Sample 3’
Species B H, D, G 100 % B
…
Pure stand Powder

Fig. 51.1 Real calibration samples versus artificial calibration samples. H harvesting, S sorting,
R recomposing, D drying, G grinding

Table 51.1 Species Species Mean (%) Range (%) Standard deviation
proportion in the calibration
samples FA 62.4 28.1–93.9 20.1
LP 33.4 5.7–66.9 17.7
TR 5.2 0–18 5.1

steps of 2 nm. The calibrations were derived using modified partial least squares
regression (Shenk and Westerhous 1991). Scatter effects were reduced using first
derivate of the spectra and multiplicative scatter correction. The standard errors of
calibration and cross validation were calculated.

51.3 Results and Discussion

FA was the main component in the calibration samples. The standard deviations for
FA and LP were comparable (Table 51.1). Variation for TR was smaller.
From a statistical point of view, the calibration based on artificial samples was far
better than the calibration based on real samples: for all three species, Standard error
of calibration (SEC) and cross validation (SECV) were at least three times higher
in the real samples (Table 51.2). Despite the excellent calibration statistics, the cali-
bration based on artificial samples failed to predict the composition of real samples.
Standard errors of prediction (SEP) were 57.3, 37.8 and 20.1 for the proportion of
FA, LP and TR respectively. Inversely, when the calibration based on real samples
was used for prediction of the artificial samples, SEP were 13.6, 11.1 and 4.2 for
the proportion of FA, LP and TR respectively. Especially for TR, these values are
only slightly higher than the SECV found for the calibration, which indicates that
374 M. Cougnon et al.

Table 51.2 Calibration statistics of calibrations based on artificial and real samples
Species Real samples Artificial samples
SEC SECV RSQ SEC SECV RSQ
FA 4.7 7.3 0.95 1.46 1.92 0.99
LP 5.0 7.5 0.92 1.55 1.79 0.99
TR 3.3 4.1 0.57 0.62 0.84 0.99
SEC standard error of calibration, SECV standard error of cross validation, RSQ R squared

the calibration based on real samples works for the prediction of the clover content
of artificial samples.
Principal component analysis indicates that although the botanical composition
of both types of calibration samples was exactly the same, the spectral information
included in the set of artificial samples was smaller than that of the set of real
samples. The artificial samples occur as a concentrated cloud, when plotted in the
space formed by the first three principal components, whereas the real samples occur
as a large cloud around the artificial samples. Surault et al. (2006) observed the same
phenomena and suggested that the spectral signature of a grass species may vary
when grown in pure stands or in mixtures with other species.
The reason for the difference between the two types of calibration samples prob-
ably lies in the lack of environmental variation in the plant material used to build
the artificial samples. The material grown in pure stands was harvested in different
plots, pooled together, and handled as a single sample when dried, ground and stored
before mixing, whereas the material for the real samples was harvested in different
plots and samples remained separated per plot from harvest until the spectra were
collected. The method of grinding, processing and storing the samples are affect-
ing the NIRS spectra (Shaffer et al. 1990). The material used to build the artificial
calibrations is homogenised and variables that affect the NIR spectra are partially
lost.
Although good calibrations for determination of white clover content based on
artificial samples were described (Chataigner et al. 2010; Locher et al. 2005), our
findings suggests there is more advantage in taking the time and the effort to work
with real samples, that represent all the variables which would affect the NIR spectra,
rather than creating a lot of artificial samples. The results shown here are preliminary,
further research is needed to understand exactly why the calibration based on artificial
samples fails to predict the botanical composition of real samples.

51.4 Conclusion

In our experiments, calibrations based on artificial samples failed to predict the

botanical composition of real samples, which is the aim of building a calibration.
Based on our preliminary work, we recommend a calibration strategy based on fewer
but more diverse hand sorted samples, rather than making a lot of artificial samples
that contain relatively few spectral information.
51 NIRS Calibration Strategies for the Botanical Composition of Grass-Clover Mixtures 375

References

Chataigner F, Surault F, Huyghe C, Jullier B (2010) Determination of botanical composition in

multispecies forage mixtures by near infrared reflectance spectroscopy. In: Huyghe C (ed) Sus-
tainable use of genetic diversity in forage and turf breeding, Springer, Heidelberg, pp 199–203
Coleman S, Barton F, Meyer R (1985) The use of near-infrared reflectance spectroscopy to predict
species composition of forage mixtures. Crop Sci 25:834–837
Locher F, Heuwinkel H, Gutser R, Schmidhalter U (2005) Development of near infrared reflectance
spectroscopy calibrations to estimate legume content of multispecies legume-grass mixtures.
Agron J 97:11–17
Petersen J, Barton F, Windham W, Hoveland C (1987) Botanical composition definition of tall
fescue-white clover mixtures by near infrared reflectance spectroscopy. Crop Sci 27:1077–1080
Pittman W, Piacitelli C, Aiken G, Barton F (1991) Botanical composition of tropical grass-legume
pastures estimated with near infrared reflectance spectroscopy. Agron J 83:103–107
Shaffer J, Jung G, Shenk J, Abrams S (1990) Estimation of the botanical composition in
alfalfa/ryegrass mixtures by near infrared spectroscopy. Agron J 82:669–673
Shenk J, Westerhaus M (1991) Population structuring of near infrared spectra and modified partial
least square regression. Crop Sci 31:1548–1555
Surault F, Briand M,Veron R. Huyghe C (2006) NIRS calibration equations to determine species con-
tribution in grassland swards. In: Lloveras J (ed) Grassland science in Europe Vol 11, 598–600
Wachendorf M, Ingwersen B, Taube F (1999) Prediction of the clover content of red clover and white
clover-grass mixtures by near-infrared reflectance spectroscopy. Grass Forage Sci 54:87–90
Chapter 52
Comparison of LOCAL and GLOBAL
Calibration Models to Predict Ryegrass Quality
Using Near Infrared Reflectance Spectroscopy

G. A. Burns, T. J. Gilliland, D. Grogan and P. O’Kiely

Abstract GLOBAL calibration models using near infrared reflectance spectroscopy

have been successfully developed for determining the quality of forages. The use of
LOCAL calibration models has been shown to improve the accuracy of large calibra-
tion sets by selecting a representative sub-section from within the complete dataset to
form a unique calibration for each individual sample. GLOBAL and LOCAL calibra-
tion models (n = 2,076) were developed to predict four quality attributes (buffering
capacity, crude protein, in vitro dry matter digestibility and water soluble carbo-
hydrates) using three ryegrass species (perennial, Italian and hybrid ryegrass). The
GLOBAL approach produced accurate calibration models with R2 > 0.86. The max-
imum number of samples was altered in the LOCAL approach (10–250) to assess
the optimal number of ‘local’ samples required. GLOBAL calibrations were more
accurate than LOCAL calibrations for all quality traits except buffering capacity
which was of similar accuracy. This research recommends the use of GLOBAL cali-
bration models for the assessment of the quality of ryegrasses in the national variety
evaluation scheme in Ireland.

52.1 Introduction

Near infrared reflectance spectroscopy (NIRS) is commonly used to assess the quality
of forages in breeding programs and national variety evaluation schemes (e.g. Burns
et al. 2011). NIRS is a secondary technique that requires calibration models to form
predictions of quality traits based on their near infrared spectra. A GLOBAL approach

P. O’Kiely ()
AGRIC, Teagasc, Grange, Dunsany, Co. Meath, Ireland
e-mail: [email protected]
G. A. Burns
AGRIC, Teagasc, Grange, Dunsany, Co. Meath, Ireland
School of Biological Sciences, Queen’s University Belfast, Belfast, Northern Ireland
e-mail: [email protected]
T. J. Gilliland
AFBI, Crossnacreevy, Co. Down, Northern Ireland
D. Grogan
DAFM, Backweston, Co. Kildare, Ireland

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 377

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_52, © Springer Science+Business Media Dordrecht 2013
378 G. A. Burns et al.

Table 52.1 Summary of the reference values of four quality traits for three ryegrass species
n Mean Minimum Maximum SD
−1
In vitro dry matter digestibility (g kg )
Perennial ryegrass 1836 801 629 890 36.9
Italian ryegrass 137 748 576 870 73.7
Hybrid ryegrass 103 773 587 886 70.7
Water-soluble carbohydrate (g kg−1 DM)
Perennial ryegrass 1836 180 50.4 376 51.9
Italian ryegrass 137 207 92.9 430 74.4
Hybrid ryegrass 103 212 71.6 413 70.4
Crude protein (g kg−1 DM)
Perennial ryegrass 1836 150 57.4 266 36.8
Italian ryegrass 137 128 73.7 267 40.5
Hybrid ryegrass 103 133 71.4 241 43.4
Buffering capacity (mEq kg−1 DM)
Perennial ryegrass 1836 437 215 736 90.8
Italian ryegrass 137 360 188 559 87.5
Hybrid ryegrass 103 368 207 616 83.6
SD standard deviation

is usually employed in which the full calibration set is used to form these calibration
models. The calibration set in GLOBAL models should ideally encompass as much
variation as the model will face in practice (Berzaghi et al. 2000). However, this can
potentially lead to a trade-off between the accuracy and robustness of calibration
models (Shenk et al. 1997). Shenk et al. (1997) developed the LOCAL regression
algorithm that selects a calibration set of spectrally similar samples from within the
full calibration set, to the sample of interest. Berzaghi et al. (2000) showed LOCAL
calibration models produced lower standard errors of predictions than a GLOBAL
approach that encompassed several for ages. This was attributed to the non-linear
relationship between the spectra and quality trait values which can exist in large
databases, reducing the accuracy of GLOBAL models. The LOCAL algorithm selects
representative sub-sets from within the full database that allow for modeling of non-
linear relationships while maintaining heterogeneity in the calibration set (Shenk et al.
1997). The aim of this research was to compare GLOBAL calibration models (Burns
et al. 2011) that encompassed three ryegrass species, perennial (Lolium perenne L.),
Italian (Lolium multiflorum Lam.) and hybrid ryegrass (Lolium boucheanum Kunth)
with LOCAL calibration models for the prediction of four quality traits of ryegrasses
in a national variety evaluation scheme.

52.2 Materials and Methods

52.2.1 Field Trials

Field trials were carried out at the Grass and Clover Variety Evaluation Unit at
Backweston, Co. Kildare, Ireland (53◦ 26 N, 06◦ 30 W). Trial management
and protocol are described in detail by Grogan and Gilliland (2010). A c. 300 g
52 Comparison of LOCAL and GLOBAL Calibration Codels to Predict Ryegrass . . . 379

Table 52.2 Summary of optimal modified partial least squares regression calibration models for
predicting four ryegrass quality traits from Burns et al. (2011)
Validation
2
Quality trait n Mean SD SEC R SEP SEP (%) R2
−1
BC (mEq kg DM) 1985 428 93.1 20.3 0.952 21.2 4.95 0.953
CP (g kg−1 DM) 1941 149 37.1 5.0 0.982 5.3 3.56 0.980
DMD (g kg−1 ) 1986 798 43.1 15.9 0.864 16.0 2.01 0.845
WSC (g kg−1 DM) 1941 182 53.0 10.4 0.961 11.4 6.26 0.956
BC buffering capacity, CP crude protein, DMD In vitro dry matter digestibility, WSC water-soluble
carbohydrate, SD standard deviation, SEC standard error of calibration, SEP standard error of
prediction, SEP % standard error of prediction as percentage of the mean

sub-sample was collected from each harvested plot for analysis. Two thousand sev-
enty six ryegrass samples (1,836 perennial, 137 Italian and 103 hybrid ryegrass:
Table 52.1) were selected for developing the calibration models. These represented
the range of species, ploidy, maturity and development stage over five harvest years
occurring at this site. Each sample was oven dried at 80 ◦ C for 16 h and then milled
using a Retsch mill (1 mm sieve) prior to NIRS analysis.

52.2.2 NIRS Analysis

Absorbance (log 1/reflectance) was measured on a NIRsystems 6500 or standardized

NIRsystems XDS (Foss UK Ltd., Warrington, UK) at 2 nm intervals over the wave-
length range 1,100–2,500 nm. Each sample was subsequently analyzed for in vitro
dry matter digestibility (DMD), water soluble carbohydrate (WSC) concentration,
crude protein (CP) concentration and buffering capacity using reference methods
(Burns et al. 2011). GLOBAL calibrations were developed using modified partial
least square regression as reported by Burns et al. (2011) and the optimal calibra-
tion models are summarized in Table 52.2. A standard normal variate and detrend
(Barnes et al. 1989) and a 1,4,4,1 curve smoothing technique were applied to the
spectra prior to calibration development. The LOCAL calibration technique used a
maximum number of spectrally similar samples (k) and these were selected as 10,
25, 50, 100, 150, 200, 250.

52.3 Results and Discussion

52.3.1 GLOBAL Calibration

The GLOBAL equation (Table 52.2) produced results with R2 greater than 0.95 for
WSC concentration, CP concentration and buffering capacity. In vitro DMD had a
lower R2 than the other three traits. The lower R2 for in vitro DMD in comparison with
other quality traits is similar to other GLOBAL calibrations (Jafari 2003) as this is a
chemically diverse assemblage influenced by animal factors that cannot be measured
in the spectrum of forage (Coleman and Moore 2003). In addition the relatively low
380 G. A. Burns et al.

Table 52.3 Regression

k R2 SEP SEP(c)
statistics of four quality traits
using a LOCAL calibration Buffering capacity (mEq kg−1 DM)
with (k = 10, 25, 50, 75, 100, 10 0.867 34.09 34.08
150, 200, 250) local samples 25 0.917 26.91 26.91
50 0.935 23.80 23.79
100 0.949 21.05 21.05
150 0.952 20.35 20.36
200 0.953 20.35 20.24
250 0.952 20.36 20.36
Crude protein (g kg−1 DM)
10 0.946 8.82 8.82
25 0.960 7.58 7.58
50 0.966 7.04 7.03
100 0.969 6.68 6.67
150 0.970 6.61 6.61
200 0.968 6.77 6.76
250 0.966 7.01 7.00
In vitro dry matter digestibility (g kg−1 )
10 0.710 24.82 24.82
25 0.739 23.42 23.43
50 0.759 22.47 22.48
100 0.785 21.26 21.27
150 0.791 20.95 20.95
200 0.792 20.90 20.90
250 0.789 21.09 21.09
Water-soluble carbohydrate (g kg−1 DM)
10 0.893 18.14 18.15
25 0.917 15.99 16.00
50 0.929 14.85 14.84
100 0.934 14.30 14.30
150 0.934 14.28 14.28
200 0.933 14.40 14.41
250 0.931 14.55 14.56
k maximum number of local samples in cali-
bration model, SEP standard error of prediction,
SEP(c) standard error of prediction of calibration

standard deviation of perennial ryegrass, relative to Italian and hybrid ryegrass may
indicate a clustering of values closer to the mean. As perennial ryegrass contributes
a large proportion of the calibration set (0.88) this may be a limiting factor in the
accuracy of the GLOBAL in vitro DMD model.

52.3.2 LOCAL Calibration

As the maximum number of samples (k) selected for use in the LOCAL calibration
increases a similar pattern emerged for all four quality traits (Table 52.3); initially
52 Comparison of LOCAL and GLOBAL Calibration Codels to Predict Ryegrass . . . 381

there was a sharp increase in R2 associated with decreasing standard error of pre-
diction. R2 increases at a decreasing rate between 50–150 samples followed by a
plateau in accuracy as k increased further. The selection of the optimal number of
samples was based on the criteria of high R2 and low standard error of prediction.
The optimal number of samples for CP and WSC concentration was k ≈ 150 and
for in vitro DMD and buffering capacity it was k ≈ 200 when developing calibration
models using the LOCAL algorithm.
Using the optimal number of samples, the LOCAL calibration produced a similar
level of accuracy to the GLOBAL calibration model for buffering capacity (SEP 20.4
vs. 20.3). However the accuracy of the LOCAL calibration models for in vitro DMD,
WSC and CP concentration was inferior to their respective GLOBAL models.
Berzaghi et al. (2000) showed the improvement of LOCAL algorithms using a
calibration set of forages that encompassed a larger range of feeds (hay, corn silage,
haylage, small grain silage and total mixed ration) than for the current calibration set
(perennial, Italian and hybrid ryegrass). It is likely that bias existed between these
categories of forage and non-linear relationships were present due to the large range in
composition which could have potentially limited the performance of the GLOBAL
calibration models. Similar levels of bias might not exist between monocultures of
three ryegrass species in the present study and the use of a GLOBAL calibration may
be more appropriate when most variation is encompassed with little bias between
factors in the model.

52.4 Conclusions

Initially increasing the maximum number of samples selected for calibration using
the LOCAL algorithm decreased the standard error of prediction for all four quality
traits. The decrease in standard error of prediction continued at a decreasing rate
as more locally selected samples were used in the calibration set and as more local
samples were used the decrease in standard error plateaued.
The GLOBAL calibrations produced more accurate equations for in vitro DMD,
WSC and CP concentration and were of similar accuracy for buffering capacity.
From this set of results the continued use of GLOBAL calibrations is recommended
for the accurate and robust analysis of the quality of ryegrasses of the Irish national
variety evaluation scheme.

Acknowledgments The authors acknowledge the funding of the Department of Agriculture, Food
and the Marine (DAFM) research stimulus fund 07 526 to carry out the research.

References

Barnes RJ, Dhanoa MS et al (1989) Standard normal variate transformation and de-trending of
near-infrared diffuse reflectance spectra. Appl Spectrosc 43(5):772–777
382 G. A. Burns et al.

Berzaghi P, Shenk JS et al (2000) LOCAL prediction with near infrared multi-product databases.
J Near Infrared Spectrosc 8(1):1–9
Burns GA, Gilliland TJ et al (2011) Using chemometrics to develop near infrared spectroscopy
(NIRS) calibrations to predict ryegrass quality. Agricultural Research Forum 2011 Tullamore,
Ireland, p 135
Coleman SW, Moore JE (2003) Feed quality and animal performance. Field Crops Res 84(1–2):
17–29
Grogan D, Gilliland TJ (2010) A review of perennial ryegrass variety evaluation in Ireland. Grasses
for the Future. Perennial ryegrasses: current and future genetic potential. In: O’Donovan M. and
Hennessy D (eds) Moorepark, Teagasc, pp 99–118
Jafari A, Connolly V et al (2003) A note on estimation of quality parameters in perennial ryegrass
by near infrared reflectance spectroscopy. Irish J Agric Food Res 42(2):293–299
Shenk JS, Westerhaus MO et al (1997) Investigation of a LOCAL calibration procedure for near
infrared instruments. J Near Infrared Spectrosc 5(4):223–232
Chapter 53
Grass for Biogas Production—Anaerobic
Methane Production from Five Common
Grassland Species at Sequential Stages
of Maturity

K. O’Riordan, J. McEniry, T. Woodcock, C. King and P. O’Kiely

Abstract Grassland biomass represents the most significant feedstock resource in

Ireland, accounting for approximately 91 % of the 4.3 million hectares of agricul-
tural land. Grass can be an excellent energy crop and may be classified as a high
yielding (up to 15 t dry matter ha−1 a−1 ), low input perennial crop. Consequently,
grass will be a dominant feedstock for anaerobic digestion (AD) on Irish farms. This
study investigated the effects of stage of maturity of five grass species on methane
production using dried, milled samples in a small-scale (160 ml), high-throughput
batch digestion test. Five common grass species (perennial ryegrass, Italian ryegrass,
cocksfoot, timothy and tall fescue) were grown in field plots (with three replicate
blocks) under a high nitrogen fertiliser input (125 kg N ha−1 ) and harvested at five
sequential dates (fortnightly from 12 May to 7 July; n = 75 plots) in the primary
growth. Of the five grass species investigated, average total CH4 production was
highest (P < 0.01) for the perennial ryegrass. On average, the rate of digestion
decreased (P < 0.001) with increasing plant maturity. Although total CH4 produc-
tion decreased numerically with advancing plant maturity, this difference was not
significant (P > 0.05).

53.1 Introduction

Grassland biomass represents the most significant feedstock resource in Ireland,

accounting for approximately 91 % of the 4.3 million hectares of agricultural land.
Grass can be an excellent energy crop and may be classified as a high yielding (up
to 15 t dry matter ha−1 a−1 ), low input perennial crop. Consequently, grass will be
a dominant feedstock for anaerobic digestion (AD) on Irish farms.

P. O’Kiely () · J. McEniry · C. King

Animal & Grassland Research and Innovation Centre, Teagasc, Grange, Dunsany,
Co. Meath, Ireland
e-mail: [email protected]
K. O’Riordan · T. Woodcock
The Bioresources Research Centre, University College Dublin, Belfield, Dublin 4, Ireland
e-mail: [email protected]

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 383

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1_53, © Springer Science+Business Media Dordrecht 2013
384 K. O’Riordan et al.

Grasslands are usually managed to enhance the production and quality of herbage.
Two of the main management factors affecting herbage chemical composition are
grass species and stage of maturity at harvest. The physiology of different grass
species can impose differences in adaptability, productivity and chemical compo-
sition, which may impact on their methane production potential. However, there is
an absence of detailed information on the specific methane yield of different grass
species grown under similar management conditions. One study by Seppala et al.
(2009) reported that there was no significant difference between the specific methane
yields of cocksfoot (318 L CH4 kg−1 volatile solids (VS)), tall fescue (314 L CH4
kg−1 VS) and timothy (311 L CH4 kg−1 VS) grown under boreal conditions.
In general, advancing maturity of grass from the vegetative to the inflorescence
growth stage is characterized by an increase in fibre components (Stefanon et al. 1996)
and a decrease in digestibility (Ballard et al. 1990). Herbage nutritive value for animal
production is generally considered to decline with advancing plant maturity and a
delayed harvest can have a significant negative impact on herbage quality. Delaying
the harvest date also has an important negative influence on methane production from
AD (Prochnow et al. 2009).
This study investigated the effects of stage of maturity of five grass species on
methane production using dried, milled samples in a small-scale (160 ml), high-
throughput batch digestion test.

53.2 Materials and Methods

Five common grass species, perennial ryegrass (PRG; Lolium perenne L. var. Gan-
dalf), Italian ryegrass (IRG; Lolium multiflorum Lam. var. Prospect), cocksfoot
(Dactylis glomerata L. var. Pizza), timothy (Phleum pratense L. var. Erecta) and
tall fescue (Festuca arundinacea Schreb var. Fuego) were grown in field plots (with
three replicate blocks) under a high nitrogen fertiliser input (125 kg N ha−1 ) and
harvested at five sequential dates (fortnightly from 12 May to 7 July; Harvests 1–5;
n = 75) in the primary growth. On each harvest date, the appropriate plots were
harvested to a 6 cm stubble height and a representative sample of each herbage was
oven dried at 40 ◦ C for 48 h before being milled (Wiley mill, 1 mm screen). Dried,
milled samples were used for the determination of neutral detergent fibre (NDF) as
described by McEniry et al. (2006).
Replicate samples were also analysed for methane production in 160 ml batch
digestion tests, according to VDI guideline 4630 (2006). Briefly, substrate and inocu-
lum were added to the 160 ml incubation bottles at a VS inoculum to substrate ratio of
2:1 and at a finalVS concentration of 10 g kg−1 (i.e. 0.7 gVS per bottle; 0.47 substrate:
0.23 inoculum). The inoculum (pH = 7.98; 4 g DM kg−1 , 2 g VS kg−1 ) was sourced
from a cattle slurry digester at the Agri-Food and Biosciences Institute in Hillsbor-
ough, Northern Ireland. Micro- and macro- mineral solutions were added to ensure
that nutrient conditions in the bottles were not limiting and sodium hydrogen carbon-
ate was added to act as a buffer system (3.5 g L−1 ). Water was added to each bottle to
adjust the final volume to 70 ml. Six replicate blank (no grass substrate added) and
53 Grass for Biogas Production—Anaerobic Methane Production . . . 385

800

700

600
NDF (g kg-1 DM)

500

400

300

200

100

0
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

Perennial ryegrass Italian ryegrass Timothy Cocksfoot Tall fescue

Harvest date x grass species

Fig. 53.1 Effect of harvest date and grass species on neutral detergent fibre (NDF) concentration
(SEM 8.9; P < 0.001)

cellulose (reference sample to assess the biological activity of the inoculum) controls
were also included. The final pH in the incubation bottles was adjusted to 7.2 before
bottles were flushed with N2 for 1 min and sealed with butyl rubber stoppers and
aluminum crimp seals. Bottles were incubated at 38 ◦ C and hand-mixed daily.
Using a detachable pressure transducer, the gas headspace pressure inside each
bottle was recorded after 2, 5, 7, 9, 12, 15 and 19 days incubation. The total
amount of gas produced was estimated using the following equation: Gas production
(ml) = (V h /Pa ) × Pt ; where V h is the headspace volume (ml), Pa is the atmospheric
pressure (hPa) and Pt is the gas headspace pressure (hPa). Following determination
of gas volume, a 0.8 ml sample of gas was used to determine CH4 concentration
by gas chromatography (Purcell et al. 2011). Analysis of the results included the
following steps: (a) headspace correction for gas values on day 2, (b) subtraction of
the volume of gas produced by the inoculum (i.e. blank) from the volume produced
in the batch digestion test with substrate and inoculum and (c) normalising the gas
volume to standard conditions (i.e. dry gas, 273 K, 1,013 hPa).
Data were analysed as a split-plot design using the Proc MIXED procedure of
SAS, Version 9.1.2 (SAS 2004) with harvest date as the main plot and grass species
as the sub-plot, and accounting for repeated measures effect of sampling day.

53.3 Results

In general, herbage NDF concentration increased (P < 0.001) with advancing plant
maturity (Fig. 53.1). Of the five grass species investigated, the NDF concentration
was lowest (P < 0.001) for the IRG followed by the PRG.
386 K. O’Riordan et al.

300

Harvest 1 Harvest 2 Harvest 3 Harvest 4 Harvest 5

250
Cumulative CH 4 (L kg-1 VSadded )

200

150

100

50

0
0 2 4 6 8 10 12 14 16 18 20

Time (days)

Fig. 53.2 Effect of harvest date on cumulative methane production in batch digestion tests over a
19 day incubation period (averaged across grass species; SEM 4.4; P < 0.001)

On average, 0.69 of the CH4 produced over the 19 day incubation period was
produced by day 7 of the batch digestion test (Fig. 53.2). Of the five grass species in-
vestigated, average daily CH4 production (P < 0.001; data not shown) and total CH4
production (P < 0.01) over the 19 day incubation period were highest for the PRG,
followed by the IRG, timothy, tall fescue and cocksfoot (Fig. 53.3). The specific
methane yield varied from 192–259 CH4 kg−1 VSadded . The early harvest PRG (Har-
vest 1–12 May) had the highest specific methane yield (259 L CH4 kg−1 VSadded ),
while similarly high values were observed for the Harvest 2 PRG (257 L CH4 kg−1
VSadded ) and Harvest 1 IRG (258 L CH4 kg−1 VSadded ). The lowest specific methane
yield (192 L CH4 kg−1 VSadded ) was recorded for the late harvest cocksfoot (Har-
vest 5–7 July). On average the apparent rate of digestion decreased (P < 0.001) with
advancing harvest date (Fig. 53.2). Although total CH4 production decreased nu-
merically with advancing harvest date, this difference was not significant (P > 0.5).
This trend of decreasing CH4 production with advancing plant maturity was similar
(P > 0.05) across the five grass species investigated.
Cellulose was used as the reference substrate in this study producing 267 L CH4
kg−1 VSadded , which represents 0.63 of the maximum theoretical yield following
complete degradation.

53.4 Discussion

As a plant matures the proportion of cell wall components (e.g. cellulose, hemicellu-
lose and lignin) increases, while the proportion of cell contents decreases (Stefanon
et al. 1996; Ugherughe 1986). These changes particularly reflect the general decrease
53 Grass for Biogas Production—Anaerobic Methane Production . . . 387

300

250
Total CH 4 (L kg-1 VSadded )

200

150

100

50

0
1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

Perennial ryegrass Italian ryegrass Timothy Cocksfoot Tall fescue

Harvest date x grass species

Fig. 53.3 Effect of harvest date and grass species on total CH4 production (over 19 days incubation;
SEM 14.2, P > 0.5)

in plant leaf to stem ratio and the increasing cell wall content within the stems. Since
this process is accompanied by increased lignification within the cell wall fraction,
there is an overall decrease in herbage digestibility resulting in a slower rate of di-
gestion. Seppala et al. (2009), in a study investigating the effect of various harvesting
dates and cultivation years on the methane production potential of four grass species,
reported that the average specific methane yield of the first harvest of all grasses
(345 L CH4 kg−1 VS) was higher than the second harvest (311 L CH4 kg−1 VS).
In temperate grassland regions, PRG is preferred for animal production because
of its high digestibility and water soluble carbohydrate content, and reduced fibre
concentration. These factors also combined in the current study to produce the highest
methane yield in batch digestion tests.
The lower than expected CH4 yield with the cellulose control suggests either that
the activity of the inoculum was sub-optimal or that the duration of the assay was
too short. This is further supported by the lower CH4 yields than that reported for
grass samples (range 298–467 L CH4 kg−1 VS; Braun et al. 2010).

53.5 Conclusions

Of the five grass species investigated, PRG may be most desirable as a feedstock for
biogas production. The rate of digestion decreased with advancing plant maturity.
Although the total CH4 produced also decreased numerically with advancing plant
maturity, this trend did not reach statistical significance.
388 K. O’Riordan et al.

Acknowledgments Funding for this research was provided under the National Development Plan
through the Research Stimulus Fund (#RSF 07 557), administered by the Department of Agriculture,
Fisheries & Food, Ireland.

References

Ballard RA, Simpson RJ, Pearce GR (1990) Losses of the digestible components of annual ryegrass
(Lolium rigidum gaudin) during senescence. Aust J Agric Res 41:719–731
Braun R, Weiland P, Wellinger A (2010) Biogas from energy crop digestion: International Energy
Association, bioenergy task 37. http://www.iea-biogas.net/_download/energycrop_def_Low_
Res.pdf. Accessed 17 August 2011
McEniry J, O’Kiely P, Clipson NJW, Forristal PD, Doyle EM (2006) The microbiological and
chemical composition of baled and precision-chop silages on a sample of farms in County
Meath. Irish J Agr Res 45:73–83
Prochnow A, Heiermann M, Plochl M, Linke B, Idler C, Amon T, Hobbs PJ (2009) Bioenergy from
permanent grassland—a review: 1. Biogas. Bioresource Technol 100:4931–4944
Purcell PJ, O’Brien M, Boland TM, O’Kiely P (2011) In vitro rumen methane output of perennial
ryegrass samples prepared by freeze drying or thermal drying (40 ◦ C). Anim Feed Sci Technol
(166–167):175–182
SAS (2004) SAS for windows (Version 9.1.2). Statistical Analysis System Institute Inc., Cary, NC,
USA
Seppala M, Paavola T, Lehtomaki A, Rintala J (2009) Biogas production from boreal herbaceous
grasses—specific methane yield and methane yield per hectare. Bioresource Technol 100:2952–
2985
Stefanon B, Pell AN, Schofield P (1996) Effect of maturity on digestion kinetics of water-soluble
and water-insoluble fractions of alfalfa and brome hay. J Anim Sci 74:1104–1115
Ugherughe PO (1986) Relationship between digestibility of Bromus inermis plant parts. J Agron
Crop Sci 157:136–143
VDI 4630 (2006) Fermentation of organic materials—characterisation of the substrate, sampling,
collection of material data, fermentation tests. The association of German engineers, Dusseldorf,
Germany
Author Index

A D
Abberton M. T., 47 D’Ottavio P., 327
Alaux L., 203 De Boever J., 181
Allegrucci S., 69 de Cauwer B., 3
Allen D., 239 De Dauw K., 147
Anderson W.F., 231 De Riek J., 181
Andjelkovic S., 275, 269 De Rosa L., 249
Annicchiarico P., 249, 289, 341 Dehmer K.J., 301
Antanasović S., 327, 353 del Carmen Lagunes Espinoza L., 197
Aper J., 3 Delić D., 333
Azhaguvel P., 115 Della Marianna G., 255, 307
Devine T.E., 353
B Diekmann K., 141
Babic S., 275 Doležel J., 105, 115
Baert J., 63, 181, 209, 261, 295, 371 Ðordević V., 353
Baird J.H., 115 Doyle E.M., 347
Ball T., 209 Duller S.J., 239
Barre P., 203 Durand J.-L., 121
Barth S., 55, 115, 141
E
Bartoš J., 105, 115
Ellison N.W., 225
Bocian A., 189
Erić P., 327
Boland T.M., 315
Boller B., 29, 37, 209
F
Bourgoin T., 121 Farrell M.S., 171
Brazauskas G., 81 Ferradini N., 69, 155
Burns G.A., 377 Feuerstein U., 209
Byrne S.L., 115 Fievez V., 181
Flajoulot S., 203
C
Cagaš B., 209, 219 G
Carroni A.M., 341 Gay A., 209
Černoch V., 115 Ghesquiere A., 63, 121, 171, 295
Číhalíková J., 105 Gilliland T.J., 377
Cnops G., 161 Godwin I.D., 231
Cougnon M., 3, 147, 371 Gouzy J., 203
Ćupina B., 327, 353 Grogan D., 323, 377
Czembor E., 209 Gusmeroli F., 255, 307

S. Barth, D. Milbourne (eds.), Breeding Strategies for Sustainable Forage 389

and Turf Grass Improvement,
DOI 10.1007/978-94-007-4555-1, © Springer Science+Business Media Dordrecht 2013
390 Author Index

H Maksimović S., 333

Hartmann S., 209 Manzanares C., 55
Havránková M., 105 Marczak Ł., 189
Hayes R. C., 47, 55 Martín Osorio V.E., 225
Haygarth P.M., 171 Mathews R., 239
Hirakawa H., 29 McEniry J., 365, 383
Hodkinson T.R., 141 McEvoy M., 315
Huguet T., 197 Meusnier I., 203
Humphreys M.W., 89, 171 Mihailović V., 353
Huttner E., 121 Mikić A., 327, 353
Milenkovic J., 275
I Muylle H., 63, 261
Isobe S., 29
N
J Navrátil P., 105
Jakesova H., 209 Nehrlich S., 301
Jevtić G., 269 Nicolia A., 69, 155
Jewell M.C., 231
Jones N., 75 O
Julier B., 197, 203 O’Brien M., 315
Jurczyk B., 189 O’Donovan S. A., 89
O’Kiely P., 315, 347, 365, 377, 383
K O’Riordan K., 383
Kilian A., 115, 121 Obianugba G., 261
King C., 365, 383
Kingston-Smith A. H., 89 P
Klima M., 209, 115 Pašakinskienė I., 81
Klimenko I., 29 Pecetti L., 249, 255, 307
Knotová D., 283 Pelikán J., 283
Kölliker R., 27, 115 Perić V., 353
Kölliker S., 29 Persson C., 209
Kopecká J., 105 Pietraszek W., 209
Kopecký D., 105, 115 Pitzalis M., 341
Kosmala A., 75, 189 Pogrzeba M., 245
Krautzer B., 209 Poinsard L., 209
Krstić Ð., 327, 353 Posselt U.K., 209
Ksiażczyk T., 75, 97 Proietti S., 249
Kubaláková M., 105 Prokopiuk K., 245
Kuzmanović D., 333 Purcell P.J., 315

L
Lambrides C.J., 231 Q
Leenheer H., 209 Quitté Y., 209
Loch D.S., 231
Lovatt J. A., 47 R
Lübberstedt T., 81 Raab S., 283
Lugic Z., 275, 269 Radovic J., 275, 269, 333
Lukaszewski A.J., 115 Rana J. C., 29
Lynch J.P., 347 Rapacz M., 189
Rasulić N., 333
M Reheul D., 3, 147, 371
MacDuff J., 239 Richardson M., 365
Macháè R., 359 Riday H., 21
Macleod C. J. A., 171 Rognli O.A., 115
Author Index 391

Rohde A., 131, 161 V

Roldan-Ruiz I., 131, 161, 261 Van Bockstaele E., 161
Romani M., 209, 255, 307, 341 Van Dingenen J., 161
Rosellini D., 69, 155 Van Minnebruggen A., 161
Ruda P., 341 Van Nes M., 209
Russi L., 209 Van Parijs F., 261
Ruttink T., 131 Van Waes C., 181, 261, 371
Rybka K., 245 Vandewalle M., 181
Vasic T., 269, 275
S Verhelst J., 147
Šafář J., 105 Vermeulen E., 131
Saha M., 115 Veronesi F., 69, 155
Salis M., 341 Vorhölter F.-J., 37
Sandve S.R., 115
Vrána J., 105
Sato S., 29
Vymyslický T., 283
Schubiger F. X., 209
Schulze S., 209
Sharma T. R., 29 W
Shirasawa K., 29 Whalley W. R., 171
Simić A., 333 Wichmann F., 37
Šimková H., 105 Widmer F., 37
Sokolovic D., 275 Willner E., 209, 301
Sokolović G., 269 Wolfe K.H., 141
Spoleto P., 255, 307 Wolters L., 209
Srebrić M., 353 Woodcock T., 383
Stajković-Srbinović O., 333
Sterck L., 131
Z
Stewart A.V., 225
Stočes Š., 105 Żurek G., 245
Studer B., 55 Zwierzykowska E., 75, 97
Sulas C., 341 Zwierzykowski Z., 75, 97, 189
Svobodová M., 219

T
Tabata S., 29
Taciak M., 75
Tardin M.C., 209
Thorogood D., 55
Tomaszewski C., 115
Tosca A., 255, 307
Turner L.B., 171