TREES ON THE FARM

Assessing the Adoption Potential of

Agroforestry Practices in Africa
 TREES ON THE FARM
Assessing the Adoption Potential of
Agroforestry Practices in Africa

Edited by

S. FRANZEL
International Centre for Research in Agroforestry (ICRAF),
Nairobi, Kenya

S.J. SCHERR
Agricultural and Resource Economics Department, University
of Maryland, College Park, USA

CABI Publishing
in association with the
International Centre for Research in Agroforestry
 CABI Publishing is a division of CAB International
CABI Publishing CABI Publishing
CAB International 10 E 40th Street
Wallingford Suite 3203
Oxon OX10 8DE New York, NY 10016
UK USA

Tel: +44 (0)1491 832111 Tel: +1 212 481 7018

Fax: +44 (0)1491 833508 Fax: +1 212 686 7993
Email: cabi@[Link] Email: cabi-nao@[Link]
Web site: [Link]

© CAB International 2002. All rights reserved. No part of this publication may be
reproduced in any form or by any means, electronically, mechanically, by photocopying,
recording or otherwise, without the prior permission of the copyright owners.
Published in association with ICRAF, PO Box 30677, Nairobi, Kenya.

A catalogue record for this book is available from the British Library, London, UK.

Library of Congress Cataloging-in-Publication Data

Trees on the farm : assessing the adoption potential of agroforestry practices in Africa /
edited by S. Franzel and S.J. Scherr.
p. cm.
Includes bibliographical references (p. ).
ISBN 0-85199-561-6 (alk. paper)
1. Agroforestry systems--Africa. 2. Agroforestry--Technology transfer--Africa. I.
Franzel, Steven Charles. II. Scherr, Sara J.

S494.5.A45 T74 2002

634.99096--dc21
2001043834
ISBN 0 85199 561 6

Typeset in Adobe Garamond by Columns Design Ltd, Reading

Printed and bound in the UK by Biddles Ltd, Guildford and King’s Lynn
 Contents

Contributors vii
Foreword ix
P.A. Sanchez
Acknowledgements x
1. Introduction 1
S. Franzel and S.J. Scherr
2. Methods for Assessing Agroforestry Adoption Potential 11
S. Franzel, S.J. Scherr, R. Coe, P.J.M. Cooper and F. Place
3. Assessing the Adoption Potential of Improved Fallows in
Eastern Zambia 37
S. Franzel, D. Phiri and F. Kwesiga
4. The Adoption Potential of Short Rotation Improved Tree Fallows:
Evidence from Western Kenya 65
R.A. Swinkels, S. Franzel, K.D. Shepherd, E. Ohlsson and J.K. Ndufa
5. Assessing the Adoption Potential of Hedgerow Intercropping for
Improving Soil Fertility, Western Kenya 89
R.A. Swinkels, K.D. Shepherd, S. Franzel, J.K. Ndufa, E. Ohlsson
and H. Sjogren
6. Farmer-designed Agroforestry Trials: Farmers’ Experiences in Western
Kenya 111
S. Franzel, J.K. Ndufa, O.C. Obonyo, T.E. Bekele and R. Coe
7. Calliandra calothyrsus: Assessing the Early Stages of Adoption of a
Fodder Shrub in the Highlands of Central Kenya 125
S. Franzel, H.K. Arimi and F.M. Murithi
8. Promoting New Agroforestry Technologies: Policy Lessons from
On-farm Research 145
S.J. Scherr and S. Franzel

v
vi Contents

9. Assessing Adoption Potential: Lessons Learned and Future Directions 169

S. Franzel and S.J. Scherr
Glossary 185
Index 189
 Contributors

H.K. Arimi, National Agricultural Research Centre, Muguga, PO Box 30148,

Nairobi, Kenya
T.E. Bekele, Institute of Biodiversity, Conservation and Research, PO Box 2003,
Addis Ababa, Ethiopia
R. Coe, International Centre for Research in Agroforestry (ICRAF), PO Box
30677, Gigiri, Nairobi, Kenya
P.J.M. Cooper, International Development Research Centre, PO Box 8500,
Ottawa, Ontario, Canada K1G 3H9
S. Franzel, International Centre for Research in Agroforestry (ICRAF), PO Box
30677, Gigiri, Nairobi, Kenya
F. Kwesiga, SADC-ICRAF Regional Agroforestry Programme, PO Box MP 128,
Mount Pleasant, Harare, Zimbabwe
F.M. Murithi, Kenya Agricultural Research Institute, PO Box 58137, Nairobi, Kenya
J.K. Ndufa, Kenya Forestry Research Institute (KEFRI), Maseno Agroforestry
Research Centre, PO Box 25199, Kisumu, Kenya
C.O. Obonyo, Kenya Forestry Research Institute (KEFRI), Maseno Agroforestry
Research Centre, PO Box 25199, Kisumu, Kenya
E. Ohlsson, Sida/SAREC, SE-105 25 Stockholm, Sweden
D. Phiri, World Vision Integrated Agroforestry Project, PO Box 510948, Nairobi,
Kenya
F. Place, International Centre for Research in Agroforestry (ICRAF), PO Box
30677, Gigiri, Nairobi, Kenya
S.J. Scherr, Agricultural and Resource Economics Department, University of
Maryland, College Park, MD 20742, USA

vii
viii Contributors

K.D. Shepherd, International Centre for Research in Agroforestry (ICRAF), PO

Box 30677, Gigiri, Nairobi, Kenya
H. Sjogren, Swedish University of Agricultural Sciences, Uppsala, Sweden
R.A. Swinkels, The World Bank, 8th Floor, 63 Ly Thai To Street, Hanoi 20, Vietnam
 Foreword

This is a book that pushes forwards the emerging science of agroforestry. There are
several syntheses on the biophysical aspects of agroforestry but little is available on
the socioeconomics of agroforestry and its adoption by farmers, following rigorous
scientific methods. This book fills such a gap by assessing the adoption potential of
selected agroforestry practices, describing the appropriate methodologies and draw-
ing lessons for improving the effectiveness of the research–development continuum.
Five case studies are described and analysed in this book. Four of them are success-
ful and one – hedgerow intercropping – was promoted before a rigorous biophysical
assessment was made.
This book demonstrates how farmers in selected areas have tested and adapted
these practices, incorporated them into their farming systems and improved their
welfare and incomes. It also provides examples of scaling up within the areas where
testing took place: farmer-to-farmer dissemination, partnerships among researchers,
extension services, and non-governmental organizations (NGOs) to promote scal-
ing up.
The key finding was that agroforestry reduces the risks that farmers face from
input markets by investing small amounts of land and labour, rather than spending
cash on expensive inputs, for improving soil fertility, feeding their livestock, or pro-
viding wood for fuel and construction.
There were also important methodological lessons, that add further value to
this book. The authors should be congratulated for advancing the science of agro-
forestry.
Pedro A. Sanchez
Director General of ICRAF
Nairobi, Kenya

ix
 Acknowledgements

The editors are grateful to the International Centre for Research in Agroforestry for
contributing towards the costs of publishing this book. We also thank the ICRAF
staff in Maseno and Embu, Kenya, and Chipata, Zambia, and our partners at these
sites: the Kenya Agricultural Research Institute, the Kenya Forestry Research
Institute, and the Ministry of Agriculture Food, and Fisheries, Zambia. We also
wish to thank the many colleagues in the extension services and non-governmental
organizations who participated in the work reported in this book. Finally, we thank
the many farmers who participated in the surveys and on-farm trials, especially
those who shared with us their agroforestry innovations. They are too many to
mention, but we would like to thank, in particular, Mr Harrison Chongwe
(Chivungwe, Zambia), Ms Zelina Mwanza (Chipata, Zambia) Ms Jennipher Zulu
(Kalunga, Zambia), Mr Michael Mwaniki (Ngandori, Kenya), Ms Purity Njagi
(Embu, Kenya) and Mr Samuel Nyaga (Kagahari North, Kenya).
We also gratefully acknowledge the International Food Policy Research
Institute, Washington, DC, for hosting the senior editor on a 9-month sabbatical to
work on this book; thanks to Peter Hazell, Director, Environment and Production
Technology Division, IFPRI, for his support.
Chapter 4 of this volume is a revised version of an article appearing in
Agricultural Systems, Vol. 55, used with permission from Elsevier Science: Swinkels,
R., Franzel, S. Shepherd, K. Ohlsson, E. and Ndufa, J. (1997) The economics of
short rotation improved fallows: evidence from areas of high population density in
western Kenya, pp. 99–121.
Chapter 5 draws upon material appearing in an article, Adoption potential of
hedge-row intercropping in the maize-based cropping systems of the highlands of
Western Kenya, Experimental Agriculture, Vol. 33, 1997, used with permission
from Cambridge University Press:
Part 1. Shepherd, K., Ndufa, J.K., Ohlsson, E., Sjogren, H., and Swinkels, R.,
1997. Background and agronomic evaluation, pp. 201–210.
Part 2. Swinkels, R. and Franzel, S.. Economic and farmers evaluation, pp.
211–223.
We are also grateful to the many colleagues who assisted in the work and
reviewed chapters; their names are listed in the acknowledgements at the end of
each chapter.
 Introduction
S. FRANZEL1 AND S.J. SCHERR2
1International Centre for Research in Agroforestry, PO Box 30677,
1
Gigiri, Nairobi, Kenya; 2Agricultural and Resource Economics
Department, University of Maryland, College Park,
MD 20742, USA

Background
Agricultural land use and management present major development challenges
throughout sub-Saharan Africa. The area under cultivation has expanded notably,
total yields are rising, and there is large-scale conversion from fallow-based cropping
systems to continuous cultivation. None the less, per capita food production has
declined by about 2% per year since 1960 (World Bank, 1996), and constraints on
growth in agricultural sectors, which remain prominent in most African economies,
are an important factor explaining a 1% per year decline in per capita incomes
between 1983 and 1993 (Cleaver and Schreiber, 1994; World Bank, 1994).
Environmental problems associated with agricultural production have also
become a major concern. With the marked expansion and intensification of farm-
ing, total forested area in Africa declined by 50 million hectares during the 1980s
(Dembner, 1991), reducing the availability of wood products for fuel and construc-
tion, degrading range resources, and exposing vulnerable soils to degradation. In
many areas, particularly in the densely populated highlands and in drylands, soil
degradation due to inadequate agricultural practices and nutrient depletion threat-
ens long-term productive potential (Scherr and Yadav, 1995; Buresh et al., 1997).
Agroforestry is defined as a dynamic, ecologically based, natural resource man-
agement system that, through the integration of trees on farms and in the agricul-
tural landscape, diversifies and sustains production for increased social, economic
and environmental benefits for land users at all levels (Leakey, 1996). Agroforestry
has considerable potential to contribute towards solving some of these problems.
Nitrogen-fixing trees, as substitutes or complements for chemical fertilizer, can
increase smallholder incomes, conserve foreign exchange and improve regional food
security. By providing a supply of fuelwood from the farm, agroforestry can help
reduce pressure on forests and communal woodlands. Moreover, agroforestry trees
can supply farm households with a wide range of other products, including food,
medicine, livestock feed, and timber for home use and sale. Other services that trees
provide, such as boundary markers, windbreaks, soil erosion barriers, beauty and
shade, are difficult to quantify but are none the less of substantial importance to
farm families and for natural resource protection.

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 1
2 S. Franzel and S.J. Scherr

Indeed, indigenous agroforestry practices are already widespread in Africa, as is

the planting and protection of many tree species introduced during the colonial
period and later through forestry and agroforestry extension projects (Le Houerou,
1987; Kerkhof et al., 1990; Warner, 1993). However, most are suited to low-intensity
agricultural systems, and function well below their potential productivity as a result of
using poorly adapted species, provenances and management systems. Over the past 20
years, research and extension systems working in Africa have sought to improve the
productivity of existing systems and develop new practices to meet new land-use chal-
lenges and opportunities (Scherr and Müller, 1990; Kang, 1993; Cooper et al., 1996).

Assessment of Agroforestry from the Farmers’ Perspective

There is now a substantial body of scientific literature on the biophysical features

and potentials of agroforestry practices in Africa (Cooper et al., 1996; Sanchez,
1996). However, these provide only a first indication of their suitability for wide-
spread promotion, and their likely economic, social and environmental contribu-
tions. There is, unfortunately, almost no information available quantifying the
contribution of agroforestry to household income, food security and welfare. This
gap is especially surprising given the widespread use of agroforestry in development
projects and extension programmes; as early as 1989, Scherr and Müller (1991)
were able to identify 60 development projects promoting agroforestry in Africa.
But most agroforestry research has been conducted at research stations or
research plots near development project sites; few studies are available on the per-
formance of agroforestry practices under farmer-managed conditions. Even fewer
quantify the financial returns of the practices1 or examine their advantages and dis-
advantages, as perceived by farmers. For example, Swinkels et al. (Chapter 4) report
finding only two studies on the testing of improved fallows for increasing soil fertil-
ity under farmer management, and neither included a financial analysis. They note
that hedgerow intercropping (also called alley cropping or alley farming) for
improving soil fertility is more widely tested, but the few farmer-managed trials
conducted were mostly limited to the humid and subhumid tropics of west Africa.
Moreover, assessment of adoption potential has been problematic in many
hedgerow intercropping trials and analyses: targeting was inappropriate as the farm-
ers’ perceived priority problem was not low soil fertility; farmers’ participation was
obtained through the provision of material incentives which were likely to bias their
views and actions; and there was inadequate monitoring of socioeconomic variables
(Whittome, 1994).

1 The economics profession distinguishes ‘financial analysis’ from ‘economic analysis’. The
former takes the perspective of the individual farmer and values inputs and outputs at
prices farmers face. By contrast, the latter is defined from the perspective of society as a
whole; market prices of inputs and outputs are corrected if these do not reflect their real
economic values to society (Gittinger, 1982). For example, if the fertilizer price was
subsidized, financial analysis would use the subsidized price and economic analysis
would use the unsubsidized price.
 Introduction 3

A literature review by Swinkels and Scherr (1991) found 68 economic studies

of agroforestry systems in Africa, from 17 countries, but most were ex ante analyses,
or models based on research station data or assumptions not grounded in farmer
experience, or general economic descriptions of the technology or market condi-
tions. Scherr and Müller (1991) found only a few extension projects in Africa evalu-
ating the financial impact of agroforestry systems being promoted. This situation is
not unique to Africa; it reflects a global paucity of research on the economics and
adoptability of agroforestry practices in developing countries. The only two com-
parative studies assessing the economics of agroforestry practices are Current et al.
(1995) for Central America and the Caribbean and Sullivan et al. (1992) for
selected practices in various parts of the world.
Social and financial analyses of agroforestry practices are important for several
reasons. African governments, non-governmental organizations (NGOs) and
donors need information on the performance of these practices and how they con-
tribute to household welfare and to development objectives, in order to assess
whether and how they should continue investing in their development and dissemi-
nation. Data are needed on the financial and non-financial benefits of the practices,
what works where and why, why some farmers within specific communities adopt
and others do not, and who within the family reaps the benefits.
Researchers and development practitioners also need information on how
farmers are using and modifying the practices and what problems they encounter,
so they can develop technologies and practices that better meet farmers’ needs and
circumstances. Similarly, policy makers need information on the influence of policy
factors on the adoptability and performance of agroforestry practices. Different
incentives, institutional mechanisms and policies could be implemented to facilitate
agroforestry adoption.
Finally, and perhaps of greatest importance, farmers themselves need informa-
tion on the performance of agroforestry practices, so they can make more informed
decisions on whether to test and adopt them. Such information is especially impor-
tant because many agroforestry practices require considerable resources, skills and
time to bear fruit. In some areas, farmer groups and organizations are emerging to
lead adaptive research, and can use this type of information in planning their agro-
forestry activities.

Objectives

This volume has two primary objectives. First, it assesses the adoption potential of
several new agroforestry practices that researchers, development practitioners and
farmers are currently testing in Africa. The adoption potential of a practice is
defined as its feasibility, profitability and acceptability, as viewed from the farmers’
perspective.
Secondly, the volume draws lessons for improving the effectiveness and effi-
ciency of the process of developing, modifying and disseminating new agroforestry
practices. These include recommendations to: (i) researchers on technology design
4 S. Franzel and S.J. Scherr

and the implementation of on-farm trials; (ii) development organizations concern-

ing the establishment of networks for adaptive research and extension; and (iii) pol-
icy makers on institutional mechanisms and incentives for promoting agroforestry.
Feedback to research and extension is also critical for improving and promoting
practices and for assessing impact, based on farmers’ experiences in managing and
using agroforestry practices. Farmers’ experiences include their assessments, prefer-
ences, innovations and their problems, and how they cope with them.
The volume also presents methods that can be used to assess the adoption
potential of agroforestry practices during the on-farm testing stage and just after,
when farmers are deciding whether or not to adopt the practice. An extensive body
of literature exists about methods for on-farm experimentation (e.g. Stroud, 1993),
ex ante financial analysis (e.g. CIMMYT, 1988) and ex post analysis of agricultural
technology adoption (e.g. CIMMYT Economics Program, 1993), especially for
annual crops technology. But little is available on the assessment of adoption poten-
tial during that critical period when farmers are making adoption and design deci-
sions. Nor have these methods been widely adapted to the special needs of
agroforestry practices, where management is often more complex and the period of
farmer testing is much longer than in conventional cropping systems. Whereas a
farmer can harvest and evaluate a new crop variety after a few months, agroforestry
practices often take 3–6 years before benefits are realized. The approach illustrated
in this volume is promising for the evaluation of other sustainable agriculture and
natural resource management practices with similarly long gestation periods and
complex choices about components and management.
The book’s primary audience is development practitioners and researchers
working throughout the Third World who are interested in evaluating the perfor-
mance of agroforestry practices. The book will interest persons from a wide range of
disciplines in the social and biophysical sciences, involved in agricultural technology
development and the promotion of sustainable agricultural systems.

Conceptual Premises

The approach and methods described in the book are based upon four conceptual
premises. First, we assume that a systems approach is required in assessing adoption
potential. Rural households operate complex farming systems, allocating their lim-
ited resources among many enterprises in a manner determined by their priorities,
preferences and their biophysical and socioeconomic circumstances (Collinson,
1981; Scherr, 1997). This system complexity has several implications for researchers
and development practitioners.
To begin with, they must understand the farming system and work closely with
farmers to select and develop appropriate technologies. Since African farming sys-
tems are particularly heterogeneous, regarding both biophysical variables (e.g. soils)
and socioeconomic variables (e.g. market access), some degree of targeting of new
practices is usually required. At the same time, research has shown that when mov-
ing beyond simple interventions, such as varietal improvements, researchers and
 Introduction 5

extensionists need to present farmers with a ‘basket of options’ from which they can
choose the practices that are most suitable to them (Chambers et al., 1987).
Indeed, small-scale farmers are rarely able to manage any single enterprise in
the ‘optimal’ manner prescribed by researchers. Rather, they make compromises in
the management of individual enterprises in order to reduce risk, alleviate con-
straints and increase the productivity of the whole household livelihood system.
Thus, researchers and development practitioners must be willing to test low-cost,
simple-to-implement practices that improve on farmers’ existing practices. Some
complex technologies may have higher potential returns, but pose major risks and
are unlikely to be adopted until after other marginal changes have been introduced
successfully (Byerlee and Hesse de Polanco, 1986).
African farming systems are also quite dynamic, undergoing significant long-
term changes in crop components, commercialization, input use and cropping inten-
sity as populations grow and markets expand (Ruthenberg, 1971; Boserup, 1981).
Given the relatively long life cycle of many agroforestry systems, improved technolo-
gies must be appropriate for farming systems of the future, not just the present.
The second concept is that participatory research approaches are needed to
ensure that farmers play a leading role in problem diagnosis, testing and evaluation
of new practices (Chambers et al., 1987; Okali et al., 1994). New methods in the
late 1980s and 1990s helped researchers and other change agents to play a facilita-
tive role in the on-farm research process, helping farmers to carry out their own
appraisals, design their own experiments and conduct their own evaluations. This
volume, for example, highlights farmers’ innovations, in experiments they designed
and implemented themselves, that were later adopted by farmers on a wide scale.
This approach can thus help substantially to streamline and focus the research
process, rather than requiring that researchers predict on their own which of hun-
dreds of possible tree species and dozens of technologies deserve development and
dissemination under different farm conditions and different landscape niches.
The third concept highlighted in this book is that financial analysis and farmer
assessment are both needed to complement biophysical assessment of a technology.
Whereas financial analysis shows the economic attractiveness of a practice, farmer
assessment highlights its advantages and disadvantages as perceived by the farmer,
and is especially useful after farmers have had a chance to experiment with the prac-
tice. Farmer assessments may highlight the suitability of a product for use by a
farmer (e.g. as fuelwood or construction wood), or growth characteristics of a tree
that a farmer may like or dislike (e.g. interference with crop production or provi-
sion of a boundary marking), or socioeconomic constraints that inhibit the use of a
practice, such as unsuitable land tenure or lack of labour during the peak work
period.
Farmer assessment is especially important in agroforestry, since tree products
often have multiple uses; new practices frequently involve significant management
changes for the farmers, and the technology testing period is long. Also, in areas
where markets are weak and it is difficult to value inputs such as labour and out-
puts such as fuelwood, farmer assessment may be more important than financial
analysis for determining the advantage of a practice. Systematic analysis of farmers’
6 S. Franzel and S.J. Scherr

assessments of practices are rare; they have been considered by some researchers to
be too ‘soft’ and subjective. However, methods are available from sociology and
anthropology for quantifying data on farmers’ preferences, and subjecting
hypotheses about those preferences to tests of statistical inference. Examples of
these methods include hierarchical decision-tree modelling (Chapter 5) and matrix
rating (Chapter 6).
The fourth concept is that the interaction of actors made possible through the
on-farm research process itself can strengthen the technology development process,
and accelerate dissemination of new information, by compressing in time the con-
ventional sequential information transfer pattern. Benefits range from learning key
lessons about how to communicate ideas about a technology and market it to farm-
ers (i.e. simultaneous rather than sequential development of ‘extension messages’);
identifying technology design tips for getting around farmer constraints with the
technology (i.e. compressing the training of extensionists for effectively adapting
the technology to farmer conditions); and recruiting farmer ‘champions’ of the new
technology (i.e. getting an early start with farmer-to-farmer diffusion). On-farm
research provides a defined ‘space’ for interaction among researchers, extensionists,
farmers and community groups that would be unlikely to occur otherwise, particu-
larly when these belong to different social groups. The communication channels
opened up through on-farm research may make possible continued dialogue
between groups even after projects are completed.

Overview

This book is composed of three parts. Chapter 2 provides an overview of the methods
used in the studies. It defines the information needed to assess adoption potential,
and explains different techniques for collecting and analysing data on those variables,
including participatory on-farm trials, financial analysis and farmer assessment.
There follow five case studies assessing the adoption potential of specific agro-
forestry practices: hedgerow intercropping for improving soil fertility in western
Kenya, improved tree fallows for improving soil fertility in eastern Zambia and
western Kenya, multipurpose agroforestry trees in western Kenya, and fodder trees
for dairy cows in central Kenya (Table 1.1). The studies are from three contrasting
zones – a humid highland area in central Kenya located close to major markets, a
humid highland area in western Kenya with poorer soils and farther from markets,
and a subhumid plateau area in eastern Zambia far from major markets (Table 1.2).
The Kenyan study areas have very high population densities, while the density is
low in the Zambia study area.
The five case studies examine the feasibility, profitability and acceptability of
the practices, based on farmers’ experience in testing them. All cases involve
researcher-designed, farmer-managed trials. Four of the studies also include trials
designed by farmers, that is, in which farmers test practices as they wish. Four stud-
ies also include some assessment of farmers’ post-trial experiences, that is, farmers’
preferences and actions following completion of the trials. One case, for improved
 Introduction 7

Table 1.1. Agroforestry practices examined in this volume.

Practices and tree Main problem Important
species addressed by-products Management
Improved fallows Nitrogen Fuelwood Farmers raise seedlings
(southern Africa) deficiency (sesbania) in nursery
(Sesbania sesban and in soil or direct seed
Tephrosia vogelii ) (tephrosia). They plant
in a pure stand or plant
trees in a standing
maize crop. After 2 or
more years, trees are
cut and leaves incorpo-
rated into the soil
Improved fallows Nitrogen Fuelwood Farmers direct seed into
(eastern Africa) deficiency a standing maize
(Crotalaria spp. and in soil crop. They leave trees
Tephrosia vogelii ) to grow during the sec-
ond season and cut
trees and incorporate
leaves before the
beginning of the third
season
Hedgerow intercropping Nitrogen Fuelwood, Trees are planted in
(Leucaena deficiency fodder rows, about 30 cm
leucocephala in soil, soil apart within rows and
and Calliandra erosion 4 m between rows.
calothyrsus) Hedges are pruned at
about 1 m and prunings
are incorporated into
soil as green manure
Boundary planting of Fuelwood Boundary Trees are planted on
upper-storey trees shortage, markings, boundaries or around
(Grevillea robusta and lack of shade, the homestead
Casuarina poles and beauty
junghuhniana) timber for
construction
Fodder trees (Calliandra Fodder Fuelwood, Fodder shrubs are
calothyrsus) shortages soil erosion planted in hedges on
in dairy control, boundaries and along
enterprises boundary contours. Beginning in
markings the second year, they
are pruned at about
1 m and prunings are
fed to livestock
8 S. Franzel and S.J. Scherr

Table 1.2. Main characteristics of the study areas examined in this book.
Features of study areas Eastern Zambia Western Kenya Central Kenya
Agroforestry technologies Improved tree Improved tree Fodder trees
examined fallows fallows,
hedgerow
intercropping,
multipurpose
agroforestry
trees
Altitude (m) 900–1200 1500 1300–1800
Rainfall Unimodal, Bimodal, Bimodal,
1000 mm 1600–1800 mm 1200–1500 mm
Soil type Alfisols and Nitosols Nitosols
Luvisols
Population density (km−2) 25–40 300–1000 450–700
Access to markets Low Medium High
Main crops Maize, Maize, vegetables Coffee, maize,
groundnuts beans
Livestock types Zebu cattle, Zebu cattle, goats Improved dairy
goats cattle
Area cultivated (ha) 1.2–3.2 0.5–1.5 1–2
Sources: Chapters 3, 4 and 7.

fallows in Zambia, examines the role of an on-farm research and dissemination net-
work, composed of farmer groups, NGOs, and research and development organiza-
tions, in promoting the practice.
Finally, common themes emerging from the case studies are examined. Chapter
8 concerns the role of policies, institutional mechanisms and farmer incentives in
promoting agroforestry. In the concluding chapter, the results are set in the context
of existing information, the prospects for different practices are assessed, and rec-
ommendations are made concerning priorities for research and development work
in agroforestry.
Results and conclusions reported in this book about agroforestry potentials for
specific practices must be considered preliminary. Most on-farm experiments in
agroforestry did not begin until after 1990, some results are not yet definitive, and
only a limited number of agroecological zones and socioeconomic conditions are
considered. Also, this volume assesses agroforestry exclusively from the perspective
of the individual farmer. Relatively little attention is given to projected impacts on
the local economy or on environmental variables such as soil erosion or deforesta-
tion. Such analyses would require additional socioeconomic and biophysical data,
not yet available at most of these sites.
Still, these case studies demonstrate that agroforestry is already contributing to
the solution of some of Africa’s challenges for increasing rural incomes and food
security and mitigating land depletion. Some cases show significant social and
 Introduction 9

financial impacts. Moreover, the studies illustrate a new and dynamic approach to
generate effective and adoptable technologies, which accelerates the technology
development process itself, as well as the process of farmer adoption. The dynamic
partnerships between farmers, researchers and extensionists that are illustrated in
the case studies offer promise for developing and disseminating other sustainable
land-management practices.

References

Boserup, E. (1981) Population and Growth. Basil Dembner, S. (1991) Provisional data from the
Blackwell, Oxford. Forest Resources Assessment 1990. Unasylva
Buresh, R., Sanchez, P.A. and Calhoun, F. (eds) 164(42), 40–44.
(1997) Replenishing Soil Fertility in Africa. Gittinger, J.P. (1982) Economic Analysis of
SSSA Special Publication Number 51. Soil Agricultural Projects. Johns Hopkins
Science Society of America and American University Press, Baltimore, Maryland.
Society of Agronomy, Madison, Wisconsin. Kang, B.T. (1993) Alley cropping: past achieve-
Byerlee, D. and Hesse de Polanco, E. (1986) ments and future directions. Agroforestry
Farmers’ stepwise adoption of technological Systems 23, 141–155.
packages: evidence from the Mexican alti- Kerkhof, P., Foley, G. and Barnard, G. (1990)
plano. American Journal of Agricultural Agroforestry in Africa: a Survey of Project
Economics 68, 519–527. Experience. Panos, London.
Chambers, R., Pacey R.A. and Thrupp, L. (1987) Leakey, R. (1996) Definition of agroforestry revis-
Farmer First: Farmer Innovation and ited. Agroforestry Today 8(1), 5–7.
Agricultural Research. Intermediate Le Houerou, H.N. (1987) Indigenous shrubs and
Technology Publications, London. trees in the silvopastoral systems of Africa.
CIMMYT (1988) From Agronomic Data to In: Steppler, H.A. and Nair, P.K.R. (eds)
Agroforestry, a Decade of Development.
Farmer Recommendations: an Economics
ICRAF, Nairobi.
Training Manual. CIMMYT, Mexico.
Okali, C., Sumberg, J. and Farrington, J. (1994)
CIMMYT Economics Program (1993) The
Farmer Participatory Research. Rhetoric and
Adoption of Agricultural Technology: a Guide
Reality. Intermediate Technology
for Survey Design. CIMMYT, Mexico.
Publications, London.
Cleaver, K.M. and Schreiber, G.A. (1994)
Ruthenburg, H. (1971) Farming Systems in the
Reversing the Spiral; the Population,
Tropics. Clarendon Press, Oxford.
Agriculture, and Environment Nexus in Sub- Sanchez, P.A. (1996) Science in agroforestry.
Saharan Africa. World Bank, Washington, Agroforestry Systems 9, 259–274.
DC. Scherr, S.J. (1997) Meeting household needs:
Collinson, M.P. (1981) A low-cost approach to farmer tree-growing strategies in western
understanding small farmers. Agricultural Kenya. In: Arnold, J.E.M. and Dewees, P.
Administration 8, 433–450. (eds) Farms, Trees and Farmers: Responses to
Cooper, P.J., Leakey, R.R.B., Rao, M.R. and Agricultural Intensification. Earthscan
Reynolds, L. (1996) Agroforestry and the Publications, London.
mitigation of land degradation in the Scherr, S.J. and Müller, E. (1990) Evaluating
humid and sub-humid tropics of Africa. agroforestry interventions in extension pro-
Experimental Agriculture 32, 235–290. jects. Agroforestry Systems 11(3), 259–280.
Current, D., Lutz, E. and Scherr, S.J. (1995) Scherr, S.J. and Müller, E.U. (1991) Technology
Costs, Benefits and Farmer Adoption of impact evaluation in agroforestry projects.
Agroforestry: Project Experience in Central Agroforestry Systems 13, 235–257.
America and the Caribbean. World Bank Scherr, S.J. and Yadav, S. (1995) Land
Environment Paper No. 14. The World Degradation in the Developing World:
Bank, Washington, DC. Implications for Food, Agriculture, and the
10 S. Franzel and S.J. Scherr

Environment to 2020. Food, Agriculture Warner, K. (1993) Patterns of Farmer Tree-

and the Environment Discussion Paper 14. Growing in Eastern Africa: A Socioeconomic
International Food Policy Research Analysis. Oxford Forestry Institute and
Institute, Washington, DC. International Centre for Research in
Stroud, A. (1993) Conducting On-Farm Agroforestry, Oxford.
Experiments. CIAT, Cali, Colombia. Whittome, M.P.B. (1994) The adoption of alley-
Sullivan, G.M., Huke, S.M. and Fox, J.M. (eds) farming in Nigeria and Benin: the on-farm
(1992) Financial and Economic Analyses of experience of IITA and ILCA. PhD thesis,
Agroforestry Systems. Nitrogen Fixing Tree Department of Geography, University of
Association, Paia, Hawaii, 10–14 July Cambridge, Cambridge, UK.
1991. NFTA, Paia, Hawaii. World Bank (1994) Global Economic Prospects
Swinkels, R.A. and Scherr, S.J. (eds) (1991) and the Developing Countries. World Bank,
Economic Analysis of Agroforestry Washington, DC.
Technologies: An Annotated Bibliography. World Bank (1996) Development Indicators.
ICRAF, Nairobi. World Bank, Washington, DC.
 Methods for Assessing
Agroforestry Adoption
Potential
2
S. FRANZEL,1 S.J. SCHERR,2 R. COE,1
P.J.M. COOPER3 AND F. PLACE1
1International Centre for Research in Agroforestry, ICRAF House,

PO Box 30677, Gigiri, Nairobi, Kenya; 2Agricultural and Resource

Economics Department, University of Maryland, College Park,
MD 20742, USA; 3International Development Research Centre,
PO Box 8500, Ottawa, ON, Canada KIG 3H9

Introduction
This chapter describes methods for assessing the adoption potential of agroforestry
practices through participatory on-farm trials, and their application in the five
Kenya and Zambia case studies. We discuss briefly the evolution of adoption assess-
ments through on-farm research. We then define three types of on-farm trials and
discuss their main features, management and suitability for generating information
on adoption potential. The next section discusses the organization of participatory
on-farm trials; the following section presents specific methods for determining the
biophysical performance, profitability, feasibility/acceptability, boundary conditions
and insights about dissemination of practices tested in on-farm trials. The chapter
then discusses how this on-farm research approach fits into a new farmer-centred
model of the research–development continuum, and concludes with some thoughts
on future priorities.

Background
In conventional approaches to technology generation in the 1960s, assessment of
adoption potential focused almost exclusively on biophysical variables such as a new
crop variety’s potential to increase yield per hectare. Where technologies were fairly
simple and biophysical circumstances fairly homogeneous, as for rice varieties in the
irrigated areas of South-East Asia, the approach achieved considerable success. But
in Africa, where farming systems were often more complex, more subsistence-ori-
ented and more variable than in the irrigated areas of South-East Asia, the biophysi-

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 11
12 S. Franzel et al.

cal approach was found wanting. In the late 1970s and early 1980s, farming sys-
tems research emphasized the need to determine adoption potential based on the
priorities and circumstances of farmers (Byerlee and Collinson, 1980). Researchers
emphasized the need for testing new practices under farmers’ circumstances but
research prototypes for on-farm trials still tended to be drawn up by researchers, fol-
lowing consultation with farmers (Zandstra et al., 1981). Participatory approaches
in the late 1980s and 1990s highlighted empowering farmers to choose the tech-
nologies they wanted to test and to design and implement the research themselves
(Lightfoot, 1987; Chambers et al., 1989; Haverkort et al., 1991; Rocheleau, 1991;
Scherr, 1991a). Little emphasis was given to statistical analysis or to extrapolating
results to areas beyond the location where the research was conducted.
In the 1980s, there was considerable experimentation with adapting both
researcher-led and participatory on-farm research methods to agroforestry (Scherr,
1991b,c). During the 1990s the International Centre for Research in Agroforestry
(ICRAF) and other organizations devoted much effort to the design and testing of
methods for on-farm research for different types of practices, and with an explicit
view to understanding adoption potential. As ICRAF researchers gained experience
in on-farm research and technology development in agroforestry, their approach has
been refined. Assessment of adoption potential was observed to be multifaceted,
requiring an understanding of biophysical performance under farmers’ conditions,
profitability from the farmers’ perspective and its acceptability to farmers (in terms
of both their assessment of its value and their willingness and capacity to access the
information and resources necessary to manage it well). On-farm research should
make it possible to define the ‘boundary conditions’ for a particular practice, that is,
the biophysical and socioeconomic circumstances under which the practice is likely
to be profitable, feasible and acceptable to farmers, and thus adopted by them.1
Furthermore, it was realized that on-farm research offers researchers, extensionists,
policy makers and farmers themselves an opportunity to learn important lessons
about achieving effective dissemination of agroforestry practices, as well as feedback
on further research priorities (Table 2.1).
The evolving approach outlined in this chapter includes elements of various
on-farm research and adoption assessment approaches. During participatory
appraisals and surveys, researchers and farmers identify farmers’ problems and needs
and select practices to test in on-farm trials, some of which may be researcher-
designed and some of which may be farmer-designed. These trials, and the analyses
that researchers and farmers conduct together, form the basis for determining
whether farmers on a wider scale will adopt the practices.

1 The concept ‘boundary conditions’ complements the term ‘recommendation domain’,

commonly used in farming systems research. A recommendation domain is a roughly
homogeneous group of farmers with similar circumstances for whom we can make more
or less the same recommendation (Byerlee and Collinson, 1980). Recommendation
domains are defined in the early stages of the research process and technologies are sought
that are appropriate for them. Once a technology is found that benefits farmers at
particular sites in a recommendation domain, it is useful to try to assess that technology’s
boundary conditions.
 Assessing Agroforestry Adoption Potential 13

Table 2.1. Framework for assessing the adoption potential of an agroforestry

practice.
Factors Key questions
Biophysical performance Does the practice result in higher yields, lower vari-
ability and provide the anticipated environmental ser-
vices? Are these biophysically sustainable?
Profitability Is the practice profitable to the farmer as compared
with alternative practices? How variable are returns,
and how sensitive to changes in key parameters?
Feasibility and Do farmers have the required information and
acceptability resources, and are they willing and able to establish
and manage the practice and cope with problems
that occur? Do farmers perceive significant advan-
tages of using the technology?
Boundary conditions Under what circumstances (e.g. biophysical, house-
hold, community characteristics, market conditions)
is the practice likely to be profitable, feasible and
acceptable to farmers?
Lessons for effective What does farmer feedback suggest will help inter-
dissemination: est farmers in the practice? What type of extension
● extension support do they need most? What types of
● policy changes in institutional arrangements, public invest-
ments or market conditions would enhance the
adoption potential of the practice?
Feedback to research How do farmers modify the practice? What does
and extension farmer experience suggest are research priorities for
further modification and development of the
practice?

On-farm research has been seen as especially critical in agroforestry technology

development. This is due to the poor understanding of farmers’ agroforestry strate-
gies, lack of empirical information about on-farm agroforestry practices, agro-
forestry system complexity and variability (in terms of objectives, components,
management and ecological interactions), the longer technology cycle and period
required for farmer and researcher assessment and the lack of scientifically validated
technologies (Scherr, 1991a). Many of these same features characterize other ‘sus-
tainable agriculture’ and natural resource management practices and research chal-
lenges, with their greater spatial and temporal complexity. Thus many of the lessons
learned from on-farm agroforestry research should be highly relevant to broader
efforts in sustainable agricultural development.

Objectives of On-farm Experimentation

On-farm experimentation has several different objectives. First, it permits farmers

and researchers to work as partners in the technology development process. The
14 S. Franzel et al.

more, and the earlier, that farmers are involved in the technology development
process, the greater the probability that the practice will be adopted. On-farm trials
are important for getting farmers’ assessment of a practice, their ideas on how it
may be modified and for observing their innovations. Assessments are likely to vary
and may be associated with particular biophysical (e.g. soil type) or socioeconomic
(e.g. wealth status) circumstances. Farmers’ innovations often serve as a basis for
new research or for modifying recommendations (Stroud, 1993; van Veldhuizen et
al., 1997).
Secondly, on-farm testing is useful for evaluating the biophysical performance
of a practice under a wider range of conditions than is available on station. This is
especially important because soil type, flora and fauna on research stations are often
not representative of those found on farms in the surrounding community.
Thirdly, on-farm trials are important for obtaining realistic input–output data
for financial analysis. Financial analyses conducted on on-station experiments are
unreliable because yield response is often biased upward, because estimates of
labour use by station labourers on small plots are unrepresentative, and because
operations often differ, as when tractors instead of oxen or hoes are used for prepar-
ing land.
Fourthly, on-farm testing provides important diagnostic information about
farmers’ problems. Even if diagnostic surveys and appraisals have already been con-
ducted, researchers can still learn a great deal about farmers’ problems, preferences
and livelihood strategies from interacting with them in on-farm trials. Trials have
important advantages over surveys in that they are based on what farmers do rather
than on what they say.

Types of On-farm Trials

On-farm trials can thus provide critical information for determining the biophysical
performance, profitability and acceptability of agroforestry, i.e. adoption potential.
However, the design of a trial depends on its specific objectives.
Assessment of biophysical performance requires biophysical data on the prod-
ucts and services that the technology is planned to produce. These are likely to
change with different adaptations of the technology as might occur if farmers were
asked to manage them. To prevent such possible variation, trials designed to assess
biophysical performance should be controlled in order to replicate specific technol-
ogy designs. The trials should also be implemented in a way that farmers’ willing-
ness and ability to establish and maintain the trials does not affect the outcome.
Thus trials to assess biophysical performance need a high degree of researcher con-
trol in both design and implementation.
The assessment of profitability requires biophysical data (to estimate returns),
that must be generated from standardized experiments. However, the financial
analysis also requires realistic input estimates, of which labour poses most difficul-
ties. Realistic data can only be obtained if farmers manage the trials to their own
standards. Thus profitability objectives require trials in which researchers have con-
 Assessing Agroforestry Adoption Potential 15

siderable input into the design but farmers are responsible for implementation. The
objectives of assessing feasibility and acceptability require data on farmers’ assess-
ments and adaptations of the technology. These can only be assessed if farmers are
left to experiment with little researcher involvement.
There are many different ways of classifying on-farm trials (Okali et al., 1994).
The differing requirements of the objectives of biophysical performance, profitability
and acceptability mean it is helpful to classify trials according to the balance of
researcher and farmer involvement in their design and implementation. The classifi-
cation used in this volume involves three types of trials and draws upon Biggs (1989).

Type 1, trials designed and managed by researchers

These trials are simply on-station trials transferred to farmers’ fields. They are useful
for evaluating biophysical performance under farmers’ conditions and require the
same design rigour as on-station research with regard to treatment and control
choice, plot size, replication and statistical design. In the design stage, researchers
need to consult the farmer on the site’s homogeneity and history. If possible, they
should observe a crop in the field before establishing a trial.
Because type 1 trials take place on farmers’ fields, trial results are generally
more representative of farmers’ biophysical conditions than are on-station trials
(Shepherd et al., 1994). More accurate information may be obtained on interac-
tions between the biophysical environment and management; for example, how dif-
ferent species in an improved fallow trial compare on different soil types.
Type 1 trials are usually more expensive and more difficult to manage than on-
station trials; they often involve renting land from farmers and bringing labourers
from the station to implement them. Farmers’ assessments are an important objec-
tive of type 1 trials; as with on-station trials, it is useful to get farmers’ feedback on
the different treatments (Sperling et al., 1993; Franzel et al., 1995).
Type 1 trials were conducted on three of the technologies reported in this vol-
ume. The numbers of farmers per trial was small, 1–5, because of the relatively high
cost of conducting these trials. For the other technologies, on-station trials provided
technical information before starting farmer-managed trials.

Type 2, trials designed by researchers but managed by farmers

Here, farmers and researchers collaborate in the design and implementation of the
trial. The trial is labelled ‘researcher-designed’, because it follows the conventional
scientific approach to conducting an experiment: one or more test treatments are
laid out in adjacent plots and compared to one or more control treatments.
Researchers consult farmers on the design of the trial and each farmer agrees to fol-
low the same prototype (or chooses one of several possible prototypes), so that
results may be compared across farms. Farmers are responsible for conducting all of
the operations in the trial.
16 S. Franzel et al.

In type 2 trials, reliable biophysical data over a broad range of farm types and
circumstances are sought. The trials also facilitate the analysis of costs and returns;
inputs, such as labour, and outputs, such as crop yields, are relatively easy to mea-
sure because plot size is uniform and known. The trials are also useful for assessing
farmers’ reactions to a specific practice and its suitability to their circumstances.
Farmers are encouraged to visit each other’s trials and to conduct group field days to
assess the practice at different stages of growth.
Type 2 trials were conducted in four of the five case studies reported in this
book. In most cases, the number of farmers per type 2 trial ranged from 20 to 50.
Trials usually started off small, with fewer than 10 farmers, in order to learn from
experience. They were then modified and expanded in the following year.

Type 3, trials designed and managed by farmers

In type 3 trials, farmers are briefed about new practices through visits to field sta-
tions or on-farm trials. They then plant and experiment with the new practices as
they wish. They are not obliged to plant in plots or to include control plots.
Researchers monitor the farmers’ experiments, or a subsample of them, focusing in
particular on their assessment of the new practice and their innovations. In addi-
tion, farmer-to-farmer visits and meetings are useful so that farmers can compare
their experiences and assessments with others. Any farmers experimenting with a
new practice could be said to have a type 3 trial, regardless of whether they obtained
planting material and information from researchers, other facilitators or other farm-
ers. This ‘hands-off ’ approach, which assumes that farmers know best how to test a
new practice on their own farms, is supported by some in the literature (Lightfoot,
1987). Others emphasize training farmers to conduct trials following scientific prin-
ciples, such as replication and non-confounding of treatments (Ashby et al., 1995).
Three of the case studies in this book involve the use of type 3 trials. The num-
ber of type 3 trials was often quite high. In eastern Zambia in 1997, extension ser-
vices and NGOs were helping about 2800 farmers to test improved tree fallows
(Chapter 3). Researchers coordinated the monitoring of small samples of farmers,
60–110, depending on the task and available staff.

Suitability of Trial Types for Meeting Objectives

The suitability of the different trial types for differing objectives is summarized in
Table 2.2. Suitability involves both the appropriateness of the trial for collecting
the information and the ease with which it can be collected. Different types of tri-
als are suited to different types of analyses. Biophysical measurements are most
meaningful in type 1 and 2 trials; they are less useful in type 3 trials because each
farmer may manage the practice in a different manner. Type 2 trials are well suited
for collecting parameters (e.g. labour use) for financial analysis; such data are diffi-
cult to collect in type 3 trials because plot size and management vary. The data can
 Assessing Agroforestry Adoption Potential 17

Table 2.2. The suitability of type 1, 2 and 3 trials for meeting specific
objectives.a
Type 1 trial: Type 2 trial: Type 3 trial:
researcher- researcher- farmer-
designed, designed, designed,
researcher- farmer- farmer-
Information types managed managed managed
Biophysical response H M L
Profitability L H L
Acceptability
Feasibility L M H
Farmers’ assessment of a L H M
particular prototypeb
Farmers’ assessment of a L M H
practiceb
Other
Identifying farmer innovations 0 L H
Determining boundary conditions H H H
aH, high; M, medium or variable; L, low; 0, none. The suitability involves both the
appropriateness of the trial for collecting the information and the ease with which
the information can be collected.
bBy particular prototype, we mean a practice that is carefully defined. For example,

a prototype of improved fallows would include specific management options

such as species, time of planting, spacing, etc.

be collected in type 1 trials but will be less relevant to farmer circumstances; yield
response to new practices tends to be biased upward and labour use, measured
using labourers hired by researchers and working on small plots, is unrepresenta-
tive of farmers’ labour use.
Farmers’ assessments are more accurate in type 3 trials for several reasons.
Because farmers control the experimental process, they are likely to have more inter-
est and information about the practice. Furthermore, because farmers in type 3 tri-
als usually have less contact with researchers than farmers in other types of trials,
their views of a technology are less influenced by researchers’ views. Finally, whereas
it is often necessary to provide inputs to farmers in type 2 trials to ensure that
results are comparable across farmers, no inputs, with the possible exception of
planting material, are provided in type 3 trials. Thus farmers’ views in type 3 trials
are more likely to be sincere than in type 2 trials, where positive assessments may
simply reflect the farmers’ interest and satisfaction in obtaining free inputs. For
example, in the hedgerow intercropping trial in western Kenya (Chapter 5), 50% of
the farmers claimed that hedges increased crop yields whereas technicians noted
yield increases on only 30% of the farms; the technicians claimed that the differ-
ence was due to farmers trying to please researchers.
Finally, all three types of trials play a potentially important role in defining the
boundary conditions for the technology. Which type of trial is best depends on the
participants’ (facilitators’ and farmers’) objectives and the particular circumstances.
18 S. Franzel et al.

Continuum and sequencing of trial types

The different types are not strictly defined; rather they are best seen as points along
a continuum. For example, it is common for a trial to fit somewhere between type 2
and type 3, as in the case where farmers agree to test a specific protocol (type 2) but
over time, individuals modify their management of the trial (type 3). For example,
in the hedgerow intercropping trial in Chapter 5, farmers planted trials in a similar
manner but most later modified such variables as the intercrop, pruning height and
pruning frequency.
The types of trials are not necessarily undertaken sequentially; researchers and
farmers may decide to begin with a type 3 trial, or to simultaneously conduct two
types of trials. For example, in the case of upper-storey tree trials in western Kenya
(Chapter 6), no type 1 or type 2 trials were needed, because much was already
known about the growth of the trees in the area. Rather, farmers planted type 3 tri-
als, in order to assess the performance of the species on their farms. In Zambia,
many farmers planted type 2 and type 3 improved fallow trials in the same year
(Chapter 3). They tested a prototype in their type 2 trials and used type 3 trials
either to extend their plantings or to test a modification of the practice. Researchers
wished to assess biophysical response in the type 2 trials and to monitor farmers’
innovations in the type 3 trials. Type 2 and 3 trials often generate questions or
sharpen hypotheses about biophysical factors which can then be best evaluated
through type 1 on-farm or on-station trials. In western Kenya, several researcher-
managed trials to explore specific aspects of improved fallow function and design
were set up following farmer-managed trials (Chapter 4).

Handling complexity

‘Complexity’ involves the number and diversity of components (intercropping trees

and crops, as opposed to trees or crops in pure stand), the length of the cycle of the
technology (3+ seasons as opposed to single-season cycles), and the size of the trial
(whether it takes up more than 10% of a farmers’ cultivated area). In a trial com-
paring annual crop varieties, it is often possible to combine biophysical and socioe-
conomic objectives because, according to the above definition, the trial is not
complex. However, most agroforestry trials are complex, and thus different trial
types are needed to meet the differing objectives.
The type 1–2–3 classification system is useful for highlighting the different
objectives for conducting on-farm trials and for illustrating that different types of
trials are suitable for particular types of assessments (Table 2.2). For example,
researchers often want to use an on-farm trial to collect information on biophysi-
cal responses and farmer assessment. But these objectives are often conflicting. A
high degree of control is needed to collect accurate biophysical data whereas
farmer assessment is most valid when individual farmers are allowed to use the
practice in the manner they see fit. Researchers and farmers interested in biophysi-
cal and socioeconomic data may be better off conducting type 1 trials for biophys-
 Assessing Agroforestry Adoption Potential 19

ical data and type 3 trials for socioeconomic assessment, rather than a single type
2 trial that tries to do both. The more complex the trial or technology, the less
effective a type 2 approach is likely to be for both biophysical and socioeconomic
assessments.

Organization of Participatory On-farm Trials

Key aspects of the organization of participatory on-farm trials (principally types 2

and 3) discussed in this section include: farmer and technology selection, the village
approach to technology testing, trial backstopping and supervision, adaptive
research and dissemination teams, and provision of incentives.

Farmer and technology selection

The selection of farmers for the trials reported in this volume generally took place
through the assistance of extension staff (eastern Zambia) or farmer groups (western
Kenya). Interested farmers were asked to volunteer and participants were selected so
as to represent a range of the types of different farmers in the area, including large
and small farmers (all sites), male and female farmers (all sites), and farmers prepar-
ing land with oxen and hoe (eastern Zambia).
Providing farmers with different options to test was a key feature of the trials,
for several reasons. Different farmers had different circumstances and preferences;
farmers wanted to diversify; and any single option could have failed. For example,
in eastern Zambia, farmers selected among six improved fallow practices for their
type 2 trials (Chapter 3). In type 3 multipurpose tree trials in western Kenya, farm-
ers chose among five tree species but were encouraged to plant all five so as to per-
mit a comparison among them (Chapter 6).
To ensure appropriate analysis, interpretation and extrapolation of results, it is
important to specify or characterize the technology design in detail. Key design ele-
ments considered in the case studies include: objectives (priority outputs or envi-
ronmental services expected), site characteristics (location in the landscape, soil
type, field size), components (tree crop/provenance/variety; crop species/variety;
livestock species/breed), method of tree establishment (and re-establishment or cop-
picing), spacing and sequencing of components (including rotations), and tree and
crop management practices.

Village approach to technology testing

A common approach to on-farm technology testing has been to identify a relatively

small number of farmers in many different villages across a large area who are will-
ing to undertake experiments. This ‘scattered farmer’ approach can be useful when
key socioeconomic or biophysical factors such as farm size or soil type vary mainly
20 S. Franzel et al.

across villages. In the mid-1990s, ICRAF researchers began to experiment with the
‘village approach’ to technology testing, that concentrates efforts in a relatively few
contrasting but representative sites. The key feature of the approach is that all vil-
lagers are given equal access to information and germplasm, encouraging wider par-
ticipation. As such, it is most appropriate for type 2 or 3 trials. The advantages of
the village approach are:
● a reduction in monitoring costs per farmer through higher concentration of
farmers;
● a wider participation ensures that different household types are involved in
testing and development;
● the possibility to study inter-farm linkages and higher-scale effects (e.g. pest
and disease outbreak, income from labour hiring) which require identification
prior to wide dissemination;
● the mitigation of intra-village jealousies and the promotion of improved inter-
action with researchers;
● the involvement of village-based organizations, such as farmer groups, in the
testing and dissemination process; and
● the spread of information across farms within the village is facilitated.
There is one disadvantage to the village approach: the more or less equal distribu-
tion of information and high participation rates make the study of diffusion
processes within the village more difficult.
With a limited number of villages, it is important that they represent a large
percentage of the target population (e.g. representing major agroecological zones).
With the large number of households testing technologies, researchers need to
select a smaller number of households for detailed monitoring. These should repre-
sent different types of households, for example, poor, moderate and well-off house-
holds; male- and female-headed households; and those using oxen and hoe for
land preparation. With concentration of testing in a relatively small number of
sites, it is important to establish a communications strategy as part of the research
and for diffusion. Effective communication is needed within the community,
between researchers and the community, and, most important, from the village to
other villages.

Trial supervision and backstopping

At all sites, facilitators provided technical backstopping to farmers in types 2 and 3

trials, and helped lay out type 2 trials. In eastern Zambia, facilitators included
researchers, extensionists and NGO staff. In central Kenya, researchers and exten-
sionists were involved; and in western Kenya, until the mid-1990s, only researchers
were involved. Facilitators also conducted the monitoring surveys during and fol-
lowing the trials. In most cases research technicians collected information; extension
staff often participated (central Kenya and Zambia), or enumerators were hired
(western Kenya).
 Assessing Agroforestry Adoption Potential 21

Adaptive research and dissemination teams

Participatory on-farm experimentation and dissemination are better viewed as

points along a research–development continuum than as separate activities. After
all, when farmers participate in on-farm trials some degree of technology dissemina-
tion always takes place. Similarly, when a new technology is disseminated, each
farmer trying the technology for the first time can be said to be experimenting with
it. Therefore, at all case study sites, research, extension, NGOs and farmer groups
have established partnerships called ‘adaptive research and dissemination teams’
(ICRAF, 1997). The teams plan, implement and evaluate on-farm research, training
and dissemination activities (Cooper, 1999). Extension and NGO staff have much
to offer and can benefit greatly from participating in on-farm research. Their
involvement reduces the costs of the research and their knowledge of local circum-
stances improves the design and quality of the research. They also benefit from
greater interaction with researchers and are likely to be more knowledgeable about a
practice if they are involved with farmers in its development in on-farm trials.
Similarly, it is important for researchers to be involved in disseminating practices, in
order to assess adoption and impact, to obtain feedback from farmers and to iden-
tify issues for further research. One of the most important impacts of the teams is
that all partners develop a sense of involvement, enthusiasm and ownership of
promising innovations. A critical task of the team is to define clearly the roles and
responsibilities of the different actors in on-farm research and dissemination.
The most developed adaptive research and dissemination team discussed in this
volume is in eastern Zambia (Chapter 3). Seventy-five representatives of research,
extension, NGOs and farmer groups meet 1–2 times per year to review progress
and plan activities for testing and disseminating improved fallows and other soil fer-
tility measures. The number of farmer-experimenters has increased from fewer than
10 in 1993–1994 to about 200 in 1994–1995, and to about 3000 in 1997–1998.
The mechanisms in place for disseminating the practice and for providing feedback
on performance will contribute greatly to the impact that improved fallows achieve
in Zambia and elsewhere.

Provision of incentives

Farmers did not receive any financial or material incentives for participating in the
trials. Farmers conduct their own research; there should thus not be any need to
provide them with incentives to participate in collaborative trials with researchers.
Moreover, giving incentives has four important disadvantages. It promotes a depen-
dency relationship between facilitator and farmer, instead of a partnership; it sets an
unfortunate precedent for other facilitators who come later and do not have the
resources for offering incentives; it may create conflict, because in most programmes
incentives cannot be offered to all farmers; and it biases farmers’ assessments, as
they may tend to give more positive opinions in the hopes of obtaining more free
inputs.
22 S. Franzel et al.

However, most practitioners would agree that it is necessary, and even ethical,
to provide free of charge the experimental variable, such as seeds or seedlings for a
new tree species being tested. Unfortunately, this may bias farmers’ assessment.
Some practitioners have also found that farmers value their visits as a social benefit
and may give positive assessments merely to encourage them to continue their visits
(Chapter 5). Where the farmer’s role expands far beyond that of a trial participant,
to active involvement in dissemination, demonstration, etc., which demands signifi-
cant time unrelated to on-farm experimentation, it may be appropriate to treat and
remunerate such people as staff or consultants. Such arrangements are especially rel-
evant where there are no extension agents.

Assessing Adoption Potential

This section focuses on methods for assessing biophysical performance, profitability,

acceptability and boundary conditions. The use of on-farm trials to draw lessons
for extension, policy and research is also described. Criteria for selecting which
methods to use are also reviewed. The methods used in each case study for assessing
profitability and acceptability are shown in Table 2.3.

Biophysical performance

Assessing the biophysical performance of a technology on-farm uses much the same
general methods as used in analysis of on-station trials (see, for example, Little and
Hills, 1972; Mutsaers et al., 1997). The products and services of the technologies
are measured and compared among different options. Great care must be taken in
assessing whether all sites should be included in a particular analysis, especially
when data are from trials managed by farmers. For example, in an analysis of an
improved fallows trial, it would be important to exclude sites where a trial plot was
shaded by adjacent shrubs, but the control plot was not.
Often the objectives in on-farm assessment include identification of environ-
ment (e.g. soil type) by technology interactions. Analysis then requires determining
how the differences between technology options change with changing environ-
ment (Hildebrand and Russell, 1996). It is therefore necessary to measure the key
environment variables. These key variables must be identified before the trial starts,
as many (e.g. rainfall or frost incidence) cannot be measured after the trial has
ended.
The most difficult issue in assessing biophysical performance is ensuring that
the comparisons being made are representative of those that farmers would make.
For example, on-station hedgerow intercropping trials showed a strong fertility
response to the technology; type 2, on-farm trials did not, probably because farmers
were unable to prune the hedges in time to avoid competition between the trees
and adjacent crops (Chapter 5). On the other hand, yield responses obtained in
type 1 improved fallows trials in Zambia were only 20% higher than those obtained
Table 2.3. Types of assessments conducted in analyses of the profitability and acceptability of selected agroforestry practices.
Profitability Acceptability
Break- Enter- Evaluation Survey of Risk Farmer
even Partial prise Farm Resource of quality farmer assess- Monitoring Matrix assessment Decision
Practices analysis budget budget model budget of practice problems ment expansion ranking survey trees
Improved fallows,
eastern Zambia * * * * * * * * *
Improved fallows,
western Kenya * * * * * * * *
Hedgerow inter-
cropping, western
Kenya * * * * * * * * *
Agroforestry trees,
western Kenya * * * * *
Fodder trees,
Assessing Agroforestry Adoption Potential

central Kenya * * * * * * * *
23
24 S. Franzel et al.

in type 2 and type 3 trials, indicating that the type 1 trials were probably not man-
aged much differently than the type 2 and type 3 trials (Kwesiga et al., 1999). In
type 1 and 2 trials, researchers need to ensure that the site, treatments compared,
and trial management are each similar to those of the farmers for whom the practice
is targeted.
Long-term monitoring of trials is likely to be required to assess the biophysical
sustainability of different practices, in terms of maintaining soil quality, managing
pests and diseases, etc. Assessing the sustainability of a practice involves identifying
key elements that will be needed over the long term, at least 20 years, to ensure that
the practice will remain feasible, profitable and acceptable to farmers. For example,
high-yielding maize following improved fallows in eastern Zambia will draw down
stocks of soil phosphorus. Thus it is likely that over the long term, improved fallows
will have to be supplemented by phosphorus fertilizer in order to sustain high maize
yields. The sustainability of an agroforestry practice may involve a range of different
variables, such as seed production and distribution, soil nutrient balance, and pest
and disease management.

Profitability

Profitability issues can usefully be divided into three categories. The first concerns
whether the financial net benefits of the new practice are greater than for alternative
practices, including those that farmers currently use. Secondly, it is important to
assess the variability of benefits across farmers and seasons and the sensitivity of the
results to changes in key parameters. Thirdly, benefits are appraised relative to total
household income in order to assess their potential for contributing to improved
household welfare.
Greater financial benefits may arise through increased biophysical productivity
or through reduced input costs. Biophysical productivity and financial net benefits
were assessed in the case studies by comparing results on treatment plots with those
on control plots, which represented farmers’ current practices. For example,
researchers assessed the impact of hedgerow intercropping by comparing crop yields
and net benefits on hedge plots with those on plots without hedges (Chapter 5).
Where it was not possible to assess productivity responses, as in the case study of
improved fallows in western Kenya (Chapter 4), the yield increases required to
break even, that is, to cover the costs of planting and maintaining the practice
under different assumptions, were calculated. Break-even analysis was also con-
ducted in three other case studies to show the minimum returns needed to cover
the costs of establishing and maintaining the practice (Chapters 3, 5 and 7). Break-
even analysis thus provides useful information about profitability long before the
yield response of a practice is known.
In all cases, financial analyses were based on the actual costs and returns that
farmers face. Partial budgets were drawn up for those practices that had limited
impacts on the costs and returns of an enterprise, as in the case of Calliandra
calothyrsus as a fodder for dairy cows (Chapter 7). A partial budget is a technique
 Assessing Agroforestry Adoption Potential 25

for assessing the benefits and costs of a practice relative to not using the practice. It
thus takes into account only those changes in costs and returns that result directly
from using the new practice (Upton, 1987).
Where a practice was expected to have substantial financial effects, enterprise
budgets were used. Here, all of the enterprises’ costs and returns for a single period,
a growing season or a year, were calculated. Prices were collected from local markets.
Detailed information on labour use among participating farm households was col-
lected using a range of methods, including farmers’ recall just after a task was com-
pleted and monitoring of work rates through observation (Franzel, 1996).
Collecting data on labour use in on-farm trials is problematic: monitoring farmers
while they are in the field introduces an important ‘observer’ bias, while asking
farmers to recall how much time they spent at a specific task is often unreliable. We
found that interviewing farmers in the field just after a task was completed was the
best way to collect data on labour use. Moreover, we found that only researchers
could collect accurate labour data; the task was too complex to be relegated to enu-
merators. Finally, not all farmers could provide accurate data on labour used; the
sample must be purposive based on the confidence that the researcher has in the
quality of data being provided.
Net returns to farmers’ production factors (land, labour and capital) are calcu-
lated by subtracting purchased inputs from the production value. After subtracting
farmers’ capital inputs, which are generally minor, the net returns are allocated
among farmers’ land and labour by valuing one factor at its opportunity cost and by
attributing the remainder to the other factor. This permits a calculation of the net
returns to land, which is relevant for farmers whose most scarce resource is land,
and the net returns to labour, relevant for those who lack labour. Net returns to
capital for agroforestry practices were often extremely high or infinite because little
or no capital was used in implementing them. This finding explained the attractive-
ness of many of the options as the alternatives, such as fertilizer to improve crop
yields or dairy meal concentrate to increase milk yields, were very expensive for
farmers.
Data for a single period are usually inadequate for evaluating agroforestry.
Therefore, cost–benefit analyses, also called investment appraisals (Upton, 1987),
were developed in most of the case studies for estimating resource inputs, costs and
benefits over the lifetime of the investments (2–6 years). Average values for costs
and returns across the sample were used to compute net present values. Also, net
present values were calculated for each farmer, based on his/her particular costs and
returns. This latter method permitted a better understanding of the variation across
farms in returns, and thus the risk of the practices.
To assess the actual changes in annual income generated by a new practice,
farm models were developed in which the farm was partitioned to contain specified
portions of land devoted to each phase (corresponding to a season or year) of the
technology. For example, in the model of improved fallows in Zambia (Chapter 3),
the farm was assumed to have equal portions of area in each of the practice’s four
phases: planting of the improved fallow (year 1), maturing of the fallow (year 2),
the first post-fallow maize crop (year 3), and the second post-fallow maize crop
26 S. Franzel et al.

(year 4). The net returns of this farm were compared to two other farms having the
same amount of labour (the main constraining resource), one planting fertilized
maize continuously without fallow and the other planting unfertilized maize con-
tinuously without fallow. The model was thus useful for estimating the impact of
improved fallows on annual net farm income and maize production.

Acceptability and feasibility

To assess the adoptability of a technology, it must be analysed from the farmers’ per-
spective, which means much more than its biophysical performance and profitabil-
ity (Scherr, 1995). In particular, it must be feasible for farmers to plant and
maintain, and it must be considered acceptable and desirable. Feasibility and
acceptability are grouped together because the feasibility of a technology is depen-
dent on its perceived value. For example, labour bottlenecks that appear when farm-
ers attach a low value to an activity may disappear when the farmers’ perception of
the value increases.

Feasibility
Farmers’ ability to plant and maintain agroforestry technologies depends on three
factors: available resources (land, labour and capital), whether they have the
required information and skills, and whether they are able to cope with any prob-
lems that arise. Several tools are available for assessing the feasibility of a practice in
on-farm trials. Resource budgets may be assembled to compare the availability of
the resource with the needs of the practice. For example, in Chapter 7, the number
of calliandra trees that can be planted on boundaries and contours is compared to
the numbers needed to feed a cow. In Chapter 5, household labour use in different
weeks of the year is compared to the labour requirements of hedgerow intercrop-
ping. If the peak period for labour use of a new practice coincides with that of the
household, farmers may have difficulty implementing the practice.
Another means for assessing feasibility is to evaluate the general biophysical per-
formance of the technology as planted and maintained by the farmer. This assessment
may involve quantitative data, such as survival rates of seedlings planted, or qualita-
tive ratings, such as technicians’ assessment of the amount of biomass in an improved
fallow. Both are used in assessing the feasibility of improved fallows in Zambia
(Chapter 3). Where performance is not as high as expected, technicians and farmers
are asked their opinions as to what went wrong. Possible reasons might include:
● inappropriate biophysical conditions, such as soil type;
● lack of labour or other resources;
● lack of information or skills;
● incompatibility of gender with a particular task, as in cultures where females do
not prune trees; or
● inability to cope with a problem that arose, such as pests, diseases or inade-
quate rainfall.
 Assessing Agroforestry Adoption Potential 27

Finally, farmers may be interviewed in informal or questionnaire surveys about

the problems they experienced implementing the trial. Their ranking of the impor-
tance and frequency of problems provides useful information on feasibility
(Chapter 3).

Acceptability
Acceptability includes profitability, feasibility and a range of criteria that are diffi-
cult to quantify, such as risk, general compatibility with farmers’ values and farmers’
valuation of benefits, such as a tree’s ornamental value or its use as a boundary
marker.
Risk was assessed by: (i) measuring variability in the returns of individual farm-
ers; (ii) conducting minimum returns analysis (CIMMYT, 1988), in which the
average of the lowest 25% of the net benefits of each treatment are compared; and
(iii) by conducting informal interviews with farmers. The interviews were particu-
larly useful for understanding farmers’ views towards risk. For example, in Zambia,
farmers stated that even though fertilizer could be very profitable in a year of good
rainfall, the risk of losing their investment in a year of poor rainfall prevented them
from using the technology. Sensitivity analysis was conducted to assess the effect on
net present value of changes in key parameters, such as prices of inputs and outputs,
changes in input–output coefficients and changes in the discount rate. By assessing
the effects of likely future market patterns on these sensitive parameters, the eco-
nomic sustainability of the practice can be evaluated.
Asking farmers whether a practice was acceptable did not prove to be very use-
ful; nearly all farmers gave positive assessments, probably because they felt that criti-
cizing a practice would be insulting to the researcher. Rather, acceptability is best
ascertained by monitoring whether farmers continue using and expand their use of
a practice, and whether neighbouring farmers take it up. Important indicators of
acceptability include the numbers of times farmers expanded their planted area, the
numbers of trees planted and area planted per expansion, and the numbers of farm-
ers to whom the original experimenters gave or sold planting material. Farmers’
expansion of practices was assessed in four of the case studies and farmer-to-farmer
dissemination was assessed in three of the case studies.
But using expansion or adoption as a proxy for acceptability is problematic for
three reasons. First, in some cases, farmers may be interested in expanding but
unable to do so because they lack access to critical information or inputs. Secondly,
farmers may expand use of a practice not because they like it but because they
expect to receive other benefits, such as free inputs or employment. Thirdly, agro-
forestry practices take a long time to evaluate and it was reasonable to assume that a
farmer needed to experience the full cycle of a technology (4–5 years in the case of
improved fallows in Zambia) before deciding whether to continue using it. Any
expansion that took place before the end of the cycle could arguably have been
called an expansion in testing rather than an indication of acceptability.
Assessments of farmers’ preferences among alternative options can provide use-
ful feedback for research and extension, especially when they are quantified. For
28 S. Franzel et al.

example, in western Kenya, farmers used an indigenous board game, bao, to score
upper-storey trees on criteria important to them (Chapter 6). In this variant of
matrix ranking (Ashby, 1990), branches of each tree were laid out on the ground.
For each criterion, farmers rated the tree by putting one to five seeds in the pocket
next to each branch – five being a high rating and one being a low rating. In con-
trast with questionnaires, which farmers find tedious, the bao game can be used for
collecting quantitative data on farmer assessments in an accurate, entertaining, yet
statistically rigorous manner. It also allows farmers to assess their ranking visually
and perhaps, on reflection, make changes (Franzel, 2001).
Farmer assessment surveys may be implemented to find out what farmers per-
ceive as the advantages and disadvantages of a practice. These surveys may be
informal, involving individual or group interviews, or formal, using a question-
naire administered by technicians. As with matrix ranking, they can be conducted
at various times to find out farmers’ assessments at each stage of the practice.
Informal and formal farmer assessment surveys were conducted in all five of the
case studies.
Hierarchical decision trees may be used to model complex decisions, such as
whether or not to expand the use of hedgerow intercropping in western Kenya
(Chapter 5). This method is useful for explaining the decisions that farmers make,
by breaking them down into a series of subdecisions and mapping each farmer’s
decision path along the branches of the tree. Questions needed for decision trees
can be included in both informal and formal surveys (Gladwin, 1989).
Farmer workshops are also an important means to find out farmers’ views on
the technologies and their potential impacts (Kristjanson et al., 2002). To facilitate
the exchange of information, farmers divide into small working groups, each
addressing a specific issue. The workshops provide information on important effects
of practices, ‘invisible effects’, such as secondary effects on other enterprises, indica-
tors that farmers would use to evaluate the impact of adoption and clarification of
possible constraints to adoption. Whereas in many cases the information provided
by farmers in such workshops is what researchers might have anticipated, it is possi-
ble that important new information may be obtained. For example, a key finding in
the Zambia workshop was that many farmers intended to use improved fallows not
so much to increase the total amount of maize they produced, but rather to increase
maize yields and reduce the area they devoted to maize, freeing up land for growing
cash crops (Chapter 3). Farmer workshops can also be an important means for
farmers to discuss issues related to new practices, exchange opinions and lessons,
and come to consensus or clarify their differences.

Defining boundary conditions

The boundary conditions of a practice are defined by identifying the variables that
are most important in determining who will and will not use the practice.
Information on variables affecting biophysical performance, profitability, feasibility
 Assessing Agroforestry Adoption Potential 29

and acceptability are thus critical. The variables should be fairly easy to identify;
otherwise, they will not be useful in distinguishing among farmers or areas.
Biophysical variables used for assessing boundary conditions in the case studies
examined in this book included altitude (a proxy for temperature), rainfall, soil type
and depth, and soil nutrient status. Critical socioeconomic variables included
wealth, gender and farm size. The two groups of variables were found to be useful
in different ways. Biophysical boundary conditions were often used to exclude a
component or practice from particular areas. For example, calliandra did not per-
form well on acidic soils. Socioeconomic boundary conditions, on the other hand,
were used mainly to inform researchers, extensionists and farmers about the appro-
priateness of choices. For example, the finding that improved fallows have a higher
adoption potential among males than females is important for identifying and alle-
viating the constraints that women face or for understanding the need to identify
alternative technologies for them (Chapter 3).
Some boundary conditions may be assessed through secondary data, as when it
is known that a particular tree species does not perform well outside a certain alti-
tude range. Modelling is also useful, as when a financial analysis shows that a prac-
tice is profitable only when the opportunity cost of labour is above a certain level
(Chapter 4). But in most cases, assessments must be based on empirical data con-
cerning where the practice performs well and who adopts it. Biophysical variables
often vary spatially; type 1 trials are especially useful for assessing the biophysical
boundary conditions of a practice (Chapter 3). Type 2 and type 3 trials are less reli-
able because variation in management practices across site may confound the results
on biophysical performance.
The farm and household characteristics that were examined most frequently in
the case studies, for their association with testing and continued use of a practice,
included gender, household type, wealth level, farm size, soil type and soil nutrient
status. These were investigated by testing the statistical association between individ-
ual variables and performance, as in Zambia, where farmers used a wealth ranking
exercise to classify themselves into different wealth categories. They found that
high-income households tended to use improved tree fallows more than low-
income households (Chapter 3). In western Kenya, multiple regression was used to
assess the relative importance of selected variables affecting farmers’ preferences
among upper-storey trees (Chapter 6). The small number of farmers that could be
monitored in a type 2 trial, usually fewer than 50, limited the degree to which fac-
tors affecting adoption potential could be examined rigorously.

Drawing lessons for effective dissemination and diffusion

The multiyear dissemination, testing and evaluation of agroforestry practices with

farmers was found to provide valuable lessons for the subsequent design of effective
promotion efforts by extension organizations. Farmers’ own innovations were iden-
tified in on-farm trials that can increase productivity and adoption potential. For
30 S. Franzel et al.

example, low-cost tree establishment methods such as bare-rooted seedlings signifi-

cantly reduced production costs and increased profitability. In establishing
improved fallows, intercropping the trees with crops during the first year may be
more acceptable to farmers than planting them in pure stands (Chapter 3).
Several methods are available for identifying ‘best-bet’ modifications of practices:
● participatory informal appraisals with farmers to determine their assessment of
practices and what changes they propose (Ashby, 1990);
● visits to farmers’ fields to identify modifications that farmers have tested;
● screening new practices and species from other sites for their relevance to farm-
ers’ needs and circumstances; and
● ex ante quantitative data on costs and returns of different modifications and
practices.
By working closely with farmers during the technology development process,
extensionists can identify effective ‘selling points’ for the new technologies, for dif-
ferent groups of farmers. The process of identifying boundary conditions helps with
the formulation of extension materials and development of strategies for targeting
different groups. The advantages of individual versus group or village approaches
can be evaluated under different circumstances, prior to the launching of major dis-
semination initiatives.

Drawing lessons for policy support

Public policies and institutions can influence the feasibility, profitability and accept-
ability of agroforestry practices (Chapter 8). For example, public subsidies and seed
distribution networks affect the price of seedlings, which in turn influences the scale
of tree-planting by farmers. Land tenure legislation and local land-use regulations
influence farmers’ selection of sites and niches for tree planting. The design and tar-
geting of agroforestry extension programmes influences the dissemination of infor-
mation critical for farmer adoption.
On-farm research, particularly large-sample, type 3 trials, can provide both
valuable diagnostic information about possible policy constraints and benefits, and
also the opportunity to assess particular policy instruments on an experimental basis
(such as a new subsidy, product marketing approach, type of farmer organization or
a locally agreed land-use regulation). Depending upon the issue and the sample size,
various methods can be used to obtain policy-relevant insights:
1. Direct observation and participant feedback about policy or institutional factors
that constrain adoption or expansion of agroforestry. An example is the identifica-
tion of the constraints to on-farm seed production for fodder trees and the need for
seed bank development, described in Chapter 7.
2. Regression analysis of factors associated with adoption, to assess the importance
of existing policy and institutional factors (for example, plot tenure status) and to
identify factors that could be affected by policy action.
 Assessing Agroforestry Adoption Potential 31

3. Sensitivity analyses from cost–benefit studies can be used to estimate the magni-
tude of the potential impact of policy changes (such as planting subsidies) on vari-
ous types of farmer costs and benefits (Chapter 5).
4. Formal or informal evaluation of pilot policy interventions or institutional inno-
vations. Examples include the pilot programmes for community-based extension in
western Kenya (Chapter 4), and institutional mechanisms for research–extension
linkages in Zambia (Chapter 3).

Feedback to research

The close interaction between researchers and farmers during on-farm research, and
the opportunities for direct observation by researchers, can provide an excellent
learning opportunity to identify research problems and priorities. Formal surveys to
assess the frequency and severity of important problems can also help researchers to
set priorities among problems and among different potential solutions to a given
problem (Tripp and Wooley, 1989). All five of the case studies report on feedback
to research, which often led to trials and results that alleviated or solved the original
problems. For example, farmers’ main problem with Sesbania sesban in improved
fallows in western Kenya, that it germinated poorly when direct seeded, led directly
to trials that were successful in selecting species with higher germination rates.

Selection of methods

The approach presented in this chapter for assessing adoption potential is eclectic in
that it draws on a range of methods, including the formal surveys of economists,
research trials of biophysical scientists, the participatory techniques of anthropolo-
gists, and the informal experiments and assessments that farmers themselves make.
The case studies do not all use the same techniques for assessing adoption potential
(Table 2.3). Rather, the selection of activities was driven by critical information
gaps, identified jointly by researchers, extensionists and farmers, in technology
design and in understanding boundary conditions. The choice of methods thus
depended on several factors:
1. The practice’s resource requirements. Improved fallows and hedgerow intercrop-
ping had relatively high labour requirements. Thus researchers decided to measure
the practice’s labour requirements and compare them with the seasonal and total
labour requirements of the household. In contrast, upper-storey trees planted on
farm boundaries have low labour requirements, so a formal assessment of labour
requirements relative to total labour use was not viewed as critical.
2. The practice’s impact on farming enterprises. Enterprise budgets are needed to
assess profitability when a new practice has an important impact on the costs,
returns, resource requirements and management of an enterprise. Thus, enterprise
budgets were drawn up for improved fallows and hedgerow intercropping. But for
32 S. Franzel et al.

practices that have less impact, such as substituting calliandra leaves for a purchased
protein concentrate in a dairy enterprise, a partial budget sufficed for determining
profitability.
3. The size of the sample. Where the number of farmers testing a practice were few
(for example, only 20 farmers tested improved fallows in western Kenya), tests of
association between farmers’ circumstances and use of the practice could be con-
ducted, but the results were not usually statistically convincing.
4. Farmers’ experience with the practice. Where farmers had not yet received the ben-
efits of the practices, as with upper-storey trees in Chapter 6, it was not possible to
assess factors affecting uptake. But farmers’ preferences among the different trees
could be examined.
5. Availability of staff and resources. Lack of human resources and a change of per-
sonnel and research priorities prevented the monitoring of uptake of improved fal-
lows in western Kenya, following the end of the experiments in the early 1990s.
6. Nature of the practice. Where a practice involves choosing among several alterna-
tives, as among five agroforestry species (Chapter 6), matrix ranking was a useful
tool. It was not appropriate when a practice involved a single prototype, as with
improved fallows in western Kenya (Chapter 4).

The Research–Development Continuum

The on-farm research experiences in Kenya and Zambia demonstrate that there are
multiple sources of innovation in agroforestry – formal sector researchers, farming
tradition, farmer-innovators, extensionist-innovators. Through shared experiences
in on-farm research studies, their complementary strengths can be effectively
exploited and integrated, at reasonable cost. Figure 2.1 presents a farmer-centred
model that is evolving in western Kenya and eastern Zambia. Instead of a linear
sequence whereby technology is developed by researchers, then passed to extension-
ists and finally to farmers, in the on-farm research-centred model there is continual
interaction among these groups throughout the process. Input from farmers and
extensionists is provided early on, opportunities for early extensionist and farmer
innovation and adaptation are encouraged, and implementation on farmers’ fields –
and hence potential for farmer-to-farmer diffusion – begins much earlier in time.
Moreover, building a coalition of organizations to conduct on-farm research and
dissemination together is vastly more effective and efficient than leaving each to
work independently on only one element.
Researchers and extensionists working together in particular sites can jointly
develop locally appropriate extension recommendations for use by other extension
programmes in the region. They can, meanwhile, report regularly to this broader
community – as well as senior staff in their respective organizations – on progress in
technology development and problems encountered, which can then be jointly
addressed. Within this structure, technical specialists can collaborate more effec-
tively with, and support, initiatives by existing or newly formed farmer-experi-
menter groups than can conventional research or extension organizations.
 Assessing Agroforestry Adoption Potential 33

Appraisals and diagnostic

surveys to identify priorities
for agroforestry research

Agroforestry Controlled trials needed to

options available determine management
for farmer strategies and measure
evaluation key processes
Farmer feedback Refinement of
identifies need problem
for further survey description
work needed

Participatory on-farm
research to assess the Promising
options identified Researcher-controlled
adoption potential:
trials, both on-station
• biophysical response
and in farmers’ fields
• profitability
• acceptability
Farmer feedback
indicates need for more
detailed research
Biophysical
Interaction between performance
adoption potential assessed in multi-
and socioeconomic Improved targeting location trials
factors assessed of research

Boundary conditions
determined

Targeted dissemination of Targeted dissemination of

a range of agroforestry a range of agroforestry
choices/options choices/options
Dissemination
facilitates wider
testing by farmers Dissemination
activities

Widespread adoption and

impact

Adoption and impact studies identifying

second-generation research issues associated with
scaling-up and new opportunities for agroforestry

Fig. 2.1. Flow diagram of decisions and activities in farmer-centred agroforestry

research and extension.
34 S. Franzel et al.

Conclusion

The experiences of ICRAF and its partners in Kenya and Zambia demonstrate the
importance of assessing the adoption potential of agroforestry practices. First, such
assessments improve the efficiency of the technology development and dissemina-
tion process, by feeding back information on farmers’ problems, modifications and
preferences to research, extension staff and policy makers. Secondly, the assessments
help document the progress made in disseminating new practices, demonstrating
the impact of investing in technology development and dissemination. Thirdly,
because the activities are conducted with partner institutions, they facilitate inter-
disciplinary and inter-institutional cooperation. Finally, the assessments help to
identify the factors contributing to successful technology development programmes
as well as the constraints limiting achievements.
Future assessments need to take advantage of farmers’ increased experience
with agroforestry practices. Analyses of social, economic, biophysical and ecological
impacts will thus be possible at community, landscape and regional scales.
Improvements in the development of spatially explicit databases and models should
permit the use of geographic information systems for assessing the boundary condi-
tions of new technologies. Efforts are also needed to hand over many of the activi-
ties in assessing adoption potential to local institutions, such as farmer groups and
organizations. The greater control they have over assessing adoption potential, the
more responsive technology generation activities will be to their needs.

References
Ashby, J.A. (1990) Evaluating Technology With Cooper, P.J.M. (1999) Agroforestry: learning as
Farmers: a Handbook. CIAT Publication we go in Africa. In: Martin, G.J., Agama, A.
No. 187. CIAT, Cali, Colombia. and Leakey, R. (eds) Cultivating Trees. People
Ashby, J., Gracia, T., Guerrero, M., Quiros, C., and Plants Handbook, 5. UNESCO, Paris,
Roa, J. and Beltran, J. (1995) France, pp. 30–33.
Institutionalizing Farmer Participation in Franzel, S. (1996) Collecting Data on Labour Use
Adaptive Technology Testing With the ‘CIAL’. in On-Farm Trials. Training Note. ICRAF,
Overseas Development Institute Network Nairobi.
Paper 57, London. Franzel, S. (2001) Use of an indigenous board
Biggs, S. (1989) Resource-Poor Farmer game, ‘bao’, for assessing farmers’ prefer-
Participation in Research. ISNAR, The ences among alternative technologies. In:
Hague, Netherlands. Peters, G.H. and Pingali, P. (eds) Tomorrow’s
Byerlee, D. and Collinson, M.P. (1980) Planning Agriculture: Incentives, Institutions,
Technologies Appropriate to Farmers. CIM- Infrastructure, and Innovations. Proceedings
MYT, Mexico. of the International Agricultural Economics
Chambers, R., Pacey, A. and Thrupp, L.A. (1989) Association Conference, 13–18 August
Farmer First: Farmer Innovation and 2000, Berlin, Germany. Ashgate, UK.
Agricultural Research. Intermediate Franzel, S., Hitimana, L. and Akyeampong, E.
Technology Publications, London. (1995) Farmer participation in on-station
CIMMYT (1988) From Agronomic Data to tree species selection for agroforestry: a case
Farmer Recommendations: an Economics study from Burundi. Experimental
Training Manual. CIMMYT, Mexico. Agriculture 31, 27–38.
 Assessing Agroforestry Adoption Potential 35

Gladwin, C.H. (1989) Ethnographic Decision Tree Scherr, S.J. (1991b) On-farm research: the chal-
Modeling. Sage, Newbury Park, California. lenges of agroforestry. Agroforestry Systems
Haverkort, B., van der Kamp, J. and Waters- 15(1–2), 95–110.
Bayer, A. (eds) (1991) Joining Farmers’ Scherr, S.J. (1991c) Methods for Participatory On-
Experiments: Experiences in Participatory Farm Agroforestry Research: Summary
Technology Development. Intermediate Proceedings of an International Workshop.
Technology Publications, London. International Center for Research in
Hildebrand, P.E. and Russell, J.T. (1996) Agroforestry, Nairobi.
Adaptability Analysis. A Method for the Scherr, S.J. (1995) Economic analysis of agro-
Design, Analysis and Interpretation of On- forestry systems: the farmers’ perspective.
Farm Research-Extension. Iowa State In: Current, D., Lutz, E. and Scherr, S.J.
University Press, Ames, Iowa. (eds) Costs, Benefits and Farmer Adoption of
ICRAF (1997) Annual Report 1996. Nairobi. Agroforestry: Project Experience in Central
Kristjanson, P., Place, F., Franzel, S. and Thornton, America and the Caribbean. World Bank
P.K. (2002) Assessing research impact on Environment Paper No. 14. The World
poverty: the importance of farmers’ perspec- Bank, Washington, DC, pp. 28–44.
tives. Agricultural Systems (in press). Shepherd, K., Swinkels, R. and Jama, B. (1994)
Kwesiga, F.R., Franzel, S., Place, F., Phiri, D. and A question of management: the pros and
Simwanza, C.P. (1999) Sesbania sesban cons of farmer- and researcher-managed tri-
improved fallows in eastern Zambia: their als. Agroforestry Today 6(4), 3–7.
inception, development, and farmer enthu- Sperling, L., Loevinsohn, M.E. and Ntabomvura,
siasm. Agroforestry Systems 47, 49–66. B. (1993) Rethinking the farmers’ role in
Lightfoot, C. (1987) Indigenous research and plant breeding: local bean experts and on-
on-farm trials. Agricultural Administration station selection in Rwanda. Experimental
24, 1–11. Agriculture 29(4), 509–518.
Little, T.M. and Hills, F.J. (1972) Statistical Stroud, A. (1993) Conducting On-Farm
Methods in Agricultural Research. University Experiments. CIAT, Cali, Colombia.
of California Press, Davis, USA. Tripp, R. and Wooley, J. (1989) The Planning
Mutsaers, H.J.W., Weber, G.K., Walker, P. and Stage of On-Farm Research: Identifying
Fisher, N.M. (1997) A Field Guide for On- Factors for Experimentation. CIMMYT/
Farm Experimentation. IITA/CTA/ISNAR, CIAT, Mexico; Cali, Colombia.
International Institute of Tropical Upton, M. (1987) African Farm Management.
Agriculture, Ibadan, Nigeria. Cambridge University Press, Cambridge,
Okali, C., Sumberg, J. and Farrington, J. (1994) UK.
Farmer Participatory Research. Rhetoric and van Veldhuizen, L., Waters-Bayer, A. and de
Reality. Intermediate Technology Zeeuw, H. (1997) Developing Technology
Publications, London. with Farmers: a Trainers Guide for
Rocheleau, D.E. (1991) Participatory research in Participatory Learning. Zed Books,
agroforestry: learning from experience and London.
expanding our repertoire. Agroforestry Zandstra, H.G., Price, E.C., Litsinger, J.A. and
Systems 15, 111–138. Morris, R.A. (1981) A Methodology for On-
Scherr, S.J. (ed.) (1991a) On-farm agroforestry Farm Cropping Systems Research.
research. Agroforestry Systems Special Issue International Rice Research Institute, Los
15(1–2), 2–3. Baños, Philippines.
 Assessing the Adoption
Potential of Improved
Fallows in Eastern Zambia
3
S. FRANZEL,1 D. PHIRI2 AND F. KWESIGA3
1International Centre for Research in Agroforestry, ICRAF House,
PO Box 30677, Gigiri, Nairobi, Kenya; 2World Vision Integrated
Agroforestry Project, PO Box 510948, Nairobi, Kenya; 3SADC-ICRAF
Regional Agroforestry Programme, PO Box MP 128, Mount Pleasant,
Harare, Zimbabwe

Summary

Declining soil fertility is a key problem faced by farmers in eastern Zambia. This chap-
ter assesses farmers’ experiences of testing improved tree fallows in participatory on-
farm trials to increase soil fertility. It also highlights the development of an adaptive
research and dissemination network of institutions and farmer groups for testing and
disseminating improved fallows. Sesbania sesban and Tephrosia vogelii performed well,
but Cajanus cajan was discontinued because it was browsed heavily by livestock. The
economic analysis compared a 2-year improved fallow, followed by maize cropped for
3 years, with fertilized and unfertilized continuously cropped maize. Over a 5-year
period, farmers used 11% less labour on the improved fallow plot than on unfertilized
maize, but harvested 83% more maize. Improved fallows had higher returns to land
and to labour than continuously cropped unfertilized maize; returns compared to fer-
tilized maize were mixed. Farmer interest is strong, as the number of farmers planting
improved fallows has increased from under 20 in 1993–1994 to roughly 10,000 in
2000. Key elements contributing to the progress made thus far include: (i) effective
diagnosis of farmers’ problems and screening of potential solutions; (ii) farmer partici-
pation in the early stages of testing of improved fallows; (iii) testing of a range of man-
agement options by farmers and researchers, and encouraging farmers to innovate; and
(iv) development of an adaptive research and dissemination network.

Introduction
Land depletion and declining soil fertility are viewed increasingly as critical prob-
lems affecting agricultural productivity and human welfare in tropical Africa

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 37
38 S. Franzel et al.

(Cooper et al., 1996; Sanchez et al., 1997). Much attention has focused on biologi-
cal technologies, such as agroforestry, for solving these problems because of the
increasing costs of inorganic fertilizers, their limited availability and the need to
increase soil organic matter and improve soil structure. Improved tree fallows, the
enrichment of natural fallows with trees to improve soil fertility, are a promising
agroforestry practice for small-scale farmers and are being tested at many sites
throughout the tropics.
There is considerable evidence from many areas that farmers use leguminous
trees on fallow plots for improving soil fertility (Raintree and Warner, 1986). But
there is little evidence that farmers have benefited from researchers’ efforts to build
upon the performance of farmers’ traditional practices, or from efforts to introduce
improved tree fallows in areas where farmers are unfamiliar with them.
In 1987, following farmer surveys that highlighted the soil fertility problem
(Zambia/ICRAF, 1988), the Zambia/ICRAF Agroforestry Research Project began
research at Msekera Research Station, Chipata, Eastern Province, Zambia, on improved
fallows to increase crop production. In on-station trials, improved fallows using
Sesbania sesban seedlings transplanted from nurseries greatly increased maize yields
(Kwesiga and Coe, 1994; Kwesiga et al., 1999). Moreover, economic analysis showed
that 2-year improved fallows were likely to be profitable for small-scale farmers (Place et
al., 1995). Species that can be established directly from seed, such as Tephrosia vogelii
and Cajanus cajan, have also been tested on station, with encouraging results (Kwesiga
et al., 1995). Farmer experimentation began in 1992–1993 and 1993–1994 with five
farmers, and involved 204 farmers in 1994–1995 and about 3000 in 1996–1997.
The objective of this chapter is to assess the adoption potential of improved
tree fallows in Eastern Province, based on farmers’ experiences. We examine:
1. Feasibility: the degree to which farmers are able to manage the technology, that
is, whether they have the required information, are able to plant and maintain the
fallows, and cope with any problems that arise.
2. Profitability: whether the financial benefits obtained are greater than the costs
incurred and how the benefits compare with farmers’ alternative practices.
3. Farmer interest: farmers’ views and actions to expand testing and use of the practice.
4. Institutional support: particularly the adaptive research and dissemination net-
work that has emerged for evaluating and extending improved fallows.
First, the study area is described and the design of the trials and monitoring surveys
are examined. Next, results from the trials are discussed and the strategy for institu-
tional support is examined. Finally, conclusions are drawn.

Methods

Study area

The plateau area of eastern Zambia is characterized by a flat to gently rolling land-
scape and altitudes ranging from 900 to 1200 m (Fig. 3.1). Seasonally waterlogged,
 Assessing Improved Fallows Adoption Potential 39

Kasama

Northern
Luapula

North Copper
Western belt
Chipata
Eastern
Kabwe
Central
Lusaka
Western Lusaka
N

Southern

100 0 100 200 kilometers

Fig. 3.1. Eastern Province, Zambia.

low-lying areas, known locally as dambos, are also common. The main soil types are
loamy sand or sand Alfisols, interspersed with clay and loam Luvisols. The Alfisols
are well-drained and relatively fertile but have low water and nutrient-holding
capacities (Zambia/ICRAF, 1988; Raussen et al., 1995). Rainfall averages about
1000 mm year−1, with about 85% falling in 4 months, December–March. Rainfall
is highly variable; the area received less than 600 mm in 2 of 8 years between 1990
and 1997. The growing season lasts for about 140–155 days. Average air tempera-
tures range from 15 to 18°C during June–July to 21–26°C in September–October
(Zambia/ICRAF, 1988).
Population density varies between 25 and 40 persons km−2. About half of the
farmers practice ox cultivation, the others cultivate by hand-hoe. Average cropped
land ranges from 1.1–1.6 ha for hoe cultivators to 2.3–4.3 ha for ox cultivators.
The two groups are mixed amongst each other and grow similar crops, though ox
cultivators tend to use more purchased inputs. Maize is the most important crop
accounting for about 60–80% of total cultivated area. Other crops include sun-
flower, groundnuts and cotton. Average numbers of cattle per household range from
1.5 to 3, depending on the district, and goats are also common. The main ethnic
groups are the Chewa and the Ngoni. Rural households are concentrated in village
settlements of up to 100 homesteads, a legacy of government-sponsored village
regrouping programmes (Zambia/ICRAF, 1988; ARPT, 1991; Celis and Hollaman,
1991; Jha and Hojjati, 1993; FSRT, 1995; Peterson et al., 2000).
40 S. Franzel et al.

Farmer surveys have identified declining soil fertility as one of farmers’ main
perceived problems (Zambia/ICRAF, 1988; ARPT, 1991). Nitrogen deficiency was
judged to be the most important problem responsible for low maize yields.
Increased pressure on land has reduced fallow periods, farmers’ main method for
maintaining soil fertility, to 1–3 years. Many farmers even practise continuous crop-
ping while having land in ‘reserve fallow’ that they have never cultivated
(Zambia/ICRAF, 1988; Peterson et al., 2000). Fertilizer use was common during
the 1980s but the cessation of subsidies caused the ratio between the price of nitro-
gen and the price of maize to increase from 3.1 in 1986/87 to 11.3 in 1995/96.
Fertilizer use in Zambia declined by 70% between 1987/88 and 1995/96 (Howard
et al., 1997) and the decline in the smallholder sector was even greater.

Trial design

In 1992/1993 and 1993/1994, five farmers established type 31 (farmer-designed,

farmer-managed) improved fallow trials, using sesbania seed or bare-rooted or pot-
ted seedlings obtained from researchers. The objective of the trials was to give farm-
ers the opportunity to experiment with improved fallows, to assess their ability to
establish and maintain improved fallows, and to obtain their assessment of the prac-
tice. Farmers were advised on how to manage improved fallows and they each
designed their own trial, planting the trees where and how they wished on their
farms. Spacing averaged 1 m × 1 m (10,000 trees ha−1) and plot size ranged from
400 m2 to 800 m2. In the third year of the trial, farmers were given fertilizer to use
on an adjacent plot of 200 m2, so as to be able to compare the post-fallow maize
yield with yields of fertilized maize. A non-fertilized maize plot was also identified
for the comparison.
In 1994/95, 157 farmers planted type 2 (researcher-designed, farmer-managed)
trials. The farmers selected one of six improved fallow technologies to plant: sesba-
nia, tephrosia or cajanus as well as the same three species intercropped with maize
during fallow establishment in the first year and then allowed to grow into a pure
stand fallow in the second year. Sesbania was planted using bare-rooted seedlings;
tephrosia and cajanus were planted by direct seeding. Bare-rooted seedlings were
inoculated with rhizobia from the already established stands of sesbania at farmer-
training centres or on farm. The treatments included:
● the improved fallow plot (pure stand or intercropped with maize) in a 400 m2
plot – trees were planted at a density of 0.9 m × 0.75 m or 14,763 trees ha−1,
whether in pure stand or intercropped with hybrid maize;
● unfertilized maize in a 200 m2 plot;

1 Type 1 on-farm trials are researcher-designed and researcher-managed and are similar to
trials done on station. Type 2 trials are researcher-designed and farmer-managed. Type 3
trials are farmer-designed, farmer-managed, that is, farmers experiment as they wish and
thus may not have control plots (see Chapter 2).
 Assessing Improved Fallows Adoption Potential 41

● maize fertilized at the recommended level, 112–40–20 kg N–P–K per hectare,

in a 200 m2 plot.
Maize was planted at the recommended density of 75 cm × 30 cm, whether in pure
stand or intercropped, giving a density of 44,444 plants ha−1. When intercropped,
trees and maize were planted in the same rows, to permit space for weeding between
the rows. The project supplied hybrid maize seed and fertilizer for the maize plots
and farmers supplied labour for all of the operations. Farmers were advised to man-
age the plots uniformly in order to ensure that the results among them would be
comparable. They agreed to continue growing maize on the maize plots for 5 years.2
On the improved fallow plots, farmers agreed to maintain the trees for 2 years, and
to then cut them down, incorporate the leafy biomass, and plant maize in the third,
fourth and fifth years.
In 1995/96, 35 type 2 trials were added, primarily to assess the effects of differ-
ences among seasons on the improved fallows. The sample was stratified by technol-
ogy and included 12 farmers with sesbania in pure stand, 13 with tephrosia in pure
stand, 4 with intercropped sesbania and 6 with intercropped tephrosia. Cajanus was
omitted because of the browsing problems the previous year.
Type 3 trials, in which farmers experimented on their own with improved fal-
lows, expanded greatly, from 37 farmers in 1994/95 to about 800 farmers in
1995/96 and about 3000 in 1998. Farmers received no incentives for experiment-
ing with improved fallows.
Overall, the on-farm experimentation programme was designed to provide
farmers with different options for testing. Table 3.1 shows that by 2000, five of
eight variants of the technologies were being used by farmers; three had been
rejected.

Monitoring and analysis

Monitoring surveys were conducted by a researcher, a technician and collaborating

extension staff. Data were collected on labour use and other costs from selected type
2 and type 3 farmers; these were supplemented by data from other farmers, local
markets and secondary sources (Appendix 3.1). Enterprise costs and returns were
drawn up for each farm and used to calculate net present values per hectare to assess
returns to land (in which household labour is valued) and net returns per work day
to assess returns to labour (in which household labour is not valued). The analysis
covered a period of 5 years: 2 years of fallow and the 3 subsequent years for which it
is assumed that maize yields would be affected. Maize yields following sesbania fal-
lows were available for five farmers for 1996 and seven farmers for 1997. Average
data on costs were used in each individual farmers’ budget; maize yield figures were
specific to each farm. Since data on maize yields during the second and third years

2 Maize yields in the control plots were approximately equal to average maize yields in the
area.
 42

Table 3.1. Improved fallow practices tested by farmers.

Sesbania sesban Tephrosia vogelii Cajanus cajan Gliricidia sepium
Potted seedling Bare-root seedling Direct seeded Direct seeded
Bare-root seedling,
Year Pure stand Pure stand Intercropped Pure stand Intercropped Pure stand Intercropped intercropped
91–92 *
92–93 * * * * *
93–94 * * * * *
94–95 * * * * * *
95–96 * * * *
S. Franzel et al.

96–97 * * * *
97–98 * * * * *
98–99 * * * * *
99–00 * * * * *
 Assessing Improved Fallows Adoption Potential 43

following improved fallows were not available from on-farm trials, data on the per-
centage decline in the maize yield response from on-station trials, 30% for the sec-
ond year and 60% for the third year, were used in the analysis. Where cost was a
function of yield, as in the case of harvesting labour, costs were adjusted in relation
to yield. Sensitivity analysis was conducted to show the effects of changes in para-
meters on the results of the economic analysis. A semi-structured survey was con-
ducted following the first post-fallow maize harvest, to assess farmers’ experiences
and opinions.
Farm models using the Microsoft Excel Program were drawn up to assess the
impact of adopting improved fallows on maize income. Models were drawn up for
the same three scenarios as for the enterprise budgets: farms that adopt improved
fallows, farms that cultivate unfertilized maize and those with fertilized maize.
For the 157 type 2 farmers planting in 1994/95, four subsamples were sur-
veyed during the first 2 years:
1. Three months after the 1994/95 planting, a survey assessed survival rates and
problems inhibiting tree survival and growth. The survey included 110 of the 157
farmers; sample farmers were chosen on the basis of their accessibility.
2. Six months after planting, maize yields were measured on 20 farms to assess the
effects of intercropping trees with maize on maize yields.
3. One year after planting, a questionnaire survey was conducted of the 68 farmers
who had achieved high survival rates, that is, rates of over 60% in the first survival
survey, and were thus expected to have successful improved fallows.3 The sample
was stratified by technology. Farmers were interviewed concerning their manage-
ment of the improved fallows and problems encountered.
4. Two years after planting, 31 farmers with low survival rates, that is, rates below
60% 6 months after planting, were interviewed to find out whether they had con-
tinued with their experiments and whether they had started new ones.
The 35 type 2 farmers planting in 1995/96 were surveyed 6 months after
planting, using a questionnaire, to assess their experiences and problems they
encountered.
A random sample of 65 of the roughly 800 type 3 farmers planting in 1995/96
was surveyed 6 months after planting, using a questionnaire, to assess their experi-
ences.
For assessing the uptake of improved fallows by gender and wealth group, a
census of households was conducted in 1998 in four villages where substantial pro-
portions of farmers were planting improved fallows. Community members con-
ducted a ‘wealth ranking exercise’, in which they defined the different wealth groups
– well off, fairly well off, poor and very poor – and classified households into the
groups (Phiri et al., 2001).

3 We estimated that for sesbania a minimum survival rate of 60%, 6 months after planting,
would ensure a survival rate of at least 30% at the time of cutting the trees, and that 30%
was the minimum required to obtain benefits. For tephrosia and cajanus, the minimum is
higher, about 40%, because these species do not have as wide a canopy.
44 S. Franzel et al.

Results

Feasibility

In the five type 3 trials planted in 1992/93 and 1993/94, the farmers represented a
range of types of farmers found in the area: three were male, while two were female
heads of households; two were considered to be of high income, one middle
income, and two low income. Two farmers used oxen for land preparation while
three used hoes, and three were on Alfisols while two had Luvisols. Four of the five
farmers were able to establish dense sesbania fallows. Lack of weeding and bush
fires, which are common during the dry season, constrained tree growth on the fifth
farm. One female farmer made an important modification in the technology, inter-
cropping trees with maize during the first year of establishment, that was later tested
by researchers and is now used on a wide scale by many farmers.
In the type 2 trials planted in 1994/95, sesbania was the most popular species,
planted by 42% of the farmers. Tephrosia was planted by 34% and cajanus by 22%.
Most (83%) planted their trees in pure stand whereas 17% intercropped them with
maize during the season of establishment.
Survival rates were used in this and other trials as rough indicators of the feasi-
bility of planting improved fallows. Tree survival and growth were hampered by the
season’s low rainfall, 580 mm. Overall mean survival rates 3 months after planting
were highest for tephrosia, 74% (standard deviation (SD) = 27) in pure stands.
Survival rates for cajanus averaged 68% (SD = 19), and for sesbania, 58% (SD = 28).
The higher survival rates of tephrosia and cajanus were associated with the relative
ease in ‘gapping up’ directly seeded species as compared with bare-rooted seedlings,
which become either in short supply or too large for successful transplanting as the
season advances.
Farmers cited dry spells of up to 2 weeks early in the season as the principal
reason for low survival (Table 3.2). Other important problems affecting survival
and growth, noted by farmers and technicians, included weed competition, live-

Table 3.2. Reasons for low survival of tree seedlings on 33 farms with survival
rates lower than 60%, 3 months after planting, 1996.a
Number of farmers
Problem Sesbania Tephrosia Cajanus Total
Dry spells 12 1 3 16
Weeds 6 0 0 6
Browsing 1 3 1 5
Poor germination 0 1 4 5
Seedlings remaining 5 0 0 5
too long in nursery
Number of farmers 20 5 8 33
aReasons as given by research staff or farmers. In some cases more than one
response was given, thus numbers in columns do not sum to column totals.
 Assessing Improved Fallows Adoption Potential 45

stock browsing, low germination (particularly with cajanus), and sesbania seedlings
remaining too long in the nursery (in part, because of the late start of the rain).
Lack of experience also affected performance, as this was the first time that most of
the farmers had ever planted tree nurseries or transplanted bare-rooted seedlings.
In the survey of farmers with high survival rates, two-thirds had gapped up or
replanted seedlings. Fifteen per cent did so twice. As is common in years of poor
rainfall, many farmers had to replant their maize as well.
During the 1995/96 dry season (March–November), there was a significant
reduction in the mean survival rates for all species because of moisture stress and
uncontrolled communal grazing. A paired comparison of survival rates 3 months
and 1 year after planting showed that sesbania ranked highest in its ability to with-
stand the long dry season. On 19 farms that had high survival rates on pure stands
of sesbania, survival dropped from 81% at 3 months to 63% after 1 year. Tephrosia
survival dropped from 91% to 51% (15 farms) and cajanus from 72% to 21% (13
farms). Because cajanus was so heavily browsed, it was discontinued from on-farm
trials the following year. The differences in the reduction in survival rates among
the species were significant at P = 0.01.
Intercropping appeared to reduce tree survival and maize yields, probably
because of competition for resources, especially for moisture in a dry year. Seedling
survival rates 3 months after planting in intercropped fields were 9–13 percentage
points lower than for seedlings planted in pure stands, depending on the species.
The proportion of farmers achieving survival rates over 60% was generally higher
for pure stands than for intercropped trees (Table 3.3). Furthermore, the dry season
had a greater negative impact on intercropped trees than those in pure stands. For
example, whereas the survival rate of tephrosia in pure stand declined by 44%
between 3 and 12 months, intercropped tephrosia declined by 59%. Trees planted
in pure stand grew more vigorously because of the lack of competition for resources
by maize, and were thus better able to develop a deep rooting system, necessary to
withstand the long dry season. Moreover, intercropped trees were more susceptible
to trampling by livestock, which feed on the crop residues remaining after harvest.

Table 3.3. Proportion of farmers with survival rates greater than 60%, 3 months
after planting, 1996.
Species and % farmers with % farmers with Number of
arrangement survival rates >60% survival rates <60% farmers
Sesbania pure 50 50 40
Tephrosia pure 74 26 23
Cajanus pure 72 28 9
Sesbania intercropped 40 60 18
with maize
Tephrosia intercropped 100 0 10
with maize
Cajanus intercropped 44 56 9
with maize
Total 62 38 110
46 S. Franzel et al.

Data from on-farm trials also suggested that intercropping reduced maize
yields during the season of establishment by 18–28%, depending on the species.
But because of high variability, differences in maize yields between pure stand and
intercropping were not significant for any of the tree species.
As 1995/96 was a season of average rainfall, survival rates were considerably
higher than in 1994/95. Overall, 80% of the type 2 farmers planting pure-stand
sesbania in 1995/96 and 84% growing pure-stand tephrosia achieved high survival
rates, that is, rates over 60%. Intercropping did not appear to affect survival nega-
tively, as three of four intercropping sesbania and five of six intercropping tephrosia
also achieved acceptable survival rates. Only 46% needed to replant or gap up.
There was also a marked reduction in the numbers of farmers reporting prob-
lems. Twelve of 19 tephrosia growers and 3 of 16 sesbania growers reported that
they did not experience any problems. The mesoplatys beetle (Mesoplatys ochroptera)
was the most frequently cited problem affecting sesbania. Our observations indi-
cated that, in most cases, trees once established were able to recover fully after a bee-
tle attack, even without being sprayed. No other problem was mentioned by more
than four farmers. All but one farmer weeded their plots; half weeded two or three
times.
The results in the type 3 trials during the same year were somewhat less
favourable. Ninety-five per cent of the tephrosia growers and 61% of sesbania grow-
ers had survival rates over 50%. Nursery management and transplanting techniques
were suboptimal, probably because sesbania growers in type 3 trials received less
training than those in type 2 trials. Farmers with pure stands of sesbania and
tephrosia had slightly higher survival rates than those who intercropped. The prob-
lems reported by type 3 farmers were similar to those reported by type 2 farmers.
Performance was also assessed by soil type. Survival rates after 6 months did
not vary between Luvisols and Alfisols. But mean survival rates after 12 months,
that is, after the long dry season, were significantly lower on the Alfisols for
tephrosia (41% versus 75%, P < 0.05), because Alfisols are lighter than Luvisols and
retain less moisture. Tephrosia and cajanus were less able than sesbania to survive
the long dry season because they are direct seeded and are thus less able to develop
an extensive root system. Moreover, cajanus and, to a lesser extent, tephrosia were
subject to browsing during the dry season.
The last management task, cutting the trees, was not a serious problem for
farmers for three reasons. First, the trees were cut during the dry season before land
preparation begins, a slack period for farmers. Secondly, sesbania and, in particular,
tephrosia were easier to fell than other trees. Thirdly, females traditionally clear bush
fallows, so the task of cutting trees is not new to them, as it is in other countries.
The above analyses assess feasibility of improved fallows on very small plots;
but will farmers be able to manage them on a scale needed to maintain fertility on
their entire farms? A family manually cultivating 1.2 ha needs to plant 0.27 ha
improved fallow each year, assuming they crop for 3 years following each 2-year fal-
 Assessing Improved Fallows Adoption Potential 47

70
8
2 7
60
6
50
7
Work days

40
3 Improved
fallow
30
54 55 52 Current
46
20 38
30
10

0
Sept. Oct. Nov. Dec. Jan. Feb.

Fig. 3.2. Extra monthly labour required for practising improved fallows (0.27 ha),
as compared to current labour used for maize production, eastern Zambia.

low.4 Planting 0.27 ha in pure-stand sesbania increases labour use by 35 work days,
5–16% per month during the September–February peak period (Fig. 3.2).
Fortunately, the extra labour is distributed fairly equally among the peak months;
however, it is still a substantial amount. The other improved fallow options, pure-
stand tephrosia, intercropped sesbania, or intercropped tephrosia have much lower
labour requirements, increasing labour use by 22, 16 and 3 days per 0.27 ha,
respectively (Fig. 3.3). Intercropped tephrosia, the least labour-intensive option,
increases labour use during the peak months by less than 2%.

Profitability

In the economic analysis, the benefits of improved fallows, relative to continuously

cropped maize, were: labour saved in years 1 and 2 because maize was not planted,
fuelwood production in year 2, increases in maize yields in years 3–5, and reduced land
preparation and weeding costs in the first post-fallow maize crop. Added costs included
sesbania seed, labour for establishing the nursery, transplanting and maintaining the
fallow, and labour for harvesting and threshing the increased maize produced.
Maize yields following the improved fallows averaged 3.6 t ha−1, as compared
to yields of 1.0 t ha−1 for continuous, unfertilized maize and 4.4 t ha−1 for continu-
ous, fertilized maize (Table 3.4). There was no significant difference in the response
to improved fallows or to fertilizer between the 2 years surveyed, 1996 and 1997.
The post-fallow plot out-yielded the unfertilized plot on all 12 farms, and the fertil-
ized plot on four of the 12 farms.
4 It is assumed that the family owns 1.35 ha and uses 119 work days cultivating 1.2 ha
maize. Using sesbania improved fallows and, allowing for the use of only 119 work days,
the family could cultivate 0.81 ha maize and plant 0.27 ha in improved fallows each year.
48 S. Franzel et al.

40
36
35

30
Work days/0.27 ha

25
22
20
16
15

10

5
3
0
Sesbania Tephrosia Sesbania Tephrosia
pure stand pure stand intercropped intercropped

Fig. 3.3. Extra labour (work days/0.27 ha) for different improved fallows options.

Results of the economic analysis of the 12 farms, using average values across
farms, are summarized in Table 3.5; the detailed budgets for improved sesbania fal-
lows and fertilized and unfertilized maize are shown in Appendix 3.1. Over a 5-year
period, a hectare of improved fallows required 11% less labour than a hectare of
unfertilized maize and 32% less labour than fertilized maize. Relative to unfertilized
maize, the improved fallow increases total maize production per hectare over the 5-
year period by 77%, even though it does not produce maize during the first 2 years
of the fallow. But fertilized maize gives the highest 5-year maize yield, 2.5 times that
of improved fallows. The value of fuelwood produced in the fallow was low, only
about 3% of the value of maize following the improved fallow.5
For economic data, two scenarios are presented, one using prices for a year fol-
lowing a bumper harvest when prices were low (1996) and one following a poor har-
vest when prices were high (1998). Values in both years are expressed in 1998 US
dollars, taking into account inflation between 1996 and 1998. In the analysis of
returns to land using 1996 prices, net present values (NPVs) per hectare for fertilized
maize were over 30% higher than those of improved fallows; both were much higher
than for unfertilized maize. Six of the 12 farmers obtained higher NPVs for
improved fallows than for fertilized maize; 11 obtained higher NPVs for improved
fallows than for unfertilized maize. Using 1998 prices, NPVs for fertilized maize were
over double those for improved fallows, because of the much higher maize prices.
A main disadvantage of improved fallows relative to continuous maize is that
farmers have to wait until after the fallow to recoup their investment; in continuous
maize farmers earn positive net benefits in the first year. The payback period, that is,

5 The value of sesbania wood varies: in some areas, farmers burn the wood in the field to get
rid of it, whereas in other areas, they carry it to the homestead to use as fuelwood.
 Assessing Improved Fallows Adoption Potential 49

Table 3.4. Maize yield (kg ha−1) following 2-year Sesbania sesban improved fal-
lows, as compared to yields in continuous unfertilized and fertilized maize, type 2
and type 3 trials, 1996 and 1997.
II. Maize
I. Continuous following III. Continuous
unfertilized improved fertilized
maize fallow maize Ratio: II/I
Phiri Tikozenji 440 3000 2860 6.8
Isaac Phiri 1310 4720 3630 3.6
Whyson Mbewe 2200 5010 5100 2.3
Harrison Chogwe 960 3790 4580 3.9
Maine Mwale 700 3560 4150 5.1
Lazarus Mwanza 970 2760 6570 2.8
Peniyas Tembo 190 1420 2820 7.5
[Link] 1300 2300 5100 1.8
Z. Mwanza 300 4400 3700 14.7
M. Jere 1100 3500 4200 3.2
P. Nthani 800 4800 4200 6.0
J. Zulu 1300 4400 5700 3.4
Mean 964 3638 4384 3.8
SD 548 1108 1108
aFallow period was for 3 years.

the period required for improved fallows to yield higher cumulative net present val-
ues than unfertilized maize, was 3 years for 10 of the 12 farmers. This indicates that
even without residual maize yield increases during the second and third post-fallow
maize harvests, improved fallows were still more profitable than unfertilized maize
for 10 of the 12 farmers.
Assessing returns to labour is more relevant to most Zambian farmers than
returns to land, because labour tends to be scarcer than land. On returns to labour,
improved fallows outperformed unfertilized maize by a wide margin and fertilized
maize narrowly, using average values across the 12 farms and 1996 prices (Table
3.5). Improved fallows gave higher net returns to labour than for unfertilized maize
on 11 of the 12 farms and higher net returns to labour than for fertilized maize on
8 of the 12 farms. Even assuming no maize yield response to improved fallows in
year 4 and year 5, returns to labour on improved fallows were higher than those for
unfertilized maize on 10 of 12 farms. Using 1998 prices, fertilized maize had higher
returns to labour than improved fallows. In summary, improved fallows had much
higher returns to land and labour than unfertilized maize but lower returns to land
than fertilized maize. On returns to labour, the improved fallows did better using
1996 prices while fertilized maize did better using 1998 prices.
The performance of improved fallows relative to continuous, unfertilized maize
is fairly stable under a wide range of possible changes in parameters (Table 3.6). For
example, improved fallows have returns to land and labour at least double those of
50 S. Franzel et al.

Table 3.5. Labour requirements, maize production, and returns to land and
labour of Sesbania sesban improved fallows and continuously cropped maize
over a 5-year period, using an average farm budget.a
Returns to land: Returns to labour:
Work Tons net present value net returns
days maize (US$ ha−1) (US$) per work day
Option ha−1 ha−1 1996 1998 1996 1998
Continuous 499 4.8 6 6 0.47 0.79
unfertilized maize
Improved 2-year 441 8.5 170 215 1.11 1.64
sesbania fallow
Continuous fertilized 645 21.9 229 544 1.04 2.18
maize
aMeans of values from individual budgets of the 12 trial farmers were used.
Monetary values for 1996 and 1998 are in 1998 constant US dollars adjusted for
inflation. Details on budgets and coefficients are provided in Appendix 3.1.

unfertilized maize under most tested changes, including a 500 kg decline in post-
fallow maize yields, and 50% increases or decreases in the discount rate, and the
prices of fertilizer and labour. An increase in post-fallow maize yield of only 1.1 t
ha−1 is needed in the third year to cover the costs of establishing and maintaining
the fallow, relative to unfertilized maize, in terms of returns to land or labour.
In contrast, the performance of improved fallows relative to continuous, fertil-
ized maize is sensitive to changes in key parameters (Table 3.6). Increases in maize
prices (such as between 1996 and 1998) raise the returns to fertilized maize at a
much faster rate than they raise the returns to improved fallows. Similarly, the rela-
tive profitability of the two practices is highly sensitive to the price of fertilizer;
reductions in fertilizer price greatly increase the profitability of fertilized maize rela-
tive to improved fallows. Changes in the discount rate and in the cost of labour and
seedlings have little effect on the performance of improved fallows relative to fertil-
ized maize.
The risk of drought is critical for farmers in Zambia; unfortunately the effects
of drought in the season following an improved fallow cannot be assessed using the
data collected for this study. But there are four reasons why improved fallows are
likely to be much less risky than fertilized maize. First, in the event of a complete
crop failure, a farmer using fertilizer would lose his investment in fertilizer, US$154
ha−1, whereas a farmer with improved fallow would lose his investment in planting
and maintaining the trees, only about US$90 ha−1 (using 1998 prices). In addition,
both farmers would lose their investment in growing maize that year. Secondly,
whereas nearly all of a farmer’s investment in fertilizer is in cash terms, improved
fallows require little or no cash input. The opportunity cost of cash is extremely
high, and in the case of the farmer buying fertilizer on credit, loss of the maize crop
may result in substantial losses in productive capacity in order to repay the loan.
Thirdly, the benefits of improved fallow are likely to be spread over a 3-year period,
 Assessing Improved Fallows Adoption Potential 51

Table 3.6. Sensitivity analysis showing the effects of changes in parameters.a

Continuous Continuous
unfertilized maize Improved fallows fertilized maize
Returns Returns Returns Returns Returns Returns
to land to labour to land to labour to land to labour
Base analysis 6 0.47 170 1.11 229 1.04
Maize price + 50% 114 0.83 330 1.73 718 2.31
Maize price − 50% −101 0.11 10 0.49 –260 –0.23
Labour price + 50% −61 0.47 117 1.14 142 1.04
Labour price − 50% 73 0.47 223 1.05 315 1.04
Discount rate 30% 5 0.47 119 1.02 186 1.04
instead of 20%
Discount rate 10% 8 0.47 246 1.20 290 1.04
instead of 20%
Seedling cost + 50% 6 0.47 164 1.06 229 1.04
Seedling cost − 50% 6 0.47 176 1.16 229 1.04
Fertilizer price + 50% 6 0.47 170 1.11 –11 0.42
Fertilizer price − 50% 6 0.47 170 1.11 469 1.66
Yield response to 6 0.47 209 1.25 229 1.04
improved fallows
+ 500 kg ha−1
Yield response to 6 0.47 131 0.96 229 1.04
improved fallows
− 500 kg ha−1
No response to 6 0.47 72 0.89 229 1.04
improved fallow
(years 4–5)
No response to 6 0.47 138 1.09 229 1.04
improved fallow
(year 5)
aValues are based on 1996 prices expressed in 1998 US dollars.

whereas those of nitrogen fertilizer take place in a single year. Thus in the above
case, where a farmer’s crop fails in the first post-fallow season, there is likely to be a
substantial response the following year. Fourthly, improved fallows appear to
improve the soil structure and organic matter content of the soil, thus enhancing
the soil’s ability to retain moisture during drought years.
The above analysis of profitability examines returns per hectare; but how will
adoption of improved fallows affect farm income once they have been incorporated
into the farming system? A farm household cultivating manually and having 1.4 ha
and 120 work days available for cultivating maize would earn US$262 year−1 using
fertilized maize, US$225 year−1 growing improved fallows, and only US$95 year−1
cultivating continuous maize without fertilizer (Table 3.7). Even if there is no resid-
ual effect on maize yields in the third year following improved fallows, earnings are
still twice as high as on unfertilized maize.
 52

Table 3.7. Farm models comparing net returns to labour per year of a 1.4 ha farm practising Sesbania sesban improved fallows with
farms cultivating continuous maize, with and without fertilizer.a
Farm practising improved fallows (farm adds 0.28 ha of Farm with unfertilized maize Farm with fertilized maize
improved fallow per year) (1.2 ha cultivated) (0.92 ha cultivated)
Work Maize Net Work Maize Net Work Maize Net
Area days production returns days production returns days production returns
Crop (ha) year−1 (kg year−1) (US$ year−1) Crop (year−1) (kg year−1) (US$ year−1) Crop year−1 (kg year−1) (US$ year−1)
Fallow, 1st year 0.28 31 0 0 Maize 120 1157 95 Maize 120 4077 262
Fallow, 2nd year 0.28 1 0 1
Maize 1st post- 0.28 7 1026 97
fallow
Maize 2nd post- 0.28 31 800 75
fallow
S. Franzel et al.

Maize 3rd post- 0.28 30 573 52

fallow
Total 1.4 120 2390 225
Net returns to labour if maize in the 3rd post- 195
fallow season yields the same as on the farm
with unfertilized maize
aHousehold is assumed to have only 120 work days available during the cropping season for maize production; the amount needed to
cultivate 1.2 ha maize manually without using fertilizer. Costs and returns are from Appendix 3.1. Improved fallows are 2 years in length
and are followed by 3 years of maize crops.
 Assessing Improved Fallows Adoption Potential 53

Farmer interest and innovation

Determining farmers’ interest in improved fallows is problematic, as 5 years are

required for a full cycle of the technology, that is, 2 years of fallow and 3 years of
cropping. Moreover, farmers often state that they like a technology, even when they
do not, because they hope to obtain material or social benefits from interacting with
facilitators, or because of cultural taboos against criticism. Nevertheless, there is
considerable evidence that farmer interest in improved fallows is extremely strong.
First, there has been a rapid expansion in the number of farmers testing and using
the technology: the number of type 3 trials, for which farmers receive no incentive
for planting the trees, rose from 37 in 1994–1995 to about 2800 in 1996–1997. By
2000, roughly 10,000 farmers in Eastern Province were planting improved fallows.
Secondly, most farmers planting the improved fallows continue to plant them: in a
study of 100 farmers planting improved fallows in 1996–1997 or earlier, 71% con-
tinued to plant them over the following three seasons (Keil, 2001). Thirdly, even
farmers with low survival rates have continued testing improved fallows on parts of
their plots where survival was higher (71% of farmers with low survival rates) or
have planted new improved fallow plots the following year (32%).
The main intention of farmers in using the technology is undoubtedly to
increase returns to land and labour. But it is not clear how they will actually incor-
porate the technology into their farming system. For example, in workshops to dis-
cuss the potential impact of improved fallows, some expressed interest in rotating
improved fallows around their farm, so that their supply of maize would be stable
and all areas of their farm would benefit from the fallows. Some intended to use the
technology to reduce their area cropped to maize, so as to be able to devote more
land to cash crops (Place, 1997).
Fuelwood availability varies considerably within the province and thus farmers’
interest in using the trees for fuelwood is variable. Some claimed that the larger ses-
bania trees were good substitutes for the fuelwood they collected from off the farm.
Others mentioned that the quality of the wood was poor and that it was useful only
as kindling. Tephrosia and cajanus produce lower-quality fuelwood than sesbania.
Farmers’ preferences between sesbania and tephrosia vary considerably. Maize
yields following sesbania fallows are generally higher than those following tephrosia
fallows, but some farmers prefer tephrosia because it can be direct seeded and thus
requires less labour than sesbania. In type 2 trials in 1994/95, 42% chose sesbania
and 30% tephrosia; in type 3 trials the following year 64% chose sesbania and 32%,
tephrosia. By 2000, tephrosia had surpassed sesbania; 85% of farmers with
improved fallows were planting tephrosia while 69% were planting sesbania (Keil,
2001).
Farmers’ innovations of the improved fallow technology have been one of the
main elements contributing to its success. Two of the main technological options,
bare-root seedlings instead of potted seedlings, and intercropping instead of plant-
ing in pure stands, were innovations that farmers introduced in type 3 trials in the
early 1990s. In these trials, farmers were given potted seedlings grown at farmer-
training centres, but to reduce the cost of transporting them to their farms, they
54 S. Franzel et al.

removed the seedlings from the pots and carried them ‘bare-rooted’ in basins. When
farmers’ plantings of these seedlings proved successful, researchers conducted type 1
trials to compare the performance of bare-rooted seedlings, grown in raised
seedbeds, with potted seedlings. They found no significant difference in perfor-
mance and as potted seedlings were much more costly to produce, they were phased
out (Kwesiga et al., 1999).
The farmers’ second main innovation, intercropping during the year of tree
establishment, was also tested later in on-farm trials. The trials found that inter-
cropping reduces maize yields and tree growth during the year of establishment, but
many farmers prefer it because it economizes on land and labour use relative to
planting in pure tree stands. Intercropping appears to be increasing; the percentage
of farmers practising it rose from 17% during the planting of 1994/95 type 2 trials
to 42% in the 1995/96 type 3 trials.
Researchers have also noted several other farmer innovations. One farmer
planted seedlings into a bush fallow without preparing the land first. Another
planted sesbania seedlings behind the ox-plough; as the plough moved along an
adjacent furrow, it covered the seedling roots with soil. Several farmers gapped up
their sesbania fields with seedlings planted 1 year after the first planting. Several
planted sesbania at weeding time into parts of fields where maize was performing
poorly. Some also tested the effect of improved fallows on crops other than maize,
such as sunflower and groundnuts.

Gender and wealth

The data support the hypothesis that females were able to manage improved fallows
as well as males. In the type 3 trial planted in 1995/96, half of the participants were
females and they had somewhat higher survival rates for sesbania than males. For
example, 47% of females and 29% of males had survival rates for sesbania of over
75%, 6 months after planting. For tephrosia, males had somewhat higher survival
rates. Males and females reported similar problems with similar frequency and did
not differ in the number of times they weeded their trees. But females had smaller
plot sizes, 332 m2, as compared to 679 m2 for males. Since the same percentage of
males and females stated that they had obtained enough planting material, it
appears that females wanted smaller plots than males.
In the four villages where a census was conducted on the use of improved fal-
lows in 1998, 32% of the male-headed households and 23% of the female-headed
households had planted improved fallows. There was no significant difference across
villages between the two proportions. Moreover, whereas single females are often
disadvantaged relative to female heads of household whose husbands live away
(Bonnard and Scherr, 1994), the data showed that the same proportions of these
two groups were testing the technology. In fact, Peterson (1999) found a higher
proportion of single females planting improved fallows than married females; the
latter needed permission from their husbands and thus were sometimes prevented
from testing the practice.
 Assessing Improved Fallows Adoption Potential 55

As could be expected, there was an association between wealth level and plant-
ing improved fallows (log linear model, P < 0.08). As wealth status declined, the
proportion of farmers planting improved fallows also declined. Whereas 53% of the
well-off farmers planted fallows, 40% of the fairly well off, 22% of the poor and
16% of the very poor planted fallows (Phiri et al., 2001). Interestingly, though, the
proportion of farmers continuing to plant improved fallows after their first planting
did not appear to vary by wealth status; 59% of the farmers in the well-off and
fairly well-off groups expanded whereas 58% of the farmers in the poor and very
poor groups expanded (Peterson, 1999)

Institutional Support

In some areas of Eastern Province, the planting of improved fallows has spread
almost spontaneously, that is, largely through the efforts of farmers without much
support from researchers or extension staff. For example, during the dry season,
1995, a lorry load of 78 farmers arrived unannounced at Msekera Research Station.
The farmers came from Kapinde, a village bordering a village having on-farm trials,
and they had hired the truck to come to the station to learn about improved fal-
lows. The farmers were members of self-help groups and were accompanied by their
camp officer. Project staff gave them a tour of the station and nearby on-farm trials.
The farmers were given sesbania seed and instructions on raised-bed nursery meth-
ods to produce bare-rooted seedlings. Several months passed without contact but in
December, project staff went to visit the village, arriving unannounced. The camp
officer quickly assembled some of the group leaders and accompanied Chipata staff
to the nursery. It was well-managed, weed-free and well-watered, with 40,000
seedlings ready for planting. Seventy-one farmers planted improved fallows using
seedlings from the nurseries.
But in most cases, substantial support, that is, training and planting material, is
required. In the above case, Kapinde farmers had significantly lower survival rates
than in nearby camps, probably because of less training and experience. The
Zambia/ICRAF project has helped facilitate the establishment of an informal net-
work to conduct adaptive research, training and facilitate dissemination of
improved fallows (Fig. 3.4). The network has two functions: to provide coordinated
and analytical mechanisms for participatory monitoring and evaluation of on-farm
research in improved fallows, and to act as a catalytic and action-oriented group for
the widespread dissemination of the technology. The network began when the pro-
ject started supplying planting material, training, and information to extension ser-
vices, development projects, NGOs and farmer groups that wanted to help their
members test improved fallows. In exchange, these organizations provided the pro-
ject with feedback on the performance of the technology.
The network is based on the principle that adaptive research and extension are
really two sides of the same coin; once on-farm research has confirmed that a tech-
nology has adoption potential, dissemination is already beginning. Researchers need
to be involved to obtain feedback from farmers and extension staff on problems and
 56

er + Farmer v
Farm isits
FARMERS AND FARMER GROUPS

NGOs with own GRZ Camp Org. working through Gov. of Zambia extension service
extension staff officers
NGOs Donor projects GRZ

Man a
German Lutheran

g
Reform Catholic Finnida Danish Adaptive Soil cons.

e
Tech. World
Service Federation GRZ Block Church Church project research Agroforestry

ment
office planning extension

Visits
team project

of an
to res
d fee
GRZ

d
V

Fe
b
ing

Visits to research stati

earch
a

ed
Adaptive research planning GRZ District

ba
c
train isits to
teams in other provinces

k
[Link]

e
ck fr- o

,
statio

Ad on
n
res
Copper- Luapula Central Western ls,

training materials, Visit

vic pe
era rainin
ia

ini
m on
Data from on-farm trial

rf
er

rt
belt at

Genand t earch

,
m

tra orm
s
in

s
Key

a
g

ppo mat

e
farm
in e
in s anc ials

ng nce
Fe Adv Other organizations testing improved
S. Franzel et al.

eria
i

l su g tation
ed

,p
tra nce sist g tr GRZ Prov.

,
ls
d a s n fallows outside Eastern Province

trials
ba ce, t
ra an w a si

s, feedback
ck agric. office

ed robl
in g llo t, vi t

ons seed, training, allow

n in a por per o r r t s

em
pr on p ing i
, p u pp po

s
ob e
lem rfo see Tra su s ral su ss re Other organizations testing improved
s rma d ne re fallows in Eastern Province

s to other areasances,
nc Ge t prog
e, n
que
Fre
ICRAF- General support GRZ Director of Extension service, Ministry of Agriculture
ZAMBIA TEAM Frequent Agric. Lusaka
eed Priority areas requiring additional support
ing, s ce,
v ic e, train erforman GRZ Government of the Republic of Zambia
Ad on p
back lems
Feed prob

Fig. 3.4. Institutional links between ICRAF/Zambia team, farmers and other organizations testing improved fallows.
 Assessing Improved Fallows Adoption Potential 57

to identify researchable issues. Moreover, the more extension staff become involved
in on-farm research, the more knowledgeable and enthusiastic they will be in
extending the practice. Their involvement helps to save scarce research resources
and improves the feedback to research.
The extension service in Zambia is a full partner in the on-farm research. In
fact, about half of the type 2 trials were laid out by extension staff in the absence of
researchers. Extension staff also play an important role in supporting the village
nurseries and in monitoring the trials. They view the trials as joint research–exten-
sion work. Relations are also excellent at higher levels. Throughout the system, the
managing of on-farm trials is seen as a normal duty of extension and NGO staff,
rather than a burden imposed on them from outside. Development projects pro-
vide some incentives to extension staff, such as bicycles and lunch allowances,
which facilitate institutional linkages and raise the effectiveness of the extension
staff. That only one researcher and technician from the Zambia/ICRAF project
were involved in the establishment and monitoring of the hundreds of on-farm tri-
als in the mid-1990s attests to the strength of the network.
Development projects, NGOs and community-based organizations are also
active members of the network. The World Vision International Zambia Integrated
Agroforestry Project (financed by the United States Agency for International
Development) plays a major role in promoting improved fallows. Farmer groups
are also important members of the network. The groups (often called clubs) are
generally composed of farmers in a single village; sometimes they are limited to a
specific group such as females or youths. Most have several different self-help activ-
ities, but some were formed specifically to promote improved fallows. The groups
manage nurseries, distribute seed and seedlings, and exchange knowledge, training
and experience on improved fallows.
The first meeting of the network was hosted by the Zambia/ICRAF project in
April 1996 and was attended by 75 representatives of extension services, projects,
NGOs and farmer groups (FSRT/SCAFE, 1997). Participants planned for the
wider testing of improved fallows, reviewed the problems and the state of knowl-
edge about them, and developed a draft extension manual. The network meets 1–2
times yearly and meetings are chaired by the Provincial Agricultural Officer, who
leads the government’s agricultural programmes in the province. Representatives of
extension and NGOs present the progress they have made and problems encoun-
tered in helping farmers use agroforestry. Farmers present their experiences and
researchers report on the results of on-station and on-farm trials. Participants also
plan training exercises and germplasm production and distribution.

Conclusion

Improved fallows have generated a high degree of enthusiasm among farmers,

extension staff, NGOs and researchers. But several critical biophysical, socioeco-
nomic, institutional and policy issues remain. On the biophysical side, the farmers’
most important problem was poor tree growth and the low tree survival rate. Poor
58 S. Franzel et al.

rainfall was the main contributing factor, followed by poor nursery and transplant-
ing techniques and attacks by the mesoplatys beetle. Despite poor rainfall in some
years, the quality of fallow stands will likely improve as farmer’s knowledge of nurs-
ery and tree management increases. But sustained training of extension staff and
farmers is needed to ensure progress in improving survival rates. Efforts to address
the beetle problem focus on improving tree management, especially early planting,
so that trees will be better able to withstand the attacks, and selecting sesbania
provenances that the beetles do not attack. Researchers at Makoka Research
Station, Malawi, have identified one such provenance.
Type 1 trials established across Zambia and three other countries in southern
Africa confirmed that sesbania does not perform well on sandy soils, because of
nematode attacks, or on shallow soils, because of mortality during the dry season
(ICRAF, 1996). These problems, along with that of mesoplatys, highlight the need
to screen new species for improved fallows. Gliricidia sepium is the most promising
new species; farmers like it because it can be coppiced, that is, it regrows after cut-
ting. Thus it is not necessary to replant it after using it in an improved fallow.
Several socioeconomic issues are also critical. Research is needed to monitor
farmers’ uptake of improved fallows and to assess their appropriateness for different
types of farmers in the area. Researchers need to monitor carefully how those farm-
ers testing the practice differ from those not testing and, eventually, how adopters
differ from those rejecting the practice. The most important variables influencing
adoption appear to be wealth level and gender.
Preliminary assessments presented in this chapter suggest that improved fallows
are a gender-neutral technology; women are planting them about as frequently as
men. Although wealth is associated with planting improving fallows, the fact that
about one-sixth of the poor and very poor farmers are planting them suggests that
there are no absolute constraints preventing these groups from doing so. More work
is needed on understanding why some poor households participate while others do
not. For example, participation in farmer groups may be an important factor.
An additional variable, labour availability at the time of planting, also needs to
be assessed. Labour availability does not appear to be constraining farmers from
testing improved fallows, but it may be important in limiting the area a farmer
allocates to the practice. Farmers’ apparent lack of concern about devoting labour
to planting trees during their peak period of labour use reflects their high expecta-
tions concerning the benefits of the practice. Nevertheless, increased attention is
needed for reducing labour requirements of improved fallows through such tech-
niques as coppicing species, direct seeding and intercropping trees with maize dur-
ing the year of establishment.
Several other socioeconomic issues need to be assessed. The economics of
alternative improved fallow practices (e.g. sesbania versus tephrosia; pure stand ver-
sus intercropping; trees versus herbaceous legumes) needs to be examined. Also, as
farmers expand their use of improved fallows, impact assessment is important to
determine the financial and environmental effects of the technology at the farm,
household and village scales. For example, will widespread adoption of improved
fallows reduce deforestation, that is, tree felling in the woodlands for fuelwood?
 Assessing Improved Fallows Adoption Potential 59

Will farmers use improved fallows to reduce their area under maize so as to grow
more cash crops? Researchers have conducted two village-level workshops to find
out farmers’ views on potential impacts and are planning participatory monitoring
of the uptake of improved fallows and their impact in selected villages (Place,
1997).
On the institutional side, priority needs to be given to developing sustainable
systems of producing and distributing seed; in 1996, it was necessary to import 7
tons of sesbania seed from Kenya to meet seed demand. Seed orchards are now in
place and the area is self-sufficient in seed, but production and distribution are
largely through NGOs, ICRAF and other organizations, the private sector and
farmer organizations play less of a role. The advantages and disadvantages of collec-
tive and private nurseries also need to be examined. Large, collective nurseries facil-
itate the training of farmers, but dividing labour duties and distributing seedlings
have proved problematic. Farmer groups need to be strengthened to take over some
of the functions of research and extension (Ashby and Sperling, 1995). Such devo-
lution not only reduces the costs of research and extension; it permits increased
scale of testing and more appropriate research to meet farmers’ needs. For example,
group members could decide on research issues they want to investigate (e.g. inter-
cropping versus pure stand) and could allocate responsibilities among members. At
the end of the experimental cycle, the testers could draw their own conclusions and
present them to the community.
Policy research can help identify how communities can control the free grazing
of livestock to limit browsing damage to improved fallows, as well as to off-season
gardens and woodlots. Several communities are currently trying to regulate brows-
ing; their progress and lessons learned need to be monitored (FSRT/SCAFE,
1997). An alternative means to addressing the grazing problem is to identify tree
species that are not palatable to livestock. Also, if improved fallows increase the
quantity of produce marketed, marketing constraints, already severe, may be exac-
erbated. With market liberalization in the early 1990s, parastatal marketing of
maize and cash crops has greatly declined and the private sector has not filled the
gap.
In spite of the problems, much progress has been made in developing and dis-
seminating improved fallows and it is useful to identify key elements that have con-
tributed to the achievements thus far. First, diagnostic surveys were effective in
identifying farmers’ problems and drawing up research projects that could help
solve them and fit farmer circumstances. Secondly, farmers participated in the
design and testing of improved fallows at an early stage and their feedback was crit-
ical in modifying the technology. The early experimenters also played an important
role, hosting visits from other farmers and training them. Thirdly, researchers and
farmers tested a wide range of management options, not just a single prototype,
and encouraged farmers to modify and innovate. Fourthly, an adaptive research and
dissemination network was critical for mobilizing research, extension, NGOs and
farmers, and for testing and extending the technology in new areas. Emphasis on
these four elements can help promote success in technology development and dis-
semination efforts elsewhere.
60 S. Franzel et al.

Acknowledgements

The authors are grateful to Stanislas Phiri for assisting in the collection of the data,
to Josefynne Miingi for help in analysing the data, and to Thomas Raussen and Sara
Scherr for reviewing earlier drafts of the manuscript.

References
ARPT (1991) Report of an informal survey on ogy environment right: lessons from maize
degraded lands in Eastern Province, development in Southern Africa. Paper pre-
Zambia. Adaptive Research Planning Team, sented at the Meeting of the International
Department of Agriculture, Chipata. Association of Agricultural Economics,
Ashby, J. and Sperling, L. (1995) 10–16 August, Sacramento, CA.
Institutionalizing participatory, client-dri- ICRAF (1996) Annual Report, 1995. ICRAF,
ven research and technology development Nairobi.
in agriculture. Development and Change 26, Jha, D. and Hojjati, B. (1993) Fertilizer Use on
753–770. Smallholder Farms in Eastern Province,
Bonnard, P. and Scherr, S. (1994) Within gender Zambia. Research Report No. 94,
differences in tree management: is gender International Food Policy Research
distinction a reliable concept? Agroforestry Institute, Washington, DC.
Systems 25, 71–93. Keil, A. (2001) Improved fallows using legumi-
Celis, R. and Holleman, C. (1991) The effects of nous trees in eastern Zambia: do initial
adopting technology on the household use testers adopt the technology? MSc thesis,
of labour. In: Celis, R., Milimo, J.T. and Institute of Rural Development, Georg-
Wanmali, S. (eds) Adopting Improved Farm August-University of Goettingen, Germany.
Technology: a Study of Smallholder Farmers in Kwesiga, F.R. and Coe, R. (1994) The effect of
Eastern Province, Zambia. IFPRI, short rotation Sesbania sesban planted fal-
Washington, DC. lows on maize yield. Forest Ecology and
Cooper, P.J., Leakey, R.R.B., Rao, M.R. and Management 64, 199–208.
Reynolds, L. (1996) Agroforestry and the Kwesiga, F.R., Phiri, D., Mwanza, S. and
mitigation of land degradation in the Simwanza, P.C. (1995) Zambia/ICRAF
humid and sub-humid tropics of Africa. Agroforestry Research Project (1995) Annual
Experimental Agriculture 32, 235–290. Report. AFRENA Report No. 98. ICRAF,
DOA (1991) A Handbook for Agricultural Nairobi.
Extension Workers. Department of Kwesiga, F.R., Franzel, S., Place, F., Phiri, D. and
Agriculture, Eastern Province, Chipata. Simwanza, C.P. (1999) Sesbania sesban
FSRT (1995) Annual Report 1994/95. Farming improved fallows in eastern Zambia: their
Systems Research Team – Eastern Province. inception, development, and farmer enthu-
Ministry of Agriculture, Food, and siasm. Agroforestry Systems 47, 49–66.
Fisheries, Chipata. Peterson, J.S. (1999) Kubweletza Nthaka: ethno-
FSRT/SCAFE (Farming Systems Research Team graphic decision trees and improved fallows
– Eastern Province, and Soil Conservation in the Eastern Province of Zambia.
and Agroforestry Extension Programme – University of Florida/Gender and Soil
Eastern Province) (1997) Proceedings of the Fertility in Africa Collaborative Research
Consultative Workshop on Integrated Soil Support Program, Gainesville, Florida.
Fertility Management on Small Scale Farms Peterson, J.S., Tembo, L., Kawimbe, C. and
in Zambia’s Eastern Province, 13–19 October Mwang’amba (2000) The Zambia
1996, Chipata, Zambia. Integrated Agroforestry Project baseline sur-
Howard, J.A., Rubey, L. and Crawford, E.W. vey: lessons learned in Eastern Province
(1997) Getting technology and the technol- Zambia. World Vision International/
 Assessing Improved Fallows Adoption Potential 61

University of Florida/ICRAF/Ministry of Dambos in Eastern Province. Ministry of

Agriculture, Gainesville, Florida. Agriculture, Food, and Fisheries, Chipata.
Phiri, D., Franzel, S., Mafongoya, P., Jere, I., Sanchez, P.A., Shepherd, K.D., Soule, M.J.,
Katanga, R. and Phiri, S. (2001) Who is Place, F.M., Buresh, R.J., Izac, A.N., Uzo
using the new technology? A case study of Mokwunye, A., Kwesiga, F.R., Nderitu,
the association of wealth status and gender C.G. and Woomer, P.L. (1997) Soil fertility
with the planting of improved tree fallows in replenishment in Africa: an investment in
Eastern Province, Zambia. ICRAF, Nairobi. natural resource capital. In: Buresh, R.J.,
Place, F. (1997) A methodology for a participa- Sanchez, P.A. and Calhoun, F.G. (eds)
tory approach to impact assessment: impli- Replenishing Soil Fertility in Africa. SSSA
cations from village workshops in Eastern Special Publication Number 51. Soil
Province, Zambia. ICRAF, Nairobi. Science Society of America and American
Place, F., Mwanza, S. and Kwesiga, F.R. (1995) A Society of Agronomy, Madison, Wisconsin,
cost–benefit analysis of improved fallows in pp. 1–46.
Eastern Province, Zambia. ICRAF, Nairobi. Zambia/ICRAF (1988) Agroforestry Research
Raintree, J. and Warner, K. (1986) Agroforestry Project for the Maize-Livestock System in the
pathways for the intensification of shifting Unimodal Upland Plateau in Eastern
cultivation. Agroforestry Systems 4, 39–54. Province, Zambia. ARENA Report No. 10.
Raussen, T., Daka, A.E. and Bangwe, L. (1995) ICRAF. Nairobi, Kenya.
Appendix 3.1. Cost–benefit analysis of improved fallow and cropping options (ZK ha−1).
62

Maize cropping without fertilizer Two-year sesbania fallow Maize cropping with fertilizer
Year 1 Year 2 Year 3 Year 4 Year 5 Year 1 Year 2 Year 3 Year 4 Year 5 Year 1 Year 2 Year 3 Year 4 Year 5
Costs
Cash costs
Maize seed 22.24 22.24 22.24 22.24 22.24 22.24 22.24 22.24 22.24 22.24 22.24 22.24 22.24
Nursery costs 2.93
Fertilizer 0.00 0.00 0.00 142.82 142.82 142.82 142.82 142.82
Fertilizer transport 0.00 0.00 0.00 6.40 6.40 6.40 6.40 6.40
Total cash costs 22.24 22.24 22.24 22.24 22.24 2.93 22.24 22.24 22.24 171.46 171.46 171.46 171.46 171.46
Labour
Tree nursery 10.52
Land preparation 12.00 12.00 12.00 12.00 12.00 12.00 9.00 12.00 12.00 12.00 12.00 12.00 12.00 12.00
Ridging 4.00 4.00 4.00 4.00 4.00 4.00 3.00 4.00 4.00 4.00 4.00 4.00 4.00 4.00
Planting maize 2.00 2.00 2.00 2.00 2.00 0.00 2.00 2.00 2.00 2.80 2.80 2.80 2.80 2.80
Planting trees 0.00 0.00 0.00 0.00 0.00 11.24 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
1st weeding 8.00 8.00 8.00 8.00 8.00 8.00 6.00 8.00 8.00 10.00 10.00 10.00 10.00 10.00
S. Franzel et al.

2nd weeding 4.00 4.00 4.00 4.00 4.00 4.00 3.00 4.00 4.00 6.00 6.00 6.00 6.00 6.00
Tree cutting 0.00 0.00 0.00 0.00 0.00 0.00 2.08 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Harvesting maize 5.96 5.96 5.96 5.96 5.96 0.00 0.00 8.88 8.00 7.13 9.58 9.58 9.58 9.58 9.58
Maize shelling 3.96 3.96 3.96 3.96 3.96 0.00 0.00 6.88 6.00 5.13 7.58 7.58 7.58 7.58 7.58
Total labour costs 39.92 39.92 39.92 39.92 39.92 49.76 2.08 38.76 44.01 42.26 51.96 51.96 51.96 51.96 51.96
Total costs 62.16 62.16 62.16 62.16 62.16 52.68 2.08 61.00 66.25 64.50 223.42 223.42 223.42 223.42 223.42
Labour work days 99.8 99.8 99.8 99.8 99.8 124.4 5.2 96.9 110.0 105.6 129.9 129.9 129.9 129.9 129.9
Benefits
Maize 63.74 63.74 63.74 63.74 63.74 0.00 0.00 257.63 199.47 141.30 304.11 304.11 304.11 304.11 304.11
Fuelwood 0.00 0.00 0.00 0.00 0.00 0.00 6.40 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Total benefits 63.74 63.74 63.74 63.74 63.74 0.00 6.40 257.63 199.47 141.30 304.11 304.11 304.11 304.11 304.11
Net benefit to labour 41.50 41.50 41.50 41.50 41.50 −2.93 6.40 235.39 177.23 119.06 132.66 132.66 132.66 132.66 132.66
Net return to 0.42 0.42 0.42 0.42 0.42 −0.02 1.23 2.43 1.61 1.13 1.02 1.02 1.02 1.02 1.02
labour day−1
 Net benefits 1.58 1.58 1.58 1.58 1.58 −52.68 4.32 196.63 133.22 76.80 80.70 80.70 80.70 80.70 80.70
Work days 499.0 442.2 649.5
Net present value 4.74 168.00 241.33
Discounted days 0.24 0.21 0.31
Discounted net 124.12 271.54 396.72
benefit to labour
Discounted net 0.42 1.05 1.02
benefit/discounted
days
Quantity of maize 5 t/5 years 9 t/5 years 23 t/5 years
Prices are from local markets for the 1996 cropping season. Exchange rate: US$1.00 = 1250 Zambian Kwacha (ZK), 1996.
Cash costs
Maize seed: seed rate of 20 kg ha−1. Cost: 1340 ZK kg−1.
Nursery cash costs: total costs per seedling, including cash and labour costs, is 1.4 ZK, median from cost analysis of eight farmer nurseries. Mean
cost was 1.9 ZK, SD = 1.2. It is assumed that 12,000 seedlings are raised in order to achieve a density of 10,000 seedlings ha−1 in the field. Nursery cash
costs accounted for 22% of the total cost of the nursery and included rent of land in the valley bottom and purchase of a watering can.
Fertilizer: the recommended rate is 112–40–20 kg of N–P2O5–K2O per hectare. In 1996, it required 200 kg of D compound, purchased, at 459 ZK kg−1,
and 200 kg of urea, purchased at 433 ZK kg−1. 1998 prices were 580 ZK kg−1 and 520 ZK kg−1, respectively.
Fertilizer transport: estimated at 1000 ZK 50 kg bag−1, from Chipata to farm in 1996 and 1350 ZK/bag in 1998.
Labour: labour data for maize cultivation are assembled from DOA (1991), Kwesiga et al. (1995) and Place et al. (1995) and from survey farmers.
Labour data concerning trees are from surveyed farmers.
Labour cost: costed at 500 ZK work day−1 in 1996. A work day is assumed to involve 7 hours of work. Hiring labour is not common; reported wage rates
Assessing Improved Fallows Adoption Potential

were highly variable. 500 ZK day−1 represents the approximate average returns per labour in maize production for 1996, that is, the value of labour at which
a farmer growing maize without fertilizer breaks even. In 1998, this value was about 1300 ZK work day−1.
Nursery: see ‘nursery cash costs’ above. Activities included collecting and threshing seeds, constructing beds, collecting sand, compost and soil, planting,
covering with grass, watering, weeding, digging out the seedlings, and transporting them to the field. Mean number of work days required to produce
12,000 seedlings, sufficient to plant and gap up one hectare, was 26.8. (SD 22.7).
Land preparation and ridging: 30 and 10 work days ha−1, respectively. They are 25% less during the year after the improved fallow, according to esti-
mates of trial farmers.
63
Appendix 3.1. Continued.
64

Planting maize: 5 work days ha−1. When applying fertilizer, 7 work days ha−1.
Planting trees: 420 trees per day, median of data from 12 farmers (mean = 499, SD = 424).
Weeding: assumed to be the same for trees as for maize, as claimed by farmers. Weeding requirements decline by 25% during the year after the
improved fallow, according to estimates of trial farmers. Weeding requirements are assumed to increase 33% with fertilizer use.
Harvesting and post-harvest: labour varies with quantity. A yield of 1 t ha−1 requires 15 work days for harvesting and 10 days for post-harvest activities
(shelling and transportation). A yield of 4.6 t ha−1 is estimated to require 60% more harvest labour and 90% more post-harvest labour.
Benefits
Eleven of the 12 trial farmers had 2-year fallows; one had a 3-year fallow. For the purpose of comparison with the other sample farms for drawing up
enterprise budgets, we assumed that Phiri had a 2-year fallow. This assumption increased the net present values in the table above by 1% and the net
benefit day−1 by 1%.
Maize: yields are from the 12 trial farmers for the season following the improved fallow and are compared with yields on continuously cropped adjacent
fields, with and without fertilizer (Table 3.4). For the continuously cropped maize fields, yields are assumed to be constant over the 5-year period (964
kg ha−1 without fertilizer and 4384 kg ha−1 with fertilizer). Maize yields following the improved fallows are 3638 kg ha−1 for the first post-fallow season
(Table 3.4), 2836 kg ha−1 for the second, and 2034 kg ha−1 for the third season. The latter two figures are based on a 30% and 60% reduction in
response, as obtained in on-station trials. The maize price was 83 ZK kg−1, the estimated farm-gate price during the harvest period, 1996. The 1998
price was 167 ZK kg−1.
S. Franzel et al.

Fuelwood: fuelwood is not normally sold; yield is estimated at 4 t ha−1 and price at 2000 ZK t−1.
Discount rate: 20%.
 The Adoption Potential of
Short Rotation Improved
Tree Fallows: Evidence
4
from Western Kenya
R.A. SWINKELS,1 S. FRANZEL,2 K.D. SHEPHERD,2
E. OHLSSON3 AND J.K. NDUFA4
1TheWorld Bank, 8th Floor, 63 Ly Thai To Street, Hanoi, Vietnam;
2International
Centre for Research in Agroforestry, ICRAF House,
PO Box 30677, Gigiri, Nairobi, Kenya; 3Sida/SAREC, SE-105
25 Stockholm, Sweden; 4Kenya Forestry Research Institute (KEFRI),
Maseno Agroforestry Research Centre,
PO Box 25199, Kisumu, Kenya

Summary
This study assesses the economics of improved fallow in the densely populated areas
of western Kenya, where soils are nutrient depleted. A formal survey of 71 randomly
selected farmers showed that half fallow 10–50% of their land for at least one season,
mainly for soil fertility restoration. An improved-fallow prototype, established by
direct seeding Sesbania sesban, an indigenous tree, into the maize crop preceding the
fallow, was subjected to an economic analysis, based on 20 farmer-managed trials
where the technology was tested. One would expect intuitively that the yield of maize
grown after a one-season fallow would need to be at least double that obtained before
the fallow, in order to compensate for the lost production during the fallow period.
Instead, the break-even yield increase following the one-season fallow, compared to
continuous cropping, was only 21% of the long rains yield of 600 kg ha−1 for the
base scenario. It was relatively low, because the foregone maize yield during the fallow
was compensated by savings in crop labour. Improved fallow is a promising technique
for reclaiming depleted land, especially for households with access to off-farm income
or having low labour-to-land ratios. The farm trials facilitated a realistic economic
analysis and farmers’ input into the design of the technology to help focus research on
improving the practice. Insights from this trial led to a new phase of on-farm
research, with researcher-designed, researcher-managed trials to screen potential
improved fallow species, researcher-designed, farmer-managed trials on a range of
design variables, a large network of farmer-designed, farmer-managed trials, and a
rapidly expanding extension programme promoting improved fallows.

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 65
66 R.A. Swinkels et al.

Introduction
Soil nutrient depletion is a worldwide problem affecting 135 million ha, mostly in
South America and Africa (Oldeman et al., 1991). In the East African highlands,
soil depletion is a common feature of food-crop oriented small-scale farms, and has
led to low labour and land productivity. Rapid population growth has resulted in an
increase in continuous cropping of farmers’ fields, while little use is made of organic
or inorganic fertilizers. Organic inputs, such as manure, are limited by their avail-
ability, and the use of inorganic fertilizers is constrained by the unreliable returns to
recommended ‘packages’ of hybrid seeds and fertilizers (Ruthenberg, 1980;
Anderson, 1992) and lack of access to capital (Hoekstra and Corbett, 1995). As off-
farm income has become relatively important for farm households in eastern and
southern Africa (Low, 1988), technologies are required that not only are able to
recover depleted soils and are appropriate to the suboptimal production conditions
of small-scale farmers, but also offer attractive returns per work day.
Work in Zambia has shown that improved fallows (enrichment of natural fal-
lows with trees or shrubs to improve soil fertility) have the potential to increase crop
yields, while also providing fuelwood (Chapter 3). Results have been positive using
Sesbania sesban, a rapidly growing (though short-lived), nitrogen-fixing tree that
seeds profusely. It is indigenous throughout eastern and southern Africa; in western
Kenya it is often found in cropland and in fallowed plots (Bradley, 1991). Farmers
sometimes scatter sesbania seed in cropland for fuelwood production and as a
means for soil enrichment (Bradley, 1991; Scherr, 1993).
Uses of sesbania species as green-manure crops have been documented exten-
sively (Evans and Rotar, 1987). In on-station trials in western Kenya, sesbania is
reported to produce up to 9 t ha−1 of dry leafy biomass per 6-month season (Onim
et al., 1990), equivalent to about 252 kg of nitrogen. However, few studies have
examined improved fallow for soil fertility improvement under farmer manage-
ment, particularly in areas of high population density, where one would think they
have little potential (Hoefsloot et al., 1993; Versteeg and Koudokpon, 1993).
The objectives of this work were to: (i) assess the extent and role of fallowing in
the densely populated farming systems of the area; (ii) obtain farmers’ evaluations of
improved fallow and determine problems farmers had in managing them; (iii) assess
the economics of improved fallow in terms of yield increase following the fallow
required to cover the costs; and (iv) assess possible management options based on
observation of farmers’ experimentation with the technology. As there had been no
earlier research on improved fallow in the area or in a similar environment, the
work was important in guiding decisions regarding further research on the technol-
ogy. Assessing yield response following the fallows was originally an objective, but
this was not possible due to high managerial variability: farmers often planted test
and control plots at different times in different crop densities with varying amounts
of manure and weeding labour. Determining yield response would have required
controlling non-experimental variables, which was inimical to the objective of
assessing farmer management of the technology. Therefore, yield response was mea-
sured in separate researcher-managed trials under controlled conditions.
 Short Rotation Improved Tree Fallows 67

Study Area
The study area is located around the town of Maseno in western Kenya (0°00 N
34° 35 E) (Fig. 4.1). The area lies on the equator, and includes adjacent portions of
Siaya, Vihiga and Kisumu Districts, i.e. Yala, Emuhaya, Winam and Maseno
Divisions. These represent humid parts of the food-crop-based land-use system of
western Kenya. The area has high agricultural potential (high rainfall, well-struc-
tured soils) but the land is nutrient depleted (Shepherd et al., 1996). The altitude is
about 1500 m above sea level and the mean annual temperature is 21.0°C. Rainfall
is bimodal, averaging 1600–1800 mm year−1 and divided over the long rains season,

Turkana
Mandera

Marsabit

zoia
nsn
Bun Tra Sambaru
gom
a Isiolo
Baringo
Ua ish
G

Busia
sin u

Kakamag Laikia Meru

Nandi
Si

Meru
ay
a

Na Njeri ag a
o

kur Kiriny Garisa

rich

u Njeri K Embu
So Mu ir
i Em
Ke

Ny uth Kisii
Kia Kia
ran nyaga bu
an ga Tana
za mb m
u b River
u
Nairobi M Kitui
ac
ha
ko
s
Kajiado Lamu

Key
Provinces
Districts Kilifi
Taita

Mombasa
Kwale

Fig. 4.1. Study area in western Kenya.

68 R.A. Swinkels et al.

from March to July, and the short rains season, from August to November. During
the trial period, rainfall varied within the research area and was slightly lower than
usual in 1991 (1350–1500 mm) and about the average in 1992 (1400–2050 mm).
The landscape is undulating with slopes of 2–8%. Main soil types are Ferralsols,
Acrisols and Nitisols with pH 5 to 7, and moderate to high base. Nitrogen and
phosphate deficiencies in soils are widespread (Shepherd et al., 1996; Sanchez et al.,
1997).
Farm size varies between 0.5 and 2.0 ha, with a median of 1.2 ha (David and
Swinkels, 1994), and population densities range from 300 to over 1000 persons
km−2. Maize, often intercropped with beans, dominates the cropping pattern. Cash
crops are rare: only 14% sell either sugarcane, French beans, coffee or tea. Maize
yields are low: about 700–1200 kg ha−1. Most households (74%) own a few local
Zebu cattle (median is four cattle). Land is privately owned. About half of the
heads of households are involved in off-farm money-making ventures. Roughly half
of all households are female-headed. Use of hired labour is variable; some house-
holds hire labour for almost all their farm work, while others use only household
labour. Less than one-third use mineral fertilizer (David and Swinkels, 1994).

Methods

The methods used include a formal survey of a randomly selected sample of farm-
ers, a participatory rapid appraisal of farmers with dense stands of sesbania fallows,
and a participatory on-farm trial in which interested members of self-help groups
tested improved fallows.

Formal surveys and participatory rapid appraisals

In 1991, informal discussions with farmers were conducted regarding their meth-
ods of soil fertility maintenance, including existing fallowing practices. A formal
survey of 71 randomly selected farmers using a structured interview schedule was
conducted in 1992. Its aim was to verify hypotheses developed from the earlier
informal discussions with farmers. The methods of maintaining soil fertility were
then correlated with key parameters, such as household characteristics, farm size
and labour available for agricultural work.1 Fallowing was defined as leaving land,
which is normally cropped, temporarily uncultivated. A participatory rapid
appraisal was conducted in 1993 among six farmers who had dense stands of sesba-
nia, to assess the role of these stands in their existing fallowing practices. Semi-
structured interview techniques were used to stimulate free expression of farmers’
views.

1 This variable was calculated by adjusting household size for age composition and the
reported primary activity of each member. It is only indicative of available labour, as
households may also hire labour.
 Short Rotation Improved Tree Fallows 69

Trial design

Farmers were selected from self-help groups during 1991; after discussions with the
group members, the most interested ones were invited to participate (Ndufa et al.,
1995). Twelve planted a trial in 1991 and eight others planted in 1992. Researchers
provided farmers with tree seed, some seedlings for gapping up, and advice on
planting method. The researchers marked out plots, but farmers conducted all oper-
ations. The trial thus began as a ‘type 2’ (researcher-designed, farmer-managed) trial
(Chapter 2).
The trial included two adjacent plots managed by each farmer: (i) an improved
fallow rotation, maize→improved fallow→maize (6 months per phase); and (ii) a
control plot under continuous maize or maize→unimproved fallow without
sesbania→maize, depending on the farmers’ choice (unimproved fallow without
sesbania is the most common fallow practice in the area). On the improved-fallow
plots, the prototype recommendation was to: (i) establish sesbania rows by direct
seeding as a relay crop between 3-week-old maize during the long rains season; (ii)
fallow during the subsequent short rains season; and (iii) cut down the trees just
before the next long rains season, incorporate the leafy biomass, and plant a maize
crop. The improved-fallow prototype helps farmers to achieve a higher density of
sesbania in their fallow fields.
The sesbania were planted in continuous lines 4 m apart in 1991, but 2 m
apart in 1992. A linear planting arrangement was chosen to facilitate crop cultiva-
tion. Farmers were given about 150 g of tree seed (110 seeds g−1) per 100 m of ses-
bania line. Sizes of plots were variable: 300–900 m2. Just before the start of the long
rains following the fallow, farmers were given the option of cutting the trees at
ground level to kill them, or cutting at 0.5–1 m height to maximize regrowth. The
latter option would turn the tree lines into hedges, but these would not be long-
lived as sesbania at this altitude does not withstand frequent removal of all its leaves
(Yamoah and Burleigh, 1990; ICRAF, 1992).
Informal interviews with the trial farmers were conducted each season to deter-
mine methods and criteria farmers used to evaluate improved fallow, to obtain their
assessments and to examine the problems they encountered. We also monitored
their willingness to repeat the improved fallow by offering them seed during 1992
and 1993. Formal questionnaire surveys at the end of the trial period were used to
test hypotheses developed from the informal interviews, to quantify farmers’ per-
ceptions on yield responses, and to assess household characteristics so as to deter-
mine whether they were representative of the population in the area (David and
Swinkels, 1994).
Data on labour inputs for planting and managing the trees were collected by
monitoring work rates through observation. Crop input–output data were collected
from 150 maize plots, including those of the trial farmers as well as an additional
50 farmers with other on-farm trials in the area, using farmers’ recall just after a task
was completed (Chapter 5). Prices were collected from six local markets. Yields of
tree foliage and wood at the time of tree harvest were determined from quadrat
samples.
70 R.A. Swinkels et al.

Economic analysis

Enterprise budgets were developed for the maize with the improved fallow proto-
type, for continuous maize, the most common practice on the control plots, and for
maize with an unimproved fallow without sesbania. Intercropped beans and ground-
nuts were omitted in the economic analysis as their yields were low and formed only
a minor part of the production value. As this regards an economic analysis from the
farmers’ point of view, we incorporate the inputs, outputs and prices that the farmer
faces. Net returns to farmers’ production factors (land, labour and capital) were cal-
culated by subtracting purchased inputs from the production value. After subtracting
farmers’ capital inputs, which were minor, the net returns were allocated among
farmers’ land and labour by valuing one factor at its opportunity cost and by
attributing the remainder to the other factor. This enabled us to calculate the net
returns to land, which is relevant for farmers whose most scarce resource is land, and
the net returns to labour, relevant for those who lack household labour.
As comparing crop yields between test and control plots was not meaningful,
due to a high managerial variability, we are not able to report crop yield responses
here. Instead, we calculated the break-even maize yield relative to continuous crop-
ping, that is, the yield increase over continuous maize that is required to cover the
costs of the improved fallow (including the maize yield foregone during the fallow
period and including the cost of postponing consumption, i.e. the social time pref-
erence rate, estimated at 20%). We also calculated the break-even maize yield
increase relative to an unimproved fallow. This is the extra yield required over the
yield after an unimproved fallow, to cover the cost of improved fallow. Equations
were developed for the calculation of the break-even yield increase for the compari-
son with continuous maize, based on an analysis of returns to land (Appendix 4.1);
for the comparison with unimproved fallow; and for both types of comparisons
based on an analysis of returns to labour. Only the first season after the fallow is
included in the analysis and thus residual effects of the improved fallow are not
considered. The analysis is therefore likely to underestimate benefits. Furthermore,
negative effects of the trees on adjacent fields during the fallow and positive effects
on adjacent fields after the fallow are not considered.
The economic analysis incorporates two maize yield scenarios: low and high.
To represent severely depleted soils the average 25th percentile seasonal yields for
1991 and 1992 were taken from the sample of 150 maize plots (600 kg ha−1 in the
long rains and 400 kg ha−1 in the short rains). To represent less depleted soils the
average 75th percentile sample yields were taken (1900 kg ha−1 and 1200 kg ha−1).
For each yield scenario, yields under continuous maize were assumed to remain
constant for the three seasons. Under unimproved fallow, yields were assumed to
remain the same before and after the one-season fallow. These assumptions allow an
assessment of the break-even yield increase of improved fallow relative to the two
alternative systems. Crop labour inputs were about 50% higher for the high-yield
scenario compared to the low-yield scenario.
In situations of high population density where land is scarce and farms are
small, the marginal product of labour is usually lower than the wage rate (Ellis,
 Short Rotation Improved Tree Fallows 71

1988). Therefore, we valued cropping labour according to three different scenarios:

25, 50 and 75% of the hired labour price. The first scenario would be applicable
where labour is abundant and the farmer is usually not able to find off-farm work;
the third one is applicable where the farmer is usually able to do off-farm work, but
incurs some transaction costs in finding it (De Janvry et al., 1991). The second sce-
nario represents an intermediate situation.
Land was valued at its rental rate of about US$33 ha−1 year−1, but as tenancy
and land transactions are uncommon,2 this is only indicative of its opportunity
cost. In the analysis we therefore incorporate two other land values: (i) double the
rental rate, approximating the opportunity cost of land based on the relatively high
purchase price (US$660 ha−1); and (ii) half the rental rate, to estimate the opportu-
nity cost of severely depleted land.
For each yield scenario, the average of the 1991 and 1992 medians of other
inputs, such as quantities of maize seed, farmyard manure and fertilizer, were used.
Sensitivity analysis was conducted to measure the effects of changes in some of the
base assumptions on the break-even maize yield increase.

Results

Formal survey results

Farm sizes are small: the mean was 1.7 ha (SD = 1.6 ha; median = 1.2 ha).
Labour–land ratios were variable: the mean was 3.1 equivalents of adult full-time
workers per ha (SD = 5.1; median = 1.8).
More than half the farmers (52%) mentioned that they fallow some of their
land periodically, about 10–50% of total farm land at a time. During the short rains
season of 1992, on average 32–47% of farmers’ cropland was under fallow, depend-
ing on the district. The length of the fallow varies between one season (24% of
those who fallow), 1 year (35%) and 2 or more years (41%). As expected, farmers
who fallow have on average significantly larger farms (2.2 versus 1.2 ha; P < 0.01).
However, fallowing is not limited to the larger farms as 31% of households with
farms smaller than the median also fallow periodically. Labour–land ratios were not
significantly lower than for farmers who do not fallow.
Of the farmers who fallow, 84% mentioned that they did so to restore soil fer-
tility, while 51% mentioned shortages of labour or cash to hire labour. Other rea-
sons mentioned by less than 5% of farmers included need for grazing land/fodder,
lack of seeds and need for fuelwood (farmers were allowed to give more than one
answer). Farmers that mentioned lack of labour as a reason for fallowing had lower
labour–land ratios than the rest (1.4 versus 3.6 adult full-time workers; P < 0.08). A
higher proportion of farmers with off-farm income mentioned that they fallow peri-

2 Tenancy is infrequent because of enforcement problems and lack of credit, while land
transactions are rare due to high land prices and lack of credit (Collier, 1989). Usually land
is obtained through inheritance.
72 R.A. Swinkels et al.

odically, compared to those without off-farm income (68% versus 32%; P < 0.003).
No association was found between fallowing and gender or age of head of house-
hold.
Three-quarters of the farmers reported that they had some sesbania growing in
their crop land; tree density ranged between 20 and 200 trees ha−1. The proportion
of farmers varied between 33% and 94%, depending on the district. A fifth of the
farmers claimed they sometimes scatter sesbania seeds in the cropland after land
preparation.

Results of participatory rapid appraisal

Among the six farmers with dense sesbania stands, sesbania usually grows naturally
in cropland. In case the population of sesbania drops below desirable levels, farmers
may broadcast sesbania seeds. During land preparation or weeding, farmers some-
times remove sesbania trees or branches to prevent excessive shading of crops. The
farmers’ main reason for growing sesbania is for fuelwood for domestic use, but
they are aware of its beneficial effect on soil fertility, although the impact is limited
as tree densities are low.
When land is left fallow, the sesbanias are allowed to grow tall and seed, which
may result in a dense stand. After a fallow, all sesbania trees are cut down, the land
is prepared and a maize crop planted. None of the farmers interviewed had sown
sesbania during the season prior to a fallow or at the start of a fallow in order to
encourage a dense stand.

Trial results

The average farm size of trial farmers was 1.2 ha (SD = 1.4 ha). Seven of the 20 trial
households were female-headed; ten farms were managed by a female, five by a cou-
ple, three by a male and two by permanent labourers. These results are in line with
findings for the broader population of the area (David and Swinkels, 1994).
Tree establishment failed on five of the 20 farms because of washing away of
seed during heavy rains (four farms) and poor weeding (one farm). In the remaining
15 trials, the mean number of trees germinating was 13 m−1 (SD = 7), leading to
tree stands of 32,500 trees ha−1 for trees in rows 4 m apart and 65,000 trees ha−1 for
trees in rows 2 m apart. Farmers did not do any extra weeding of the trees, apart
from the normal weeding of the maize. Tree height 6 months after sowing was
extremely variable, ranging between 0.5 and 2.5 m, with a mean of 1.3 m.
Of the 15 remaining trials, another five failed in the course of the first year
after establishment. Causes included browsing by dik-diks (Madoqua kirki) (three
farms, all in areas of low population density where such antelopes are still com-
mon), browsing by the farmer’s cow (two farms), farmer leaving the area (one farm)
and trees accidentally uprooted during land preparation (one farm). On some
farms, there were more than one cause of failure.
 Short Rotation Improved Tree Fallows 73

Trial management was highly variable. In four of the remaining ten trials, the
farmer followed the prototype recommendation for the test plot of fallowing for
one season and cutting down the trees about 1 year after sowing. Quantities of dry
leafy biomass on three of these four farms for which data were available were 9000
kg ha−1, 3700 kg ha−1 and 700 kg ha−1. Assuming 18 g nitrogen kg−1 leafy biomass
(Shepherd et al., 1996), the leafy biomass provided about 162, 67 and 13 kg nitro-
gen ha−1. Dry weights of wood were 7900 kg ha−1, 3400 kg ha−1 and 500 kg ha−1,
and the amounts of nitrogen removed in the wood (assuming 7.3 g nitrogen kg−1
wood) were 58, 25 and 4 kg ha−1. Thus about one-third of the nitrogen in the
standing biomass was removed from the field. Tree yields were lower on the third
farm partly because trees were planted in rows 4 m apart, compared to 2 m apart on
the other two farms.
Five other farmers cut down the trees at 1.5–3 years after establishment. Three
of these fallowed only during the last season before cutting, cropping in between
the tree rows during the other seasons. They side-pruned the trees to prevent shad-
ing. One other farmer fallowed for two seasons, and another one for three seasons,
before harvesting the trees. Tree biomass data were available only for the latter
farmer; he harvested about 2400 kg ha−1 of dry leaves and 2500 kg ha−1 of dry
wood (trees in rows 4 m apart). One other farmer still had not cut down the trees 3
years after sowing. Only 1 of the 10 farmers fallowed the control plot at least once
during the trial period. The rest planted maize (eight farmers) or sweet potato (one
farmer). Average long-rains maize yields before the fallow on 6 of the 10 success-
fully established trials for which data were available was 1073 kg ha−1 (SD = 667).
Data on long-rains crop yields in the trial plot after the fallow were only available
for three farmers: their mean maize yield was 658 kg ha−1 (SD = 238). In the control
plot, long-rains maize yields during the trial period were, on average, 995 kg ha−1
(SD = 610; n =7) in the long rains and 633 kg ha−1 (SD = 356; n = 19) in the short
rains.
Farmers’ evaluations of the ten successfully established improved fallows were
positive. All of them stated that, in the establishment season, the young sesbania
seedlings had no influence on maize yields and weeding requirements. However, the
four farmers who cut down the trees within 1 year after sowing mentioned that
regrowth of the remaining tree stumps made land preparation and weeding more
difficult. In contrast, the five farmers who cut down their trees 1.5–3 years after
sowing saw a much higher tree mortality and said that the remaining tree stumps
had no effect on labour requirements for land preparation or weeding. The three
farmers who intercropped the trees for more than one season before fallowing,
observed that after the first season the crops were negatively effected by shading of
the trees.
Six of eight farmers for whom data were available claimed that the improved
fallow increased crop yields; two mentioned the trees had made no difference. Three
farmers noted the trees had caused a decrease in weeds and two reported that the
trees helped curb soil erosion. Five asked for more sesbania seed and four planted a
new improved fallow on their own.
74 R.A. Swinkels et al.

Economic analysis

Extra costs of improved fallow, compared to continuous cropping, consisted of tree

seeds, labour for establishing and removing the trees, and the maize yield foregone
during the fallow; extra benefits included crop labour and other crop inputs saved
during the fallow, fuelwood and the increased maize yield after the fallow (Table
4.1). Improved fallow required less labour than continuous cropping: over three
seasons the reduction was 83 work days ha−1, or 21% under the low-yield scenario
(Table 4.1). Thus, the higher the opportunity cost of labour, the more profitable is
the improved fallow. The extra costs of improved fallow compared to an unim-
proved fallow consisted of planting and cutting down the trees, minus the cost of
clearing the natural fallow, which is estimated to require 5.8 work days ha−1 (one-
third of the labour required to cut down an improved fallow).

Returns to land

The break-even maize yield increase required to cover the costs of improved fallow,
as compared to continuous cropping, was calculated from Equations (4.1) and (4.2)
(Appendix 4.1). Using values from Table 4.1 to calculate (Equation 4.2) we get:
41 S
275 – 8.6 Lm YS
BEYIland = − L
+ PL L
+ 1.2 L
Y Y Y
where,
BEYIland = the break-even yield increase in the long rains following the fallow (% of
long-rains yield under continuous cropping) for the analysis of returns to land, and
Y L = maize yield in the long rains under continuous cropping (kg ha−1),
P L = opportunity cost of labour (US$ day−1),
LmS = maize labour input in the short rains (work days ha−1), and

Y S = maize yield in the short rains (kg ha−1).

Under the low-yield scenario, the break-even yield increase required to cover
the costs of improved fallow was 21% (Table 4.1), and ranged between −5% and
47% of the long-rains yield of 600 kg ha−1, depending on the opportunity cost of
labour (Table 4.2). In the high-yield scenario, the break-even yield increase for
improved fallow was between 31% and 59% (Table 4.2). Thus at higher base yields,
improved fallow becomes less attractive (Fig. 4.2), reflecting the higher maize yield
foregone while fallowing. Under a high-yield scenario, the break-even yield increase
also becomes less sensitive to changes in the opportunity cost of labour compared to
the low-yield scenario (Table 4.2).
A higher opportunity cost of labour reduced the break-even maize yield increase
because higher labour costs were foregone during fallowing, increasing the attractive-
ness of improved fallows (Table 4.2; Fig. 4.3). In the low-yield scenario, the break-
even yield increase is negative when the opportunity cost of labour is higher than
US$0.59 day−1 (70% of the wage rate) (Table 4.2). This is because at that labour cost,
the present value of net returns per hectare of continuous maize is negative (Fig. 4.2).
Table 4.1. Enterprise budget showing the 21% maize yield increase, following the improved fallow, required under the base analysis to
cover the costs of adopting it, relative to continuous cropping, based on returns to land (US$ ha−1).a
Improved fallow Continuous maize
Year 1 Year 2 Year 1 Year 2
Costs and returnsb Season 1 Season 2 Season 3 Season 1 Season 2 Season 3
Returns (US$ ha−1)
maize yield 82 0 99 82 55 82
fuelwood 0 0 10 0 0 0
Costs (excluding labour) (US$ ha−1)
maize seed 4 0 4 4 4 4
tree seed 8 0 0 0 0 0
Net returns to farmers’ land, labour and capital 74 0 109 82 55 82
(US$ ha−1)
Net returns to farmers’ land and labour 70 0 105 78 51 78
(US$ ha−1)
Present value of net returns to farmers’ land 131 161
and labour (US$ ha−1)
Labour input (work days ha−1)
Short Rotation Improved Tree Fallows

maize 136.4 0 136.4 136.4 118.2 136.4

tree sowing 17.5 0 0 0 0 0
cutting trees 0 0 17.5 0 0 0
Total labour 153.9 0 153.9 136.4 118.2 136.4
Present value of labour input (work days ha−1) 235.1 306.9
Present value of net returns to land (US$ ha−1) 33 33
aMaize yield increases are expressed as percentages of long-rains yield. The base analysis uses the low maize yield scenario: 600 kg
ha−1 in the long rains and 400 kg ha−1 in the short rains; opportunity cost of labour is 50% (US$0.42 day−1) of wage rate. 1991 prices, 1
US$ = 30 Kenya Shillings.
75
Table 4.1. Continued.
76

bCosts and returns:

Maize price = US$0.14 kg−1 (market price at planting time and harvesting time).
Fuelwood yields varied between 3 and 10 tons dry wood ha−1, but as fuelwood is not particularly scarce, we value only the first ton;
estimated value is one-third of the market price of US$0.03 kg−1.
Maize seed rate is 32 kg ha−1.
Tree seed rate is 6.1 kg ha−1 of sesbania seeds, valued at the price of US$1.3 kg−1, at which it is bought by NGOs in the area.
Net returns to land, labour and capital is production value minus purchased inputs; only tree seeds are assumed to be purchased.
Net returns to land and labour: farmers’ capital inputs are assumed to consist only of own maize seeds; value of farm tools is negligi-
ble and their depreciation costs are not included.
Present value uses a discount rate of 20%.
Labour input on maize: in western Kenya a working day in agriculture consists of 5.5 hours.
Labour input for tree sowing: 13.1 work days ha−1 for sowing plus 4.4 work days ha−1 for gapping up a few weeks later.
Labour input for cutting trees consists of two cuts: one at the end of the fallow and one cut of regrowth during the cropping season.
R.A. Swinkels et al.
Table 4.2. Break-even yield increase (expressed in percentage of long-rains yield) required to cover the costs of improved fallow
under different assumptions, comparisons and types of analysis.
Analysis of returns to land
Low opportunity High opportunity
cost of labourd cost of laboure
Maize yielda Comparisonb Equationc % kg ha−1 % kg ha−1
low IF vs CM 0.731 – 1.239 PL (1) 47% 282 −5% −30
high IF vs CM 0.736 – 0.679 PL (2) 59% 1121 31% 589
low IF vs UF 0.005 + 0.389 PL (3) 9% 54 25% 150
high IF vs UF 0.002 + 0.123 PL (4) 3% 57 8% 152
Analysis of returns to labour
Low opportunity High opportunity
Equation cost of landf cost of landg
low IF vs CM 0.090 + 0.0060R LR + 0.0034R SR (5) 17% 102 42% 252
high IF vs CM −0.047 + 0.0033R LR + 0.0018R SR (6) −2% −38 13% 247
low IF vs UF 0.186 – 0.0027R LR – 0.0015R SR (7) 15% 90 4% 76
high IF vs UF 0.166 – 0.0009R LR – 0.0003R SR (8) 16% 96 12% 228
aLow-yields scenario is the 25th percentile sample maize yield: 600 kg ha−1 in the long rains and 400 kg ha−1 in the short rains.

Corresponding crop labour inputs are 136.4 and 118.2 work days ha−1, respectively. High yields are the 75th percentile sample maize
Short Rotation Improved Tree Fallows

yields: 1900 kg ha−1 in the long rains and 1200 kg ha−1 in the short rains. Corresponding crop labour inputs are 200.0 and 181.8 work
days ha−1, respectively.
bIF = improved fallow, CM = continuous maize, UF = unimproved fallow.
cP = opportunity cost of labour in US$ day−1; R LR, R SR = opportunity cost of land in the long rains and short rains, respectively, in US$ ha−1.
L
dUS$0.21 per work day, i.e. 25% of wage rate.
eUS$0.63 per work day, i.e. 75% of wage rate.
fHalf the rental rate, i.e. US$16.5 ha−1 year−1 (US$10 ha−1 in the long rains and US$6.5 ha−1 in the short rains).
gDouble the rental rate, i.e. US$66 ha−1 year−1 (US$40 ha−1 in the long rains and US$26 ha−1); buying price for land is about US$660

ha−1; if the interest rate is 20%, in a perfect market rent should be about US$132 ha−1, showing that the land market is highly imperfect
77

and land rent relatively cheap.

 78 R.A. Swinkels et al.

Break-even yield increase (% of long-rains yield)

900 50
45
Break-even yield increase (kg ha–1)

800

700 40
35
600
30
500
25
400
20
300
15
200 10
100 5
0 0
0 500 1000 1500 2000 2500 3000 3500
Base maize yield (kg ha–1 year–1)

kg ha–1 (returns to land) kg ha–1 (returns to labour)

% (returns to land) % (returns to labour)

Fig. 4.2. Break-even yield increase of improved fallow as a function of base maize
yield. For the returns-to-land analysis, labour was valued at 50% of the wage rate;
for the returns-to-labour analysis, land was valued at the rental rate.

The percentage break-even yield increase was not greatly affected by changes in
the value of most of the economic parameters (Table 4.3). However, using potted
seedlings instead of direct seeding raised the break-even maize yield increase from
between −5% and 47% to between 126% and 147%. An increase in the length of
the improved fallow, or an increase in the firewood price, also greatly affected the
results. Compared to the unimproved fallow prototype, the break-even yield
increase for improved fallow (Equations 4.3 and 4.4 in Appendix 4.1) is low under
all assumptions (Table 4.2).

Returns to labour

As improved fallow required less labour than continuous cropping, the break-even
maize yield for improved fallow, compared to continuous maize, was generally
lower when the analysis was based on returns to labour than when on returns to
land (Table 4.2). Furthermore, the break-even yield increase, when expressed in kg
ha−1, is hardly affected by a change in yield scenario (Fig. 4.2). The outcome is
highly sensitive to changes in the fuelwood price and establishment method
(Table 4.3).
 Short Rotation Improved Tree Fallows 79

Present value of net returns to land (US$ ha–1) 200

150

100

50

–50

Yield increase improved fallow

Improved fallow; labour valued at 25% of wage rate

Continuous maize; labour valued at 25% of wage rate
Improved fallow; labour valued at 50% of wage rate
Continuous maize; labour valued at 50% of wage rate
Improved fallow; labour valued at 75% of wage rate
Continuous maize; labour valued at 75% of wage rate
Break-even yield increase with increasing opportunity cost of labour

Fig. 4.3. Effect of opportunity cost of labour on break-even maize yield increase of
improved fallow, relative to continuous cropping (low-yield scenario).

Discussion
Extent of fallowing

The survey finding that periodic fallowing for soil-fertility restoration is a common
practice in western Kenya is in conflict with the theories and empirical findings of
others concerning areas of high population density. For example, Boserup (1965)
claims that as population increases, fallowing is replaced by continuous cropping.
Ruthenberg (1980) regards continuous cropping on impoverished soils as the final
steady state in the land-use development process, although fallowing may occur as a
consequence of labour shortages and the need for grazing. The survey results sug-
80 R.A. Swinkels et al.

gest that labour shortages were not the main reason for fallowing. Instead, it was
depleted land often in combination with the availability of off-farm income that
caused farmers to fallow. As farmers with off-farm income are likely to have a higher
opportunity cost of labour, they are more likely to conclude that cropping depleted
land is uneconomical to them, compared to those with no off-farm income.
Moreover, off-farm income also makes it easier for them to forego crop production
during the fallow period, as they can purchase food. Off-farm employment of
household members leading to more extensive land use has been observed else-
where, e.g. in Zambia (Low, 1988), in Java, Indonesia (Nibbering, 1991), in
Central Province of Kenya and in north-west India (Dewees and Saxena, 1995).
The potential beneficial effect of this extensification process on soil recovery in the
Java uplands has been suggested by Barbier (1990) and Nibbering (1991).
Before our survey, Maseno agroforestry researchers had not considered
improved fallow as relevant to the area, because it was thought that farmers do not

Table 4.3. Sensitivity analysis showing the effect of changes in assumptions on

the break-even maize yield increase (expressed as a percentage of long-rains
yield) of improved fallow for the low-yield scenarioa and relative to continuous
cropping.
Returns to land Returns to labour
opportunity cost opportunity cost
of labourb of landc
Other assumptions Low High Low High
‘Base assumptions’ 47% −5% 17% 42%
Maize price + 50% 55% 27% 13% 29%
Maize price − 50% 21% −67% 23% 73%
Discount rate 50% instead of 20% 61% 4% 20% 52%
Valuing 10 tons DM ha−1 of −59% −100% −92% −67%
fuelwood instead of 1 ton ha−1
Establishment with potted 147% 126% 158% 168%
seedlings instead of direct
seeding
3 seasons fallow instead of 170% −45% −67% 36%
1 seasond
a25th percentile sample maize yield: 600 kg ha−1 in the long rains and 400 kg
ha−1 in the short rains; corresponding labour inputs are 136.4 and 118.2 work
days ha−1, respectively.
bFor the low scenario, US$0.21 day−1, i.e. 25% of the wage rate. For the high

scenario, US$0.63 day−1, i.e.75% of the wage rate.

cFor the low scenario, half the rental rate, i.e. US$16.5 ha−1 year−1 (US$10 ha−1

in the long rains and US$6.5 ha−1 in the short rains). For the high scenario, dou-
ble the rental rate is used, i.e. US$66 ha−1 year−1 (US$40 ha−1 in the long rains
and US$26 ha−1 in the short rains).
dIncreased fuelwood production compared to a one-season fallow not valued;

cutting back labour is assumed to be the same as with one-season fallow.

 Short Rotation Improved Tree Fallows 81

fallow their land. As a result of the survey showing the prevalence of fallowing and
the on-farm trials demonstrating the feasibility of improved fallow, the technology
became one of the main research foci of the Maseno Agroforestry Research Centre.
The existing occasional dense sesbania stands in fallow land usually originate
from wildlings. It is not clear why farmers in western Kenya had not previously
thought of deliberately planting sesbania in high densities just before a fallow.
Some farmers indicated that they did not consider it because sesbania is a common
indigenous tree in the area, and that they instead look for ‘new’ technologies to
solve their problems.

Feasibility and farmers’ evaluation of improved fallow

The on-farm trial showed that improved fallow to restore depleted land appears
promising under farmers’ conditions. Almost half of the farmers with good tree
establishment planted a new improved fallow on their own. Farmers on degraded
soils in densely populated areas of Benin gave similar positive evaluations of fallows
improved with Mucuna pruriens (Versteeg and Koudokpon, 1993).

Costs and break-even yield increase

With low yields equivalent to the quartile for the area, maize growing is only prof-
itable when opportunity costs of labour are below 70% of the rural wage rate (Fig.
4.3). As hiring of labour is common in the area, even among farmers with low
yields, this means that some farmers grow maize at a loss and would financially
gain by leaving land fallow. That they do cultivate maize even when it is not prof-
itable is probably associated with: (i) the cultural value that a household should be
self-sufficient in maize; (ii) the high maize price during pre-harvest months, relative
to just after the harvest, the frequent need to purchase maize at this time and the
poorly functioning credit markets that make it difficult for farmers to finance such
purchases; and (iii) risk management to cope with the occasional seasons when
maize is only available at extremely high prices. Thus improved fallow technology
may also have a role in reclaiming depleted land that is currently being continu-
ously cropped, as farmers are often making a financial loss in such situations.
Improved fallow need not threaten food self-sufficiency of the household in any
given season, as a farmer would likely have only a portion of land under fallow at a
time and the fallow rotates within the farm. However, improved fallow will not
substantially increase food production at a farm scale, unless yield responses
(including the residual effects) are much larger than the lost production during the
fallow.
Direct seeding and relay cropping were two key features that reduced the
labour requirements of improved fallows and made them relatively inexpensive, as
compared to using potted seedlings and planting trees at the beginning of the fal-
low period. Direct seeding reduces the break-even yield increase by at least two-
82 R.A. Swinkels et al.

thirds compared to potted seedlings (Table 4.3). Relay cropping the trees into
maize greatly reduces the extra weeding requirement of the trees. In addition, relay
cropping permits tree growth to be extended for two seasons while crops are being
fallowed for only one season. However, improved fallows do have to be planned in
advance and therefore cannot be adopted in cases where unpredictable labour
shortages occur at planting time and prevent a farmer from planting part of his
land. Access to off-farm income or remittances may increase farmers’ ability to
practice improved fallows, depending on the reliability of the food market, that is,
whether or not food is available in the market at a reasonable price.
One would intuitively expect that the yield of maize grown after a fallow
would need to be at least double that obtained before the fallow in order to com-
pensate for the lost production during the fallow period. Thus, one would expect
the break-even yield increase to be at least 100%. However, the results of the eco-
nomic analysis show that the break-even yield increase required was much less than
that under almost all assumptions. The main reason is that the foregone maize
yield during the fallow is partly compensated by the savings in crop labour and
other crop inputs. Thus the break-even maize yield increase of improved fallow
depends on the amount and opportunity cost of the saved labour and the amount
and value of the crop yield foregone. The lower the base maize yield is, and the
higher the opportunity cost of the household labour, the more attractive improved
fallows become. This confirms results from available theoretical economic models
for the optimal private and social utilization of soil. For example, Barbier (1990)
shows that only when soil is severely degraded do soil restoration measures become
economical for a farmer that maximizes long-term income. Krautkraemer (1994)
demonstrates that recurring cycles of cropping and fallow cycles can be an optimal
soil management strategy because there are seasonal ‘fixed costs’ in farming that can
be avoided by not farming. Such ‘fixed costs’ include the minimum amount of
labour required before any yield is obtained, e.g. land preparation labour. Incurring
such costs can result in an annual return to continuous cropping which is less than
the average return from a cycle of farming and fallow.

Factors affecting adoption potential

The analysis shows that improved fallow is especially attractive to farmers on

depleted land and in areas where the opportunity cost of labour is high. It is espe-
cially suited to farmers with off-farm income because they can afford to fallow land
and are more likely to be interested in saving labour than full-time farmers.
Improved fallow may qualify as an ‘induced innovation’, i.e. an innovation that
reflects farmers’ changing resource scarcities and uses less of resources that have
become scarcer (and thus expensive) (Hayami and Ruttan, 1985). Improved fallow
is also likely to be more profitable for higher-value crops, such as kale (Brassica oler-
acea), than for maize, assuming a similar biophysical response.
The technology seems to be particularly beneficial for high-income house-
holds, as they have a higher opportunity cost of labour. However, high-income
 Short Rotation Improved Tree Fallows 83

households may prefer to put more emphasis on the use of other improved soil
management practices, such as fertilizer. When labour is scarcer than land,
improved fallow requires only low amounts of yield increase to become more prof-
itable than continuous cropping (rows 5–8 of Table 4.2; Fig. 4.2). Thus, the scarcer
labour is relative to land for a household, the lower the required break-even yield
increase. Therefore, improved fallows may be of greatest benefits to poor house-
holds that have little labour available per unit of land.
Because improved fallow with sesbania provides only small quantities of phos-
phorus, it needs to be supplemented with phosphorus fertilizers on phosphorus-
deficient soils, which are common in western Kenya (Sanchez et al., 1997).
On-farm research and dissemination of rock phosphate, available from Tanzania,
has begun in western Kenya (Niang et al., 1999)

Management options and implications for research

The surveys and trials reported in this chapter have confirmed the high degree of
fallowing in the densely populated farming systems of western Kenya and the
potential of improved tree fallows to improve farmers’ incomes. Two examples
highlight how the trials have helped focus further research on problematic aspects
of improved fallows:
1. The variable success rates farmers had with tree establishment prompted
researchers to initiate a new trial evaluating different seeding methods for sesbania,
as well as to look for new tree species with a larger seed size to decrease the chance
of seeds being washed away by rain.
2. The high variability in tree growth triggered a new trial investigating the effect
of early phosphorus applications on sesbania tree growth. Research is also needed
to assess the role of inorganic fertilizer as an alternative or complement to improved
fallow.
The trials have also demonstrated various management options of the technol-
ogy. In 1991, one farmer doubled the tree density we had initially recommended,
planting in rows 2 m apart instead of 4 m. As this led to a much faster biomass
accumulation, we adopted this tree spacing for our recommended prototype in
1992 and for later researcher-managed trials. Farmers’ experiences also showed that:
(i) species planted to improve fallows should be easy to uproot; and (ii) improved
fallow using palatable tree species are not suitable in areas where wild antelopes are
common or where farmers’ animals graze freely.
In summary, the surveys and on-farm trials on improved fallows have proven
to be effective means of demonstrating the potential of the technology to increase
returns to farmers’ resources, the socioeconomic limits to adoption of improved fal-
lows, and how research may alleviate the main constraints and improve the produc-
tivity and adoptability of the technology. Following completion of this study in
1994, an expanded programme of tree species screening for improved fallows was
initiated in researcher-designed, on-farm trials. Those showing consistently high
84 R.A. Swinkels et al.

biomass production and crop response, Crotalaria grahamiana, Crotalaria

ochroleuca, Tephrosia vogelii and sesbania were advanced to type 2 and 3 trials
(Niang et al., 1996, 1998). Concurrently, new type 1 on-farm trials were established
to further evaluate fallow performance and management. By 2000, an extension
programme was rapidly expanding and several thousand farmers were planting
improved fallows (Niang et al., 1999; Place et al., 2002).

Acknowledgements

We thank field technicians L. Anjeho, A. Otieno, J. Onoo and S. Amoko, and field
assistants F. Ochuka, J. Otiende, J. Ounza, C. Omondi, M. Waswa, V. Atieno, N.
Odhiambo, E. Olesi and M. Ayiemba. The support of the Kenya Forestry Research
Institute, in particular Daniel Nyamai, is appreciated. Henk Moll provided sugges-
tions for presentation of the results. Comments on earlier drafts were received from
Eve Crowley, Dirk Hoekstra, Bashir Jama, Jan-Willem Nibbering, Sylla
Pahladsingh, Frank Place and Meredith Soule. Financial support was provided by
the Rockefeller Foundation, while DGIS of The Netherlands Government sup-
ported the first author and the Swedish International Development Association
financed the fourth author.

References

Anderson, J.R. (1992) Difficulties in African agri- missing markets: some paradoxes explained.
cultural systems enhancement: ten hypothe- The Economic Journal 101, 1400–1417.
ses. Agricultural Systems 38, 387–409. Dewees, P.A. and Saxena, N.C. (1995) Tree
Barbier, E.B. (1990) The farm-level economics of planting and household land and labour
soil conservation: the uplands of Java. Land allocation: case studies from Kenya and
Economics 66, 199–211. India. In: Arnold, J.E.M. and Dewees, P.A.
Boserup, E. (1965) The Conditions of Agricultural (eds) Tree Management in Farmer Strategies:
Growth: the Economics of Agrarian Change Responses to Agricultural Intensification.
Under Population Pressure. Aldine Oxford University Press, Oxford,
Publishing Company, Chicago, Illinois. pp. 198–241.
Bradley, P.N. (1991) Woodfuel, Women and Ellis, F. (1988) Peasant Economics: Farm
Woodlots, Vol. 1, The Foundation of a Households and Agrarian Development. Wye
Woodfuel Development Strategy for East Studies in Agricultural and Rural
Africa. Macmillan, London. Development. Cambridge University Press,
Collier, P. (1989) Contractual constraints on Cambridge.
labour exchange in rural Kenya. Evans, D.O. and Rotar, P.P. (1987) Sesbania in
International Labor Review 128, 745–768. Agriculture. Westview Press, Boulder,
David, S. and Swinkels, R.A. (1994) Socio- Colorado.
Economic Characteristics of Households Hayami, Y. and Ruttan, V. (1985) Agricultural
Engaged in Agroforestry Technology Testing in Development: an International Perspective.
Western Kenya. AFRENA Report No. 78. Johns Hopkins University Press, Baltimore,
International Centre for Research in Maryland.
Agroforestry, Nairobi, Kenya. Hoefsloot, H., van der Pol, F. and Roeleveld, L.
De Janvry, A., Fafchamps, M. and Sadoulet, E. (1993) Jachères Améliorées: Options Pour Le
(1991) Peasant household behaviour with Développement Des Systèmes De Production
 Short Rotation Improved Tree Fallows 85

En Afrique De l’Ouest. Royal Tropical Oldeman, L.R., van Engelen, V.W.P. and Pulles,
Institute (KIT), Amsterdam. J.H.M. (1991) The extent of human-
Hoekstra, D. and Corbett, J. (1995) Sustainable induced soil degradation. In: Oldeman,
Agricultural Growth for the Highlands of L.R., Hakkeling, R.T.A. and Sombroek,
East and Central Africa: Prospects to 2020. W.G. (eds) World Map of the Status of
ICRAF, Nairobi, Kenya. Human-induced Soil Degradation.
ICRAF (1992) Annual Report 1991. International Soil Reference and
International Centre for Research in Information Centre, Wageningen,
Agroforestry, Nairobi. Netherlands and UNEP, New York.
Krautkraemer, J.A. (1994) Population growth, Onim, J.F.M., Otieno, K. and Dzowela, B.
soil fertility, and agricultural intensifica- (1990) The role of sesbania as multipur-
tion. Journal of Development Economics 44, pose trees in small-scale farms in western
403–428. Kenya. In: Macklin, B. and Evans, D.O.
Low, A. (1988) Farm-household economics and (eds) Perennial Sesbania Species in
the design and impact of biological research Agroforestry Systems, Proceedings of a work-
in southern Africa. Agricultural shop held in Nairobi, 27–31 March 1989.
Administration and Extension 29, 23–34. The Nitrogen Fixing Tree Association,
Ndufa, J.K., Ohlsson, E., Swinkels, R.A. and Hawaii, pp. 167–179.
Shepherd, K.D. (1995) Participatory Place, F., Franzel, S., DeWolf, J., Rommelse, R.,
research methods for agroforestry technol- Kwesiga, F., Niang A. and Jama, B. (2002)
ogy development in western Kenya. In:
Agroforestry for soil fertility replenishment:
Design Implementation and Analysis of On-
evidence on adoption processes in Kenya
Farm Trials: An Assessment of Field
and Zambia. In: Barrett, C.B., Place, F. and
Experience. Proceedings of an international
Aboud, A.A. (eds) Natural Resource
workshop, 28 June–2 July 1993, Arusha,
Management in African Agriculture:
Tanzania. Royal Tropical Institute,
Understanding and Improving Current
Amsterdam, pp. 187–204.
Practices. CAB International, Wallingford,
Niang, A., Amadalo, B., Gathumbi, S.M.,
UK.
Otieno, J.H.O., Obonyo, C.O. and
Ruthenberg, H. (1980) Farming Systems in the
Obonyo, E. (1996) AFRENA Project
Tropics, 3rd edn. Clarendon Press, Oxford,
Maseno, Kenya. Progress and Technical
Report 1996. AFRENA Working Paper No. UK.
110. ICRAF, Nairobi, Kenya. Sanchez, P.A., Shepherd, K.D., Soule, M.J.,
Niang, A., Amandalo, B., Kaleb, M., Obonyo, Place, F.M., Mokwunye, A.U., Buresh,
C.O. and Simone, R. (1998) AFRENA R.J., Kwesiga, F.R., Izac, A.N., Ndiritu,
Project, Maseno, Kenya. Progress and C.G. and Woomer, P.L. (1997) Soil fertility
Technical Report 1997. AFRENA Working replenishment in Africa: an investment in
Paper No. 122. ICRAF, Nairobi, Kenya. natural resource capital. In: Buresh, R.J.
Niang, A., De Wolf, J., Gathumbi, S. and and Sanchez, P.A (eds) Replenishing Soil
Amandalo, B. (1999) The potential of Fertility in Africa, SSSA Special
short duration improved fallows with Publication. Madison, Wisconsin.
selected trees and shrubs for crop produc- Scherr, S.J. (1993) Evaluation of agroforestry
tivity enhancement in the densely popu- practices over time in the crop-livestock
lated highlands of Western Kenya. Paper system of western Kenya. In: Dvorák, K.A.
presented at the workshop: La Jachère en (ed.) Social Science Research for Agricultural
Afrique Tropicale. Rôles, Aménagements, Technology Development: Spatial and
Alternatives, Dakar, 13–16 April. CORAF- Temporal Dimensions. CAB International,
ISRA-ORSTOM. Wallingford, UK, pp. 118–144.
Nibbering, J.W. (1991) Crisis and resilience in Shepherd, K.D., Ohlsson, E., Okalebo, J.R. and
upland land use in Java. In: Hardjono, J. Ndufa, J.K. (1996) Potential impact of
(ed.) Indonesia: Resources, Ecology and agroforestry on soil nutrient balances at the
Environment. Oxford University Press, New farm scale in the East African Highlands.
York, pp. 104–132. Fertiliser Research 44, 87–99.
86 R.A. Swinkels et al.

Versteeg, M.N. and Koudokpon, V. (1993) Yamoah, C.F. and Burleigh, J.R. (1990) Alley
Participative farmer testing of four low cropping Sesbania sesban (L.) Merill with
external input technologies, to address soil food crops in the highland region of
fertility decline in Mono Province (Benin). Rwanda. Agroforestry Systems 10, 169–181.
Agricultural Systems 42, 265–276.
 Short Rotation Improved Tree Fallows 87

Appendix 4.1 Derivation of the Break-even Equations

Returns to land

In the analysis of returns to land, labour was valued but land was not.

Improved fallow compared to continuous maize

In order to break even, the extra costs per unit of land of improved fallow should
equal the extra benefits:
PtsSt + PLLts + PmY S PLLc P S S + PLLm
S P F + PmBEYIlandY L
+ = m m + f (4.1)
1+r (1 + r)2 1+r (1 + r)2
where,
Pts = price of tree seed (US$ kg−1),
St = amount of tree seed required to plant the improved fallow (kg ha−1),
PL = opportunity cost of labour (US$ day−1),
Lts = amount of labour required for tree sowing (work days ha−1),
Pm = market price of maize (US$ kg−1),
Y S = maize yield in the short rains (kg ha−1),
r = social time preference rate (discount rate),
Lc = amount of labour required to cut back the trees at the end of the fallow (work
days ha−1),
SmS = amount of maize seed used in the short rains (kg ha−1),

LmS = maize labour input in the short rains (work days ha−1),

Pf = opportunity cost of sun-dried firewood (US$ kg−1),

F = amount of dry firewood (kg ha−1),
BEYIland = the break-even yield increase in the long rains after the fallow (% of
long rains yield under continuous cropping) for the analysis of returns to land, and
Y L = maize yield in the long rains under continuous cropping (kg ha−1).
The break-even yield increase can then be derived:
(1 + r)(PtsSt + PLLts + PmY S − PmSm
S −P LS ) (P L − P F )
BEYIland = L m + L c f . (4.2)
PmY L PmY L

Improved fallow compared to unimproved fallow

As above, in order to break even, the extra costs per unit of land of improved fallow
should equal the extra benefits:
(PtsSt + PLLts) PL (P F + PmBEYIlandY L) (4.3)
+ L c = f
1+r (1 + r)2 (1 + r)2
88 R.A. Swinkels et al.

from (Equation 4.3) follows:

−Pf F + (1 + r)(PtsSt + PLLts) + PLLc
BEYIland = (4.4)
PmY L

Returns to labour

In the analysis of returns to labour, land was valued, but labour was not; instead the
net returns were divided by the amount of labour. Thus, for improved fallow com-
pared to continuous maize, to break even:
present value of net returns improved fallow present value of net returns continuous maize
=
present value labour input improved fallow present value labour input continuous maize.

Or, for the comparison with unimproved fallow:

present value of net returns improved fallow present value of net returns unimproved fallow
=
present value labour input improved fallow present value labour input unimproved fallow.
 Assessing the Adoption
Potential of Hedgerow
Intercropping for
5
Improving Soil Fertility,
Western Kenya
R.A. SWINKELS,1 K.D. SHEPHERD,2 S. FRANZEL,2
J.K. NDUFA,3 E. OHLSSON4 AND H. SJOGREN4
1The
World Bank in Vietnam, 8th Floor, 63 Ly Thai To Street, Hanoi,
Vietnam; 2International Centre for Research in Agroforestry, I
CRAF House, PO Box 30677, Gigiri, Nairobi, Kenya; 3Kenya Forestry
Research Institute (KEFRI), Maseno Agroforestry Research Centre,
Maseno, PO Box 25199, Kisumu, Kenya; 4Sida/SAREC, SE-105 25
Stockholm, Sweden

Summary
The biophysical and socioeconomic performance of hedgerow intercropping for soil
fertility improvement was assessed in a farmer-participatory trial in western Kenya
over 3 years. Farmers successfully established dense hedgerows but planting and
pruning considerably increased labour use during the busiest period of the year.
Women did not generally prune the hedges. The yields of hedgerow prunings of
Leucaena leucocephala and Calliandra calothyrsus (1–4 t ha−1 year−1, n = 24) were
low compared to potentials in the region (4–8 t ha−1). The hedgerows had no effect
on grain yield over five seasons. But they reduced slopes from 7.2 to 4.5% within
alleys (P < 0.01) and thus were likely to have decreased soil erosion.
The average extra maize yield required each year, beginning in the second year,
to cover the added cost of hedgerow intercropping was 10.5% (SD = 5.5%) when
based on returns to land, and 17.5% (SD = 6.5) based on returns to labour.
Although about half the farmers claimed that hedges improved crop yields, after 3
years of experimentation, only about a fifth planted additional hedges and only
14% did so to improve soil fertility. It thus appears that the potential for its adop-
tion as a soil fertility practice in this area is low. Hedgerow intercropping appears to
have greater adoption potential if its aim is to provide feed for an intensive dairy
operation or for curbing soil erosion. Control plots were of little use as farmers were
more interested in contrasting test-plot yields with past yields than comparing the

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 89
90 R.A. Swinkels et al.

test and control plots. Different types of trials may be required to obtain reliable bio-
physical data on yield response and socioeconomic data on farmer assessment of com-
plex agroforestry technologies.

Introduction
There is a considerable body of research showing that crop yields can be increased
significantly by hedgerow intercropping (also called alley cropping or alley farming),
a technology in which crops are grown between rows of nitrogen-fixing trees and
the trees’ leafy biomass is applied to the crops (Kang et al., 1990; Kang, 1993). But
there is little scientific data on the performance of hedgerow intercropping for soil
fertility improvement under farmer-managed conditions in the tropics, despite con-
siderable extension (Müller and Scherr, 1989; Carter, 1995). Potential difficulties in
conducting on-farm trials with complex agroforestry technologies have been
reviewed by Shepherd and Roger (1991). Earlier analyses of the adoption potential
of hedgerow intercropping were based mainly on ex ante analysis of on-station trial
results (Swinkels and Scherr, 1991). Few on-farm trials or social or economic analy-
ses have been reported, and these are mostly from subhumid to humid areas in west
Africa (Whittome, 1994; Dvorak, 1996). Even these trials had several limitations
for the assessment of adoption potential: (i) targeting was inappropriate as often
farmers’ priority problem was not low soil fertility; (ii) farmers’ participation was
obtained through the provision of incentives, such as free fertilizer and improved
crop material; and (iii) there was limited monitoring to establish labour require-
ments and crop and economic performance relative to existing systems (Whittome,
1994). Better farmer participation was obtained where soil fertility decline was per-
ceived by farmers as a serious problem (Versteeg and Koudokpon, 1993). The lim-
ited adoption of hedgerow intercropping that has occurred has been for soil
conservation on sloping land (particularly in the Asian/Pacific region) and for fod-
der production in intensive dairy systems, or because of indirect benefits provided
by the projects themselves (Whittome, 1994; Carter, 1995).
This study uses 3 years’ results from a farmer-participatory trial to assess the
adoption potential of hedgerow intercropping, primarily for soil fertility improve-
ment, in a subsistence-level crop/livestock farming system in the humid highlands
of western Kenya. The components were:
1. Biophysical response:1 What were the effects of the trees on slope and crop
yields?
2. Feasibility: Are farmers able to manage the technology? Do they have the
required information and are they able to plant and maintain the hedges, and to
cope with any problems that arise?

1 Several agronomic substudies were conducted to assess the degree of competition between
hedges and adjacent maize rows, to confirm maize yield responsiveness to applied
nutrients and tree prunings, and to test the response of hedge biomass production to
nitrogen and phosphorus applications. These studies are reported in Shepherd et al. (1997).
 Adoption Potential of Hedgerow Intercropping 91

3. Profitability: Are the financial benefits obtained greater than the costs incurred?
4. Acceptability: Do farmers perceive that the advantages of using the technology
are greater than the disadvantages? Acceptability thus includes a range of criteria in
addition to profitability and feasibility, such as riskiness and compatibility with
other enterprises.
For each of the above components, selected farm and household characteristics were
examined to assess whether they were associated with the uptake of the technology.

Materials and Methods

Study area and background

The study area included parts of Siaya, Kisumu and Vihiga Districts, located
between 0°05S and 0°10N, to 34°31E to 34°40E, chosen to represent humid
areas of the food-crop land-use system of western Kenya which have high agricultural
potential but where land has become nutrient depleted (Fig. 4.1) (Shepherd et al.,
1992). The population density ranges from 300 to over 1000 persons km−2 in some
areas. There are two cropping seasons: the long rains from March to July and the
short rains from August to November, totalling 1500–1800 mm year−1, with a mean
annual temperature of 21.0°C. The landscape is undulating with average slopes of
2–8%, dominated by Acrisols, Ferralsols and Nitisols (Shepherd et al., 1992).
Hedgerow intercropping had been introduced to farmers in the area in 1983
by the CARE-Kenya Agroforestry Extension Project, to address soil fertility con-
straints and produce fuelwood. Several hundred farmers planted hedges of primarily
Leucaena leucocephala (leucaena) but no quantitative data on biophysical or eco-
nomic performance were available. An evaluation in 1988 found that spacing and
management were highly variable and that median tree density was very low, 2000
trees ha−1, roughly half of the density required to supply adequate nitrogen to
impact on crop yield. Not surprisingly, only one-quarter of plots showed notably
higher crop yields, according to farmers’ qualitative assessments (Scherr and Oduol,
1989; Scherr, 1995).

Farmer selection and treatment design

In participatory rapid appraisals, farmers identified soil fertility as a key problem

and expressed interest in testing hedgerow intercropping (Ohlsson et al., 1992).
Two to three farmers from 29 farmer groups in the area participated in the trial: the
groups were selected to represent the main variation in soils and ethnic groups in
the area (Ohlsson et al., 1992). The recommended prototype, which was developed
with farmers, consisted of L. leucocephala, Leucaena diversifolia, Calliandra calothyr-
sus (calliandra), or Gliricidia sepium, planted from inoculated seedlings, with
0.30 m between trees in contour-aligned rows about 4 m apart (8333 trees ha−1)
(Ohlsson et al., 1992). Crops were planted close to the trees so there was no loss of
92 R.A. Swinkels et al.

cropped area due to trees. The trial was started with 24 farmers from ten groups in
August 1990, and with a further 29 farmers, including six additional groups, in
February 1991. Farmers received advice from the project technicians on hedgerow
management throughout the reported period, but farmers made all crop and tree
management decisions.

Trial design and measurements

The trial design consisted of a control plot without hedges and an adjacent plot
with hedges on each farm. The L. leucocephala seedlings were a mixture of
Hengchun, Kisumu, Gede, Siakago and Baobab provenances (29 farms), C.
calothyrsus was Guatemala provenance (18 farms), L. diversifolia was
Veracruz/Mexico provenance (four farms), and G. sepium was established from cut-
tings from a local source (two farms). The 1990 plantings were all of L.
leucocephala. The seedlings were inoculated with compatible rhizobia in polythene
tubes in a nursery before being delivered to the farms.
Project technicians recorded the number of trees planted, the number surviving
at 6 months, and the causes of plant death on each tree plot. Crop harvest yields
were measured on the whole area of test and control plots in five consecutive sea-
sons from the short rains in 1990, on farms where annual crops were grown and the
trial maintained (n = 41). The total produce of maize cobs and pods of any inter-
crops from each plot were weighed and subsamples taken to determine oven-dry
grain weight. Farmers’ estimates of numbers of green maize cobs harvested earlier in
the season were converted to dry grain weight on the basis of average cob weights.
Crop plant populations were estimated from quadrat counts, and the incidence of
pests, weeds and diseases were scored on each plot.
Soil properties were determined in each plot (Shepherd et al., 1996). Average
slopes were measured on all farms using a clinometer. On eight of the steeper farms,
the average slope of the hedgerow plot and the slope within each alley were deter-
mined in October 1994, using a surveying level.

Social and economic analysis

Project staff conducted informal, semi-structured surveys during the trial’s first 2
years to define the methods and the criteria farmers used to evaluate hedgerow
intercropping, to obtain their assessments, and to examine the problems they
encountered (David, 1992). Formal surveys using structured questionnaires were
used to test hypotheses developed from the informal surveys and to quantify critical
parameters, including household characteristics (David and Swinkels, 1994) and
farmers’ assessments of the technology. Data on the use of labour and other inputs
were collected on each plot using two different methods: farmers’ recall just after a
task was completed (for crop management) and monitoring of work rates through
observation (for planting and cutting back trees). The number of farmers included
in the above surveys ranged from 33 to 46.
 Adoption Potential of Hedgerow Intercropping 93

To assess profitability, enterprise budgets for maize with and without hedges
were drawn up, using input and output data from the 23 farms for which complete
data sets were available. The only data not used were for the response of maize
yields to the hedgerows, although the available evidence indicated that the
hedgerows had no effect on yields (Shepherd et al., 1997). Therefore, break-even
maize yields were calculated, that is, the increase in maize yields required to pay for
the costs of establishment and maintenance of the hedges. Separate budgets were
drawn up for each farmer in order to assess the variability in costs and returns
among farms. Two types of analysis were carried out: net returns to land and net
returns to labour. In net returns to land, household labour was valued at its oppor-
tunity cost, as estimated by the hired labour price. Land was not valued; returns
were instead expressed on a per-hectare basis. This measure was relevant for farmers
whose most scarce input was land. Net returns to labour, on the other hand, were
most relevant for farmers whose most scarce input was labour. Land was valued at
its opportunity cost, as estimated by rental rates. Household labour was not valued
and returns were expressed per unit of labour. The break-even analysis was based on
a 5-year time period with maize yields increasing in the second year and remaining
constant thereafter.
Sensitivity analysis was conducted to measure the effects of changes in some of
the base parameters on the break-even maize yield increase. The riskiness of
hedgerow intercropping was also assessed, using two methods. First, the variability
in the net returns was compared between the two treatments, that is, maize with and
without hedges. Secondly, minimum returns analysis (CIMMYT, 1988) was used to
compare the average of the lowest 25% of the net benefits for the two treatments.
Three methods were used to assess acceptability. First, project staff asked farm-
ers to compare yields on the control and test plots. Secondly, the other criteria used
by farmers to evaluate the technology were identified and farmers were asked how
the technology performed by these criteria. Thirdly, their willingness to expand the
hedgerows on their farms was monitored by offering them seed in the third year
after establishment. Data on acceptability were collected using informal interviews,
farmer-to-farmer group visits, formal surveys and monitoring of expansion through
observation. A hierarchical decision tree was constructed to model farmers’ deci-
sions to expand their plantings (Gladwin, 1980). This method was useful for
explaining the decisions farmers made by breaking them down into a series of sub-
decisions and mapping farmers’ decision paths along the branches of the tree.

Results and Discussion

Farmer and seasonal rainfall characteristics

Participants were broadly representative of the range of types of farmer found in the
area; over one-quarter were female-headed, most were of average wealth, and farm
size ranged from 0.1 to 6.1 ha (Table 5.1; David and Swinkels, 1994). Eleven of the
94 R.A. Swinkels et al.

Table 5.1. Characteristics of farmers participating in the hedgerow intercropping

trial (n = 45).
Variable Min. Median Max. Frequency (%)
Farm size, all parcels (ha) 0.1 1.3 6.1 –
Household size (members) 2 6 15 –
Labour:land ratio 0.6 2.2 23.6 –
(adult equivalent workers ha−1)
Farm first cultivated (year) 1935 1977 1991 –
Ethnic group
Luhya 22
Luo 78
Household type
male 72
female de jure 13
female de facto 15
Age of household head (years)
<40 24
40–60 52
>60 24
Education of household head
<primary 19
primary 59
>primary 22
Wealth category
<average 26
average 67
>average 7
Off-farm income (one or both 82
partners)
Use of hired labour 87
Ownership of cows 80
Ownership of improved-breed 6
dairy cows
Source: David and Swinkels, 1994.

56 participating farmers dropped out of the trial within the first year but the partici-
patory approach was otherwise successful in maintaining farmers’ interest in the trial.
The seasonal total rainfall at Maseno for each of the seven seasons from the
short rains 1990 did not differ by more than 25% from the long-term average val-
ues (1960–1993; 895 mm in long rains, and 774 mm in short rains).

Tree establishment and early growth

The size of plots planted with hedges ranged from 270 to 2010 m2 with a median
of 790 m2 (n = 42). Tree density, which ranged from 3660 to 10,040 with a median
 Adoption Potential of Hedgerow Intercropping 95

of 6680 trees ha−1, was negatively correlated with plot size (r = −0.45, P < 0.01).
This indicated that the potential benefits of the technology, resulting from biomass
production, would decrease with increasing plot size. The proportion of the origi-
nally planted trees surviving at 6 months after planting (MAP) ranged from 0.33 to
1.00 with a median of 0.91. Termite damage was observed to be the primary cause
of tree mortality on 30 farms, whereas other causes were important on only a few
farms: moles, 4; drought, 3; uprooting, 1; erosion, 1; and fire, 1. Browsing, mostly
by Kirk’s dik-dik (Madoqua kirki), occurred on ten farms, primarily those near
uncultivated land.
Survival was not related to browsing, but there was a higher frequency of low
survival rates in leucaena (40% farms had <80% survival, n = 25 farms) than in cal-
liandra (6% farms had <80% survival, n = 17). The frequency of termites as the
main cause of damage was similar for the two species (68% leucaena, 76% callian-
dra), but the survival differences could have been due to differences in rainfall dur-
ing the 2 months after planting. This was lower in the short rains of 1990 when
most (n = 22) of the leucaena was planted (302 mm) than in the long rains of 1991
when all of the calliandra was planted (537 mm).
At 6 months after planting, median tree height was 97 cm (range 34–140) for
leucaena (12 farms) and 152 cm (range 90–196) for calliandra (12 farms), while
median stem basal diameter was 1.4 cm (range 1.0–1.8) and 1.5 cm (range
0.8–2.1), respectively. The median early growth rates were substantially (about
40%) lower than those obtained with the same provenances in researcher-managed
experiments at the Maseno Agroforestry Research Centre (Heineman et al., 1990).
However, trees in the on-station experiments received 25 g diammonium phosphate
per tree at transplanting and were established with well-weeded and phosphate-fer-
tilized beans for the first two seasons.

Tree yield

Farmers cut the hedges between four and seven times during the 2 years of moni-
toring of hedge biomass on 24 farms. There was no significant effect of species,
provenance, season or year on tree total biomass yield and dry matter partitioning
(P > 0.05). However, total biomass was significantly greater (P < 0.001) at the first
cut (1.68 t ha−1) than at the second cut (0.74 t ha−1) in the season. Woody stem
comprised 33% of the total biomass at the first cut but only 4% at the second cut,
reflecting the younger growth at the second cut.
On the farms the annual amount of biomass returned to the soil ranged from
1.2 to 4.3 t ha−1 year−1 (median = 2.4 t ha−1 year−1) and was below on-station levels
measured over the same growth phase (about 4.4 t ha−1 year−1 for leucaena
(Hengchun) at a spacing of 3.75 × 0.25 m, and 7.3 t ha−1 year−1 for leucaena
(Hengchun) and 10.8 t ha−1 year−1 for calliandra (Guatemala) both at a spacing of
2.8 × 0.25 m (Heineman et al., 1990; Otieno et al., 1991). In experiments in the
subhumid and humid tropics with the same species and management, maximum
annual yields of prunings were 4–8 t ha−1 year−1 (Balasubramanian and Sekayange,
96 R.A. Swinkels et al.

1991; Kang, 1993). The amounts of nutrients returned in prunings were corre-
spondingly low, but on all farms the amounts are significant in relation to crop
requirements (Shepherd et al., 1997).
The growth of leucaena was affected by an infestation of the psyllid
Heteropsylla cubana from March 1993 onwards. Total biomass over the five cuts was
590 g m−1 of hedge (range 90–810) in leucaena compared with 1850 g m−1 (range
1000–2860) in calliandra; before 1993 there was no difference in yields between
the two species. Thus the potential effects of leucaena on soil fertility are likely to be
limited as long as the psyllid infestation persists.

Crop yields and terracing effect of hedges

Farmers grew a wide range of crop mixtures and fallows in the test and control
plots, making it difficult to evaluate the effects of the hedgerows on yields. For
example, in the long rains of 1992, out of 86 plots, 68 had maize; six, sorghum; 39,
beans; 18, groundnuts; and 16, weedy fallow. Farms where maize was grown as the
main crop in both plots, either as a sole crop or with grain intercrops, were selected
for analysis of yields in each season. There was no significant difference in mean
grain yields (expressed as sum of maize and intercrop grain dry weights) between
the hedgerow and the control plots in any individual season. Averaged over five sea-
sons, mean grain yields were 16% lower in the hedgerow plots than the control
plots, but the significance of this difference was marginal (P = 0.05). The lack of
evidence for an effect of the hedges on yield in the researcher-managed competition
study (Shepherd et al., 1997) suggests that the difference was due to variation in
farmer management between the two plots.
Regression models were used to test the hypothesis that selected soil and man-
agement variables (nitrogen and phosphorus concentration in the subsoil, pur-
chased inputs, and manure application) account for the variation in control plot
grain yields and the difference in grain yield between the hedgerow and control
plots. However, little of the variation could be accounted for by linear regression
with the measured variables (Shepherd et al., 1997). Apparently, variation in yield
among plots and farms was determined by a wide range of biophysical and socio-
economic factors, and could not be explained adequately by the measured variables
within the limitations of the sample size.
Average slopes ranged from 1 to 13% with a median of 5.8% (44 farms). The
hedgerows had a significant terracing effect: the mean slope within alleys was 4.5%
compared with 7.2% over the length of the hedgerow plots (P < 0.01, eight farms).

Feasibility

The hedges required planting just after the rainy season began, when farmers were
busy preparing land and planting their crops. In fact, hedgerow establishment
greatly exacerbated peak periods of labour use; when hours spent planting seedlings
 Adoption Potential of Hedgerow Intercropping 97

were added to those for the other cropping activities, labour use during the busiest
week of the year increased by 50%. Variation in labour inputs also rose; the coeffi-
cient of variation of the mean weekly labour inputs increased from 103% to 120%.
Three-quarters of the households relied solely on family labour for planting the
trees; the rest hired some casual labour.
Pruning (or cutting back) the hedges involved a single upward slash of a
machete and was a completely new task for the farmers. In most cases, they pruned
at planting time in order to reduce shading, and then pruned again about 2 months
later, during weeding (Fig. 5.1). Farmers pruned on average 240 trees h−1 (SD =
140), including spreading the mulch (Swinkels and Franzel, 1992). Pruning
required only 40–60 h season−1 ha−1 or less than 5% of total labour input for maize
cultivation. Nevertheless, it was problematic for two reasons. First, it had to be
done on a timely basis and during maize planting and weeding, which are periods
of peak labour use. When pruning was added to other cropping activities (Fig. 5.1)
the number of weeks in the cropping season in which labour requirements were
over 60 h ha−1 increased from one to three.
Over time, farmers coped better with the task of pruning. In the first season that
it needed to be done, half of the farmers delayed and noted a negative effect on yields.
However, overall, there was no significant difference between control and treatment
maize yields during the first season of pruning (Shepherd et al., 1997). In subsequent
years less than 10% of farmers delayed pruning and noted negative effects.
The second problem was that women did not generally prune the hedges. Of
the 90% of the households in which females were actively involved in managing the

80

70
hedge pruning
60 land prep
Labour (hours ha–1)

planting
50
weeding
40 harvesting

30

20

10

0
Dec Jan Feb Mar Apr May Jun Jul
Week

Fig. 5.1. Labour profile of pruning and cropping activities (long rains 1992,
hedgerow intercropping trial farmers). Cropping labour is the average of 126
maize/sorghum plots of 31 farmers. Pruning labour is the average of 31 hedge plots
of the same 31 farmers. Labour includes both household and hired labour.
98 R.A. Swinkels et al.

farm, only 19% of prunings were done by women. Only in female-headed house-
holds did females prune. Women claimed that they lacked the strength to do it and
that using a machete was a man’s task (David, 1992).
Neither browsing of trees by animals nor the effects of the hedges on other crop-
ping activities were important problems for the farmers. Browsing by wild and domes-
tic animals was reported by 8–19% of the farmers per season over the years but in
most cases only a few trees were browsed. Problems in land preparation were reported
by 25–33% of farmers, including both oxen-users, who complained that hired oxen
did not plough close enough to the hedges, and farmers preparing by hand, who men-
tioned interfering tree roots and branches. In contrast, between 5% and 15% reported
that the hedges reduced labour required for preparation, because the hedges reduced
weed incidence and softened the soil. Roughly equal numbers (10–25%) felt that the
hedges increased and reduced labour requirements for weeding.

Profitability

The costs of hedgerow intercropping were highest in the first season, when hedges
were established (Table 5.2). Seedling costs and planting labour added 78% to the
production costs of maize during this season. Thereafter, pruning was the only extra
cost on the hedge plots, adding only 3% (SD 2%) to maize production costs.
Fuelwood was harvested each year from the hedge treatment, and extra maize
from increased soil fertility was assumed to be harvested each season, beginning in
year 2. During the first year, the hedges did not affect maize yields; no evidence of
an effect was found in either the maize yield data (Shepherd et al., 1997) or in
farmers’ own assessment. The analysis (Table 5.2) showed that variability of the net
benefits was high. The extra maize yield required each year to cover the added costs
of hedge establishment and maintenance was 242 kg ha−1 (SD = 65 kg), or on aver-
age 10.5% (SD = 5.5%) of farmers’ mean maize yields, which were 1521 kg ha−1 (SD
= 873) in the long rains and 1180 kg ha−1 (SD = 1118) in the short rains; median
yields were 1220 kg ha−1 and 794 kg ha−1, respectively.2
Because household labour inputs were higher for hedgerow intercropping than
for conventional maize, the technology was less profitable and the break-even maize
yield increase higher in returns to labour analysis than in returns to land analysis.
An annual maize yield increase of 475 kg ha−1 (SD = 267 kg), or, on average, 17.5%
(SD = 6.5%) of farmers’ mean maize yield, was required to break even.
The break-even maize yield increase was fairly stable under a range of changes
in the assumptions and values (Table 5.3). For the returns to land analysis, the
break-even increase ranges from 6% to 21%, and in the returns to labour analysis,
from 12% to 35%. For the returns to land analysis, a change from potted seedlings
to bare-rooted seedlings, which resulted in a 60% reduction in establishment costs,

2 The average of all individual break-even yield increases is 10.5%; when the average of the
value of all individual maize yield increases is divided by the value of the average maize
production, the outcome is 8.5%.
Table 5.2. Enterprise budget showing the amount of extra maize yield required to cover the costs of adopting hedgerow intercropping
relative to conventional maize (in US$ ha−1).a
Hedgerow Conventional Hedgerow Conventional
intercropping maize intercropping maize
Costsb Mean SD Mean SD Benefitsc Mean SD Mean SD

Season 1
Seedlings 90 23 0 0 Maize 365 182 365 182
Tree planting labour 34 20 0 0 Intercrop 46 54 46 54
Crop seed 18 10 18 10
Fertilizer/manure 20 27 20 27
Crop labour 122 45 122 45
Subtotal season 1 283 80 159 65 Subtotal season 1 411 185 411 185
Season 2
Pruning labour 2 2 0 0 Fuelwood 2 3 0 0
Crop seed 14 10 14 10 Maize 272 224 272 224
Fertilizer/manure 6 14 6 14 Intercrop 24 34 24 34
Crop labour 117 51 117 51
Subtotal season 2 139 63 137 62 Subtotal season 2 298 234 296 234
Total: year 1 422 116 296 99 Total: year 1 709 316 707 316
Adoption Potential of Hedgerow Intercropping

Season 1
Pruning labour 5 3 0 0 Fuelwood 4 3 0 0
Crop seed 18 10 18 10 Maize 365 182 365 182
Fertilizer/manure 6 14 6 14 Extra maize 32 13 0 0
Crop labour 122 45 122 45 Intercrop 46 54 46 54
Subtotal season 1 164 66 159 65 Subtotal season 1 447 192 411 185
99
Table 5.2. Continued.
100

Hedgerow Conventional Hedgerow Conventional

intercropping maize intercropping maize
Costsb Mean SD Mean SD Benefitsc Mean SD Mean SD

Season 2
Pruning labour 5 3 0 0 Fuelwood 1 2 0 0
Crop seed 14 10 14 10 Maize 272 224 272 224
Fertilizer/manure 6 14 6 14 Extra maize 23 12 0 0
Crop labour 117 51 117 51 Intercrop 24 34 24 34
Subtotal season 2 142 63 137 62 Subtotal season 2 320 243 296 185
Total: year 2 276 88 271 87 Total: year 2 766 320 707 316
aIn two cases there was sorghum instead of maize. Exchange rate in 1992 was US$1 = 43 Kenya shillings.
bCosts:

Seedlings: Mean number of trees planted was 6949 ha−1 (SD = 1785); seedlings costed at US$1.16 per 100 (price in small private
nurseries).
R.A. Swinkels et al.

Tree planting: Mean tree planting rate was 33 trees h−1 (SD = 16) (including transport homestead to field), labour costed at hired
labour price.
Seed: Mean maize seed input was 32 kg ha−1 (SD = 12) in the long rains and 41 kg ha−1 (SD = 17) in the short rains; 18 out of 23
farmers planted an intercrop (mostly beans) in the long rains, while 12 did so in the short rains; bean seed input quantities were similar
to those of maize.
Fertilizer and manure: Mean fertilizer input was 17 kg ha−1 (SD = 43) in the long rains and 4 kg ha−1 (SD = 15) in the short rains; fertil-
izer was valued at US$0.37 kg−1. Farmyard manure input was 1037 kg ha−1 at 37% moisture content, costed at US$13.50 ton−1 DM.
Crop labour: This includes both hired and farm household labour; both are valued at the hired labour price of US$0.116 h−1; hired
labour accounted for about 30% of the total average labour input. Mean labour input in the long rains was 1054 h ha−1 (SD = 385) and
1007 h ha−1 (SD = 436) in the short rainy season. Labour inputs did not vary significantly between test and control plots.
 Pruning labour: Mean pruning labour input was 25 h ha−1 (SD = 14) for the first cut of the long rains, and 22 h ha−1 (SD = 18) for the
second cut that season; the first cut of the short rains took on average 30 h ha−1 (SD = 18), while the second cut that season took on
average 18 h ha−1 (SD = 16). First pruning was about 6 months after planting (14 farmers) or about 10 months after planting (nine farm-
ers).
cBenefits:

Maize: Mean maize yield was 1521 kg ha−1 (SD = 873) in the long rains and 1180 kg ha−1 (SD = 1118) in the short rains; maize market
price was US$0.25 kg−1 at the harvest period for the long rains season and US$0.20 kg−1 for the harvest period of the short rains crop.
Intercrop: Mean bean yield was 117 kg ha−1 (SD = 125) in the long rains and 71 kg ha−1 (SD = 88) in the short rains, valued at the
market price at harvest time of US$0.38 kg−1.
Fuelwood: Mean measured fuelwood yield in long rains was 518 kg dry matter ha−1 (SD = 421) for the first cut and zero for the sec-
ond cut; for the first cut of the short rains it was 127 kg ha−1 (SD = 267) and 30 kg ha−1 (SD = 139) for the second cut (12 farmers).
(Based on their sample of 24 farmers, Shepherd et al. (1997) reported a median yield of 1.05 t ha−1 of fuelwood per annum.) As fuel-
wood is not commonly traded in the area, its value was estimated at one-third (US$7.7 t−1 dry weight) of the market price.
Extra maize: The average extra maize yield required to cover the added cost of hedgerow intercropping, compared to conventional
maize, was 10.5% (SD = 5.5%) or 242 kg ha−1 year−1 (SD = 65 kg) (includes two seasons for each of 4 years, over years 2–5 of a 5-year
production period) when based on returns to land.
Adoption Potential of Hedgerow Intercropping
101
102 R.A. Swinkels et al.

reduced the maize yield required to break even from 10.5% (SD = 5.5%) to 6% (SD
= 3%). (A change to bare-root seedlings also reduced tree growth by about 60%.
Therefore, with bare-rooted seedlings, the first cut back was assumed to take place
in the second year instead of the first.) The analysis also showed that increasing or
reducing the value of fuelwood, even to zero, had little effect on the results, because
little fuelwood was generated in the hedgerow system.
On a few occasions, farmers were unable to prune due to absence or illness
and yields were greatly reduced. Therefore, it was hypothesized that the technology
was risky and that it would increase the variability in net benefits. However, the
standard deviation of net benefits per ha of the hedge plots differed very little from
those of the other plots. It was higher in one season and lower in three seasons.
Data from eight on-farm trials that were researcher-managed during one season
also showed no important difference in the variability in returns between the con-
trol plots and the plots with the hedges. Similarly, minimum returns analysis did
not show the hedge plots to be any riskier than the conventional maize plots. In
the first three seasons, minimum returns per ha were slightly lower on the hedge
plots (US$22–34 ha−1 lower) and in the fourth season US$39 higher.

Table 5.3. Sensitivity analysis, showing mean break-even maize yield increases
of hedgerow intercropping under different assumptions.
Returns to land Returns to labour
Assumption Mean SD Mean SD

Base assumptions 10.5% 5.5% 17.5% 6.5%

Land not valueda 10.5% 5.5% 19% 7%
Bare-rooted seedlings (instead of potted 6% 3% 13% 6%
seedlings) @ US$0.46 per 100
Fuelwood sold at market price (US$23 8% 5% 15% 7%
per ton dry weight)
Fuelwood has no value 12% 6% 19% 7%
Maize price 50% higher 7% 4% 12% 4%
Maize price 50% lower 21% 11% 35% 13%
Labour cost 100% higherb 15% 7% 18% 6%
Labour costs 50% lowerb 8% 5% 18% 6%
Discount rate is 50% instead of 20% 16% 9% 26% 9%
Ten-year period considered instead of five 7% 4% 13% 5%
Fallow in short rainy seasonc 48% 65% 15% 48%
(median 39%) (median 5%)
Fallow in short rainy season, fuelwood no 65% 64% 33% 50%
value (median 49%) (median 19%)
Fallow in short rainy season, fuelwood sold 6% 73% −27% 54%
at market price
aIn an analysis of returns to land, land is never valued; this assumption therefore
does not affect the outcome of the analysis of returns to land.
bIncludes farm household and hired labour.
cFallowing in short rains from year 2 onwards.
 Adoption Potential of Hedgerow Intercropping 103

On farms where there was no crop response to the hedgerows, only 6% (about
US$7 ha−1) of the average investment costs of the technology were earned back
through fuelwood production. Median returns to labour were reduced by 22%,
from US$0.37 h−1 to US$0.29 h−1.

Acceptability

Eleven (20%) of the 56 farmers that planted hedgerows dropped out of the trial dur-
ing the first year after establishment. The main reasons included land tenure prob-
lems, intra-household disputes, and poor tree establishment (Fig. 5.2). Thus, after the
first season, 45 farmers remained in the trial, and percentages given below are based
on this group. During subsequent monitoring, over 3.5 years, another two farmers
dropped out: one because of an intra-household conflict and one lost interest.
Farmers’ management practices varied considerably between the test and control
plots: crops and intercrops changed (Shepherd et al., 1997) or manure was applied to
one plot and not the other. There were two reasons for this variation. First, the size
of the trials, averaging 40% of cropped area, made it difficult for farmers to manage
the two plots in the same manner. For example, farmers often lacked the labour
required to conduct operations in a timely manner throughout the two plots.
Secondly, in interviews during the first year, we found that farmers’ evaluation
methods differed considerably from ours. Only five farmers (11%) concurred with
our approach to treat the control and test plots in the same manner and then com-
pare crop yields on the two plots to determine the impact of the hedges. Instead, 18
(40%) sought to compare the hedges with another soil fertility amendment, such as
manure or fertilizer, to see if the hedges were as effective. Further, we found that 17
(38%) did not compare test plot yields with control plot yields at all. Rather, these
farmers wanted to compare present yields with past yields on the same plot, that is,
yields before they planted the hedges. (We were unable to determine the method
that eight farmers (18%) used to assess yield impact. Farmers were permitted to give
more than one response.)
Moreover, farmers’ assessment methods changed during the course of the trial.
Comparing performance on the test and control plots was the main evaluation
method of 61% of the farmers during the first year, but only 18% during the sec-
ond through fourth year. Over the same period, comparing past and present yields
on the test plot as the favoured evaluation method increased from 38% to 67%. In
two-thirds of the cases where farmers assessed yield performance during the second
through fourth years, they cited changes in yields over time on the test plot; in only
one-fifth of the cases did they compare test and control yields. We suspect that
many farmers initially stated that they intended to assess yield impact by comparing
test and control yields because they knew that that was what the researchers
expected them to do.
The percentage of farmers claiming that hedges improved crop yields rose from
fewer than 10% during the third season (the first after pruning) to around 50% dur-
ing the third to seventh seasons. But project technicians and field assistants, using
 Interest in hedgerow intercropping for soil fertility
104

Yes (50 farmers)

Land tenure secure?
No Uprooting
3 cases
Yes (47)
Establishment successful?
No Drop out
Yes (47) 2 cases
Household decision-maker agreement?
No Uprooting
Yes (42) 3 cases
Impact on soil fertility seen
Yes (25) No (17 )

Impact worth the expansion work? Soil erosion reduced?

Yes No (20)
Yes (7) No (10)
Expand Soil erosion reduced?
5 cases Impact worth the Feed livestock with
Yes (12) No (8) expansion work hedge biomass?
R.A. Swinkels et al.

Impact worth expansion work? Feed livestock with Yes No (6) Yes (4) No
hedge biomass?
Yes (1) No (11)
Yes (1) Expand Feeding Impact worth the Don’t expand
Feed livestock with No 6 cases
Expand 1 case livestock expansion work
1 case hedge biomass? with hedge
Impact worth the Don’t expand biomass?
Yes (2) No expansion work 7 cases
Impact worth the No Yes No
Don’t expand
expansion work 9 cases Don’t expand
Yes No Yes 6 cases

Expand Don’t expand Expand Expand Don’t expand

1 case 1 case 1 case 1 case 3 cases

Fig. 5.2. Decision tree: expansion of hedgerows by trial farmers. Six farmers are excluded (four because of lack of reliable information
and two because they migrated from the area).
 Adoption Potential of Hedgerow Intercropping 105

subjective judgements based on frequent farm visits, noted yield increases on only
20–40% of the trials. The difference may be due to farmers noticing yield increases
that field assistants did not, to field assistants not taking into account management
differences between the plots, or to farmers simply trying to please researchers.
Farmers noted a number of other benefits from the hedges. By the sixth season
after planting, one-fifth were feeding some of the tree leaves to their livestock,
mostly local-breed cattle. The three farmers with improved-breed dairy cows were
feeding all of their leafy biomass. Feeding leaves to livestock was not associated with
opinions on the effectiveness of the biomass for improving crop yields. Forty per
cent mentioned fuelwood as a valuable benefit. However, quantities were small;
wood harvested after the first and second prunings of the third season provided fuel
for only 11 days (SD = 14) and 6 days (SD = 3), respectively. The proportion of
farmers reporting that the hedges curbed soil erosion increased from 20% during
the first few seasons to 50% in the seventh season. These farmers did not have
steeper sloping hedge plots than other farmers. The overall effects of the hedges
were judged to be positive by 60–64% of the farmers in the fourth to seventh sea-
sons. Others felt the positive and negative (that is, shading) effects were offsetting,
or saw no effect (13–25%). The number of farmers reporting mainly negative
effects declined from 25% in the first season to 3% in the sixth season, reflecting
their increasing ability to cut back in time.
By the fifth season, 18 (40%) of the farmers indicated their interest in expand-
ing the hedges. They were given seed and advised on raising them in nurseries for
production of bare-rooted seedlings, but over the next two seasons, only 10 (22%)
established a nursery and transplanted the seedlings. Five expanded hedges in their
crop fields while five planted on boundaries. The decision tree (Fig. 5.2) showed
that five farmers expanded mainly for improving soil fertility, three for fodder pro-
duction (all three had improved cattle breeds and regularly sold milk), and two for
controlling soil erosion.
The ten expanders differed from the rest of the farmers in several important
respects. They were more likely to be households where a male or a couple managed
the farm (P < 0.02), to be visited regularly by extension agents (P < 0.01), to have a
cash crop such as coffee or tea (P < 0.003), and to manage their hedges well,
according to ratings by technicians (P < 0.07). There was some evidence that
expanders had less depleted soils than others, based on analyses of the soil of four
expanders and 28 non-expanders. Expanders had significantly higher cation
exchange capacities (P < 0.04) and higher amounts of calcium (P < 0.01) and
exchangeable bases (P < 0.04), in topsoil, subsoil and deep soil, compared with
non-expanders. Maize yields in the test plots of expanders were significantly higher
than in the test plots of non-expanders in the first, second and third season, but not
in the fourth season after hedge establishment.
The expanders did not appear to have different levels of wealth, off-farm
income, farm size, or crop sales, to have different labour : land ratios, to be more
self-sufficient in food, to use more purchased inputs, or to be of different ages com-
pared with the non-expanders. Furthermore, the species of hedge did not influence
expansion; five grew Leucaena leucocephala and five, Calliandra calothyrsus.
106 R.A. Swinkels et al.

Discussion

Farmers were able to establish the hedges effectively, as has been shown in Nigeria
(Reynolds et al., 1991), Malawi (ICRAF, 1993) and the Philippines (Fujisaka,
1993). The low labour requirements for pruning are similar to those reported from
on-farm trials by Reynolds et al. (1991). Although pruning had to be done during
periods of peak labour use, farmers appeared to cope better with this task over time.
However, the fact that women do not generally prune hedges greatly reduces the
adoption potential of the technology among the 40–60% of the farm households in
the survey area which are headed by women.
Given the greater demands on their time, women may avoid pruning as part of
a strategy to force men to bear the risks and costs of the technology (David, 1992).
Should the hedges improve soil fertility, more women might be prepared to make
trade-offs and prune their hedges.
Reduction in the labour requirements for weeding was not an important bene-
fit, and this differs from the findings of other trials in the tropical humid lowlands
(Reynolds et al., 1991; Fernandes et al., 1993).
Break-even yield increases are fairly low, indicating that only relatively low
yield increases are required to cover the costs of the technology. That the break-even
yield increase is lower in the returns to land analysis highlights the attractiveness of
the technology in areas of high population density, small farms and plentiful labour.
Sensitivity analysis showed the importance of reducing establishment costs for
increasing the profitability of the technology. Our data do not support the hypothe-
sis that hedges increase variability in returns, although it is clear that farmers who
were unable to prune on time would suffer yield decreases.
In the end, only seven farmers (14%) expanded their hedges to improve soil
fertility or soil erosion control. Non-expansion is not necessarily a sign of non-
adoption; several farmers indicated that they appreciated the practice but only
wanted to adopt it on a portion of their field. The general lack of expansion, how-
ever, indicated that overall confidence in the technology was low.
Given the disappointing data on yield impacts and economic returns from
hedgerow intercropping, extension efforts were reduced and focused more on ero-
sion control. On-farm research was reoriented to focus on improved fallows, which
produced greater biomass, had lower establishment and overall labour costs, and
greater yield effects. New on-station trials were established at Maseno in 1994 using
calliandra, to better understand crop and tree productivity in contour plantings,
and the possible effects of phosphorus application on performance (Niang et al.,
1996).
The low adoption potential of hedgerow intercropping for improving soil fer-
tility has been noted in several other farmer-managed on-farm trials in the semiarid
areas of Kenya (David, 1995), in the subhumid plateau areas of Malawi (Minae,
1994), and in the humid lowlands of Nigeria (Whittome, 1994) and Cameroon
(Degrande and Duguma, 2000). The practice appears to have more adoption
potential if its main aim is to provide feed for an intensive dairy operation, as in
 Adoption Potential of Hedgerow Intercropping 107

Kenya (Chapter 7 of this volume; Reynolds, 1994; Reynolds and Jabbar, 1994) or
for soil erosion control, as in the Philippines (Fujisaka, 1993).
For reducing erosion, the practice appears to be most relevant to:
● areas with erosive soils on sloping land;
● areas with high rainfall, where competition for moisture between trees and
crops is minimal; and
● farmers with high labour–land ratios, because the labour requirements of the
practice are high.
But incentives may be needed to motivate farmers to adopt hedgerow intercropping
on a wide scale, if reducing erosion is the only benefit.
The trial results also had important implications for participatory on-farm
research methods. First, where trial plot size is large (for example, 40% of cropped
area) or trials are long-term (more than two seasons), farmers will find it difficult to
manage non-experimental variables in a uniform manner.
Secondly, in trials designed to facilitate farmer assessment, farmers’ own evalua-
tion methods should be factored into the design. In our trial, as in others (Reynolds
et al., 1991; CIAT, 1992; Versteeg and Koudokpon, 1993), farmers were not as
interested in the comparison between the test and the control plot as were the
researchers. If farmers are not interested in comparing test yields with control
yields, control plots are not needed. Where farmers do want to compare test and
control plots, the exact comparison should be the one they are interested in. For
example, in our trial, more farmers were interested in comparing hedges with
manure than were interested in the with- and without-hedges comparison.
Thirdly, the trials provided reliable socioeconomic data on feasibility, costs
and farmers’ expansion. Farmers’ assessment of the hedges were less reliable, as
evidenced by disparities between farmers’ and technicians’ opinions on yield
response and between farmers’ stated appreciation of the practice and their will-
ingness to plant new hedges. We believe that farmers’ assessments were biased
towards reporting more positive impacts for two reasons, because criticism of a
practice is impolite in their culture, and because of their belief that positive
assessments would help them obtain free inputs or even employment from the
project.
Finally, the findings suggest that it may not be possible to obtain reliable bio-
physical data on yield response and socioeconomic data on farmer assessment from
the same trial (Shepherd and Rodger, 1991). In order to obtain a good socioeco-
nomic evaluation, individual farmers need to use the technology in the manner
they see fit. Thus it is unlikely to be possible to compare yields between plots or
across farms. Conversely, a high degree of control of experimental and non-experi-
mental variables is required to measure biophysical response. Under such manage-
ment, farmers will not be able to assess the adoptability of a technology adequately.
Farmer-participatory trials are useful in the early stage of the development of a
technology for obtaining farmer feedback, for identifying critical constraints
affecting performance and to define realistic conditions for researcher-managed tri-
als, which are recommended for assessing biophysical performance (Shepherd and
108 R.A. Swinkels et al.

Roger, 1991). Following these, adoption potential and acceptability can then be
determined in extension-led, farmer-participatory trials, rather than through
experiments aimed at collecting biophysical and socioeconomic data of the kind
reported here.

Acknowledgements

We thank the following who assisted in conducting the research: Luke Anjeho,
Aggrey Otieno, Joseph Onoo, Stephen Amoko, Florence Ochuka, Julius Otiende,
Jairus Ounza, Cleophas Omondi, Mark Waswa, Victoria Atieno, Nehemiah
Odhiambo, Miriam Ayiemba, Soniia David and W. Mwangi Muturi. We are grate-
ful for comments on earlier drafts from Richard Coe, Dirk Hoekstra, Henk Moll,
Collins Obonyo, Sylla Pahladsingh, Frank Place and Meredith Soule. The support
of the Kenya Forestry Institute, in particular, Dr Daniel Nyamai, is appreciated. We
are grateful to Richard Coe of ICRAF for biometric support. Financial support was
provided by the Rockefeller Foundation, while the Swedish International
Development Authority supported Eva Ohlsson and Hans Sjögren, and the
Directoraat Generaal Internationale Samenwerking of The Netherlands
Government supported Rob Swinkels.

References

Balasubramanian, V. and Sekayange, L. (1991) evaluations of hedgerow intercropping

Effects of tree legumes in hedgerows on soil under semi-arid conditions. Agroforestry
fertility changes and crop performance in Systems 32, 1.
the semi-arid highlands of Rwanda. David, S. and Swinkels, R.A. (1994) Socio-
Biological Agriculture and Horticulture 8, Economic Characteristics of Households
17–32. Engaged in Agroforestry Technology Testing in
Carter, J. (1995) Alley farming: have resource- Western Kenya. AFRENA Report No. 78.
poor farmers benefited? Natural Resources International Centre for Research in
Perspectives 3, June 1995. Overseas Agroforestry, Nairobi.
Development Institute, London. Degrande, A. and Duguma, B. (2000) Adoption
CIAT (1992) Annual Report 1992. International Potential of Rotational Hedgerow
Centre for Tropical Agriculture, Cali, Intercropping in the Humid Lowlands of
Colombia. Cameroon. AGREN Network Paper No.
CIMMYT (1988) From Agronomic Data to 103, Agricultural Research and Extension
Farmer Recommendations: an Economics Network, Overseas Development Institute,
Training Manual. International Maize and London.
Wheat Improvement Centre, Mexico, DF. Dvorak, K.A. (1996) Adoption Potential of Alley
David, S. (1992) Open the Door and See All the Cropping: Final Project Report. Resource and
People: Intra-Household Processes and the Crop Management Research Monograph
Adoption of Hedgerow Intercropping. Paper No. 23. International Institute of Tropical
presented at the Rockefeller Foundation Agriculture, Ibadan, Nigeria.
Social Science Fellows Meeting, 9–13 Fernandes, E.C.M., Davey, C.B. and Nelson, L.A.
November 1992, CIMMYT, El Batan, (1993) Alley cropping on an acid soil in the
Mexico. Upper Amazon: mulch, fertilizer, and
David, S. (1995) What do farmers think? Farmer hedgerow root pruning effects. In: Ragland,
 Adoption Potential of Hedgerow Intercropping 109

J. and Lal, R. (eds) Technologies for Muturi, W.M. and Shepherd, K.D. (1992)
Sustainable Agriculture in the Tropics, Evaluation of hedgerow intercropping on
Proceedings of two international symposia. farms in western Kenya: participatory tech-
ASA Special Publication No. 56. American nology testing trials. In: On-farm Research
Society of Agronomy, Crop Science Society Project, Maseno, Kenya: Papers presented to
of America and Soil Science Society of the Zonal Conference of the Eastern and
America, Madison, Wisconsin, pp. 77–96. Central Africa Agroforestry Network,
Fujisaka, S. (1993) A case of farmer adaptation Nairobi, 6–10 July 1991. AFRENA Report
and adoption of contour hedgerows for soil No. 50. International Centre for Research
conservation. Experimental Agriculture 29, in Agroforestry, Nairobi, pp. 52–81.
97–105. Otieno, H.J.O., Heineman, A.M., Mengich,
Gladwin, C. (1980) A theory of real life choice: E.K. and Amadalu, B. (1991) AFRENA
applications to agricultural decisions. In: On-Station Project Maseno, Kenya Progress
Barlett, P.F. (ed.) Agricultural Decision Report for the Year 1990. AFRENA Report
Making: Anthropological Contributions to No. 41. International Centre for Research
Rural Development. Academic Press, New in Agroforestry, Nairobi, Kenya.
York, pp. 45–86. Reynolds, L. (1994) A review of the biophysical
Heineman, A.M., Mengich, E.K., Olang, A.D. and socio-economic basis of alley farming,
and Otieno, H.J.O. (1990) AFRENA On- and its adoption potential. Consultant
Station Project Maseno, Kenya. Progress report for the Alley Farming Network of
Report for the Period January 1988 to Africa. International Agricultural
January 1990. AFRENA Report No. 27. Consultancy, Kibworth, Leicester, UK.
International Centre for Research in Reynolds, L. and Jabbar, M. (1994) The role of
Agroforestry, Nairobi, Kenya. alley farming in African livestock produc-
ICRAF (1993) Annual Report 1992. tion. Outlook on Agriculture 23, 105–113.
International Centre for Research in Reynolds, L., di Domenico, C., Atta-Krah, A.N.
Agroforestry, Nairobi, Kenya. and Cobbina, J. (1991) Alley farming in
ICRAF (1995) Annual Report 1994. south-western Nigeria: the role of farming
International Centre for Research in systems research in technology develop-
Agroforestry, Nairobi, Kenya. ment. In: Tripp, R. (ed.) Planned Change in
Kang, B.T. (1993) Alley cropping: past achieve- Farming Systems: Progress in On-farm
ments and future directions. Agroforestry Research. John Wiley & Sons, London, pp.
Systems 23, 141–155. 85–108.
Kang, B.T., Reynolds, L. and Atta-Krah, A.N. Scherr, S.J. (1995) Economic factors in farmer
(1990) Alley farming. Advances in adoption of agroforestry: patterns observed
Agronomy 43, 315–359. in western Kenya. World Development
Minae, S. (1994) Malawi–ICRAF On-Farm 23(5), 787–804.
Agroforestry Research Program Annual Scherr, S.J. and Oduol, P.A. (1989) Alley-cropping
Report, 1991–92. AFRENA Report No. 82. and Tree Borders in Crop Fields in Siaya
International Centre for Research in District, Kenya. International Council for
Agroforestry, Nairobi. Research in Agroforestry and CARE
Müller, E.U. and Scherr, S.J. (compilers) (1989) International (Kenya), Nairobi.
Technology Monitoring and Evaluation in Shepherd, K.D. and Roger, J.H. (1991)
Agroforestry Projects: An Annotated Approaches to On-farm Testing and
Bibliography. International Centre for Evaluation of Agroforestry Technology.
Research in Agroforestry, Nairobi. Working Paper No. 67. International Centre
Niang, A.I., Amadalo, B.A., Gathumbi, S.M., for Research in Agroforestry, Nairobi.
Otieno, J.H.O., Obonyo, C.O. and Shepherd, K.D., Swinkels, R.A., Muturi, W.M.,
Obonyo, E. (1996) KEFRI/KARI/ICRAF- Ohlsson, E. and Ndufa, J.K. (1992)
AFRENA-Maseno Project Progress Report Evaluation of hedgerow intercropping on
September 1995–September 1996. AFRENA farms in western Kenya: diagnosis of tech-
Report No. 110. International Centre for nology potential and farmers’ interest. On-
Research in Agroforestry, Nairobi, Kenya. Farm Research Project, Maseno, Kenya:
Ohlsson, E., Ndufa, J.K., Swinkels, R.A., Papers presented to the Zonal Conference of
110 R.A. Swinkels et al.

the Eastern and Central Africa Agroforestry of the East and Central African AFRENA
Network, Nairobi, 6–10 July 1991. Workshop, 22–26 June 1992, Kigali,
AFRENA Report No. 50. International Rwanda. AFRENA Report No. 58.
Centre for Research in Agroforestry, International Centre for Research in
Nairobi, pp. 1–29. Agroforestry, Nairobi, Kenya, pp. 84–85.
Shepherd, K.D., Ohlsson, E., Okalebo, J.R. and Swinkels, R.A. and Scherr, S.J. (1991) Economic
Ndufa, J.K. (1996) Potential impact of Analysis of Agroforestry Technologies: An
agroforestry on soil nutrient balances at the Annotated Bibliography. International
farm scale in the East African Highlands. Centre for Research in Agroforestry,
Fertilizer Research 44, 97–99. Nairobi, Kenya.
Shepherd, K.D., Ndufa, J.K., Ohlsson, E., Versteeg, M.N. and Koudokpon, V. (1993)
Sjogren, H. and Swinkels, R. (1997) Participative farmer testing of four low
Adoption potential of hedgerow intercrop- external input technologies, to address soil
ping in maize-based cropping systems in fertility decline in Mono Province (Benin).
the highlands of western Kenya. Part 1. Agricultural Systems 42, 265–276.
Background and agronomic evaluation. Whittome, M.P.B. (1994) The adoption of alley
Experimental Agriculture 33, 197–209. farming in Nigeria and Benin: the on-farm
Swinkels, R.A. and Franzel, S. (1992) Analysis of experience of IITA and ILCA. PhD thesis,
factors influencing labour productivity in Department of Geography, University of
hedge pruning. In: Hoekstra, D. and Cambridge, Cambridge, UK.
Beniest, J. (compilers) Summary Proceedings
 Farmer-designed
Agroforestry Trials: Farmers’
Experiences in
6
Western Kenya
S. FRANZEL,1 J.K. NDUFA,2 O.C. OBONYO,2
T.E. BEKELE3 AND R. COE1
1International
Centre for Research in Agroforestry, PO Box 30677,
Gigiri, Nairobi, Kenya; 2Kenya Forestry Research Institute (KEFRI),
Maseno Agroforestry Research Centre, Maseno, PO Box 25199,
Kisumu, Kenya; 3Institute of Biodiversity, Conservation and Research, PO
Box 2003, Addis Ababa, Ethiopia

Summary
In farmer-designed tree trials, farmers experiment on their own with new tree
species, planting them where and how they wish on their farms. In western Kenya,
50 farmers (half female) planted and evaluated five different species: two upper-
storey species (Grevillea robusta and Casuarina junghuhniana), primarily for timber
and pole production, and three shrub species (Leucaena leucocephala, Leucaena
diversifolia, Calliandra calothyrsus), mainly for improving soil fertility. Results
revealed important criteria that farmers use in evaluating tree species. For example,
casuarina was widely appreciated for its ornamental value, despite its relatively poor
ratings on survival and growth. The findings also highlighted the importance of
testing new species under farmer conditions, as their ranking on growth was differ-
ent than in on-station trials. Farmers appreciated the upper-storey species but most
did not find the shrub species to be effective in improving soil fertility. The trials
also provided important information on farmers’ management problems, preferred
niches for tree planting, and intended uses of tree products.

Introduction

In most African countries, farmers have little input into decisions about which tree
species are made available to them. Rather, scientists or extension services generally
make the decisions – screening new species in on-station trials or from available lit-

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 111
112 S. Franzel et al.

erature and evaluating them according to well-known criteria such as growth and
form. In order to enhance farmers’ role in species introduction, scientists of the
Kenya Forestry Research Institute and the International Centre for Research in
Agroforestry and members of 13 self-help groups initiated farmer-designed trials in
western Kenya, in which farmers evaluated new tree species on their own farms.
Farmers’ assessments of the trees were an important input into the decision on
whether to disseminate them.
Farmer-designed trials, also called collegial trials (Biggs, 1989) or type 3 trials
(see Chapter 2), are an important means for increasing farmers’ role in the technol-
ogy development process. In these trials, farmers experiment on their own with new
technologies or components and, together with researchers, they monitor their
experiences (Haverkort et al., 1991). Because each farmer controls the testing proce-
dure on his/her farm, farmer-designed trials are not seen as being appropriate for
the collection of quantitative data, either biophysical or socioeconomic, or for the
formal testing of hypotheses. Thus they are often dismissed as being ‘soft’ research
and are rarely reported on in refereed journals. Nevertheless, farmer-designed trials
are important for obtaining farmers’ assessments and for understanding the process
by which farmers test new practices and incorporate them into their farming sys-
tems. As illustrated in this study, quantitative and statistically valid data on prefer-
ences can be collected by asking farmers to rate alternative species across different
criteria (Fransella and Bannister, 1977).
Study objectives were to: (i) examine how farmers in western Kenya experi-
ment with new tree species, that is, where they plant them, how they manage them,
how they use them, and the problems they encounter; (ii) elicit farmers’ evaluation
criteria and evaluation of the species across criteria; (iii) assess how the species per-
form on farmers’ fields and under farmer management, as compared to on-station;
(iv) associate farmers’ assessment of species with selected farm and household char-
acteristics; and (v) recommend which new species should be made available to farm-
ers by development institutions.

Methods

The trials were conducted in a maize-based land-use system, encompassing parts of

Siaya, Kisumu and Vihiga Districts, in western Kenya (Fig. 4.1). Altitudes range
from 1450 m to 1650 m and rainfall is bimodal, averaging 1500 mm to 2000 mm
year−1. The landscape is undulating, with slopes of 2–10%. The main types of soil
include Ferralsols, Acrisols and Nitisols, and are generally infertile with a low pH
(5–7). Most farms range in size from 0.5 ha to 2.0 ha and population densities are
400 to 1200 persons km−2. Maize accounts for over half of the cultivated area and
the use of purchased inputs, such as seed and fertilizer, is low. Most households own
cattle, predominantly of the local breed. About 45% of the households are female-
headed. Eucalyptus spp. (mainly Eucalyptus saligna) is the most common type of tree
in the area and is widely grown for poles, for domestic use and for sale, and for fuel-
wood, for domestic use (David and Swinkels, 1994).
 Farmer-designed Agroforestry Trials 113

The collaborating farmer groups selected 50 farmers who were interested in

testing new tree species on their farms. The selected farms included the range of
types of farms found in the area; over half were female-headed, median farm size
was 1.1 ha, and two-thirds had cattle. Technicians provided farmers with informa-
tion about each species (e.g. growth characteristics and end uses) and, following
group discussions, farmers decided to test three shrub species (Leucaena
leucocephala, Leucaena diversifolia, Calliandra calothyrsus) and two upper-storey
species (Grevillea robusta (Meru) and Casuarina junghuhniana (Australia)). The
shrub species were intended to improve soil fertility through application of prun-
ings to crops, and to provide livestock feed during the dry season, thus addressing
two of farmers’ most important problems in the area (Minae and Akyeampong,
1988). The upper-storey species were intended to provide timber and poles, two
highly preferred tree products that farmers use for both domestic use and cash sales
(Muturi and Franzel, 1992). The shrub species were new to the farmers; before
planting, they were able to view them in on-farm hedgerow intercropping trials.
Farmers were familiar with grevillea and casuarina but lacked information about
them as they were not widely planted in the area.
Planting took place during the short rains season, September through
November, 1991. Researchers raised trees in nurseries and delivered an average of
273 trees, 29–103 per species, to each farmer, depending on their availability.
Technicians demonstrated planting methods, but all planting and management of
the species was done by the farmers themselves. Rainfall during the season of plant-
ing was 20% lower than the 30-year average.
Trials were monitored using both informal and formal survey methods.
Informal interviews were conducted during researchers’ visits to the trials. Group
visits were also organized in which farmers visited one another’s trials and discussed
trial performance. Formal surveys using structured questionnaires were conducted
3, 15, 30 and 52 months after planting; the number of farmers interviewed ranged
from 30 to 43. The questions concerned farm and household characteristics, tree
management, preferred end uses and problems encountered. Some biophysical data
were collected; survival rates were measured 3 months after planting and tree height
and root collar diameter 30 months after planting. In the latest survey, 52 months
after planting, researchers rated farmers’ level of interest in the trees as high,
medium or low, according to the level of management and perceived benefits. A
random sample of 22 neighbours of participating farmers was also interviewed to
assess whether they had planted any of the trees in the trial (Bekele, 1997).
Two methods were used to get farmers’ ratings of different trees. In informal
visits, farmers were asked to explain the advantages and disadvantages of various
species on their farm and the criteria behind their preferences (Bruin et al., 1994;
Franzel, 2001). In subsequent visits, farmers used a local board game, bao, to rate
species according to the criteria by moving seeds among pockets of the board
(Franzel et al., 1995; Franzel, 2001). Farmers rated a species on a particular criterion,
e.g. growth rate, by placing 1–5 seeds (5 represented a high rating) in a pocket next
to a branch of that species. E. saligna was also used in the bao game ratings as a con-
trol for the rating of upper-storey species, although it was not planted in the trial.
114 S. Franzel et al.

The influence of various farm and household characteristics on farmers’ ratings

of their interest in planting different species was assessed using a linear logistic
model for ordered category response data (Collet, 1991). The variables considered
included wealth level, farm size, off-farm income, ethnic group, age, gender, district
and livestock ownership.

Results

Tree establishment, management and intended uses

About two-thirds of the farmers planted all of their seedlings; the others gave away or
sold some. Most farmers planted the shrub species in hedgerows in their crop fields,
others planted on external or internal boundaries, on terraces, or in fallow land.
Grevillea tended to be planted on external boundaries whereas casuarina was equally
distributed among the homestead, internal and external boundaries. Only a few farm-
ers planted any of the five species in woodlots or scattered in their fields (Table 6.1).
Survival rates 3 months after planting were over 70% for all species except
casuarina, which had a rate of 46%. According to farmers, principal causes of mor-
tality were termites, drought and browsing (both by wild and domestic animals).
Each of the above causes was mentioned by 11–16% of the farmers.
Over 80% mixed their shrub species in the field; this was partly because, in
many cases, the farmers, could not distinguish between them. About 40% mixed
grevillea and casuarina, even though they are easy to distinguish.
Farmers planted the shrubs at spacings of 1.0–1.7 m in hedgerows. These are
much wider than the recommended 0.5 m. Spacings on boundaries for grevillea

Table 6.1. Niche where farmers planted existing upper-storey trees and the test
species (% of farmers responding).
Existing
upper-
storey Grevillea Casuarina Leucaena Leucaena Calliandra
Niche treesa robusta junghuhniana leucocephala diversifolia calothyrsus

Hedgerows 0 0 0 63 59 70
in crops
External 44 50 29 9 9 6
boundary
Internal 75 36 29 6 6 9
boundary
Homestead 92 13 29 2 0 2
Otherb 83 12 20 18 20 18

Data from 43 farmers. Column percentages do not sum to 100 because a farmer
can plant in more than one niche.
aFrom Muturi and Franzel (1992).
bOther niches include scattered in cropland, woodlots and pasture.
 Farmer-designed Agroforestry Trials 115

were 2.3 m (SD = 1.9) and for casuarina, 3.2 m (SD = 2.9), as compared to the rec-
ommended 1 m (to be later thinned to 2 m). About two-thirds said they made their
own decision about the spacing, one-third said they based the decision on advice
from others. Farmers justified their wide spacings on the basis that they had
received few trees and were interested in maximizing the area covered.
Most of the farmers intended to use the shrub species for improving soil fertility
(Table 6.2); other reasons expressed by over 40% of the farmers were to produce fuel-
wood and fodder. Grevillea and casuarina were being planted primarily for timber
and poles. Only 29% claimed that one of the two main uses of grevillea would be for
firewood, and there were no differences in responses among males and females.

Farmers’ evaluation of tree growth

Farmers used a wide range of criteria to evaluate the growth of the trees, including
survival, rapidity of growth, compatibility with crops, and resistance to pests,
drought and browsing. Fifteen months after planting, farmers rated grevillea and
calliandra highest on survival and growth (ratings of 3.8–4.5 out of 5); casuarina
was rated lowest (1.7–2.6) (Table 6.3). These data were in line with the actual data
on survival and growth collected by the technicians. On resistance to termites, only
casuarina received a low rating (1.6).
Two other growth characteristics were also rated. Grevillea and calliandra
received high ratings (4.5 and 4.0 respectively) on drought resistance, casuarina was
lowest (2.3). Grevillea and eucalyptus were the only species with high ratings on
resistance to browsing (4.1 and 4.0, respectively); casuarina was rated low (2.6), in
part because it was frequently planted around the homestead, where livestock were
common.
Thirty months after planting, grevillea and eucalyptus were rated high on
rapidity of growth (4.4 and 4.3, respectively); casuarina was rated medium (3.2)
(Table 6.4). In fact, growth data from the farms showed grevillea outperforming

Table 6.2. Farmers’ intended uses for the trees they planted in the farmer-
designed trial (% of farmers responding).
Grevillea Casuarina Leucaena Leucaena Calliandra
Use robusta junghuhniana leucocephala diversifolia calothyrsus
Firewood 29 16 52 48 52
Fodder 0 0 43 41 39
Soil conservation 0 0 11 11 7
Soil fertility 0 0 72 73 77
Timber 93 71 0 0 2
Poles 48 43 0 0 0
Shade/beauty 9 25 0 2 2
Data from 43 farmers. Column percentages do not sum to 100 because a farmer
can have more than one use for a tree.
116 S. Franzel et al.

Table 6.3. Farmers’ mean rating of species, using the bao game, on criteria
important to farmers, 15 months after planting (standard deviation in parentheses).
Termite Drought Browsing
Species Survival Growth resistance resistance resistance
Grevillea robusta 3.9 (1.4) 3.8 (1.3) 3.4 (1.5) 4.5 (1.2) 4.1 (1.6)
Casuarina 1.7 (1.2) 2.6 (1.5) 1.6 (1.4) 2.3 (1.6) 2.6 (1.9)
junghuhniana
Leucaena 3.9 (1.1) 3.5 (1.3) 3.6 (1.6) 3.5 (1.4) 2.5 (1.6)
leucocephala
Leucaena 3.9 (1.1) 3.5 (1.3) 3.6 (1.6) 3.5 (1.4) 2.5 (1.6)
diversifolia
Calliandra 4.2 (1.2) 4.5 (1.0) 3.9 (1.5) 4.0 (1.4) 2.7 (1.7)
calothyrsus
Eucalyptus spp. 3.5 (1.7) 3.3 (1.7) 2.6 (1.4) 3.6 (1.4) 4.0 (1.7)
Data from 37 farmers. The rating of 1 to 5 refers to the score in number of seeds
the farmers gave to a tree on a particular criterion. A rating of 5 was excellent, a
rating of 1, poor.

casuarina (Table 6.5). Among the shrub species, calliandra received high ratings on
biomass production (4.9); L. leucocephala rated lowest (3.4) primarily because of
attacks by psyllids (Heteropsylla cubana). On compatibility with crops, casuarina
and grevillea rated highest (4.5 and 4.0, respectively) and eucalyptus lowest (1.4).
Ratings for the shrub species were intermediate (3.3–3.8) but their high standard
deviations reflected the varying opinions of farmers, as well as differences in prun-
ing management.

Table 6.4. Farmers’ mean ratings of species, using the bao game, on growth
characteristics, intended uses and preference for future planting, 30 months after
planting (standard deviation in parentheses).
% farmers
preferring
Biomass Compatibility for future
Species Growth production with crops Fodder Firewood planting

Grevillea robusta 4.4 (0.9) – 4.0 (1.3) – 4.1 (1.0) 73

Casuarina 3.2 (1.1) – 4.5 (0.7) – – 46
junghuhniana
Leucaena – 3.4 (0.8) 3.8 (1.8) 4.0 (1.4) 3.8 (1.0) 29
leucocephala
Leucaena – 3.7 (0.9) 3.6 (1.6) 3.4 (1.5) 3.8 (0.7) 24
diversifolia
Calliandra – 4.9 (0.2) 3.3 (1.8) 4.1 (1.3) 4.1 (1.1) 41
calothyrsus
Eucalyptus spp. 4.3 (1.0) – 1.4 (0.9) – 3.6 (1.2) 27

Data from 37 farmers. The rating of 1 to 5 refers to the score in number of seeds the farm-
ers gave to a tree on a particular criterion. A rating of 5 was excellent, a rating of 1, poor.
 Farmer-designed Agroforestry Trials 117

Table 6.5. Mean height and root collar diameter (standard deviations in paren-
theses) for species in farmer-designed trials and on-station trials, 30 months
after planting.
Casuarina Ratio:
Grevillea robusta junghuhniana casuarina/grevillea
Height Root collar Height Root collar Height Root collar
District (cm) diameter (cm) (cm) diameter (cm) (cm) diameter (cm)

On-station 678 (36) 8.05 (0.56) 694 (87) 9.14 (0.94) 1.02 1.14
On-farm 382 (138)** 4.9 (1.80)** 416 (135)* 3.5 (1.5)** 1.09 0.71
Ratio: 1.77 1.64 1.67 2.61
on-station/
on-farm

For grevillea, 84 accessions from local collections were tested in the on-station
trial; for casuarina, 20.
** and * denote significance at the 1% and 5% level, respectively, between on-
station and on-farm parameters.

Comparison of tree growth on-farm and on-station

Both grevillea and casuarina were grown in on-station trials near the survey area.
Thirty months after planting, heights and root collar diameters (RCDs) were signif-
icantly higher on-station than on-farm (Table 6.5), probably because of the higher
level of management, including application of phosphorus fertilizer, better weeding
and less competition from adjacent crops. Root collar diameters of grevillea in on-
station trials were 1.6 times greater than those in on-farm trials, the corresponding
ratio for casuarina was 2.6 times. In the on-station trial, casuarina outperformed
grevillea in RCD (14% higher, significant at P < 10%). In the on-farm trial, the
RCD of casuarina was only 71% that of grevillea. However, the difference was not
significant due to the high variability (coefficients of variation ranged from 32% to
43%) in the on-farm trial results.

Farmers’ use and evaluation of trees

By the 30th month after planting, farmers had coppiced the shrub species about
seven times and thus had considerable experience using the leafy and woody bio-
mass. Ninety per cent used the leaves as a mulch, 81% used the branches for fire-
wood and 46% used the leaves for fodder. Three farmers (7%) also sold leaves to
other farmers to use as fodder. Fifty-six per cent claimed that their main intention
in using the shrubs was to improve soil fertility; 17% cited fodder.
Farmers’ perceptions of the effect of the shrub species on their crop yields var-
ied considerably. For calliandra, the best-performing shrub species in terms of bio-
mass production, nine farmers (39% of those using the species primarily to improve
soil fertility) saw a positive effect, while 61% saw no effect or were not sure. Of the
17 (41%) who had a strong interest in planting calliandra in the future, only seven
118 S. Franzel et al.

intended to use it primarily for improving their soil fertility. Only about one-third
of the farmers who used the leucaenas for improving their crop yields claimed that
it was effective.
Farmers’ experiences using the shrubs as a fodder were also varied. Thirty per
cent of those who had tried using them claimed that they were of low palatability to
cattle; 50% gave them a high rating. None of the farmers who rated them low had
tried to mix the shrubs with other fodder and gradually increase their proportion, a
recommended strategy to get livestock accustomed to new forage species. Of the 17
farmers with a strong preference for planting calliandra in the future, six wanted to
use it primarily for fodder.
Concerning firewood, grevillea and calliandra received the highest ratings: gre-
villea because its fire lasts long (the wood is of high density) and calliandra because
it makes a strong fire (has high calorific value) and dries quickly. The leucaenas and
eucalyptus also received fairly high ratings. Farmers were unable to rate casuarina
because they had not yet used it.
No one had yet used the upper-storey species for timber or poles, their main
intended uses. However, farmers had some knowledge of eucalyptus and grevillea
for timber and poles. Concerning use as timber, grevillea was rated high (4.8)
because the wood is durable, workable and does not crack easily. Eucalyptus also
received good ratings (4.0) because it is weevil resistant, but it is difficult to work
with and cracks easily. For pole production, eucalyptus rated high (4.9) because of
its straightness, its hard, durable wood, and its resistance to termite attack. Grevillea
received medium ratings (3.2) because the wood is less resistant to weevils and rots
quickly if used as a ground post or for roofing.
Farmers’ preferences for future plantings were viewed as a composite of all the
criteria and thus an overall rating of the tree. Seventy-three per cent expressed a
strong interest (rating of 5) for planting grevillea, 46% for casuarina, and 41% for
calliandra. Casuarina, while performing relatively poorly on growth, was preferred
for its beauty and the pleasant sound it made when the wind was blowing. Fewer
than 30% expressed interest in planting the leucaenas or eucalyptus. The leucaenas
were not preferred because of their poor growth, relative to calliandra, and eucalyp-
tus, because of its poor compatibility with crops.
The farm and household characteristics examined were not very useful in
explaining farmers’ ratings of their interest in planting a species. Using the linear
logistic model, only the district emerged as a significant variable (t = 4.3; P =
0.000) affecting the rating of calliandra. The district was probably a good proxy
for biophysical performance, as Kisumu District, where ratings were particularly
low, was a low-potential area and calliandra performed relatively poorly. The find-
ings, t values and significance levels were similar for the other two shrub species.
No variables were found that affected the ratings of grevillea, in part because there
was little variation in the ratings. There was considerable variation in the ratings of
casuarina, but no variable was found that influenced the ratings. The inability of
the model to explain variation in ratings is probably due to the limited range of
variables (for example, soil nutrient status and tree growth were not included) and
the small sample size.
 Farmer-designed Agroforestry Trials 119

Fifty-two months after planting, 63% of the farmers showed a high interest in
planting upper-storey trees; only 30% in hedgerow intercropping. Sixty-three per
cent showed a high interest in grevillea, 47% in calliandra and 43% in casuarina.
Less than 10% showed a high interest in the leucaenas. Uptake by neighbours was
low; only 3 of 22 neighbours sampled had planted grevillea since the start of the
survey, one had planted calliandra, and one, casuarina. Lack of seed or seedlings was
the main reason that farmers cited for not planting (Bekele, 1997).

Discussion

The farmer-designed trials reported in this chapter served four important func-
tions. First, they provided important information about farmers’ management
problems, preferred niches for tree planting and intended end uses of tree prod-
ucts. The main problems reducing tree survival were termites, drought and brows-
ing. External boundaries were the most important niche for the planting of
upper-storey trees, whereas they ranked fourth among existing trees (Table 6.1).
This finding suggests that farmers consider the other three niches (homesteads,
internal boundaries and in crop fields) to be fairly saturated, and that external
boundaries represent the most important niche for future plantings. Casuarina was
the species most likely to be planted around the homestead, reflecting its utility to
farmers as an ornamental tree.
Shrub species were planted mainly in hedges in crop fields, as their major pur-
pose was to increase soil fertility. Many farmers were influenced by the experiences
of their fellow group members, who had planted shrub species in hedgerow inter-
cropping trials.
Upper-storey trees were planted primarily for timber and poles. These results
confirm findings from previous surveys that firewood is not a particularly scarce
product in the area and that it is considered, by both men and women, as a by-
product rather than as a main product from tree harvests (Bradley, 1991). In con-
trast, farmers in central Kenya plant grevillea mainly for firewood (Tyndall, 1996).
Secondly, the trials were important for determining farmers’ evaluation criteria
and for obtaining farmers’ assessment of each species across the criteria. Six growth
characteristics and six end uses were found to be important. Most were well known
to researchers but one, the importance of ornamental value, came as a surprise. Its
importance was highlighted by the finding that many farmers were interested in
planting casuarina as an ornamental despite its relatively poor ratings on other crite-
ria. The results show the usefulness of the trial in providing researchers with infor-
mation on the criteria farmers use to evaluate trees.
Concerning other upper-storey trees, grevillea outperformed eucalyptus on
every growth characteristic except rapidity of growth, and outperformed casuarina
on every one except compatibility with crops. Judgements on uses of the upper-
storey species for poles and timber are pending, as farmers have not yet harvested
them.
120 S. Franzel et al.

Among the shrub species, calliandra outperformed the leucaenas on all criteria
except compatibility with crops. Thirty-nine per cent of the farmers using calliandra
to improve soil fertility found that it was effective in this respect. This result is simi-
lar to the findings from the researcher-designed, farmer-managed hedgerow inter-
cropping trial in the same area, which showed that about 50% of the farmers
reported positive effects (see Chapter 5).
Thirdly, comparisons between on-farm and on-station trials showed that
growth rates of both upper-storey species were much higher in the on-station trials,
probably reflecting higher soil fertility and more effective weed and termite control
measures. Even though casuarina outperformed grevillea in on-station trials, the
reverse was true on farms. In the on-farm trials, casuarina was more susceptible to
stress factors (drought, and possibly lack of inoculant in the soil) and pests (termites
and browsing) and had lower survival rates and slower growth than grevillea. The
results demonstrate the importance of assessing the performance of species under
farmer management and farm conditions; rankings of species on growth in on-sta-
tion trials will not necessarily hold under farmers’ conditions. They also suggest that
on-farm research is needed to assess the use of fertilizer in growing high-value trees
to reduce the time to harvesting the products.
Concerning farmer recommendations, the trial is on-going – the upper-storey
species have not yet been harvested. Moreover, only one provenance of each of the
species was used in this trial; there is high variation among provenances of a species
and thus recommendations concerning one provenance may not be relevant for
others (Simons et al., 1994). On the basis of the findings, and those of other on-sta-
tion and on-farm trials, several interim recommendations can be made.
Calliandra can be recommended for distribution as a fodder tree, based on this
trial and results presented in Chapter 7 and in van der Veen (1993). But many live-
stock do not take to eating it on the first try; thus, care has to be taken to introduce
it gradually into the diet, by mixing it with other feeds. The other two shrub species
should not be recommended – Leucaena leucocephala because of its susceptibility to
psyllid damage, and Leucaena diversifolia because of low leafy biomass production.
None of the shrub species should be recommended for improving soil fertility as
evidence on their effectiveness is lacking, even when the shrubs are planted at the
recommended density (see Chapter 5). Moreover, some of the farmers who claimed
that the trees were improving their soil fertility may have been doing so because
they believed such a response would please the researchers.
Casuarina may also be recommended, based on its preference as an ornamental
and its use for construction poles in eastern Kenya (Jama et al., 1989). However,
farmers need to be warned about the slower growth of this species than of grevillea
and eucalyptus, its low survival rate and its susceptibility to termite damage, brows-
ing and moisture stress. Grevillea may be recommended, based on its widespread
use in central Kenya. Research there has shown that grevillea trees planted along
boundaries offer net benefits from fuelwood and timber that are as high as the
maize they displace. In addition, they provide other difficult-to-quantify benefits,
such as boundary marking, windbreak, erosion control and improved microclimate
(Tyndall, 1996).
 Farmer-designed Agroforestry Trials 121

Researchers need to continue monitoring the trial, especially with respect to

end uses of casuarina, and farmers’ and neighbours’ experience with, and interest in,
expanding production of the test species.

Conclusion

Farmer-designed trials are an effective means for finding out how farmers manage
trees, for determining their planting criteria and preferences, for assessing tree per-
formance under farmers’ conditions and for helping to decide which species should
be recommended to farmers. Because few biophysical data are collected, the cost of
conducting such trials is relatively low. Also, most of the data on farmer assessment
may be taken during slack periods, such as following the crop harvest. Farmer-
designed trials are thus a relatively low-cost method to help increase farmer partici-
pation in species introduction and agroforestry tree research, enhancing the
effectiveness of research systems to meet farmers’ needs and improve their welfare.
The trial reported on in this chapter also provided important feedback on
researchable problems. For example, the poor performance of upper-storey species
on farms, relative to on-station trials, has led researchers to implement on-farm trials
to assess the growth response of trees to phosphorus fertilizer (Niang et al., 1996).
Farmer-designed trials are also an important tool for extension programmes to
help farmers and extensionists decide which species should be promoted in a specific
area. In 1994, farmer-designed trials were formally institutionalized in Kenya’s Forestry
Master Plan and the Forestry Extension Services Division was given the mandate to
conduct them (Franzel et al., 1996). Farmer-designed trials, as described in this chap-
ter, in central Kenya (Holding et al., 1995) and in Eastern Province (O’Neill et al.,
1994) are giving farmers major responsibility in screening new species for extension
programmes. But two factors impede the technology development and dissemination
process. First, linkages between research and extension are weak; information exchange
on appropriate technology and farmers’ problems is scant. Secondly, germplasm is
lacking; farmers cannot get seed for 5–10 years from the grevillea and casuarina trees
that they planted, and there are no other sources of seed. As discussed in Chapter 9,
researchers, extensionists and non-governmental organizations are establishing an
adaptive research and dissemination network to facilitate linkages between research
and extension and to ensure that germplasm and information are made available to
farmers. This will greatly strengthen linkages and improve the effectiveness of the tech-
nology development and dissemination process for agroforestry in western Kenya.

Acknowledgements

The authors are grateful to the following for reviewing earlier versions of this chap-
ter: Elias Ayuk, Peter Cooper, Francis Esegu, Roger Leakey, Frank Place, Anthony
Simons and Rob Swinkels. Crispus Kamanga is gratefully acknowledged for biomet-
rics assistance.
122 S. Franzel et al.

References
Bekele, T. (1997) Adoption of hedgerow inter- trees – linking research and extension.
cropping and two upper-storey tree species Agroforestry Today 8(3), 19–21.
in western Kenya. Ethiopian MSc in Haverkort, B., van der Kamp, J. and Waters-
Forestry Programme Thesis Works. Report Bayer, A. (1991) Joining Farmers’
No. 1996:1 Faculty of Forestry. Swedish Experiments. Experiences in Participatory
University of Agricultural Sciences, Technology Development. Intermediate
Skinnskatteberg, Sweden. Technology Publications, London.
Biggs, S. (1989) Resource-poor Farmer Partici- Holding, C., Niemi, T. and Weru, S.M. (1995)
pation in Research. International Service for On-farm Diagnostic Trials. Gilgil and
National Agricultural Research, The Hague, Kiangop Divisions. Technical Report No. 12.
Netherlands. Nakuru and Nyandarua Intensified Forestry
Bradley, P. (1991) Woodfuel, Women, and Extension Project, Nairobi.
Woodlots, Vol. 1, The Foundation of a Jama, B., Macklin, B., Rashid, K. and Getahun,
Woodfuel Development Strategy for East A. (1989) Intercropping Experiment with
Africa. Macmillan, London. Casuarina equisetifolia in Kenya. Nitrogen
Bruin, E., Ndufa, J.K. and Franzel, S. (1994) Fixing Tree Research Report, 7. Nitrogen
Farmers’ evaluation of trees in the food Fixing Tree Association, Honolulu, Hawaii.
crop-based land use system of Western Minae, S. and Akyeampong, E. (eds) (1988)
Kenya. In: Atta-Krah, K. (ed.) Agroforestry Agroforestry Potentials for the Land Use
in the Highlands of Eastern and Central Systems in the Bimodal Highlands of Eastern
Africa. Proceedings of an International Africa, Kenya. AFRENA Report No. 3,
Symposium, 6–10 September 1993, Kabale, International Centre for Research in
Uganda. International Centre for Research Agroforestry, Nairobi, Kenya.
in Agroforestry, Nairobi, pp. 28–32. Muturi, W.M. and Franzel, S. (1992) Availability
Collett, D. (1991) Modelling Binary Data. and use of wood products in the high
Chapman & Hall, London. potential zones of Western Kenya. In:
David, S. and Swinkels, R.A. (1994) Socio- Hoekstra, D. and Beniest, J. (eds) East and
economic Characteristics of Households Central African Agroforestry Research
Engaged in Agroforestry Technology Testing in Network, Proceedings of an International
Western Kenya. AFRENA Report No. 78. Symposium, 22–26 June, Kigale, Rwanda.
International Centre for Research in AFRENA Report, 58. International Centre
Agroforestry, Nairobi. for Research in Agroforestry, Nairobi,
Fransella, F. and Bannister, D. (1977) A Manual Kenya, pp. 48–50.
for Repertory Grid Technique. Academic Niang, A.I., Amadalo, B.A., Gathumbi, S.M.,
Press, New York. Otieno, J.H.O., Obonyo, C.O. and Obonyo,
Franzel, S. (2001) Use of an indigenous board E. (1996) KEFRI/KARI/ICRAF/AFRENA-
game, ‘bao’, for assessing farmers’ prefer- Maseno Project Progress Report September
ences among alternative agricultural tech- 1995–September 1996. AFRENA Report
nologies. In: Peters, G. and Pingali, P. (eds) No. 110. International Centre for Research
Tomorrow’s Agriculture: Incentives, in Agroforestry, Nairobi, Kenya.
Institutions, Infrastructure, and Innovations. O’Neill, M., Murithi, F., Nyaata, Z., Tuwei, P.
Proceedings of the 2000 International and Gachanja, S. (1994) National
Agricultural Economics Association Agricultural Research Project, Kenya
Conference, Ashgate, UK. Agricultural Research Institute. Regional
Franzel, S., Hitimana, L. and Akyeampong, E. Research Centre, Embu, Annual Report:
(1995) Farmer participation in on-station March 1993–March 1994. AFRENA
tree species selection for agroforestry: a case Report No. 81. International Centre for
study from Burundi. Experimental Research in Agroforestry, Nairobi, Kenya.
Agriculture 31, 27–38. Simons, A.J., MacQueen, D.J. and Stewart, J.L.
Franzel, S., Holding, C., Ndufa, J.K., Obonyo, (1994) Strategic concepts in the domestica-
O.C. and Weru, S.M. (1996) Farmers and tion of non-industrial trees. In: Leakey,
 Farmer-designed Agroforestry Trials 123

R.R.B. and Newton, A.C. (eds) Tropical System of Kenya. AFRENA Report No. 109.
Trees: The Potential for Domestication and International Centre for Research in
the Rebuilding of Forest Resource. Agroforestry, Nairobi, Kenya.
Proceedings of an International van der Veen, W. (1993) An economic analysis
Symposium, 23–28 August 1992, on farms of alley farming with fodder trees
Edinburgh, UK. Institute of Terrestrial on dairy cows in western Kenya. MSc the-
Ecology, HMSO, London. sis, Department of Development
Tyndall, B. (1996) The Socioeconomics of Economics, Wageningen Agricultural
Grevillea robusta within the Coffee Land-use University, Wageningen, Netherlands.
 Calliandra calothyrsus:
Assessing the Early
Stages of Adoption of a
7
Fodder Shrub in the Highlands
of Central Kenya
S. FRANZEL,1 H.K. ARIMI2 AND F.M. MURITHI3
1International Centre for Research in Agroforestry, PO Box 30677,
Gigiri, Nairobi, Kenya; 2National Agricultural Research Centre,
Muguga, PO Box 30148, Nairobi, Kenya; 3Kenya Agricultural
Research Institute, PO Box 58137, Nairobi, Kenya

Summary

The uptake of Calliandra calothyrsus as a fodder shrub by small-scale dairy farmers was
assessed several years after the shrubs were introduced to farmers in on-farm trials. There
was strong evidence that farmers were adopting the shrub. A random sample of 45
farmers had increased their average number of shrubs from 84 in their first plantings in
1991–1992 to 311 after 6–7 years. Moreover, farmer-to-farmer dissemination appeared
to be high, as 47% had harvested seed and 70% of these had given or sold seed or
seedlings to other farmers. The net benefits of using 6 kg of fresh calliandra leaves per
day as a substitute for 2 kg purchased dairy meal or as a supplement to farmers’ base
feeding regime amounted to about US$130 per cow year−1. By 2000, several thousand
farmers in central Kenya were feeding calliandra to their dairy animals. Potential benefits
from adopting calliandra or similar fodder shrub species in Kenya’s smallholder dairy
sector amounted to about US$139,000,000 year−1. Several measures were proposed to
help realize this potential: facilitating on-farm research and dissemination of information
and planting material, research to identify new fodder shrub species, and assessing the
constraints and incentives affecting fodder shrub adoption.

Introduction

The low quality and quantity of feed resources is the greatest constraint to improv-
ing the productivity of the livestock sector in sub-Saharan Africa (Winrock

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 125
126 S. Franzel et al.

International, 1992). With milk and meat imports projected to increase dramati-
cally, there is increasing interest in leguminous fodder shrubs to help solve the feed
problem, for three reasons. First, the uptake of herbaceous legumes in African
smallholder systems has been disappointing (Thomas and Sumberg, 1995), encour-
aging interest in other protein sources. Secondly, the growing popularity of agro-
forestry has helped promote research and extension activities on integrating fodder
shrubs into farming systems. Thirdly, there is widespread evidence that shrub
legumes are an important component of indigenous livestock production systems
throughout the world (Gutteridge and Shelton, 1993) – cultivation of fodder
shrubs can thus be viewed as a modification of existing systems rather than a new,
and thus more difficult to introduce, practice.
Despite the strong interest, a literature review did not reveal a single case of
widespread adoption of introduced, managed fodder shrubs by small-scale farmers in
Africa. Several studies discuss on-farm research on fodder shrubs but give little or no
indication as to whether farmers adopted and expanded their use of the technology
following the trials (Reynolds et al., 1991; Jabbar et al., 1996). This chapter exam-
ines the adoption of Calliandra calothyrsus, a leguminous fodder shrub, among small-
holder dairy producers in the highlands of central Kenya. The objectives are to:
● examine the expansion of calliandra plantings by farmers after initial testing,
and determine which farm and household characteristics are associated with
expansion;
● document farmers’ experiences and assessments in testing, managing and using
calliandra;
● determine the economic impact of calliandra from the farmers’ perspective;
● provide feedback and recommendations to research, extension and policy mak-
ers for improving and promoting calliandra and fodder shrub production in
the Kenyan highlands.
First, the study area is described. Then research on the management of calliandra
as a fodder shrub and as a feed to dairy cows is summarized. Next, the survey meth-
ods used in the research are described, followed by the results and the conclusions.

Description of the Study Area

The coffee-based land-use system of central Kenya, ranging in altitude from 1300
m to 1800 m, is located on the slopes of Mt Kenya (Fig. 7.1). Rainfall occurs in
two seasons, March–June and October–December, and averages 1200–1500 mm
annually. Soils, primarily Nitosols, are deep and of moderate to high fertility.
Population density is high, ranging from 450 to 700 persons km−2. This study was
conducted in the portion of Embu District that falls within the coffee-based system.
Farm size averages 1.9 ha (SD = 1.5), and 39% of farmers have plots away from their
homesteads. Most farmers have titles to their land, thus tenure is relatively secure.
About 18% of all households are female-headed (Minae et al., 1988; Kimenye,
1998; Murithi, 1998).
 Assessment of C. calothyrsus Adoption 127

Nairobi

Fig. 7.1. Study area in central Kenya.

The main crops are coffee, produced for cash, and maize and beans, produced
for food. Most farmers also grow napier grass (Pennisetum purpureum) for feeding
their dairy cows, and crop their fields continuously because of the shortage of land.
About 80% have improved dairy cows, 1.7 cows per family, kept in zero- or mini-
mum-grazing systems. Milk yields average about 8 kg per cow day−1 and produc-
tion is for both home consumption and sale. Forty per cent of the farmers also have
goats, averaging 3.2 per family (Minae et al., 1988; Murithi, 1998). Dairy goats are
a rapidly growing enterprise and are particularly suited for poorer households.
128 S. Franzel et al.

The main feed source for dairy cows is napier grass (P. purpureum), supple-
mented during the dry season with crop residues, such as maize and bean stover,
banana leaves and pseudostems, and indigenous fodder shrubs. Herbaceous legumes
have not been widely adopted, in part because of farmers’ inability to establish them
in mixed stands with napier grass (Paterson et al., 1996b). Forty-five per cent of the
farmers buy commercial dairy meal (nominally 16% crude protein) to supplement
their cows’ diet (Murithi, 1998). Whereas the extension service recommends 4 kg
dairy meal be fed per day, depending on the level of milk production, farmers’ feed-
ing rates are considerably lower. Farmers complain that the price ratio between
dairy meal and milk is not favourable, that they lack cash for buying dairy meal,
that its nutritive value is suspect and highly variable, and that it is difficult for them
to transport dairy meal from the market to the homestead (Patterson et al., 1996a).

Research on Calliandra Management

Research on fodder shrubs began in Kenya in the 1980s by the International

Livestock Research Institute (ILRI) and the Kenya Agricultural Research Institute
(KARI). The first on-farm trials in the Embu area were initiated by the National
Agroforestry Research Project (NAFRP), a project jointly managed by KARI, the
Kenya Forestry Research Institute and the International Centre for Research in
Agroforestry (ICRAF). In 1991, NAFRP scientists tested three promising species –
calliandra, Sesbania sesban and Leucaena leucocephala – to find out which niches
farmers preferred for planting the shrubs. Because of the limited size of farms, farm-
ers and researchers focused on integrating the shrub into the existing cropping sys-
tem rather than planting the tree in pure-stand fodder banks. In type 3
(farmer-designed, farmer-managed trials, see Chapter 2) farmers planted the shrubs
as they wished in niches of their choosing, including:
1. As hedges around the farm compound. Hedges are a common feature of home-
steads in central Kenya, and were traditionally planted to relatively unproductive,
non-browse species, to prevent free-ranging livestock from eliminating them. But
livestock are now confined and there is great potential for replacing unproductive
hedges with fodder hedges (Thijssen et al., 1993).
2. Along contour bunds and terrace edges on sloping land. They thus help con-
serve soil and, if pruned continuously, have little effect on neighbouring crops.
3. Intercropped with napier grass. Results from intercropping experiments suggest
that introducing calliandra into napier grass has little effect on napier yields (Nyaata
et al., 1998).
4. Between upper-storey trees, which are commonly planted along boundaries. The
growth of fodder trees is hardly affected by taller species, such as Grevillea robusta,
planted in the same line (NAFRP, 1993).
Pruning management has also been examined. The shrubs are first pruned for
fodder 9–12 months after planting, and pruning continues at the rate of four or five
times per year (Roothaert et al., 1998). Leafy biomass yields per year rise as pruning
 Assessment of C. calothyrsus Adoption 129

frequency decreases and cutting height increases. But adjacent crop yields are nega-
tively affected (ICRAF, 1992). The most productive compromise is probably in the
range of 4–6 prunings per year at 0.6–1 m cutting height, which would yield
roughly 1.5 kg dry matter per tree year−1 planted at 2 trees m−1 in hedges under
farmers’ conditions. Thus a farmer would need about 500 trees to feed a cow
throughout the year at a rate of 2 kg dry matter day−1, providing about 0.6 kg crude
protein. A typical farm of 1.5 ha would have available about 500 m of perimeter
and several hundred metres in each of three other niches: along terrace edges or
bunds, along internal field and homestead boundaries, and in napier grass plots.
Only 250 m of hedges would be needed to accommodate 500 trees, in order to feed
a dairy cow (Paterson et al., 1996a,c).
On-farm feeding trials have confirmed the effectiveness of calliandra as both a
supplement to the basal diet and as a substitute for dairy meal. One kilogram of dry
calliandra (24% crude protein and digestibility of 60% when fed fresh) has about
the same amount of digestible protein as 1 kg dairy meal (16% crude protein and
80% digestibility) (Paterson et al., 1996b); each increases milk production by about
0.75 kg (from 10.0 to 10.75 kg day−1) under farm conditions, but the response is
variable, depending on such factors as the health of the cow and the quantity and
quality of the basal feed (Paterson et al., 1996c). The effects of calliandra and dairy
meal were found to be additive, suggesting that the two feeds are nutritionally inter-
changeable (Paterson et al., 1996c). Unfortunately, data are not available for con-
structing a response curve to show the effect of varying quantities of calliandra on
milk production. Calliandra was also found to increase the milk production of dairy
goats (Kiruiro et al., 1999).
Calliandra seedlings are raised in nurseries and transplanted following the onset
of the rains. Experiments on seedling production have confirmed that plants may
be grown in raised seedbeds rather than by the more expensive, labourious method
of planting in polythene bags (O’Neill et al., 1997). Researchers are also conducting
studies on other shrubs species, exotic and indigenous (Roothaert et al., 1997), to
help farmers further diversify their feed sources. These species include Leucaena
trichandra, Morus alba (mulberry) and Sapium ellipticum.

Methods

Informal, semi-structured interviews were conducted in April 1995, with six farm-
ers who had tested calliandra, in order to develop an understanding of the issues
involved in adoption and to formulate hypotheses. In order to draw up a sample
frame for the formal survey, a list of farmers who had received calliandra seedlings
during, or previous to, 1993 was assembled. The year 1993 was selected as the cut-
off year in order to ensure that sample farmers had at least 2 years’ experience of
growing, and thus 1 year of feeding, calliandra. The list included:
● All of the 64 farmers who had participated in the two on-farm calliandra trials
of the National Agroforestry Research Project (NAFRP) up to 1993, including
130 S. Franzel et al.

those who had dropped out of the trials. These farmers had been selected for
the trials on the basis of their interest in testing calliandra.
● Nineteen farmers who had received seedlings from the National Dairy
Development Project (NDDP). Unfortunately, a complete list of the farmers
who received seedlings was not available, nor were the criteria for their selec-
tion known. It is likely that the list drawn up by extension staff for this survey
was biased towards farmers who were adopters.
Assessing adoption among farmers who participated in on-farm trials and spe-
cial projects is sometimes suspect, as extensive contact and incentives may bias the
farmers in favour of the technology being assessed. In this particular case, we feel
that such concerns were negligible. None of the farmers in either group received any
incentives apart from free seed and seedlings. All received some advice about callian-
dra but, as the findings show, lack of information about calliandra was an important
problem (one farmer did not know that calliandra could be fed to livestock!).
Monitoring and contact with research and extension varied; about half of the farm-
ers had completed their trials by 1993 and afterwards had little or no contact with
researchers. About one-third participated in type 2 (researcher-designed, farmer-
managed) on-farm trials during 1994 and 1995, involving feeding and pruning, and
thus had repeated contact with researchers. The NDDP ended in 1994, so the
NDDP farmers had had no contact with the project for a year or longer.
A random sample of 45 farmers from the lists was interviewed twice, in 1995
and in 1998, using a structured questionnaire. The authors conducted all interviews
themselves. In addition, three farmers who had obtained seed from interviewed
farmers were visited, to find out their experience in testing calliandra.
For the economic analysis, partial budgets were drawn up to show the effects on
net income under two scenarios: (i) using calliandra as a supplement to the normal
diet; and (ii) as a substitute for purchased dairy meal. The base analysis assumed a
1.5 ha farm with 500 shrubs and one zero-grazed dairy cow and covers a 5-year
period. The benefits included in the analysis were the effect of calliandra on milk
production (in the supplementation case) and the cash saved by not purchasing dairy
meal and interest on cash freed up (in the substitution case). Costs were those of the
seedlings and labour for planting, cutting and feeding calliandra. Average prices over
a 3-year period, 1996–1998, were used for dairy meal, milk and labour. Returns are
expressed in 1998 US dollars, adjusted for inflation. Coefficients, prices and sources
of data used in the economic analysis are shown in Table 7.1.

Results

Farm and household characteristics

Most (91%) of the households were male-headed, while four (9%) were female-
headed. Of the latter group, in three cases the husband was living away and in one,
the woman was not married. Household heads were generally middle aged; 48%
 Assessment of C. calothyrsus Adoption 131

Table 7.1. Coefficients and prices used in the economic analysis.

Items Values Data sources
Coefficients
Period of analysis 5 years Assumption
Lactation period 300 days Assumption
Days fed calliandra 365 days Assumption
Days fed dairy meal 365 days Assumption
Calliandra quantity fed 6 kg fresh (equivalent Assumption
per cow day−1 to 2 kg dry)
Dairy meal quantity fed 2 kg Assumption
per cow day−1
Milk output day−1 from 1 kg 0.75 kg Paterson et al. (1996c)
dry calliandra
Milk ouput day−1 from 1 kg 0.75 kg Paterson et al. (1996c)
dairy meal
Calliandra leafy biomass 0 kg Farmers’ experience
yield per shrub in year 1
Calliandra leafy biomass 1.5 kg (dry) Paterson et al. (1996c)
yield per shrub year−1,
years 2–5
Shrubs required to feed 1 487 Computed from above
cow year−1
Shrub survival rate 80% Survey data
Calliandra planting labour 20 shrubs h−1 Farmers
Calliandra cutting and 15 min day−1 Farmers
feeding labour
Discount rate 20% Assumption
Interest on capital freed up Capital tied up for an
by using calliandra instead average of 2 weeks,
of purchasing dairy meal 20% annual interest
rate
Prices
Dairy meal US$0.201 kg−1 Market survey, 1996–1998
Transport of dairy meal US$0.008 kg−1 Market survey, 1996–1998
Seedling cost (bare-rooted) US$0.526 per 100 R. Swinkels (unpublished
seedlings data from on-farm trial)
Labour cost US$0.118 h−1 Farmers, 1996–1998
Milk price (farm gate) US$0.296 kg−1 Farmers, 1996–1998
Annualized value of fixed cost US$2.18 Use of capital recovery
for seedling establishment for formulaa (Spencer et al.,
500 shrubs 1979)
US$1 = 59 Kenya shillings Average exchange rate,
1996–1998
aK=(rv)/(1  (1=r)−n ) where K is the annual service user cost, v is the original

(acquisition) cost of the fixed capital asset, r is the discount rate and n is the
expected life of the asset. This procedure allows both the depreciation on capital
and the opportunity cost of capital to be costed out.
132 S. Franzel et al.

were between 30 and 55 years old, with 36% older and 17% younger. Eighty-nine
per cent of the females and 77% of the males work full time on the farm. Over half
(55%) were judged to be high-income farmers, 38% were of middle income and
7%, low income.
Farm size averaged 2.3 ha (SD = 1.6; median = 1.6). Of the total, 1.8 ha was
situated around the homestead and 0.5 ha at some distance away. Main enterprises,
as ranked by the farmers themselves, were dairy, coffee, tea and maize. Ninety-six
per cent had dairy cows, averaging 1.9 cows per farm. Sixty per cent had heifers,
47% had bulls or steers and 53% had calves. Goats were owned by 58%, averaging
2.3 per farm, and 27% owned sheep. Eighty-five per cent owned poultry and 68%
had rabbits. Overall, the sample farmers appeared to have somewhat higher incomes
and were more oriented towards dairy farming than average farmers in their area.
Their farms were about 20% larger than average farms in the area.

Establishment and expansion of calliandra plantings

Farmers’ first plantings of calliandra averaged 90 shrubs, of which 84 shrubs sur-

vived. The high survival rate, 93% (SD = 13%), was consistent with data collected
in farmer-managed trials in the same area and including some of the same farmers
(NAFRP, 1993). Two-thirds were participants in NAFRP on-farm trials, 20%
obtained planting material from NDDP, and 13% from other sources. Four-fifths
of the farmers used potted seedlings to establish their first calliandra plantings, 16%
established their own nursery, and 4% direct seeded. First plantings occurred
between 1988 and 1993, with over half taking place in 1992 (Fig. 7.2).
Over four-fifths of the farmers expanded their calliandra plantings after their
first planting (Table 7.2). Over one-third expanded twice and 18%, three or four
times. As farmers expanded, the number of shrubs planted1 per expansion appeared
to increase, although because of the high variability, the differences were not statisti-
cally significant (Table 7.2). In farmers’ fourth and fifth plantings, the average num-
ber of shrubs planted was 54% higher than in their first plantings.

Table 7.2. Expansion of calliandra plantings and numbers of shrubs per plantings.
Planting No. of farmers Average no. shrubs per planting (SD)
Initial planting 45 (100%) 84 (65)
1st expansion 37 (82 %) 85 (54)
2nd expansion 16 (36%) 97 (99)
3rd and 4th expansions 8 (18%) 129 (143)

1 The number of shrubs that survived is used instead of the number that were planted,
because in many cases the farmers could not remember how many they had planted,
whereas we were able to count the numbers of surviving shrubs. As survival rates averaged
93% for the 31 cases where data were available, the differences between numbers planted
and numbers survived was probably small.
 Assessment of C. calothyrsus Adoption 133

350
311
300

250
217
198
200
Shrubs

149
150

50
100

50 17
7 7
0
88/89 90 91 92 93 94 95 96 97 98

Initial planting Expansion

Fig. 7.2. Farmer expansion following testing: numbers of calliandra shrubs planted
by 45 surveyed farmers who first planted before 1993.

By 1995, the average number of shrubs per farmer had increased from 84 (SD =
65) in their first planting to 218 (SD = 225; median = 166), an increase of over 2.5
times. The rate of increase slowed somewhat over the next 3 years; by 1998 farmers
averaged 311 shrubs (Ondieki, 1999).
There were important differences in the method of planting and source of
planting material between farmers’ successive plantings (Table 7.3). Whereas the
principal method in farmers’ first and second plantings was to use potted seedlings
obtained from projects, the most important method in the third and subsequent
plantings was to establish a nursery. Similarly, farmers’ own farms and other sources
(e.g. shrubs from friends and relatives) replaced projects as the principal source of
planting material, beginning in the third planting. Of the 65 incidents of expan-

Table 7.3. Planting method and source of planting material for successive plant-
ings of calliandra.
Source of planting material
Method (% of plantings) (% of plantings)
Direct Potted Own
Planting seed Nursery seedlings Project shrubs Other
1st 4 16 80 87 0 13
2nd 0 11 89 92 5 3
3rd 17 44 39 31 42 26
4th and 5th 0 60 40 44 44 11
134 S. Franzel et al.

sion, 33 involved planting seeds or seedlings obtained from projects, and 32, seeds
or seedlings obtained from farmers’ own farms or from other persons. By mid-1995,
36% of the farmers had established calliandra nurseries. Three-quarters of these
used seed from their own shrubs.
The niches where farmers planted calliandra were sometimes determined by
farmers and sometimes by researchers, as when an on-farm trial concerned a partic-
ular niche. Overall, the most common niches were in lines on contours (62% of
farms), intercropped with food crops or coffee (40%), and on homestead bound-
aries (35%) (Table 7.4). In plantings where farmers chose the niche, the most com-
mon choices were homestead boundaries, external boundaries and in lines on
contours. Only two farmers planted calliandra in pure-stand fodder banks, reflect-
ing farmers’ reluctance to allocate even small plots to calliandra.
There did not appear to be much association between uptake of calliandra and
selected farm and household characteristics, but the assessment was constrained by
the small size of the sample. Defining an adopter as a farmer who had expanded at
least once and had more than 100 shrubs, 73% of the sample could be termed
adopters. There was no association between adoption and farm size, wealth, size of
farm adjacent to the homestead, or number of cows. There was a tendency for
adoption to be associated with age; six of seven farmers under 30 adopted, whereas
only 9 of 15 over 55 did so. The dairy enterprise’s rank in importance among other
enterprises was significantly associated with adoption at the P < 0.10 level (chi-
square test); the higher the rank of dairy, the more likely farmers were to adopt.

Management and uses of calliandra

Pruning methods were quite variable. The most common method was to periodi-
cally cut calliandra when it reached a height of about 1.0–1.4 m (before it becomes

Table 7.4. Niches where farmers planted calliandra.

All plantings Plantings where farmer chose
(numbers and niche (numbers and
percentages of percentages of farmers
Niche sample farmers)a who chose niches) (n = 35)
In lines on contours 28 (62%) 8 (23%)
Intercropped with food crops 18 (40%) 6 (17%)
or coffee
Homestead boundary 16 (35%) 11 (31%)
Intercropped in napier 15 (33%) 2 (6%)
grass plots
External boundary 11 (24%) 10 (28%)
Internal boundary 9 (20%) 6 (17%)
aPercentages do not sum to 100 because farmers often plant in more than one

niche. In some of the on-farm trials, farmers were asked to plant in a particular
niche.
 Assessment of C. calothyrsus Adoption 135

too difficult to reach and shades neighbouring crops too much), reducing the height
to about 0.5–1 m. About 80% used pruning shears, which they already owned for
use on their coffee and tea. Thirteen per cent use a machete, claiming that the stem
was too thick for using shears. Nine per cent broke off branches by hand, primarily
in order to save time. Pruning shears are recommended because they make a cleaner
cut, thus promoting regrowth and preventing disease and damage to the shrub.
Farmers fed calliandra to a wide range of types of animals. Ninety-one per cent
fed it to dairy cows, 47% to goats and 42% to heifers. Between 5% and 20% fed to
each of the following: bulls, sheep, rabbits, calves and poultry. Sixty-nine per cent
fed dry cows as well as lactating ones. This was often because a dry cow and a lactat-
ing cow fed from the same trough and it was not practical to separate their rations.
Nearly all farmers chopped calliandra before feeding, as recommended, as opposed
to giving the branches to the cows to strip off the leaves. Over 90% mixed calliandra
with napier grass when feeding, about 44% also fed calliandra separately at times.
Like dairy meal, calliandra is often fed during milking to help keep the cow still.
Only one farmer claimed to have fed calliandra to his cows throughout the
year. On average, farmers fed their cows calliandra about one-third of the time,
because the quantities they had were not sufficient and shrub growth was low dur-
ing the dry season. Only 12% reduced cutting during the wet season in order to
have increased supplies during the dry season. Farmers were sceptical about this
strategy, because not cutting calliandra would increase its competition with crops.
Three-quarters of the farmers fed their animals within an hour after cutting, in line
with the recommendation to feed only fresh leaves (Roothaert et al., 1998). This
recommendation has since been changed; recent research shows that calliandra can
be fed fresh or dry (Stewart, 2000).
Eighty-four per cent fed dairy meal to their cows, although many said that
because of cash shortages, they did not feed continuously. Most (62%) used callian-
dra as a supplement to dairy meal, that is, they did not reduce their use of dairy
meal when they fed calliandra. On the other hand, 27% used calliandra as a com-
plete substitute for dairy meal, and 10% as a partial substitute. Eighty-eight per
cent claimed that calliandra increased their milk production and 89% claimed that
their cows found it highly palatable.
Some farmers claimed that they obtained benefits from calliandra in addition
to increases in milk production. In response to an open question, 24% per cent said
that fuelwood production was a benefit, 13% cited soil conservation, and 7% each
cited increases in milk quality, calliandra’s appearance and money saved by not hav-
ing to buy dairy meal. The only negative aspects cited were scales (18%), a minor
pest that does not appear to reduce productivity very much, and that calliandra
reduces the yield of adjacent crops (7%).
The farmers varied considerably in the way they used the seed produced by
their calliandra shrubs. Forty per cent harvested seed; those that did not cited their
lack of interest or lack of knowledge about propagation techniques. One-third of
the farmers gave seed to others; each gave to an average of 13 other farmers (this fig-
ure is skewed upwards because of two farmers who gave seeds to 110 farmers – the
median number of persons given seed was four). Two farmers sold seed or seedlings
136 S. Franzel et al.

to other farmers. Two-thirds of the farmers had left some shrubs to seed at the time
they were interviewed, indicating their strong interest in expanding calliandra pro-
duction or in distributing seed.
Interviews were conducted with three farmers who obtained seed from the
sample farmers. Two used the seed to establish nurseries in 1995. One had trans-
planted 67 shrubs from the nursery to his farm; the other was waiting for the
seedlings to reach sufficient size before transplanting. A third farmer purchased 500
seedlings in 1994 from a sample farmer and he planted them on internal and exter-
nal boundaries. He purchased an additional 100 seedlings in 1995. He also estab-
lished a nursery in late 1995, and planned to transplant seedlings to his farm in
1996. As these farmers were not randomly selected, they cannot be said to be repre-
sentative of farmers receiving seed from sample farms. But their experiences indi-
cated the high potential for dissemination beyond the original farmers.
Each person who harvested seed harvested from an average of only eight
shrubs. It is advisable to harvest from at least 30 shrubs in order to conserve the
genetic diversity of the germplasm. If the genetic base in an area is too narrow,
inbreeding causes a significant decline in productivity (Roothaert et al., 1998).
Farmers were unaware of the need to harvest from at least 30 shrubs. Some may be
reluctant to leave so many shrubs for seed because they become tall and reduce the
growth of adjacent crops.
Ninety-one per cent of the farmers indicated an interest in increasing their cal-
liandra plantings. When asked about constraints limiting expansion of calliandra,
19 (42%) claimed they lacked seedlings. Since some farmers had received seedlings
from projects two or more times, they clearly were hoping to receive seedlings again
so as not to have to develop their own nurseries. Three mentioned that they lacked
knowledge on propagation techniques.
Surprisingly, most of the farmers (58%) had never visited another farmer who
had planted calliandra. Nineteen per cent had visited other calliandra farmers as
part of extension tours and 34% had visited other calliandra farmers on their own.

Monitoring, farmer innovation and feedback

Informal monitoring takes place in which farmers and extension staff provide feed-
back to project staff and researchers on their progress and problems. In one case,
feedback on a farmer innovation has resulted in a change in extension recommenda-
tions. Farmers in Kandara Division, Maragua District, conducted experiments on
soaking calliandra seeds before planting and found that seeds soaked for 48–60 h
had higher germination rates than those soaked for the recommended 24 h.
Researchers at KARI-Embu confirmed the farmers’ findings and extension staff now
recommend the longer soaking time.
Farmers’ problems with pests and their innovations in controlling them have
also led to the design of new on-farm trials. For example, in 2001, researchers and
farmers are comparing the effectiveness of using netting and local measures (spray-
ing solutions made from tobacco, marigold, neem, hot pepper or Tephrosia vogelii)
 Assessment of C. calothyrsus Adoption 137

to control crickets, hoppers and aphids damaging seedlings in nurseries. These find-
ings demonstrate the importance of monitoring farmer innovations and feeding
them back to research and extension.

Economic analysis

Partial budgets for calliandra as a supplement to farmers’ basal feed and as a substi-
tute for dairy meal are shown in Tables 7.5 and 7.6. Shrub establishment costs
(including the costs of seedlings and planting) are modest, US$6.58 per 500
shrubs.2 Beginning in the second year, harvesting and feeding 2 kg dry calliandra
day−1 as a supplement throughout the lactation period increases milk production by
about 450 kg year−1, an increase of about 10% over base milk yields. Incremental
benefits per year after the first year are over 12 times higher than incremental costs.
Net benefits per cow year−1 after year 1 are US$120.11. Treating the establishment
costs as depreciation spread over the 5-year period, the annualized net benefit is
US$117.91 per cow year−1. The net present value (NPV), assuming a 20% discount
rate, is US$258.39.
In the partial budget assessing calliandra as a substitute for dairy meal, estab-
lishment, cutting and feeding costs are the same as in the preceding analysis. By
feeding calliandra, the farmer saves the money he would have spent buying and
transporting 730 kg dairy meal during the year. Incremental benefits per year after
the first year are over 14 times higher than incremental costs. Milk production does
not increase but net benefits are slightly higher than in the supplementation case.
The net benefits per cow year−1 after year 1 are US$141.68. The annualized net
benefit is US$139.48 per cow year−1. The NPV, assuming a 20% discount rate, is
US$300.15. Therefore, using calliandra increases farmers’ annual income by about

Table 7.5. Partial budget: extra costs and benefits of using calliandra as a sup-
plement for increasing milk production (US$).
Extra cost Extra benefits Net benefit
Year Item US$ Item US$ (US$)
1 Shrub seedlings 3.05 0
Planting labour 3.53
Subtotal 6.58 −6.58
2 Cutting/feeding labour 10.75 450 kg milk 133.07 122.32
Years 3–5 same as year 2
Net present value at 20% discount rate = US$258.39 year−1
Net benefit per year after year 1 = US$120.11
Annualized net benefit treating establishment costs as depreciation = US$117.91
Base farm model: the farm has 500 calliandra shrubs and one dairy cow. The cow
consumes a basal diet of 80 kg napier grass day−1 and produces 10 kg milk day−1.

2 US$1 = 59 Kenya shillings on average for the period 1996–1998.

138 S. Franzel et al.

Table 7.6. Partial budget: extra costs and benefits of using calliandra as a sub-
stitute for dairy meal in milk production (US$).
Extra cost Extra benefits Net benefit
Year Item US$ Item US$ (US$)
1 Shrub seedlings 3.05 0
Planting labour 3.53
Subtotal 6.58 −6.58
2 Cutting/feeding labour 10.75 Saved dairy 147.10
meal cost
Saved dairy 5.32
meal transport
Interest on 0.90
capital
freed up
Subtotal 10.75 Subtotal 152.43 141.68
Years 3–5 same as year 2
Net present value at 20% discount rate = US$300.15
Net benefit per year after year 1 = US$141.68
Annualized net benefit treating establishment costs as depreciation = US$139.48
Base farm model as in Table 7.5.

US$120–142 per cow year−1, depending on whether the farmer is supplementing or

substituting. As the average farmer owns 1.7 cows, calliandra increases farmers’
income by about US$204–241 year−1, representing an increase of about 10% in
total household income (Murithi, 1998).
Returns varied somewhat among the 3 years (1996–1998) for which economic
data were collected. For example, net benefits per cow year−1 after year 1 varied from
US$94–120 in the supplement scenario and US$107–151 in the substitute scenario.
The costs of establishing, maintaining and feeding calliandra are low. In both
the substitute and supplement scenarios, farmers recover their costs very quickly, in
the second year after planting. In order to break even, a farmer using calliandra as a
supplement needs to obtain only 0.08 kg of milk from 1.0 kg of calliandra (dry),
rather than the 0.75 kg milk kg−1 of calliandra obtained in on-farm trials and
assumed in the analysis.
Several intangible, or otherwise difficult to measure, benefits and costs have
been omitted from this analysis. Calliandra provides benefits to some farmers as
firewood, in erosion control, as a boundary marker, a fence and as an ornamental. It
also increases the butterfat content of milk, giving it a richer taste and creamier tex-
ture. When used as a supplement, calliandra may improve animal health and fertil-
ity and reduce the calving interval. Finally, several farmers noted that calliandra had
important benefits relative to dairy meal: it was available on the farm, cash was not
needed to obtain it and its nutritional content was more reliable than that of dairy
meal. These views support Haugerud’s thesis that farmers prefer enterprises and
practices that do not rely on uncertain governmental or market mechanisms
(Haugerud, 1984).
 Assessment of C. calothyrsus Adoption 139

The main costs not assessed are the opportunity cost of the land occupied by
the shrubs and the effect in reducing yields of adjacent crops. However, these are
likely to be relatively low, especially when calliandra replaces, or is added to, an exist-
ing hedge or bund, is pruned frequently, is planted between upper-storey trees, or
when calliandra hedges border on homesteads, roads, paths or external boundaries.
The sensitivity analysis suggests that the net benefits of using calliandra as a
supplement or as a substitute are fairly stable (Table 7.7). Despite the range of nega-
tive situations tested, net present values and net benefits remain positive. For exam-
ple, a 30% reduction in milk price reduces the NPV in the supplement scenario by
33%, but using calliandra would still be profitable. In the substitution scenario, a
reduction of the dairy meal price by 30% reduces the NPV by 32%. If one assumes
that 1 kg dry calliandra gives 0.5 kg milk instead of 0.75 kg milk, the NPV in the
supplement scenario declines by 37%.

Conclusions

This chapter presents considerable evidence that test farmers in the Embu area are
adopting calliandra as a fodder shrub and that farmer-to-farmer dissemination is
substantial. By 1997, it was estimated that roughly 1000 farmers had planted cal-
liandra. During 1999–2000, a project implemented through the Systemwide
Livestock Program of the Consultative Group on International Agricultural
Research helped an additional 3000 farmers to plant calliandra across seven districts
of central Kenya. This project, implemented by ICRAF, KARI and ILRI also intro-

Table 7.7. Sensitivity analysis showing the effect of changes in key parameters
on the profitability of using calliandra (US$ per cow year−1).
Calliandra as supplement Calliandra as substitute
Net present Annualized Net present Annualized
value net benefit value net benefit
Base analysis 258 118 300 139
Milk price +30% 344 158 300 139
Milk price −30% 172 78 300 139
Dairy meal price +30% 258 118 395 184
Dairy meal price −30% 258 118 205 95
Discount rate 30% 199 118 231 140
Labour cost +30% 250 114 292 136
2 kg dairy meal or 1 kg 354 162 300 139
dry calliandra gives
1 kg milk
2 kg dairy meal or 1 kg 163 73 300 139
dry calliandra gives
0.5 kg milk
Data based on Tables 7.5 and 7.6.
140 S. Franzel et al.

duced two other fodder shrubs to farmers, Leucaena trichandra and Morus alba
(mulberry), and one herbaceous legume, Desmodium intortum.
Calliandra appears to be appropriate for smallholder dairy farmers throughout
eastern Africa – it can grow at altitudes between sea level and 1900 m, requires a
minimum of 1000 mm rainfall, can withstand dry seasons up to 4 months long
and is suitable for cut-and-carry feeding systems or for grazing systems (Roothaert
et al., 1998). There are approximately 625,000 smallholder dairy farmers in Kenya
with improved cows; each has about 1.7 cows per farm (Omore et al., 1999).
Therefore, the potential benefits from adopting calliandra, as measured in this
chapter, amount to about US$139,000,000 year−1 in the Kenya smallholder dairy
sector alone.
Calliandra’s actual potential benefits are much higher. Calliandra also has
important potential in the large-scale dairy sector, which supplies 30% of Kenya’s
milk, and in the dairy goat sector, which is growing rapidly and is particularly suited
to resource-poor farmers. Moreover, the shrub is being used by dairy farmers at
numerous other sites in east and southern Africa, including Rwanda, Uganda,
Tanzania and Zimbabwe, and results are promising. But several important problems
remain; the proposed measures in the following section can help to solve these prob-
lems and enhance the benefits that calliandra can offer to African dairy farmers.

Issues for research and extension

Whereas the project has successfully expanded the use of fodder shrubs across seven
districts, it is still reaching only a small percentage of dairy farmers in these districts,
and less than 1% of Kenya’s smallholder dairy farmers. Like many agroforestry prac-
tices, fodder trees spread slowly on their own; considerable facilitation in terms of
training, information and germplasm is required. Further scaling up should focus
on institutions working in areas of the country where smallholder dairy farmers pre-
dominate. ICRAF, the Oxford Forestry Institute, the Regional Land Management
Unit and other partners are planning a project that will help the government exten-
sion services, NGOs, and farmer organizations throughout East Africa to assist
farmers to plant fodder trees.
Commercial seed production and distribution are slowly emerging in project
areas, but it is not clear if seed production will continue to grow and meet local
demand. Greater emphasis is needed on promoting community-based seed produc-
tion and distribution through a range of partners: farmer groups, individual seed
producers and private nurseries. Research is also needed to compare the effective-
ness and efficiency of different mechanisms for producing and distributing planting
material.
Four other on-farm research issues need to be addressed:
1. Greater diversification of fodder shrubs is needed to reduce the risk of pest and
disease attacks and improve feed quality. KARI-Embu has a strong on-farm research
programme for evaluating fodder trees and is increasing its emphasis on the testing
of indigenous species.
 Assessment of C. calothyrsus Adoption 141

2. Many farmers now have 5–10 years of experience feeding calliandra to their
cows; the knowledge that these farmers have gained on feeding calliandra needs to
be captured and shared.
3. More research is needed on the constraints and incentives affecting the adoption
of calliandra. For example, a survey of neighbours of calliandra adopters could be
useful for identifying the problems they have in testing and adopting calliandra.
4. Networks of farmer experimenters need to be established. During the survey
reported in this chapter, a considerable amount of farmer experimentation was
observed. For example, individual farmers were found to be experimenting on the
effect of different cutting heights, on substitutability between dairy meal and cal-
liandra, and on methods of establishment. Farmers and researchers could both gain
if farmers were able to organize themselves to conduct research on particular issues,
using farms as replicates. Farmer-experimenters working together on a topic could
meet periodically to assess progress and, at the end of the experiment, could draw
conclusions.
Finally, more research is needed for expanding smallholder dairy production.
Milk demand in Africa greatly outstrips locally available supplies. Dwindling feed
resources and high transaction costs for production and marketing limit productiv-
ity and smallholder participation (Winrock International, 1992; Staal et al., 1996).
Fodder shrubs help reduce costs and improve the productivity and profitability of
smallholder dairying. Even when dairy production is stagnant, fodder shrubs may
be attractive to farmers as substitutes for purchased concentrates. But expansion of
fodder shrubs will likely be greatest when dairy production is increasing.

Factors contributing to success and challenges

Several factors have contributed to the achievements thus far in developing and dis-
seminating fodder shrubs:
1. The demand among farmers for fodder shrubs was huge, mainly because the
shrubs save cash, farmers’ scarcest resource, and require only small amounts of land
and labour.
2. Initial on-farm trials were farmer-designed and farmer-managed, permitting
farmers to plant the trees in farm niches of their choice and to manage them as they
saw fit.
3. The project area is noted for the dynamism of its farmers, and access to markets
is fairly high, enhancing the adoption of new practices.
4. Because researchers and extensionists worked through partner organizations,
they were able to build on local organizational skills and knowledge and reach far
more farmers than would otherwise have been possible.
5. The strong partnership between researchers, extensionists and farmers in the
project facilitated the flow of information among the three.
142 S. Franzel et al.

Acknowledgements

The authors are grateful to the following for their advice and assistance: John Curry,
Peterson Mwangi, Paul Ongugo, Mick O’Neill, Rob Paterson, Ralph Roothaert,
Sara Scherr and Meredith Soule. They are also grateful to the following organiza-
tions that supported the research and development work reported in this chapter:
the Swedish International Development Cooperation Agency, the Systemwide
Livestock Programme of the CGIAR, and the Forestry Research Programme of the
United Kingdom Department for International Development.

References
Gutteridge, R.C. and Shelton, H.M. (1993) The Kaboneka, S. and Ghahamanyi, A. (1988)
scope and potential of tree legumes in agro- Agroforestry Research Project Proposal for the
forestry. Agroforestry Systems 23(2–3), Coffee Based System in the Bimodal
177–194. Highlands, Central and Eastern Provinces,
Haugerud, A. (1984) Household dynamics and Kenya. AFRENA Report No. 16.
rural political economy among Embu farm- International Centre for Research in
ers in the Kenya highlands. PhD disserta- Agroforestry, Nairobi, Kenya.
tion, Department of Anthropology, Murithi, F.M. (1998) Economic evaluation of the
Northwestern University, Chicago, Illinois. role of livestock in mixed smallholder farms
ICRAF (1992) Annual Report, 1991. of the central highlands of Kenya. PhD the-
International Centre for Research in sis, Department of Agriculture, University
Agroforestry, Nairobi. of Reading, UK.
Jabbar, M.A., Larbi, A. and Reynolds, L. (1996) NAFRP (1993) National Agroforestry Research
Alley Farming for Improving Small Ruminant Project. KARI Regional Research Centre,
Productivity in West Africa: ILRI’s Embu. Annual Report: March 1992–April
Experiences. Socioeconomics and Policy 1993. AFRENA Report No. 69.
Research Working Paper No. 20. International Centre for Research in
International Livestock Centre for Africa, Agroforestry, Nairobi.
Addis Ababa. Nyaata, O.Z., O’Neil, M.K. and Roothaert, R.L.
Kimenye, L. (1998) Assessment of technology (1998) Comparison of Leucaena leuco-
dissemination and utilization by women cephala with Calliandra calothyrsus in Napier
and men farmers. A case study of Embu and (Pennisetum purpureum) fodder banks. In:
Mbeere Districts. In: Curry, J., Kooijman, Shelton, H.M., Gutteridge, R.C., Mullen,
M. and Recke, H. (eds) Institutionalising B.F. and Bray, R.A. (eds) Leucaena-
Gender in Agricultural Research: Experiences Adaptation Quality and Farming Systems,
from Kenya. Proceedings of the Gender Proceedings of a workshop held in Hanoi,
Conference, Kenya Agricultural Research Vietnam, 9–14 February 1998. Australian
Institute 5–7 October 1998, Nairobi, pp. Centre for International Agricultural
201–204. Research, Canberra, pp. 257–260.
Kiruiro, E.M., Ouma, O. and Arimi, H.K. Omore, A., Muriuki, H., Kenyanjui, M., Owango,
(1999) The Potential for Improving Milk M. and Staal, S. (1999) The Kenyan Dairy
Production from Dual Purpose Goats by Using Sub-sector: a Rapid Appraisal. Smallholder
Calliandra calothyrsus on Smallholder Farms Ministry of Agriculture/Kenya Agricultural
of the Coffee/tea Land-use System of Embu Research Institute/International Livestock
District. Annual Report, 1999. Kenya Research Institute, Kenya.
Agricultural Research Insitute, Regional Ondieki, F. (1999) The adoption of Calliandra
Research Centre Embu, Kenya. calothyrsus in the Embu area, Kenya. MSc
Minae, S.E., Kaumi, S., Avila, M., Okorio, J., thesis, Oxford University, UK.
 Assessment of C. calothyrsus Adoption 143

O’Neill, M., Gachanja, S., Karanja, G.M., Roothaert, R., Karanja, G.M., Kariuki, I.R.,
Kariuki, I.W., Kanampiu, F., Kiruiro, E., Paterson, P., Tuwei, P., Kiruiro, E., Mugwe,
Murithi, F., Okoba, B., Nyaata, Z., J. and Franzel, S. (1998) Calliandra for
Mwangi, J., Tuwei, P., Roothaert, R. and Livestock. Technical Bulletin No. 1.
Paterson, R. (1997) National Agroforestry Regional Research Centre, Embu, Kenya.
Research Project, Kenya Agricultural Research Spencer, D.S.C., Byerlee, D. and Franzel S.
Institute, Regional Research Centre, Embu, (1979) Annual costs, returns, and seasonal
Annual Report: March 1995–March 1996, labour requirements for selected farm and
AFRENA Research Report No. 108. non-farm enterprises in rural Sierra Leone.
International Centre for Research in Working Paper No. 27. Department of
Agroforestry, Nairobi. Agricultural Economics, Michigan State
Paterson, R.T., Kiruiro, E. and Arimi, H.K. University, East Lansing, Michigan.
(1996a) The use of Calliandra calothyrsus Staal, S., Delgado, C.L. and Nicholson, C.F.
for milk production. National Agroforestry (1996) Smallholder Dairying Under
Research Project, Embu. Transactions Costs in East Africa. MSSD
Paterson, R.T., Roothaert, R. and Kariuki, I.W. Discussion Paper No. 9. IFPRI,
(1996b) Fodder trees in agroforestry: the Washington, DC.
Kenyan approach. In: Mugah, J.O. (ed.) Stewart, J.L. (2000) The DFID/FRP Project
People and Institutional Participation in ‘Investigation of factors affecting the nutri-
Agroforestry for Sustainable Development. tive value of Calliandra calothyrsus leaf as
Proceedings of the First Kenya National fodder for ruminants: overview of the pro-
Agroforestry Conference, 25–29 March ject and results from Latin America and the
1996. Kenya Forestry Research Institute, United Kingdom’. In: Mwangi, M.,
Muguga. Franzel, S. and Stewart, J.L. (eds)
Paterson, R.T., Roothaert, R., Nyaata, O.Z., Calliandra calothyrsus and Other Fodder
Akyeampong, E. and Hove, L. (1996c) Trees: Moving Forward in Research and
Experience with Calliandra calothyrsus as a Development. Proceedings of an interna-
feed for livestock in Africa. In: Evans, D.O. tional workshop, Nairobi, 6–8 June 2000.
(ed.) Proceedings of an International ICRAF, Nairobi.
Workshop on the Genus Calliandra, 23–27 Thijssen, R., Murithi, F.M. and Nyaata, O.Z.
January 1996. Winrock International, (1993) Existing Hedges on Farms in the
Bogor, Indonesia. Coffee Based Land Use System of Embu
Reynolds, L., Didomenico, C., Atta-Krah, A.N. District, Kenya. AFRENA Report No. 65.
and Cobbina, J. (1991) Alley farming in International Centre for Research in
southwest Nigeria: the role of farming sys- Agroforestry, Nairobi.
tems research in technology development. Thomas, D. and Sumberg, J. (1995) A review of
In: Tripp, R. (ed.) Planned Change in the evaluation and use of tropical forage
Farming Systems: Progress in On-Farm legumes in Sub-Saharan Africa. Agriculture,
Research. John Wiley & Sons, Chichester, Ecosystems, and the Environment 54,
UK. 151–163.
Roothaert, R., Arimi, H.K. and Kamau, E.N. Winrock International (1992) Assessment of ani-
(1997) Indigenous fodder species in Kenya mal agriculture in sub-Saharan Africa.
– assessing the wealth. Agroforestry Today Winrock International Institute for
9(3), 17–20. Agricultural Development, Arkansas, USA.
 Promoting New Agroforestry
Technologies: Policy
Lessons from On-farm
8
Research
S.J. SCHERR1 AND S. FRANZEL2
1Agriculturaland Resource Economics Department, University of
Maryland, College Park, MD 20742, USA; 2International Centre
for Research in Agroforestry, ICRAF House, PO Box 30677, Gigiri,
Nairobi, Kenya

Summary

Successful diffusion and adoption of new agroforestry practices depend not only
upon the technical performance of those practices and their ‘fit’ with farming sys-
tems, but also on the broader policy environment. Key policy factors relate to: tree
germplasm supply, agricultural input supply, markets for agroforestry products, land
and forest tenure systems and strategies and institutional arrangements for exten-
sion and research support. On-farm research during the technology development
process provides a strategic opportunity to begin evaluating policy constraints and
ways to address them. Researchers need to involve policy makers in the design and
evaluation of assessments and to communicate effectively with them. Based on data
and experience from on-farm research programmes in Kenya and Zambia, general
policy lessons are drawn, and specific policy recommendations are made for pro-
moting hedgerow intercropping, improved fallows, pole and timber trees, and fod-
der banks in the eastern and southern Africa context.

Introduction

Successful diffusion and adoption of new agricultural practices depend not only
upon the technical performance of those practices and their ‘fit’ with farming sys-
tems and management constraints, but also on the broader policy environment.
The successful dissemination of improved maize production technology in
Zimbabwe, Zambia and Malawi, for example, was due not just to varietal improve-

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 145
146 S.J. Scherr and S. Franzel

ments on farm, but also to technological innovations in processing, price and sup-
port policies, institutional and market changes, and coordination of these various
components (Howard et al., 1997).
The technology development and testing phase of on-farm research provides a
strategic opportunity to begin evaluating policy effects and devising options to deal
with policy constraints and opportunities. Research results reported in this volume
indicate that there is considerable potential for many agroforestry practices to
increase farm income, meet household needs, contribute to local economic growth
and possibly improve environmental conditions in different regions of Africa.
However, those same studies revealed the need for specific types of policy action to
achieve widespread adoption and significant aggregate impact from the new agro-
forestry practices. Key factors strongly influenced by public policy include: tree
germplasm supply, agricultural input supply, markets for agroforestry products, land
and forest tenure systems, technology design for extension and institutional
arrangements for on-farm research and extension support (Oram, 1993; Place and
Dewees, 1999). In this chapter, we review briefly key findings and insights in rela-
tion to these six policy areas, and recommend priorities for policy action.

Tree Germplasm Supply

Place and Kindt (1997) note many weaknesses in current policy frameworks
towards improved agroforestry tree germplasm supply, including the fragmentation
of institutional mandates and functions, lack of coordination of planning, lack of
forecasting demand for species, and the poor and unstable funding environment of
institutions. The case studies revealed specific issues of germplasm supply for the
various agroforestry practices being developed, related to the cost of tree establish-
ment, seedling production and seed supply.

Promote low-cost planting methods

For fodder banks in central Kenya, establishment costs represent only a modest
share of total costs, and for timber and pole trees in western Kenya, seedling quality
is a much more important consideration than cost. But for improved fallows in
western Kenya and Zambia and for hedgerow intercropping in western Kenya,
establishment costs are much more important, because tree densities are high and,
in the case of improved fallows, trees need to be replanted after only 3–5 years.
Reducing the cost and labour required for planting is thus an important means for
increasing the adoption potential of improved fallows and hedgerow intercropping.
Two strategies are to promote the use of bare-root seedlings and direct seeding.
For hedgerow intercropping in western Kenya, the economic break-even maize yield
increase drops from 10.5% to 6% by using bare-root rather than potted seedlings.
In the case of improved fallows in western Kenya, direct seeding (where feasible) is
far more profitable than the use of potted seedlings. Even in these cases, though,
 Promoting New Agroforestry Technologies 147

there is no evidence that financial subsidy policies are required to promote these
types of agroforestry practices, except perhaps during an initial trial period for farm-
ers to become familiar with the species or practice.

Promote small-scale farm seedling supply

While centralized tree nurseries offer some advantages for training, seed dissemina-
tion and quality control early in the extension process, they appear to be neither
needed nor likely to persist as important sources for farmers once broader adoption
takes place. As shown in the Zambian on-farm research experience, they also pre-
sent some tricky problems for division of labour inputs and of seedling outputs.
The alternative of using small-scale farm nurseries has been very successful.
Such nurseries have several advantages. The production of seedlings from seeds or
cuttings is a very labour-intensive process; farmers can manage nurseries as comple-
mentary activities to farming, thus keeping labour costs lower than in large-scale
nurseries where full-time hired labour is used. Also, they are more convenient for
farmers who avoid the costs of transporting seedlings long distances, and also the
damage and loss experienced during transport. Nurseries also provide additional
income opportunities for farmers. The only major disadvantage of farm nurseries is
that farmers usually require training in nursery establishment. Nevertheless, a small
amount of facilitation can go a long way. For example, a single extension facilitator
in central Kenya was able to assist 150 farmer groups, composed of about 2600
farmers, to establish 250 calliandra nurseries over an 18-month period, 1999–2000
(ILRI, 2000).
An active market mechanism has developed for seedlings of multipurpose trees
in western Kenya, through small-scale private nurseries. In the case of fodder trees
in central Kenya, there has been a rapid switch from project to farmer group and
private nurseries, some of which use farm-produced seed. The major recommenda-
tions for promoting small-scale nurseries, once farmers are aware of the potential
benefits of a practice, are: (i) to focus training on existing farmer groups and private
nurseries rather than individual farmers; (ii) to work through non-governmental
organizations (NGOs) and community-based organizations as well as government
extension services; (iii) to facilitate farmer-to-farmer training and exchange visits;
and (iv) to assist farmers to maintain quality and genetic diversity of seed stocks.
The case from central Kenya mentioned above illustrates how these recommenda-
tions can be effectively implemented. The extension facilitator was able to reach so
many farmers by collaborating closely with three departments of the Ministry of
Agriculture, two provincial administrations, one international NGO, four local
NGOs, ten community-based organizations and 150 farmer groups (ILRI, 2000).
Nearly all the groups had been formed for other purposes, such as church groups,
groups to buy water storage tanks, and groups to manage dairy goat breeding stock.
Groups managed one or more nurseries; in most cases the nursery was located in
the field of a member where water was available during the dry season, and group
members shared the labour and distributed the seedlings equally. The groups them-
148 S.J. Scherr and S. Franzel

selves paid for the exchange visits and these were useful for motivating them and
enabling them to exchange information on how nurseries and fodder trees could be
managed and used. Group nurseries are not necessarily permanent, in many cases
farmers indicated that they worked together only to attract assistance and training
and that their future plans were to start their own nurseries.

Expand the supply of high-quality seed through decentralized seed

orchards

The lack of tree seed has already become a significant constraint for expansion of
improved fallows in Zambia, pole and timber trees in western Kenya, and fodder
trees in central Kenya. The on-farm research projects are currently the major source
of seeds for farmers, either producing seed or helping to procure it from elsewhere. In
addition, farmer seed collection has become increasingly important for the more pop-
ular new species, such as calliandra in western Kenya, and for traditionally available
species, such as sesbania and grevillea in western Kenya. But a reliable seed supply
and distribution system on a much greater scale is required for large-scale adoption.
In contrast to the situation for annual crop seed, where the private sector is
becoming more active in development and replication throughout Africa (although
government agencies continue to play a critical role), there are as yet few incentives
for private sector investment. Some trees require many years before they produce
seeds, which delays investment returns and locks producers into selected products;
future markets for the variety of tree species and provenances are highly uncertain.
We suggest several policy actions to promote seed production and distribution.
First, we propose that high-quality, high-productivity seed orchards be estab-
lished earlier in the technology development process. These typically require several
years to be in full production. Past policy has been to wait until species and on-farm
research trials have been completed and superior species and varieties are selected,
before establishing seed orchards. The on-farm research experience in Kenya and
Zambia suggests that a more effective strategy is to establish seed orchards in the
technology development process, so that sufficient seed is available for large-scale
adoption by the end of the trial period (Simons, 1996). This may result ultimately
in the seed of large numbers of trees not being used. However, the cost is not high
relative to the benefits foregone from farmers’ having to wait many years until
germplasm is available.
The argument for earlier seed orchards is validated by the economic analysis pre-
sented in Table 8.1, which compares the strategy of earlier orchards with later ones,
using data from calliandra trials and seed orchards in central Kenya. On-farm fodder
tree trials began in 1991 involving three species, but, by 1993, it was clear that only
calliandra was appreciated by farmers. Seed orchards were established in 1994 and
seed became available for on-farm distribution in 1996. The question analysed in the
economic analysis is, ‘Would it have been better to start 1-ha seed orchards for each
of the three species at the beginning of the on-farm trial in 1991, so that calliandra
seed would have been available 2 years earlier, even though two of the seed orchards
Table 8.1. Net present value of establishing early seed orchards, when knowledge of which species to extend is still uncertain, versus
waiting until trials are completed (US$).
(A) Model 1: On-farm trials begin in year 1 on three species. One-hectare orchards are established in year 3 for the one species that
performs well. Adoption begins in year 5.
Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Y13 Y14 Y15
Costs
Seed and 18
inoculant
Land preparation 42
Nursery 297
Planting 85
Weeding 34 34
Harvesting and 0 678 678 678 678 678 678 678 678 678 678 678 678
post-harvest
Fencing 212
Guarding 102 102
Stand 17
management
Land rental 254 254 254 254 254 254 254 254 254 254 254 254 254
Total costs 0 0 1,060 1,068 932 932 932 932 932 932 932 932 932 932 932
Benefits 0 0 0 0 0 23,316 93,265 186,529 279,794 373,058 466,323 559,587 652,852 746,116 839,381
Net benefits 0 0 −1,060 −1,068 −932 22,384 92,332 185,597 278,861 372,126 465,390 558,655 651,919 745,184 838,448
Promoting New Agroforestry Technologies

Net present 487,754

value
Seed harvest 0 0 0 11.1 22.2 33.4 44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5
(kg ha−1)
No. farmers 0 0 0 0 359 1,076 1,435 1,435 1,435 1,435 1,435 1,435 1,435 1,435 1,435
receiving seed
No. farmers 0 0 359 1,435 2,870 4,305 5,739 7,174 8,609 10,044 11,479 12,914
earning benefits
149
Table 8.1. Continued.
150

(B) Model 2: On-farm trials begin in year 1 on three species. One-hectare orchards are established in year 1 for all three species. One
species performs well and adoption begins in year 3.
Y1 Y2 Y3 Y4 Y5 Y6 Y7 Y8 Y9 Y10 Y11 Y12 Y13 Y14 Y15
Costs
Seed and inoculant 53
Land preparation 127
Nursery 890
Planting 254
Weeding 102 102
Harvesting and 0 678 678 678 678 678 678 678 678 678 678 678 678 678 678
post-harvest
Fencing 636
Guarding 203 203
Stand 51
management
Land rental 763 763 254 254 254 254 254 254 254 254 254 254 254 254 254
Total costs 3,079 1,746 932 932 932 932 932 932 932 932 932 932 932 932 932
Benefits 0 0 0 23,316 93,265 186,529 279,794 373,058 466,323 559,587 652,852 746,116 839,381 932,645 1,025,910
Net benefits −3,079 −1,746 −932 22,384 92,332 185,597 278,861 372,126 465,390 558,655 651,919 745,184 838,448 931,713 1,024,977
S.J. Scherr and S. Franzel

Net present 839,308

value
Seed harvest 0 11.1 22.2 33.4 44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5 44.5
(kg ha1)
No. farmers 0 0 359 1,076 1,435 1,435 1,435 1,435 1,435 1,435 1,435 1,435 1,435 1,435 1,435
receiving seed
No. farmers 0 0 0 359 1,435 2,870 4,305 5,739 7,174 8,609 10,044 11,479 12,914 14,348 15,783
earning benefits
Data are based on experiences developing Calliandra calothyrsus seed orchards and disseminating calliandra in the Embu area, Kenya. 31 g of seed are needed to plant 250
trees. It is assumed that an adopting farmer plants 250 trees, which provides an increase in annual net income US$65, beginning in the second year (Chapter 7). The discount
rate used in the analysis is 20%.
 Promoting New Agroforestry Technologies 151

would have been of no use?’. The analysis shows that the net present value of the strat-
egy of early seed orchards was 72% higher than the strategy of waiting for the results
of the on-farm trial before establishing seed orchards. In fact, early seed orchards
would be justified even if 20 1-ha seed orchards were established early in order to get
a head start in disseminating, but only one of the species was ever disseminated. The
results highlight the huge benefits associated with getting a head start on disseminat-
ing new practices, as compared to the costs of establishing seed orchards.
We also propose that a range of different entities be involved in seed produc-
tion; ICRAF and partners are currently helping research institutions, NGOs, pri-
vate nurseries, farmer groups and individual farmers to establish seed orchards. In
western Kenya, ICRAF has supplied seed and information to several farmers plant-
ing calliandra orchards and has agreed to purchase seed at an agreed price during
the first 2 years of production, after which they will have to market seed on their
own (James Were, ICRAF, personal communication).
Maintaining genetic diversity is critical because, after planting some technolo-
gies, such as trees for fodder, poles or firewood, farmers may not plant again for
another 20 years. Participants in fodder on-farm trials have been encouraged to har-
vest seed from at least 30 trees, but this has been very difficult for them, as allowing
a tree to seed means it will shade an area that might otherwise be used for crops.
One option being pursued is to promote members of farmer groups to each estab-
lish 5–10 different seed trees, which are used as sources by the whole group.
The active local seed propagation and distribution of calliandra by on-farm
research participants in central Kenya suggests that a strategy of decentralized distri-
bution to trained farmers in different geographic areas could induce broad diffu-
sion. In the future, as adoption demand for seed increases, there is also potential for
farmer-managed seed orchards to develop commercial supply agreements with local
agricultural supply stores and other rural retail outlets.

Agricultural Input Supply

The market distribution networks, price and price variability and regulations affect-
ing agricultural input supply have significant impacts on the adoption and prof-
itability of the agroforestry systems evaluated here. In some cases, agroforestry
represents an option for farmers to bypass poorly functioning input markets
through direct substitution of agroforestry products or services. In other cases, these
are complementary to non-agroforestry inputs and depend for their effectiveness or
profitability on a stable supply of those inputs. Financial credit appears to be less
important for agroforestry in this region than commonly assumed.

Promote agroforestry for input substitution

In many African smallholder farming systems, working capital is a binding con-

straint to expanding production (Ellis, 1988). Thus one of the most attractive fea-
152 S.J. Scherr and S. Franzel

tures of many agroforestry systems is the provision of products that substitute for
cash inputs, for example, the substitution of improved fallows for mineral nitrogen
fertilizer or of leaf fodder for dairy meal.
In the Zambian case study, farmers especially appreciated improved fallows as
nitrogen fertilizer became unaffordable after the elimination of the parastatal distri-
bution and credit systems in the late 1980s and early 1990s. These changes caused
the ratio between maize and fertilizer (urea) price to decline from 0.27 (1985–1988)
to 0.13 (1989–1991) (FAO, 1998). There was a dramatic decline in maize area and
in the use of improved inputs, especially fertilizer, the use of which declined by 70%
between 1987/88 and 1995/96 (Howard et al., 1997). Farmers had seen the benefits
of using nitrogen fertilizer, but it was no longer affordable; they were thus excited
about the possibility of using improved fallows to supply nitrogen.
In western Kenya, fertilizer is widely available, but is used by only about 20%
of farmers and at low rates of application. The ratio between the maize and fertilizer
(urea) price ranged between 0.18 and 0.26 between 1985 and 1995, but did not fol-
low any noticeable trend (FAO, 1998). Available evidence suggests that farmers pre-
fer to spend scarce working capital on other investments, such as off-farm businesses
or education, rather than on fertilizer (Crowley et al., 1996). The recent high enthusi-
asm for planting improved fallows suggests that farmers are more willing to use
additional labour, rather than working capital, to increase the fertility of their soils.
The on-farm research studies found that while improved fallows in Zambia did
not produce yields as high as those with fertilizer, yields were far higher than in
non-fertilized fields, and returns to labour were actually highest with improved fal-
lows. Sensitivity analysis shows that a decline in fertilizer prices (or increase in
maize prices) does greatly increase the profitability of fertilized, continuous maize
relative to improved fallows, although in drought years improved fallows offer a
much lower-risk alternative.
Calliandra tree fodder is highly attractive to farmers as a substitute for dairy
meal on central Kenyan dairy farms, in part because of the poor performance of the
dairy meal market. Meal prices fluctuate sharply, because maize, the nation’s main
food staple, is the main ingredient. Also, farmers are often uncertain about the
composition, and hence quality, of the meal (Chapter 7). Many farmers face cash
constraints which make purchase difficult, and many also face difficulties in trans-
porting the meal to their farms. Financial sensitivity analysis suggests that the use of
calliandra is not very sensitive to the price of dairy meal, and these other favourable
factors should lead to a robust demand for fodder bank technology. At this time,
nearly two-thirds of the farmers using calliandra feed it together with dairy meal,
but this is, in part, because of limited calliandra supplies. The share of calliandra
can be expected to increase unless there is a very large decline in meal prices.

Ensure availability of complementary inputs

Agroforestry returns and adoption are not always more attractive where input mar-
kets are poorly developed. Chemical fertilizer can, in some cases, be complementary
 Promoting New Agroforestry Technologies 153

to soil nutrients provided by trees and shrubs, such that assured fertilizer access is
important to the productivity of agroforestry systems. Agroforestry technologies can
provide nitrogen, but phosphorus is a limiting nutrient in many farming systems,
and can only be replenished from off-farm sources (either rock phosphate or phos-
phate fertilizers) (Sanchez et al., 1997). Phosphorus deficiency will become an
important issue in Zambia as the farming system intensifies. In western Kenya,
improved fallows may not work at all in some of the areas with severe phosphorus
deficiency. In many areas, there appears to be a strong nitrogen × phosphorus inter-
action. Thus current efforts to make phosphorus more available to farmers are
essential to the long-term success of these agroforestry systems, and should be pro-
moted vigorously (Sanchez et al., 1997).

Encourage self-financing of agroforestry

The provision of credit is commonly mentioned as an important policy issue for

agroforestry and forestry, because of the presumed lag in time between initial invest-
ment in tree establishment and maintenance and the beginning of significant bene-
fit streams (Gregersen et al., 1992). The five case studies in this volume suggest that
lack of financial credit should not be a significant constraint to widespread adoption
among smallholders in Africa. This is largely due to small farm size (and hence a
small scale of operation), farmers’ incrementalist approach to tree establishment,
and their preference for strategies that reduce cash costs and reduce risks. Payback
periods for fodder trees and improved fallows were 3 years or less, since farmers
were able to intercrop annual crops with trees, or harvest some tree products, such
as fuelwood or leaf fodder, during the first and second years.
Lack of credit is certainly a constraint to smallholder use of inputs such as fer-
tilizer, where cash costs account for a high proportion of total costs of the practice.
But for all of these agroforestry practices, direct cash costs for seed or seedling pur-
chase actually account for a relatively small share of total costs, even in the year of
establishment, as most farmers rely on family labour for establishment, mainte-
nance and harvest, and can usually time these operations so that they do not con-
flict significantly with other income-earning activities. Largeholder farmers are, by
contrast, likely to require financial credit to establish agroforestry on a broad scale.
Where credit might be available, there is no evidence that subsidized rates
would increase smallholder agroforestry adoption. Financial sensitivity analysis
shows relatively low sensitivity of returns to the discount rate used in improved fal-
lows in Zambia, Kenyan fodder banks, and hedgerow intercropping. In the case of
improved fallows in western Kenya, using low estimates of the opportunity cost of
land and labour, discount rate sensitivity is low; however, the effects become quite
important if high opportunity costs are assumed. For lenders, short-term agro-
forestry systems are likely to be more attractive than long-term systems. However,
given the severe limitations on agricultural credit in these areas, it is likely that
farmers would divert any available credit to higher priorities, such as off-farm busi-
nesses or education (Crowley et al., 1996).
154 S.J. Scherr and S. Franzel

Direct subsidies to farmers, in the form of cash payments for tree planting and
maintenance might indeed accelerate adoption. In some cases, not evaluated in these
studies, they may be justified on social policy grounds. Even so, the record for sustain-
ability of agroforestry practices established through subsidies has not generally been
successful in Africa or most other developing countries (Scherr and Current, 1999).
Processing industries involved in outgrower schemes for higher-value crop and
livestock products may be encouraged to subsidize some of the cash costs of agro-
forestry investments that contribute to higher crop or livestock yields or reduce
input costs. This would most likely be provided in kind rather than in cash. For
example, in Tanzania, private tobacco companies are providing seedlings to farmers
so that the farmers can produce fuelwood to cure their tobacco (Ruvuga et al.,
1999). But in the case studies presented in this volume, where the end products are
maize, milk and construction wood, processing industries are not involved, or likely
to become involved, in subsidizing agroforestry investments.

Product Markets

In the past, agroforestry extension has often been ‘supply-driven’; that is, the
emphasis has been on increasing production levels of trees and crops, with little
attention paid to trends in product demand or price. Recent studies and experience
suggest that product market conditions and market institutions can play a critical
role in farmer adoption and the economic contribution from agroforestry (Leakey
and Newton, 1994; Dewees and Scherr, 1996). For the agroforestry practices devel-
oped and evaluated in these on-farm research trials, market conditions for both
trees and associated agricultural products had important effects.

Promote marketing of farm-grown tree products

Having a marketable component contributed significantly to farm profits and to the

incentive to expand agroforestry technologies. Farmers testing hedgerow intercrop-
ping in western Kenya did not find soil erosion control benefits sufficiently com-
pelling for enthusiastic adoption, given the lack of significant impacts on crop
yields or economically important tree by-products. Fuelwood, although economi-
cally unimportant in eastern Zambia, was an important by-product from improved
fallows in western Kenya and sensitivity analysis showed that the price of fuelwood
had a strong effect there on system profits. While multipurpose trees in western
Kenya were grown mainly for timber and poles, early subsidiary products, such as
mulch, fuelwood or fodder, provided important benefits.
Further development of new marketing channels, quality standards, efficient
marketing practices and market information systems will be essential for improving
smallholder participation and benefits from new tree product markets. Poorly devel-
oped tree product markets are a significant limitation for large-scale agroforestry
adoption. For example, for any new tree species, sawmillers must learn their uses,
 Promoting New Agroforestry Technologies 155

milling characteristics, strength, etc., and must be able to sell their product to con-
sumers who want to know what they are getting (Tyndall, 1996). The lack of qual-
ity criteria or regulation means that risks of dealing in new woods are high.
Sawmillers in central and western Kenya still rely for their supplies mostly on illegal
timber from the forest. They are not very familiar with grevillea, which is widely
grown by smallholders, and at the moment have few market connections with
smallholders (Tyndall, 1996). Casuarina, by contrast, is known by construction
people in most urban areas and does not face similar market problems.
Product market characteristics and function should be analysed not only to
predict price and supply trends, but also to estimate the nature and scale of environ-
mental effects from market development. For example, studies of farmers with
improved fallows in both Zambia and western Kenya, and multipurpose trees in
western Kenya, suggest that fuelwood and other products produced from farm-
grown trees may actually be substituting, at least on a small scale, for supplies from
natural forests and woodlands. More careful market analysis is needed to determine
the scale of impact, especially in Zambia, and whether more active public interven-
tion to promote agroforestry may be used as a forest protection policy.

Promote agroforestry to reduce risk in unstable crop and livestock

product markets

Conditions in basic grains markets have clearly affected farmers’ strategies in adopt-
ing agroforestry. In Zambia, the parastatal maize and cash crop marketing systems –
with their stable outlets and price subsidies for maize producers – collapsed in the
early 1990s, and private markets have not developed sufficiently to replace them.
Farmers in Eastern Province are concerned that the marketing system itself (as dis-
tinct from aggregate demand) will not be able to absorb big increases in basic grains
production. Thus many farmers indicated that they prefer to use improved fallows to
reduce the area under maize, allowing them to emphasize higher-value cash crop pro-
duction on their other fields. Financial sensitivity analysis shows that a future increase
in maize prices would increase returns to fertilized maize much more than returns to
improved fallows. The instability common to markets for higher-value products also
affects farmer agroforestry strategies. For example, the milk industry in Kenya
became very unstable after the Kenya Cooperative Creameries lost its monopoly, in
part due to mismanagement, and new dairies emerged. Increased risk motivated
farmers to substitute farm-produced calliandra for dairy meal to reduce cash costs.
Financial returns to calliandra are not very sensitive to changes in the milk price
itself, especially when calliandra is used as a substitute for dairy meal rather than as a
supplement. Over the long term, policy makers should seek to improve milk-market
stability and livestock health services, to strengthen the sector. This would encourage
widespread adoption of calliandra as farmers learn about its benefits, even in a lower-
risk environment. Policy analysts need to address broader sectoral reform, such as
milk-market reform or credit provision for purchasing dairy animals, in addition to
specific changes that encourage adoption of a particular agroforestry practice.
156 S.J. Scherr and S. Franzel

Promote agroforestry for high-value crops

Using trees to increase the production of higher-value crops, rather than basic
grains, may greatly increase the income effects of agroforestry adoption. In western
Kenya, researchers tested new agroforestry practices on maize, because maize
accounted for most of the cultivated area and because increasing yields of the staple
grain was seen as the way to improve food security. But many farmers preferred to
test the practices on higher-value crops because, if the agronomic response of
higher-value crops were similar to that of maize for equal inputs, then profitability
would be higher. These farmers may have been motivated by food security as well,
but their means to achieve it was to increase their cash income by growing more
high-value crops, not by increasing maize production. For example, farmers in west-
ern Kenya with good market access are improving soil fertility through the applica-
tion of mulch from Tithonia diversifolia hedges on kale (Brassica oleracea) and
tomatoes, rather than on maize plots (ICRAF, 1997). Policy makers, extensionists
and researchers should redirect more of their investment to promote agroforestry
that increases production and income from high-value crops.

Land and Forest Tenure and Regulatory Systems

Land and forest tenure and regulatory systems, including rules governing women’s
control and management of tree resources, have been widely cited as constraints to
agroforestry adoption (Bruce and Fortmann, 1989; Gregersen et al., 1992).
However, Place (1995) examined tree planting in four countries of east and south-
ern Africa and did not find tenure issues to be as constraining as supply-side issues,
such as the availability of planting material or information. On-farm research
reported in this volume did not reveal tenure issues to be especially important for
the practices studied in Kenya and Zambia, which were generally established by
farmers on privately owned or controlled crop fields. Nor did regulatory issues
emerge as constraints in any of the case studies.
Few tenure limitations at all were observed for hedgerow intercropping,
improved fallow or agroforestry tree planting in western Kenya. Only 3 of 50 farm-
ers dropped out of hedgerow intercropping trials in western Kenya because of land
tenure disputes. In central Kenya, farmers did not plant fodder trees on fields away
from the homestead. However, this was not due to tenure concerns, but to the diffi-
culty in carrying fodder from the fields back to the homesteads where the cows are
kept. Most farmers in this location do have at least one such far field.

Promote agroforestry equally to women and men

Traditional or legal rules on women’s control or management of land and trees

appeared to have little effect on the adoptability of the agroforestry practices studied
here. Only 2 of 14 women dropped out of hedgerow intercropping trials because of
 Promoting New Agroforestry Technologies 157

disagreements with their spouses. Traditional taboos against tree planting by women
appear to have broken down, or are not applied to the planting of agroforestry species.
Women in western Kenya did not participate in the pruning of hedges in
hedgerow intercropping, but this may have been a way to encourage greater labour
contributions by the men or a reflection of lower perceived benefits. There were few
problems with women clearing trees in Zambia, as they have customarily been
involved in bush clearing. However, improved fallow plots of female farmers were
significantly smaller than those of male farmers, probably because they had less
access to land and labour. In central Kenya, there are fewer female-headed house-
holds than in Zambia and western Kenya, and in male-headed households, both
men and women appeared to be involved in growing and harvesting calliandra for
fodder. However, gender issues were not evaluated systematically.

Encourage community solutions to reduce free-grazing risks to

agroforestry

In Zambia, the widespread practice of free grazing eliminated one species of

improved fallow, and is a threat to others. Researchers are now seeking to introduce
species unpalatable to livestock. Free grazing also discourages intercropping of
unpalatable trees with crops because cows feeding on maize stalks trample the trees.
ICRAF and its partners have been active in organizing workshops of local pol-
icy makers, including traditional chiefs, to address the free-grazing problem. Some
local chiefs in Zambia are seeking to enforce controlled grazing, and some local
administrators in Uganda are establishing by-laws against tree destruction by free-
grazing cattle. Researchers are monitoring these efforts to determine their success
and potential application in other sites.

Technology Design for Extension

While the case studies of fodder trees in Kenya and improved fallows in Zambia
provide some evidence of spontaneous diffusion of high-performing agroforestry
practices, all of these technologies are complex and farmers require information,
training and assistance with planting materials, if they are to achieve production
and income potentials. Hence, extension must play a significant role in large-scale
implementation. The case studies highlight several key elements for agroforestry
extension, especially tailoring recommendations to farmer and regional socioeco-
nomic conditions, the need to provide information on agroforestry management,
and involving farmers strategically in technology selection and diffusion.

Tailor recommendations to farmer and regional conditions

The case studies clearly show the need for more systematic targeting of agroforestry
practices to the socioeconomic, as well as the biophysical, conditions of the pro-
158 S.J. Scherr and S. Franzel

gramme areas, and to the varying needs of individual farmers within those areas. In
particular, the technical mix of land, labour and cash requirements affects their suit-
ability for farmers with different resource endowments.
Expansion of hedgerow intercropping in western Kenya was associated with
management by males or couples, regular technical assistance, cash crops, good
hedge management and less depleted soils. The degree of expansion was not
observed, in this case, to be associated with wealth, off-farm income, farm size,
crop sales, labour to land ratios, age or use of purchased inputs. Patterns of expan-
sion for improved fallows were quite different in western Kenya and eastern
Zambia. In the former, improved fallows were most attractive to farmers facing
labour scarcity or participating in off-farm employment and with serious fertility
problems, while in the latter the major explanatory factors appeared to be soil
water-holding capacity, wealth and labour availability. By contrast, expansion of
calliandra fodder banks in central Kenya was associated most with middle-aged,
full-time, male farmers in the middle- to high-income range, who gave priority to
dairy production.
The opportunity cost for labour is a key factor operating at both regional and
farm scales. While hedgerow intercropping seems most suitable to areas and farmers
with low opportunity costs and less degraded soils, improved fallows are suited to
areas of high opportunity costs and more depleted soils. In both technologies, the
importance of labour costs in farm profitability means that careful attention should
be paid in both research and extension to strategies for reducing labour costs (such
as direct seeding and relay cropping in improved fallows). In the case of fodder
banks, since farm profits are only minimally sensitive to labour costs, labour costs
are a less important consideration.
The evidence of inter-farm diversity argues for a ‘basket of options’ approach,
with extensionists well informed about technology performance under a range of
conditions. Greater species diversity is needed, to reduce risks and meet a wider
range of farmer needs and preferences as to niche, agroecological condition, use and
management.

Disseminate more information on agroforestry management

The case studies show how important it is for farmers to have good information
not only about species selection and tree establishment – the principal foci of
most extension programmes – but also on tree management. Different manage-
ment practices are desirable for a given agroforestry technology under different
conditions, as illustrated by differences in optimal improved fallow management
in western Kenya and eastern Zambia, and for farmers with different farm size
and commercial orientation. For example, intercropping trees with crops during
the first year of establishment of improved fallows reduces tree and crop growth,
but may be optimal for farmers with limited land and labour. Extensionists
should be able to provide to farmers information relevant for this decision
process.
 Promoting New Agroforestry Technologies 159

Strengthen farmers’ roles in technology selection and diffusion

The case studies illustrated that farmers can play a much more central role in extension
programmes, both in technology selection and design, and in the diffusion process.
Farmer evaluation played a critical role, in several sites, in the selection and adaptation
of technologies for subsequent promotion by extension services. The agroforestry
study in western Kenya used a systematic approach to farmer evaluation – with farm-
ers’ own on-farm trials as the evaluation focus – which has wide potential application
in extension. Each farmer experimented on his or her own with the trees; farmers
monitored progress in group visits and researchers used participatory techniques, such
as an adaptation of the traditional bao game, to monitor farmer preferences. Using this
tool, the popularity of certain species based on criteria unrelated to production, such
as the ornamental value of casuarina, was discovered. By evaluating farmers’ choices for
new plantings, a preference for external boundary niches was identified, which would
not have been predicted from existing patterns of farm tree location.
Farmers’ evaluation criteria for tree species and agroforestry practices may differ
from those of researchers or extensionists (Raintree, 1991; Franzel et al., 1996). For
example, in the hedgerow intercropping trials in western Kenya, farmers evaluated
yield performance by comparing current with past yields, rather than control and
test plots. They compared hedgerow intercropping performance not with control
plots using no nutrient applications (as was common in researcher’s experiments
and extensionists’ demonstration plots), but rather with other nutrient management
strategies known in the area. Such information can be valuable in designing exten-
sion information and strategies for technology introduction.
There is considerable potential for improving agroforestry management by pro-
moting active information exchange among farmer users. Over half of the farmers
experimenting with calliandra fodder in central Kenya had never even seen another
farmer growing the tree, which greatly limited their knowledge and the scope for gen-
erating and sharing local innovations, as well as their economic returns. By contrast,
the diffusion of improved fallows in Zambia and the community-wide assessment
processes in western Kenya expanded and institutionalized local knowledge about tree
management. Specific mechanisms utilized in the case studies have promising appli-
cation in extension networks, including farmer-to-farmer visits, group meetings to
exchange information and evaluate new practices, identification of farmer experts and
their integration into extension efforts, and village-based change processes.

Institutional Arrangements for On-farm Research and

Extension
Agroforestry extension has been problematic in Africa because of the lack of rigor-
ously validated and locally adapted extension messages, the general weakness and
limited resources of public extension systems, lack of agroforestry training for exten-
sionists, and the unclear assignment of responsibility for agroforestry among agri-
cultural and forestry extension institutions. Agroforestry research suffers from
160 S.J. Scherr and S. Franzel

similar institutional weaknesses, as well as a weak field presence (Kerkhof et al.,

1990; Scherr, 1990b, 1995; Scherr and Müller, 1990), and poses particular chal-
lenges for the implementation of on-farm research (Scherr, 1990b, 1991; Merrill-
Sands, 1991; Shepherd and Roger, 1991). While enthusiasm for on-farm research
within the scientific community has increased steadily, it has been seen as an expen-
sive instrument that cannot be replicated very widely. The evolution of large-scale
on-farm research efforts in Kenya and Zambia has generated insights and potential
models for the design of more robust, efficient, and farmer-responsive research and
extension programmes which address these challenges.

Institutional evolution in the study sites

At all three sites – central Kenya, western Kenya and eastern Zambia – on-farm
research in the early 1990s included both ‘researcher-led’ (type 1 and type 2) trials
and ‘farmer-led’ (type 3) trials, in which farmers experimented on their own with
new species and technologies. There was considerable exchange of information
between researchers and other organizations, such as extension services, develop-
ment projects, NGOs and farmer groups. But there was initially little collaboration
with such organizations in the management of farmer surveys or on-farm trials.
Farmers interacted with each other at group meetings and farmer-to-farmer visits
but, in general, researchers interacted with farmers on an individual basis.
Gradually, however, researchers at all three sites have strengthened working relation-
ships with other institutions, especially farmer groups and extension services, pro-
jects and non-governmental organizations involved in testing and disseminating
agroforestry practices. Different approaches evolved in the three sites.

Central Kenya
In Embu in 1993, extension staff became actively involved in the implementation
of trials and monitoring surveys. The project began to facilitate broader adoption in
1996–1997 when researchers and extension staff of three government services
helped 12 farmer groups, including 410 farmers, to start 14 calliandra nurseries.
The farmers planted, on average, about 233 trees per farm. As a result of the high
profitability and suitability of the calliandra technology, there has been some
farmer-to-farmer diffusion of seeds and management information. None the less,
the aggregate impact has been quite limited because of the geographic dispersion of
the trials, shortages of seeds, and the small numbers of farmers initially involved.
The recent emphasis on dissemination through farmer groups is designed to accel-
erate diffusion of the technology.

Western Kenya
In western Kenya, the on-farm trials reported in this volume took place in the early
1990s through women’s self-help groups. These groups helped select members to
 Promoting New Agroforestry Technologies 161

participate in the trials and hosted meetings to discuss trials and provide feedback.
No other organizations were involved in managing this first round of on-farm trials.
Further on-farm research was started in 1996 on improved fallows and biomass
transfer. This second set of trials involved several other organizations, including the
Ministry of Agriculture’s extension service, the Organic Matter Management
Network and CARE International.1 A distinguishing feature of the programme was
a village-scale approach, in which members of entire villages jointly agreed to par-
ticipate in selecting, testing and evaluating new technologies. The KARI-KEFRI-
ICRAF team is the hub and coordinator of the research. In 1996, the team helped
farmers establish 133 on-farm trials on the two technologies. Through involvement
of the other organizations, especially the public extension service, they were able to
expand to 1642 farmers in 1997 and several thousand farmers in 1999.

Eastern Zambia
Since 1996, the Zambia-ICRAF project has helped facilitate the establishment of an
informal network of organizations to conduct adaptive research, training and facili-
tate dissemination of improved fallows. The project and its collaborators helped
farmers establish 194 improved fallow on-farm trials in 1994–1995; by 1998 there
were several thousand farmers using or experimenting with the technology. The on-
farm research project included only one senior researcher and a technician. The
extension service was a full partner in the on-farm research, laying out most of the
over 200 type 2 trials, supporting village nurseries and monitoring the trials.
Managing on-farm trials was seen as a normal duty of extension and NGO staff.
Development projects provided some incentives to extension staff, such as bicycle
repair allowances. Farmer groups managed nurseries, distributed seed and seedlings,
and exchanged knowledge, training and experience on improved fallows. About 75
representatives of the dozen or so organizations, including farmer groups, met one
to two times per year between 1996 and 2000 to plan for wider testing and exten-
sion of the improved fallows, to review the problems and state of knowledge about
them, and to develop and update extension materials.

Promote integrated efforts of research and extension organizations

Experience in the three study sites demonstrates the potential benefits for farmer
adoption of integrating research and extension activities. In Zambia and western

1 CARE’s Agroforestry Extension Program was originally designed in the mid-1980s to

integrate technology research and design with extension, in response to the lack of
agroforestry practices tested or adapted to local conditions (Buck, 1990). Researcher-
controlled on-farm trials were established jointly with KEFRI and technology monitoring of
farmer-designed trials (type 3) was instituted. As a result of staffing changes, high demand
for extension services, and an institutional mandate favouring extension (rather than
research), the on-farm technology design and testing component declined. None the less,
CARE’s experience widely influenced thinking about agroforestry research and extension
linkages (Scherr, 1990a).
162 S.J. Scherr and S. Franzel

Kenya, the enthusiastic interest of the many NGO and public extension pro-
grammes, as well as spontaneous farmer-to-farmer diffusion of the improved fallow
technology, led to the evolution of an integrated on-farm research approach that
resulted in rapid dissemination of new practices. The Zambia-ICRAF project coor-
dinates the network for participatory monitoring and evaluation of on-farm
research and acts as a catalytic and action-oriented group for widespread dissemina-
tion of the technology in pilot areas. Furthermore, the Zambian network associa-
tion provides a forum for identifying policy-related problems and opportunities and
promoting action to address them. For example, at a workshop of policy makers
held in 1996 several issues emerged, including the need to control free grazing, that
stimulated local policy action.
By contrast, in central Kenya, there was an almost complete absence of interna-
tional and national NGOs involved in agricultural development, largely because the
area has a relatively high per capita income by Kenyan standards. The Ministry of
Agriculture has been active in on-farm research and extension, as have the many vil-
lage-level farmer groups and community-based organizations. Research and exten-
sion efforts have required more facilitation in central Kenya than in western Kenya
and Zambia, but have been just as effective.

Promote a village-based approach to on-farm research and extension

A second factor that distinguished the central Kenya experience was that the farm-
ers involved in on-farm research in the early 1990s were geographically separated
and unrelated. They had no opportunity to generate the synergy or collective initia-
tives that led to rapid farmer-to-farmer agroforestry learning and dissemination in
the other sites. The western Kenya and Zambia experiences demonstrate the bene-
fits of using a village approach to on-farm research and extension. The village-wide
group ensured farmer-relevant trial design, attracted farmer collaborators, facilitated
farmer evaluation of technologies, and encouraged rapid local dissemination of suc-
cessful practices. In central Kenya, the new emphasis on working with farmer
groups has led to greater farmer-to-farmer learning and increased opportunities for
the spontaneous spread of fodder-tree practices.

Informing the Policy-making Process

Effective policy analysis does not only include rigorous assessments and sound rec-
ommendations; it also requires effective communication with policy makers. An
entire discipline, agricultural extension, exists for promoting communication
between researchers, development specialists and farmers. In contrast, little attention
is given to the area of communication between researchers and policy makers. This
is indeed surprising, given researchers’ frequent difficulties in influencing policy.
Several critical lessons emerge from the case studies and from research else-
where. First, policy makers need to be involved from the start in designing and
 Promoting New Agroforestry Technologies 163

implementing policy research. Whereas farmer involvement in technology design

and research is widely accepted, policy makers are often not consulted at the
design stage. Workshops involving researchers and policy makers are an effective
means of improving consultation. At workshops in Tabora, Tanzania in 1997 and
1998, policy makers discussed and chose among various policy problems and top-
ics associated with agroforestry, food security and the environment. Proposed
interventions that required research included discouraging shifting cultivation,
promoting herders to keep their livestock from harming crops, controlling and
protecting against burning, and assessing the causes of, and reducing, deforesta-
tion (Ruvuga et al., 1999). Furthermore, policy makers provide crucial insights
about their perception of problems and opportunities, and institutional mecha-
nisms for policy implementation (Tomich et al., 1998). A key principle of market-
ing is that the more consumers are consulted and involved in the design of a
product, the more it is likely to be used. Certainly, the same principle applies to
policy research.
Further, policy research is most effective when it informs policy makers about
the implications of different options rather than simply advocating one policy over
others (Weber et al., 1988). In eastern Zambia, advocates of improved fallows often
state that livestock owners should be forced to prevent their cattle from destroying
improved fallows, on the grounds of farmers’ rights to grow trees on their plots dur-
ing the dry season. A more balanced and insightful approach to the problem would
be to bring together the various stakeholders, including cattle owners, assess the
implications of alternative policies, and inform policy makers about the implica-
tions. In eastern Zambia, this would involve assessing the effects of allowing free
grazing, and thus limiting improved fallows, on maize yields and food security.
Alternatively, how could improved fallows contribute to food security if free grazing
was curbed? And, finally, are there compromise arrangements where both livestock
owners and improved fallow planters could reduce or share the burden of protec-
tion, through, for example, a rotational grazing system?
Also, there is often considerable misunderstanding about who policy makers
are. Policy makers are often characterized as high-level persons in the capital city
when, in fact, local officials and traditional leaders in the countryside often wield
considerable power. Even if local officials and leaders do not make policies, they are
the ones responsible for implementing them, which is often at least as important! In
fact, the influence of local leaders has increased in recent years as many govern-
ments in eastern and southern Africa have decentralized authority and budgeting.
ICRAF and its partners have also found that local leaders can play an important
role as advocates of technology dissemination and in mobilizing communities to
test new practices (Raussen et al., 2001).
Finally, the packaging of policy analyses is critical as policy makers, whether
local or national, lack the time to read lengthy academic reports. Policy communi-
cations need to be brief, to the point, and targeted as specifically as possible to bod-
ies responsible for making and implementing policy. Richards and Asare (1998)
provide an example of an effective policy briefing note designed to assess incentives
for Ghanaian farmers to keep timber trees in their cocoa plots.
164 S.J. Scherr and S. Franzel

Conclusions
The scale and pace of adoption of new agroforestry practices is likely to depend, to
a greater extent than for new crop varieties, upon an improved policy environment.
Table 8.2 summarizes some key policy recommendations for promoting the four
agroforestry practices, which were identified through on-farm research programmes.
Germplasm policy must focus more on reducing the cost of planting material,
especially important for technologies such as hedgerow intercropping and improved
fallows. Large-scale seed supply and distribution networks are particularly impor-
tant for improved fallows, and small-scale tree nursery promotion is needed for
improved fallows in Zambia and timber, pole and fodder trees in Kenya. Species
diversification is critical for all technologies. Subsidies and other direct incentives
are not needed.
Agroforesters working with improved fallows should follow closely the price
and availability trends for chemical fertilizers, and understand how these will affect
farmer adoption. More studies still need to be done on farmer nutrient manage-
ment strategy under different price and risk regimes, and on the biophysical interac-
tions of tree-derived nutrients and chemical fertilizers. A similar focus and studies
are needed for the dairy meal input market and farmer use. While formal and infor-
mal financial credit markets will have indirect effects on farmers’ decisions to invest
in agroforestry, provision of production credit itself is unlikely to be a critical policy
issue for smallholders.
For improved fallows and hedgerow intercropping, the main product markets
of importance are basic grains and high-value crops. More research is needed to
understand the effectiveness of these technologies with crops other than basic
grains. In the case of fodder trees, the interactions of milk, dairy meal and maize
markets need to be better understood, and linked with overall farmer decisions
about feeding practices.
For the practices studied, there was little observed effect of tenure and regula-
tory policy factors on agroforestry adoption or farm profitability. Damage to trees
from free-grazing animals can probably be addressed through choice of suitable
tree species and planting sites, although exploratory work on changing commu-
nity-level norms and regulations is worth pursuing. Women’s participation in test-
ing and using practices appears to be as high as men’s, but further monitoring is
needed and new mechanisms to enhance women’s participation should be
explored.
The general lessons about extension strategy can be applied to all of the tech-
nologies studied. The on-farm trials already implemented can guide the targeting of
farmers and design of extension messages, but probably need to be supplemented
with additional on-farm studies in other environments. More research is needed on
the functioning of farmer information networks and community group learning
and diffusion mechanisms, as these are relevant for agroforestry practices. Such
information, at subregion or local levels, should be used to guide decisions about
the balance of extension communication effort among individuals, groups and
across communities.
Table 8.2. Policy recommendations to promote adoption of specific agroforestry practices.
Agricultural Tenure systems
Agroforestry practice Germplasm policy input supply Product markets and regulation Extension strategies
Hedgerow inter- Reduce cost/improve access None Identify market demand None Target labour-abundant
cropping (western to planting materials; increase for by-products, especially farms, sloping, erosive
Kenya) species diversification fodder and fuelwood land, better soils
Improved fallows Expand seed supply; Promote where Promote demand for None Target farmers with
(western Kenya) reduce cost/improve nitrogen fertilizer high-value crops; depleted soils, with off-
access to planting is costly or assess fuelwood farm income, or those
materials; increase unavailable; markets who cannot afford
species diversification increase supply mineral fertilizers
of phosphorus
Improved fallows Expand seed supply and Promote where Strengthen basic Encourage Support diffusion net-
(eastern Zambia) reduce costs; promote nitrogen fertilizer grain marketing community works. Target farmers
small-scale tree is costly and system norms/regulations with depleted soils and
nurseries; increase unavailable to control free those who cannot afford
species diversification grazing mineral fertilizers
Multipurpose trees Promote small-scale None Provide species None Target boundary
(western Kenya) tree nurseries; increase information for niche; promote use
seed supply for new processors and of farmer criteria for
species; increase consumers; link selecting species
species diversification smallholders to
Promoting New Agroforestry Technologies

markets
Fodder trees Expand seed supply; Promote where Stabilize milk None Support diffusion
(central Kenya) support small-scale dairy meal is markets; strengthen networks and nursery
tree nurseries and costly, low quality livestock health groups; provide man-
group seed-sharing or unavailable services agement assistance
165
166 S.J. Scherr and S. Franzel

The most suitable institutional arrangements to foster useful interaction

between agroforestry research and extension will vary, depending upon the density
and type of existing extension and farmers’ organizations and the commitment to
field-based operations by research organizations. That adaptive research and exten-
sion represent points along a continuum, rather than distinct activities, suggests
that integrating research and extension efforts should receive high priority. The
principle of linking research and extension efforts by working together in specific
sites around core on-farm research activities, rather than through centralized
bureaucratic mechanisms, is strongly endorsed.

Acknowledgements

We are grateful for the advice and input from Frank Place of ICRAF, and comments
from an anonymous IFPRI reviewer.

References
Bruce, J. and Fortmann, L. (1989) Agroforestry Priorities for Forestry and Agroforestry Policy
tenure and incentives. Paper No. 135. Land Research: Report of an International
Tenure Center, University of Wisconsin, Workshop, 9–12 July 1991, IFPRI,
Madison, Wisconsin. Washington, DC.
Buck, L. (1990) Planning agroforestry extension Howard, J.A., Rubey, L. and Crawford, E.W.
projects, the CARE International Approach (1997) Getting technology and the technol-
in Kenya. In: Budd, W.W., Ducchart, I., ogy environment right: Lessons from maize
Hardesty, L.H. and Steinder, F. (eds) development in southern Africa.
Planning for Agroforestry. Elsevier Science International Association of Agricultural
Publishers B.V., Amsterdam, pp. 101–131. Economists Conference, 10–16 August 1997,
Crowley, E.L., Soule, M.J. and Carter, S.E. Sacramento, USA.
(1996) Off-farm Income and Farming in ICRAF (1997) Annual Report 1996. ICRAF,
Western Kenya. Tropical Soil Biology and Nairobi.
Fertility Program/International Centre for ILRI (2000) CGIAR Systemwide Livestock
Research in Agroforestry, Nairobi, Kenya. Programme: Biennial Report 1999–2000.
Dewees, P.A. and Scherr, S.J. (1996) Policies and ILRI, Nairobi, Kenya
Markets for Non-timber Tree Products. Kerkhof, P., Foley, G. and Barnard, G. (1990)
Environment and Production Technology Agroforestry in Africa: A Survey of Project
Division Discussion Paper No. 16. Experience. Panos, London.
International Food Policy Research Leakey, R.B. and Newton, A.D. (1994)
Institute, Washington, DC. Domestication of Tropical Trees for Timber
Ellis, F. (1988) Peasant Economics. Cambridge and Non-timber Products. MAP Digest 17.
University Press, Cambridge. UNESCO, Paris.
FAO (1998) FAOSTAT statistics database. Merrill-Sands, D.M., Biggs, S.D., Bingen, R.J.,
[Link] Ewell, P.T., McAllister, J.L. and Poats, S.V.
Franzel, S., Jaenicke, H. and Janssen, W. (1996) (1991) Institutional considerations in
Choosing the Right Trees: Setting Priorities for strengthening on-farm client-oriented
Multipurpose Tree Improvement. ISNAR research in national agricultural research
Research Report No. 10. ISNAR, The systems: lessons from a nine-country study.
Hague. Experimental Agriculture 27(4), 343–375.
Gregersen, H., Oram, P. and Spears, J. (1992) Oram, P. (ed.) (1993) Priorities for forestry and
 Promoting New Agroforestry Technologies 167

agroforestry policy research in sub-saharan Special Publication Number 51. Soil

Africa. Summary of a regional workshop co- Science Society of America and American
sponsored by ICRAF, IFPRI and CIFOR, Society of Agronomy, Madison, Wisconsin,
30 November–4 December 1992, Nairobi, pp. 1–46.
Kenya. Scherr, S.J. (1990a) The diagnosis and design
Place, F. (1995) The role of land and tree tenure approach to agroforestry project planning
on the adoption of agroforestry technologies and implementation: examples from western
in Zambia, Burundi, Uganda, and Malawi: Kenya. In: Budd, W.W., Duchhart, I.,
a summary and synthesis. Land Tenure Hardesty, L.H. and Steiner, F. (eds)
Center, University of Wisconsin, Madison, Planning for Agroforestry. Elsevier,
Wisconsin. Amsterdam, pp. 132–160.
Place, F. and Dewees, P. (1999) Policies, incen- Scherr, S.J. (1990b) Evaluating institutional
tives, and the adoption of improved fallows. capacity for agroforestry research. In:
Agroforestry Systems 47, 323–343. Echeverria, R.G. (ed.) Methods for
Place, F. and Kindt, R. (1997) The complexities of Diagnosing Research System Constraints and
supply of tree germplasm: the need for pol- Assessing the Impact of Agricultural Research.
icy interventions. In: ICRAF, International Vol. I. Diagnosing Agricultural Research
Workshop on Policy Aspects of Tree System Constraints. International Service to
Germplasm Demand and Supply. National Agricultural Research, The Hague,
Unpublished proceedings of a workshop pp. 179–208.
organized at ICRAF, Nairobi, Kenya, 6–8 Scherr, S.J. (1991) On-farm research: the chal-
October 1997. ICRAF, Nairobi. lenges of agroforestry. Agroforestry Systems
Raintree, J.B. (1991) Socioeconomic Attributes of 15(2–3), 95–110.
Trees and Tree Planting Practices. FAO, Scherr, S.J. (1995) Economic factors in farmer
Rome. adoption of agroforestry: patterns observed
Raussen, T., Ebong, G. and Musiime, J. (2001) in western Kenya. World Development 23(5),
More effective natural resources management 787–804.
through democratically elected, decentralized Scherr, S.J. and Current, D. (1999) Incentives for
government structures in Uganda. agroforestry development: experience in
Development in Practice (11)4, pp. 460–470. Central America and the Caribbean. In:
Richards, M. and Asare, A. (1998) Incentives for Sanders, D.W., Huszar, P.C., Sombatpanit,
Ghanaian cocoa farmers to keep timber S. and Enters, T. (eds) Incentives in Soil
trees. Policy Briefing Note. Forestry Conservation: From Theory to Practice.
Research Project–UK Department for Inter- Science Publishers, Enfield, New
national Development–Overseas Develop- Hampshire, pp. 345–366.
ment Institute, London. Scherr, S.J. and Müller, E. (1990) Evaluating
Ruvuga, S.A., Otsyina, R., Kusekwa, M., Gama, agroforestry interventions in extension pro-
B., Nyadzi, G.I. and Msangi, R.B. (1999) jects. Agroforestry Systems 11(1), 23–44.
Agroforestry and Environment, Proceedings of Shepherd, K.D. and Roger, J.H. (1991)
the Second Workshop on Sensitisation and Approaches to On-farm Testing and
Participatory Planning on Agroforestry and Evaluation of Agroforestry Technology.
the Environment for Policy Makers and Working paper No. 67. ICRAF, Nairobi.
other Partners in Tabora Region, 3–4 May Simons, A.J. (1996) Seed orchards and breeding.
1999, Tumbi, Tanzania. ICRAF, Shinyanga, In: Gliricidia sepium. Genetic Resources for
Tanzania. Farmers. Tropical Forestry Paper No. 33.
Sanchez, P.A., Shepherd, K.D., Soule, M.J., Oxford Forestry Institute, Oxford, UK.
Place, F.M., Buresh, R.J., Izac, A.N., Uzo Spears, J., Oram, P., Byron, N., Scherr, S. and
Mokwunye, A., Kwesiga, F.R., Nderitu, Izac, A.M. (1994) A review of tropical
C.G. and Woomer, P.L. (1997) Soil fertility forestry and agroforestry problem areas and
replenishment in Africa: an investment in policy research needs and the planned
natural resource capital. In: Buresh, R.J., response of the CGIAR system. CIFOR
Sanchez, P.A. and Calhoun, F.G. (eds) Working Paper No. 5, CIFOR, Bogor,
Replenishing Soil Fertility in Africa. SSSA Indonesia.
168 S.J. Scherr and S. Franzel

Tomich, T.P., Thomas, D.E., Kusumanto, Y. and Paper No. 109. Agroforestry Research
van Noordwijk, M. (1998) Policy research Network for Africa. ICRAF, Nairobi.
for sustainable upland systems in southeast Weber, M.T., Staatz, J., Holtzman, J., Crawford,
Asia. ICRAF Southeast Asian Regional E. and Bernsten, R. (1988) Informing food
Research Program, Bogor, Indonesia. security decisions in Africa: empirical
Tyndall, B. (1996) The Socioeconomics of analysis and policy dialogue. American
Grevillea robusta Within the Coffee Land- Journal of Agricultural Economics 705,
use System of Kenya. AFRENA Research 1044–1052.
 Assessing Adoption Potential:
Lessons Learned and
Future Directions
9
S. FRANZEL1 AND S.J. SCHERR2
1International Centre for Research in Agroforestry, PO Box 30677,
Gigiri, Nairobi, Kenya; 2Agricultural and Resource Economics
Department, University of Maryland, College Park,
MD 20742, USA

Land degradation, low soil fertility and lack of quality livestock feed are increasingly
recognized as key problems for farmers throughout sub-Saharan Africa. The
research reported in this volume documents the experiences of farmers in testing
and assessing the potential of agroforestry practices to address these problems and
increase their incomes. The volume presents five case studies from three contrasting
zones: a humid, highland area in central Kenya located close to major markets; a
humid, highland area in western Kenya with poorer soils and further from markets;
and a subhumid plateau area in eastern Zambia with degraded soils, lower popula-
tion density, and far from major markets. Maize, the main food staple in eastern
and southern Africa, is the most important crop component of all of the agro-
forestry practices examined in the case studies.
Three case studies assess soil fertility interventions, including improved tree fal-
lows and hedgerow intercropping (Table 9.1). In eastern Zambia, the number of
farmers using improved, 2-year tree fallows rose from 20 in 1993 to about 10,000
in 2000. Main trees planted were Sesbania sesban, which requires a nursery, and
Tephrosia vogelii, which may be direct seeded. In western Kenya, farmer-managed
experiments in the early 1990s helped researchers identify a range of brief, one-sea-
son fallow species, which, by 1999, were being planted by several thousand farmers.
The most popular species were Crotalaria grahamiana, Crotalaria ochroleuca, and
tephrosia, all of which were direct seeded. Hedgerow intercropping in western
Kenya, while not contributing significantly to improving soil fertility and crop
yields in the short run (1–4 years), was shown to be an important means of reduc-
ing soil erosion, and can thus improve fertility and yields in the long run. The main
species included Calliandra calothyrsus and Leucaena leucocephala. One case study,
also from western Kenya, assesses the introduction of trees for wood production,

© CAB International 2002. Trees on the Farm (eds S. Franzel and S.J. Scherr) 169
Table 9.1. Recommendation domains: conditions most suited to selected agroforestry practices.
Other
170

countries Other
Main Payback where factors
Practices and problem Agricultural Land Labour Cash periodb practice is affecting
area addressed Ecoregion Soils intensitya requirements requirements requirements (years) being used adoption
Improved fallows, Nitrogen Subhumid Sesbania Suited for Relatively High, planting None – 3 Cameroon,
eastern Zambia deficiency areas, poorly suited medium zone high is at peak minimal Uganda,
(Sesbania sesban in soil unimodal to sandy, period. Less of Tanzania,
and Tephrosia rainfall shallow soils a problem if Malawi,
vogelii) trees are direct Zimbabwe
seeded
Improved fallows, Nitrogen Subhumid Suited for Medium Medium, only None – 2 Uganda
western Kenya deficiency areas, medium zone, trees that are minimal
(Crotalaria spp. in soil bimodal some direct seeded
and Tephrosia rainfall potential for are suitable
vogelii) high zone
Hedgerow inter- Soil erosion Humid, Suited to Suited for Minimal High, planting Low 3+ Throughout Females
cropping, western subhumid erosive, high zone, and pruning sub- may have
Kenya (Leucaena areas. Up sloping some are at peak Saharan problems
S. Franzel and S.J. Scherr

leucocephala and to 1900 m areas potential in periods. Delays Africa pruning

Calliandra in altitude medium zone in pruning may trees
calothyrsus) in Kenya reduce crop
yields
Boundary planting Fuelwood Grevillea: Suited for Minimal Low, planting is Low, only for 5+ Uganda,
of upper-storey shortage, humid and high and at peak period seedlings Rwanda
trees, western poles and subhumid medium zones but problem not
Kenya (Grevillea timber for highlands. where farmers serious as few
robusta and construction Casuarina: lack access to trees are
Casuarina and cash humid, trees off the planted
junghuhniana) subhumid and farm
semiarid
areas
Fodder trees, Fodder Humid, Susceptible Suited for Minimal, Low, planting None – 2 Uganda, Suited for
central Kenya shortages in subhumid to frost, high zone, especially is during peak minimal Rwanda, cut and
(Calliandra dairy areas. Up to waterlogging some when planted period but Burundi, carry dairy
calothyrsus) enterprises 1900 m in potential for on boundary relatively few Tanzania, systems.
altitude in medium trees are Zimbabwe May have
Kenya planted potential in
other, more
extensive
dairy
systems
aLow: farmers use natural fallows to restore fertility. Medium: most farmers fallow but fallow periods not sufficient to restore fertility. High: contin-
uous cultivation is common, most fallows are brief, that is, one or two seasons.
bThe payback period is the time required for revenue earned to cover the cost of the investment.
Assessing Adoption Potential
171
172 S. Franzel and S.J. Scherr

such as Grevillea robusta and Casuarina junghuhniana, planted along boundaries

and in other arrangements. The last case study reports on the uptake of calliandra in
the mid-1990s, which farmers planted to provide fodder for increased milk produc-
tion in central Kenya. By 2000, about 4000 smallholder dairy farmers had planted
calliandra and numbers expanded rapidly over the next 3 years.
This chapter reviews the adoption potential of the practices within and outside
the areas they were tested and assesses policy support needed to promote their adop-
tion. Next, we summarize lessons learned for improving the effectiveness and effi-
ciency of developing and disseminating new practices, focusing on the
complementarity of different types of on-farm trials, the promotion of farmer par-
ticipation, adaptation and innovation, and on creating adaptive research and dis-
semination networks. Finally, suggestions are made concerning areas for future
research.

Adoption Potential of Promising Agroforestry Practices

Benefits to farmers

Agroforestry practices, as shown in this volume, can increase farmers’ incomes and
make important contributions to solving some of their most important problems.
For example, farmers in Zambia planting improved fallows can increase their
annual net farm income from maize, their most important enterprise, by US$130
(more than twofold). In central Kenya, calliandra substituted for purchased dairy
meal increases household income by US$220 per household, an increase of about
10%. Moreover, both technologies have important by-products and services that are
difficult to quantify and value – improved fallows provide firewood and improve
the soil’s structure and organic matter content and calliandra improves milk quality,
provides firewood and helps conserve the soil.
The other two practices examined in this volume, hedgerow intercropping and
upper-storey trees, offer more modest, but still important, benefits. While the con-
tributions of hedgerow intercropping to soil fertility are minimal in the short term,
this practice is effective in curbing soil erosion and providing useful by-products,
such as fodder and fuelwood. Grevillea trees planted along boundaries in western
Kenya provide fuelwood and timber, and earn net benefits as high as the maize they
displace. In addition, the trees have other difficult-to-quantify benefits as wind-
breaks, boundary markers and in controlling erosion.
The studies also demonstrate that farmers value agroforestry practices for three
‘hidden’ reasons aside from the products and services they provide. First, agro-
forestry reduces the considerable risks that farmers face from input markets by pro-
viding home-produced substitutes for purchased inputs such as mineral fertilizer or
dairy concentrates. For example, farmers in central Kenya complain that concen-
trates are, at times, not available, are difficult to transport because of poor road net-
works, and are of unreliable quality. With home-produced calliandra, none of these
problems arise.
 Assessing Adoption Potential 173

Secondly, agroforestry practices help reduce risk from uncertain rainfall. For
example, in Zambia the benefits of improved fallows are spread over a 2–3-year
period, whereas nitrogen fertilizer provides benefits for only a single year. A farmer
experiencing a crop failure would lose her cash investment in fertilizer, US$154
ha−1, whereas a farmer planting an improved fallow would lose only her investment
in planting the trees, which amounts to one-quarter of the fertilizer cost. In addi-
tion, tree fallows improve the soil’s structure and organic matter content, thus
enhancing its ability to retain moisture during drought years.
Finally, farmers in the case studies expressed a strong preference for investing
their labour, through agroforestry practices, rather than their cash, which in most
cases is scarce or unavailable for investment. Farmers in Zambia emphasized that
the profitability of mineral fertilizer is irrelevant to them – they simply do not have
the cash to purchase it. In western Kenya, farmers prefer to plant trees on their
farms for fuel and construction wood rather than purchasing wood from outside.
Even in central Kenya, where incomes are higher and cash more available, farmers
greatly appreciate the cash that they can save by feeding calliandra as a high-protein
supplement to their cows instead of purchasing dairy meal.

Suitability of practices outside areas where they were tested

The case studies indicate the usefulness of agroforestry practices in the sites in
which they were tested. But how likely are they to be of use to farmers in other
areas? As discussed in Chapter 2, the boundary conditions of a technology are
defined by identifying the variables that are most important in determining who
will and will not use the practice. Information on variables affecting biophysical
performance, profitability and acceptability are thus critical.

Improved tree fallows

Improved tree fallows have greatest potential in areas where farmers have ‘some
experience with the struggle to maintain soil fertility’ but still have lands that they
leave fallow (Raintree and Warner, 1986). Eastern Zambia and large portions of
Zambia, Zimbabwe, Tanzania and central Malawi fit this picture. But the research
presented on improved fallows in western Kenya shows that the technology has
potential in certain densely populated areas as well; that is, where rainfall is
bimodal, permitting farmers to fallow for one season per year. The technology is
thus of relevance to farmers throughout the highlands of Ethiopia, Kenya, Tanzania
and Uganda, where most of the people of these countries live.
Type 1 trials (on-farm trials designed and managed by researchers) in four
countries of southern Africa showed that sesbania-improved fallows performed
well across a range of environments, but not on sandy soils (because of nematode
attacks), on shallow soils (because of mortality during the dry season), or in frost-
prone areas. Tephrosia appears to be more widely adapted than sesbania. Cajanus
cajan is of limited use because it is susceptible to browsing by livestock, which
174 S. Franzel and S.J. Scherr

range freely during the dry season in most of southern Africa. By 2000, improved
tree fallows were being planted by roughly 25,000 farmers in four countries in
southern Africa, two countries in eastern Africa, and Cameroon. Each of these
plantings involved ICRAF and its partners; in addition, improved fallows were
reported to be planted by farmers in North West Province, Cameroon, without
ICRAF involvement.

Hedgerow intercropping
Widely promoted for improving soil fertility, hedgerow intercropping has not been
found to increase crop yields under farmers’ conditions in eastern and southern
Africa. But it still has an important role to play in curbing soil erosion and provid-
ing important by-products, such as fuelwood and fodder. It is suited to farmers on
sloping, erosive land where population densities are fairly high and labour is avail-
able for planting and pruning the hedges. Hedgerow intercropping is only suitable
in humid and subhumid areas, because in drier areas trees and crops compete for
available moisture.

Upper-storey trees for wood production

Often planted on boundaries, these are appropriate for farmers in a range of cir-
cumstances. As forests and other off-farm sources of wood become scarce, farmers
in many areas of eastern and southern Africa have started planting trees on their
own farms, because the cost of growing trees is less than the cost of collecting wood.
But there is often a lag in this transition, primarily because farmers lack information
or planting material. Research and extension programmes throughout eastern and
southern Africa are helping farmers to plant upper-storey trees that meet their needs
and circumstances – that is, those that are compatible with crops and that provide
high-quality wood for fuel and construction. Eucalyptus spp. and grevillea are prob-
ably the two most commonly planted species for wood production, the former in
woodlots and the latter along boundaries.

Calliandra as a fodder tree

This is suitable for smallholder dairy farmers throughout eastern and southern
Africa. In Kenya, there are about 625,000 smallholder dairy farmers – with
improved breed or cross-bred dairy animals fed in zero- or minimum-grazing sys-
tems. Smallholder dairy farmers are less numerous in other countries of eastern and
southern Africa but are found around virtually every city and large town in the
region. By 1998, calliandra was being used by several thousand farmers around
seven cities of Kenya, Uganda, Zimbabwe and Tanzania, each with support from
ICRAF and partner institutions. Fodder trees are probably profitable only when fed
to improved or cross-bred dairy animals, as indigenous-breed animals convert only
small percentages of additional feed to increased milk production.
 Assessing Adoption Potential 175

Wealth level and gender

Many policy makers are particularly interested in whether poor and female farmers
can adopt income-increasing technologies such as agroforestry practices. Wealth
level can influence adoption in several ways. Poor farmers may be more risk averse,
have less access to information, have a higher discount rate and thus a shorter-term
planning horizon, and have less capacity to mobilize resources (Hoekstra, 1985;
CIMMYT, 1993). In fact, most adoption studies find strong correlations between
wealth and adoption. Results from Zambia confirm that the higher the wealth sta-
tus, the greater the use of improved fallows. But, on the other hand, about 20% of
the poor and very poor households in four sample villages planted improved fallows
during the first 4 years of testing, which suggests that there are no absolute barriers
preventing low-income farmers from doing this. Moreover, low-income farmers are
more likely to adopt improved fallows than mineral fertilizers because the fallows
require no cash input.
Female farmers provide most of the labour for African food production
(Quisumbing et al., 1995) and the percentage of households that are female-headed
ranges from 18% in central Kenya, to about 30% in Zambia and 50% in western
Kenya. One would expect that females’ use of agroforestry practices would be lower
than that of males, for two reasons. First, female household heads tend to have
lower incomes than male household heads (Quisumbing et al., 1995). Thus,
females would be less likely to test and adopt improved fallows for the reasons men-
tioned above concerning poor households. Secondly, those choosing participants for
experiments and distributing planting material, usually extension staff, tend to
select men. Thus, even if the technology itself is gender neutral, adaptive research
and dissemination programmes are often biased towards males (CIMMYT, 1993).
Data from Zambia indicate that 32% of the males and 23% of the females in
the sampled villages were planting improved fallows, and that there was no signifi-
cant difference between the two proportions. In fact, in one of the villages where
womens’ groups were particularly strong, the proportion of females using improved
fallows was higher than that of males. Moreover, whereas single females are often
disadvantaged relative to female heads of household whose husbands live away
(Bonnard and Scherr, 1994), the Zambia data showed that the same proportions of
these two groups were testing the technology. Available data thus suggest that
improved fallow practices are suitable for poor and female farmers.

Policy Support for Agroforestry Adoption

A conducive policy environment is important for the adoption of any agricultural

technology, but there are several reasons why policies are particularly critical for
agroforestry practices. First, the production and distribution of planting material for
trees is more complicated than for crops, because trees often require nurseries and
because most countries have few mechanisms for the production and distribution of
tree germplasm. Moreover, agroforestry practices tend to be more complex than
176 S. Franzel and S.J. Scherr

many agricultural practices, because they involve mixing trees with crops and
because several years are often required before they generate returns. Agroforestry
practices thus may rely on supportive policies for their development and diffusion
to a greater extent than do other agricultural practices.
The case studies in this volume suggest that germplasm policy must focus more
on reducing the cost of planting material, especially important for technologies
such as improved fallows, which require large numbers of seedlings per unit area
and over time. Small-scale tree nursery promotion is needed for improved fallows in
Zambia and for timber, pole and fodder trees in Kenya. Whereas centralized, vil-
lage-level nurseries offer some advantages for training, seed dissemination and qual-
ity control early in the extension process, individual farm nurseries appear to be
more viable over the longer term.
While the case studies suggest that an improved policy environment can help
facilitate agroforestry adoption, they do not suggest that subsidies and other direct
incentives are needed. Similarly, the provision of production credit for agroforestry
should not be an important priority for policy makers.
For the practices studied, there was little observed effect of tenure and regula-
tory policy factors on agroforestry adoption or farm profitability. In southern Africa,
free-grazing livestock damage trees during the dry season but this problem is being
addressed by using non-palatable tree species. Moreover, some exploratory work on
changing community-level norms and regulating grazing appears to be promising.

Increasing the Effectiveness of Technology Development and

Dissemination
The case studies described in this book reflect the evolution of a new approach to
accelerate the process of technology development and dissemination. This approach
has promise not only for the development of agroforestry practices, but also for
other types of agricultural and natural resource management innovations in highly
heterogeneous environments such as those found in Africa. The key elements of this
approach are:
● on-farm trials, with differing levels of researcher and farmer involvement, as a
central locus for collaborative technology development;
● farmer participation, adaptation and innovation as key inputs in technology
development; and
● adaptive research and dissemination networks.

On-farm trials as a central locus for technology development

On-farm trials greatly expand the potential contributions of research to technology

development. The case studies confirm the importance and complementarity of the
three types of on-farm trials: type 1, type 2 (trials designed by researchers and man-
 Assessing Adoption Potential 177

aged by farmers) and type 3 trials (trials designed and managed by farmers), which
are suited to different objectives. Where biophysical data are needed, type 1 trials
are generally best, because conditions can be controlled between treatments and
across farms. For data on farmer assessment and innovation, type 3 trials are pre-
ferred, because farmers plant and manage technologies in these trials as they wish.
Type 2 trials are useful for economic analysis, because farmers manage the trials and
because plot sizes are standardized, facilitating the measurement of labour use and
other inputs. The optimal sequencing of trial types depends on specific circum-
stances. For example, in Zambia, in the early 1990s, type 1 trials were used to assess
the effect of improved fallows on crop yields, and type 3 trials to monitor farmers’
assessments and innovations. In the mid-1990s, other type 1 trials were conducted
across four countries to assess biophysical constraints to adoption. In addition, type
2 trials were carried out to estimate yield response and economic returns to a range
of options under farmers’ management.
It is not always necessary to conduct all three types of trials, or to always
begin with type 1 trials. In three of the case studies, researchers started with type 3
trials, for different reasons. In the agroforestry tree trials in western Kenya, much
information was already available from on-station trials on the performance of the
five species under consideration; thus type 3 trials were started in which farmers
tested whether the species could be grown in different niches of the farm. For cal-
liandra in central Kenya, researchers started with type 3 trials as much was known
about growing the tree. Several years later, a type 2 trial was conducted to assess
the effect of calliandra on milk yields. In western Kenya, a type 2 sesbania
improved-fallow trial started before any work on the technology’s performance
had been conducted on station or on farm. Researchers were particularly inter-
ested in understanding farmers’ constraints to adoption, and feeding this informa-
tion back for further research. The trial demonstrated the potential for the
technology to increase returns and the main constraint fed back to research was
the problem of establishing sesbania when direct seeded. Type 1 trials were insti-
tuted to screen new species and, 2 years later, promising species were introduced
in type 2 and type 3 trials.
The case studies also highlight the complementarity of different research and
assessment methods. No single type of assessment, such as crop yield increases or
profitability, can alone explain farmer preference for an agroforestry practice.
Rather, an integrated assessment of biophysical performance, profitability and
acceptability is usually needed to explain farmers’ uptake of a practice. Assessments
of acceptability are particularly challenging because of agroforestry’s complexity,
that is, the length of the cycle of technologies (3+ seasons as opposed to single-sea-
son cycles) and the number and diversity of components (intercropping trees and
crops, as opposed to crops in pure stand). A range of different methods were found
to be effective for assessing farmers’ preferences during and following on-farm trials,
including monitoring farmers’ use of technologies, risk assessments, matrix ranking
and decision trees (Chapter 2).
Finally, the case studies show that although farmers have a greater degree of
control in type 3 trials, all three of the trial types need to be participatory. For
178 S. Franzel and S.J. Scherr

example, at farmer workshops to design the type 2 hedgerow intercropping trials in

western Kenya, farmers made important decisions on such experimental factors as
the spacing between trees, spacing between lines of trees and crops, and the crops to
grow between the hedges (Chapter 5).

Fostering farmer participation, adaptation and innovation

Indeed, the case studies demonstrate the importance of the participatory research
approach for improving the effectiveness of the design, implementation and evalua-
tion of on-farm research. Participatory research involves giving farmers control over
the research process, as in type 3 trials, in which farmers test new technologies as
they wish, adapting them to their specific needs and circumstances. Participatory
research also strengthens researcher–farmer partnerships, as when farmers and
researchers collaborate in the design, implementation and evaluation of a type 1 or
type 2 trial. Researchers contribute their scientific knowledge and farmers con-
tribute their indigenous technical knowledge and information about their needs,
preferences and circumstances.
A key lesson from the case studies is that farmers do not adopt practices as ini-
tially introduced; they adapt them to their needs and circumstances. Simply stated,
in agroforestry, researchers can rarely get a technology ‘right’ tinkering with it in
on-station research or even in type 1 and type 2 trials. In Zambia, for example,
farmers in type 3 trials made two important modifications in the improved fallows
they planted in the early 1990s, these modifications were then tested together by
researchers and farmers. One of the farmer modifications, bare-rooted seedlings,
was first used by a female farmer who obtained some left-over potted seedlings from
a type 1 trial and removed them from their pots so she could pack them in a basin,
which she carried on her head to her farm. Bare-rooted seedlings, grown in raised
beds, have now completely replaced the higher-cost alternative, potted seedlings. A
second modification was to intercrop trees with crops during the year of establish-
ment, instead of planting them in pure stand. This method economizes on land and
labour use relative to planting trees in pure stands and, by 1997, was used by about
42% of farmers planting improved fallows.
Similarly, in central Kenya, farmers determined the best niches for calliandra
on their own, in type 3 trials. Whereas researchers had expected that farmers would
plant calliandra in small plots, few did. Rather, farmers planted in other niches, on
internal and external boundaries, around the homestead, and intercropped with
napier grass.
The importance of farmer adaptation and innovation has several implications
for an effective technology generation programme. First, researchers need to estab-
lish a partnership with farmers in which the latter understand that their role is to
innovate, not simply to follow instructions, or worse, pretend to follow instruc-
tions while innovating in secret! In many areas, farmers are accustomed to working
with change agents who have a top-down, we-know-what-is-best-for-you attitude;
they are generally not used to working as partners with research and extension
 Assessing Adoption Potential 179

staff. Thus, a great deal of time is required to establish effective research partner-
ships with farmers.
Secondly, researchers need to present farmers with different options to test. For
example, in the type 2 trials on improved fallows in Zambia, farmers chose among
three tree species and two planting methods: pure stand and intercropping. In addi-
tion, many farmers also conducted type 3 trials in which they planted as they
wished, modifying operations such as spacing, timing of planting, or crop planted
after the fallow. The rationale for testing multiple options is fourfold. Some options
may not work; thus it is better to introduce several. Even if all work, farmers appre-
ciate diversifying as a means for coping with the risk that one of the options may
fail at some future time; for example, when a species succumbs to disease. An addi-
tional reason for testing multiple options is that different farmers have different
preferences, resource constraints and circumstances that lead them to choose differ-
ent options. For example, some farmers in Zambia prefer tephrosia, which econo-
mizes on labour but gives a lower crop yield response than sesbania. Others prefer
sesbania and are willing to invest the extra labour required in order to get a higher
return. Finally, the option approach is important for enhancing biodiversity at the
landscape level for ecological reasons, as well as at the household level for economic
reasons.
Thirdly, involving farmers early on in type 3 trials is important for encouraging
modification and for providing feedback to researchers. Many researchers prefer
testing technologies on research stations and releasing them to farmers only when
they are ‘proven’. But the experiences in this volume show that what is optimal to
researchers may not be optimal to farmers and, indeed, in the above-mentioned case
of sesbania and tephrosia in Zambia, different farmers in the same villages may find
different technologies to be optimal. Farmers should thus be encouraged to test new
technologies on their own before, or at the same time as, they are tested on research
stations. In western Kenya, a type 2 improved fallow trial was started before the
technology was tested on station. Following planting, it evolved into a type 3 trial,
as farmers modified the technology in several ways, changing such operations as
spacing, the length of fallow and the crop planted after the fallow. Several farmer
modifications, such as increasing the plant density, led to new researcher-designed
trials. At the end of the on-farm testing period, the technology was not deemed fit
for disseminating. Nevertheless, the lessons learned were critical and fed back to on-
station research, where new species were identified and a successful on-farm
research and dissemination programme was launched several years later.
Fourthly, a monitoring system needs to be in place so that researchers, exten-
sionists and other farmers can learn from innovating farmers. Our experience is that
questionnaires are of little value in identifying innovation. Rather, informal interac-
tion between researchers and farmers, and farmer-to-farmer visits, are much more
useful. Monitoring surveys involving questionnaires do have a role – they are useful
for determining the extent to which different farmers use different practices. For
example, monitoring surveys in Zambia have documented the increased use of
tephrosia, relative to sesbania, and the increase in intercropping, relative to planting
trees in pure stand.
180 S. Franzel and S.J. Scherr

Building adaptive research and dissemination networks

The on-farm research experiences in Zambia demonstrate a new paradigm of

research–extension collaboration, as explained in Chapter 2. An interactive,
rather than linear, sequential mode of on-farm experiments exploits the comple-
mentary strengths of farmers, extensionists and researchers at a reasonable cost.
Building a coalition of organizations to conduct on-farm research and dissemina-
tion together appears to be more effective and efficient than leaving each to work
independently.
Whenever a farmer tests a new technology, dissemination can be said to take
place as neighbouring farmers are exposed to it and take it up. Similarly, in a dis-
semination programme, farmers using a technology for the first time are conducting
research to see if new practices work effectively for them. That adaptive research
and extension represent points along a continuum, rather than distinct activities,
suggests that integrating research and extension efforts should receive high priority.
The case studies show the importance of integrated approaches to on-farm research
and dissemination. In eastern Zambia, for example, representatives of research,
extension, NGOs and farmer groups meet twice a year to plan, implement and
evaluate on-farm research, training and dissemination activities. There the network
has already had important impacts, including (Cooper, 1999):
● reduced cost of conducting on-farm research, as field-based extensionists and
NGOs establish and monitor on-farm trials;
● enhanced breadth of input into and relevance of the research;
● expanded range of sites under experimentation with relatively little additional
cost;
● partners increasingly well-informed on key aspects of technology options and
better placed to disseminate technologies and respond to farmer feedback;
and
● partners have developed a sense of involvement, enthusiasm and ownership of
promising innovations.
The most suitable institutional arrangements to foster useful interaction
between agroforestry research and extension will vary, depending upon the density
and type of existing extension and farmers’ organizations and the commitment to
field-based operations by research organizations. The principle of linking research
and extension efforts by working together in specific sites around core on-farm
research activities, rather than through centralized bureaucratic mechanisms, is
strongly endorsed.
Some of the lessons learned from the case studies for developing a successful
adaptive research and dissemination network are that:
1. Needs and responsibilities of each actor must be clearly defined.
2. Testing should begin with a few (five) pioneer farmers who are risk-takers.
Numbers can be greatly increased in the following seasons, based on the experiences
of the initial testers.
 Assessing Adoption Potential 181

3. A range of options within and across technologies should be tested, to minimize

the risk of any single option or technology failing.
4. The network can begin informally, without a leader or host organization, and
should evolve naturally. But participants should meet regularly (twice a year) to
review progress and plan activities.
5. On-farm trials should be jointly planned by research, extension and representa-
tives of farmer groups.
These lessons about adaptive research and dissemination networks, highlighted
in the Zambia case study (Chapter 3), can be applied to all of the technologies stud-
ied. To promote this new paradigm on a larger scale may require policy makers to
effect institutional changes in programme structure and staff incentive and perfor-
mance evaluation systems in both research and extension organizations. Staff train-
ing is also required in participatory on-farm research and dissemination methods
adapted to agroforestry.

Future Directions

The experiences reported in this book demonstrate the importance of assessing the
adoption potential of agroforestry practices and suggest that more such studies are
needed in the future, for several reasons. First, such assessments improve the effi-
ciency of the technology development and dissemination process, by feeding back
information on farmers’ problems, modifications and preferences to research, exten-
sion staff and policy makers. The importance of feedback was especially critical in
the case of improved tree fallows in western Kenya, where assessments of adoption
potential helped guide a research programme that has been successful in developing
improved fallow practices for thousands of farmers (Chapter 4). Secondly, the
assessments help document the achievements in developing and disseminating new
practices, demonstrating the impact of investing in technology development and
dissemination. For example, Chapter 7 shows that the potential benefits from
adopting calliandra as a fodder tree amount to US$139 million year−1 in Kenya’s
smallholder dairy sector. Thirdly, because the activities are conducted with partner-
institutions, they facilitate interdisciplinary and inter-institutional cooperation. The
adaptive research and dissemination network in Zambia is an excellent example of
such cooperation (Chapter 3). Finally, the assessments help to identify the factors
contributing to successful technology development programmes as well as the con-
straints limiting their achievements.
Several critical research themes emerge from all of the case studies examined in
this volume. Future assessments of adoption potential need to take advantage of
farmers’ increased experience with agroforestry practices. Whereas the analyses in
this study are primarily at the plot and household level, increased adoption will per-
mit assessment of impacts on community and regional scales. Moreover, whereas
most of the assessments in this volume concern economic and crop productivity
gains, more information is needed on social and environmental impacts. There is a
182 S. Franzel and S.J. Scherr

particular need to examine the environmental impacts of agroforestry practised on a

landscape scale, for example, on hydrological flows or patterns of sedimentation.
Research also needs to continue to establish the boundary conditions of differ-
ent agroforestry practices, focusing on the main biophysical and socioeconomic
conditions that influence uptake. Improvements in the development of spatially
explicit databases and models should permit the use of geographical information
systems for assessing the boundary conditions of new technologies. Guidelines
based on such systems can be useful for helping change agents decide which tech-
nologies should be offered to farmers in specific areas and with particular features
and circumstances. Recommendations need to be prepared, documenting not that a
particular technique is ‘best’, which is an oversimplification, but, rather, what the
advantages and disadvantages are of different options in differing biophysical and
socioeconomic circumstances. Moreover, technologies always need to be adapted by
farmers to suit local needs and circumstances, and change agents need to document
these modifications for feedback to researchers and possible dissemination to other
farmers.
In all of the case studies, the main agroforestry products were substitutes for
purchased inputs, such as mineral fertilizer, dairy meal concentrates or timber.
Agroforesters need to follow closely the price and availability trends for these substi-
tutes, and understand how these will affect farmer adoption. More studies still need
to be done on farmer nutrient management strategy under different price and risk
regimes, and on the biophysical interactions of tree-derived nutrients and chemical
fertilizers. A similar focus and studies are needed for the dairy meal input market
and farmer use.
Finally, research is needed on approaches to scaling up the benefits of agro-
forestry across large areas; once participatory research succeeds and hundreds of
farmers adopt a practice, how can these benefits be spread? Cooper and Denning
(2000) highlight key elements of scaling up the impact of agroforestry research, and
the need to compare systematically different options. For example, access to
germplasm was a common constraint in the case studies; more research is needed to
assess ways of developing decentralized, village-level germplasm production and dis-
tribution systems. Research is also needed to increase the number of tree species
used in each practice, as a means of enhancing biodiversity as well as helping farm-
ers to diversify in order to reduce risk. Scientists also need to reduce the costs of
establishing trees, such as through direct seeding and bare-root seedlings as substi-
tutes for potted seedlings.
Research on dissemination pathways is critical, and should focus on the func-
tioning of farmer information networks and community group learning and diffu-
sion mechanisms. Such information, at subregion or local levels, should be used to
guide decisions about the content and balance of extension communication efforts
among individuals, groups and across communities. Efforts are also needed to hand
over many of the activities in assessing adoption potential to local institutions, such
as farmer groups and community-based, non-governmental organizations. The
greater control farmers have over assessing adoption potential, the more responsive
technology generation activities will be to their needs.
 Assessing Adoption Potential 183

References
Bonnard, P. and Scherr, S. (1994) Within gender Report of the Agroforestry Dissemination
differences in tree management: is gender Workshop, 14–15 September 1999, ICRAF,
distinction a reliable concept? Agroforestry Nairobi, Kenya.
Systems 25, 71–93. Hoekstra, D.A. (1985) Choosing the discount
CIMMYT Economics Program (1993) The rate for analysing agroforestry systems/tech-
Adoption of Agricultural Technology: a Guide nologies from private economic viewpoint.
for Survey Design. CIMMYT (International Forest Ecology and Management 10,
Maize and Wheat Improvement Center), 177–183.
Mexico, D.F. Quisumbing, A.R., Brown, L.R., Feldstein, H.S.,
Cooper, P.J.M. (1999) Agroforestry: learning as
Haddad, L. and Pena, C. (1995) Women:
we go in Africa. In: Martin, G., Agama, A.
the Key to Food Security. Food Policy Report.
and Leakey, R. (eds) Cultivating Forests: the
International Food Policy Research
Evolution of Agroforestry Systems, Plant and
Institute, Washington, DC.
People Handbook, Issue 5. UNESCO, Paris,
pp. 30–33. Raintree, J. and Warner, K. (1986) Agroforestry
Cooper, P.J.M. and Denning, G.L. (eds) (2000) pathways for the intensification of shifting
Scaling up the Impact of Agroforestry Research. cultivation. Agroforestry Systems 4, 39–54.
 Glossary

adoption potential: The feasibility, profitability and acceptability of a practice, as

viewed from the farmers’ perspective.
agroforestry: A dynamic, ecologically based, natural resource management system
that, through the integration of trees on farms and in the agricultural landscape,
diversifies and sustains production for increased social, economic and environmen-
tal benefits for land users.
biomass: The weight of material produced by living plants. It is expressed in terms
of dry weight or fresh weight.
boundary conditions: The biophysical and socioeconomic circumstances that
affect the adoption potential of a practice. The boundary conditions of using cal-
liandra as a fodder tree may include secure land tenure, altitudes between sea level
and 1900 m in the tropics, and rainfall above 1000 mm.
break-even analysis: The calculation of the increase in net returns required to cover
the costs of establishment and maintenance of a practice. In practices aimed at
increasing crop yields, the break-even yield increase may be calculated, that is, the
yield increase required to cover costs.
contour: A line joining all places at the same altitude.
coppice: Cutting certain tree species close to ground level to produce new shoots
from the stump. Not all trees may be coppiced.
crown: The canopy or top of a single tree that carries its main branches and leaves
at the top of a stem.
direct seeding: Sowing seeds directly where they are to develop into mature plants.
discount rate: The rate at which 1 unit at a future date is converted to assess its
value today. Discount rates are usually applied to monetary values but can also be
applied to other values, such as units of labour.
economic analysis: Analysis of costs and returns from the perspective of society, as
contrasted with the perspective of the individual farmer. Inputs and outputs are

185
186 Glossary

valued at prices society incurs. For example, if the government subsidizes the price of
fertilizer to farmers, then its price in economic analysis would be the unsubsidized
price.
exotic: A plant growing anywhere outside its natural range.
fallow land: Land that is temporarily uncultivated.
financial analysis: Analysis of costs and returns from the perspective of the individ-
ual farmer. Inputs and outputs are valued at prices farmers face and profitability
measures represent the returns farmers would earn.
hedgerow intercropping: The practice of growing annual crops in the spaces
between rows of trees or hedgerows. This is sometimes called alley cropping or alley
farming.
herbaceous: A plant that is not woody.
improved tree fallow: Enrichment of natural fallows with trees or shrubs to
improve soil fertility.
indigenous: Native to a specified area, not introduced.
intercropping: Growing two or more crops in the same field at the same time in a
mixture.
net present value: A criterion for evaluating alternative investments that uses the
discounting formulae for a stream of costs and returns over a specified time period
to value the projected cash flows for each alternative at one point in time. In this
fashion, the net present value criterion directly accounts for the timing and magni-
tude of the projected cash flows.
niche: A place on a farm or in the landscape that is suited to a particular plant.
nitrogen fixing: Having the ability to convert nitrogen in the air into a form that
can be used by plants.
provenance: The place in which any stand of trees is growing or from which a
stock of seed originates.
pruning: The process of cutting back the growth of plants.
returns to labour: Returns (that is, earnings) expressed per unit of labour, such as
per work day or per hour. This measure is especially relevant for farmers whose
most scarce input is labour.
returns to land: Returns (that is, earnings) expressed per unit of land. This measure
is especially relevant for farmers whose most scarce input is land.
rhizobia: Type of bacteria with the capacity to invade the roots of certain species of
leguminous plants and to fix atmospheric nitrogen that is subsequently taken up by
the host plant.
 Glossary 187

root collar diameter: The diameter at the root collar, that is, the transition zone
between the root and the shoot of a plant.
shrub: A descriptive term, not subject to strict definition, referring to a perennial
plant with wooden stem or trunk. It may be distinguished from a tree in that it is
smaller and usually divided into separate stems from near the ground.
tree: A perennial plant with wooden stem or trunk.
type 1 trial: A trial that is designed and managed (implemented) by researchers on
farmers’ fields. It is similar to an on-station trial except that it takes place on farm-
ers’ fields.
type 2 trial: A trial that is designed by researchers, in consultation with farmers,
and implemented or managed by farmers.
type 3 trial: A trial that farmers design in order to test a new practice as they wish.
The trial may not include a control plot, or uniform plots laid out side-by-side,
unless the farmer wishes.
zero grazing: A livestock production system in which the animals are fed in pens or
other confined areas and are not permitted to graze.
 Index

acceptability of agroforestry practices 3, 6, alley cropping or farming see hedgerow

12–17, 23, 24, 29, 33 intercropping
of fodder shrubs, central Kenya anthropology 6, 31
136–137, 139–140
of hedgerow intercropping, western
Kenya 91, 93, 103–106 banana 128
bao game 28, 113, 116, 159
of improved fallows, western Kenya 69,
bare-rooted seedlings 40, 42, 44, 45,
73, 81
53–54, 55, 98, 102, 105, 129, 146,
of improved fallows, Zambia 53
178
of trees in farmer-designed agroforestry beans 96, 127, 128
trials, western Kenya Benin 81
see also expansion of use of agroforestry biophysical performance of agroforestry
practices practices
Acrisols 68, 91, 112 in agroforestry tree trials, western Kenya
adaptive research and dissemination 111, 113, 116, 118, 120
networks 4, 8, 21, 32, 55, 57, 59, assessment of 14–15, 17, 18, 22, 24,
121, 161, 180–181 26, 28, 33
adoption in fodder shrubs, central Kenya 131
definition of 134 in hedgerow intercropping, western
factors affecting adoption of agroforestry Kenya 90, 91, 93, 94–96, 98, 102,
trees, western Kenya 107
factors affecting adoption of fodder in improved fallows, western Kenya 66,
shrubs, central Kenya 134, 141 72, 73, 82
in improved fallows, Zambia 38, 44–49
factors affecting adoption of hedgerow
boundaries, planting on farm 1, 5, 7, 27,
intercropping, western Kenya 15
31, 105, 114, 128, 171, 178
factors affecting adoption of improved boundary conditions 12, 13, 17, 31, 33,
fallows, western Kenya 82–83 34, 119, 120, 129, 134, 138, 139, 159,
factors affecting adoption of improved 173, 182
fallows, Zambia 54–55, 58 Brassica oleracea (kale) 82, 156
agroforestry tree trials, farmer-designed, break-even analysis 23–24, 65, 70, 87, 89,
western Kenya 18, 19, 111–121, 93, 98–102, 102, 146
155, 156, 159, 169–170, 172–173, browsing 40, 44, 45, 59, 72, 83, 95, 98,
177 114–116, 119, 120
Alfisols 39, 44, 46 Burundi 171

189
190 Index

Cajanus cajan 38, 42, 44– 45, 53, 173 economic analysis 2
calcium 105 see also financial analysis
Calliandra calothyrsus 7, 26, 29, 32, 89, Emuhaya 67
91–92, 95–96, 105–106, 111–121, enterprise budget see financial analysis
125–144, 148–152, 155, 157–158, environmental impact and services 8, 13,
169–171 58, 96, 146, 155, 181
Cameroon 106, 170–171, 173 establishment of agroforestry trees 46
CARE International 161 Ethiopia 173
CARE-Kenya Agroforestry Extension Eucalyptus spp. 112–113, 116, 173
Project 91 exchange rate 100, 131
expansion of use of agroforestry practices
Casuarina junghuhniana 7, 111–121, 155,
27, 38, 53, 58, 119
173
agroforestry trees, western Kenya 119
cation exchange capacities 105
fodder shrubs, central Kenya 125–126,
cattle 39, 68, 112, 113
132–133, 136, 140
see also dairy cows hedgerow intercropping, western Kenya
churches 56, 147 93, 104, 106, 158
coffee 68, 105, 126, 127, 135 improved fallows, western Kenya 73,
community-based organizations 57, 147 83–84
see also non-governmental organizations improved fallows, Zambia 37, 53, 55
complexity of agroforestry practices 13, extension staff and programmes 2–3, 5–6,
18, 90, 175 9, 12, 13, 20, 21, 30, 33–34, 145, 147,
contours 26, 106, 128, 134 151, 173, 176–182
cost–benefit analysis see financial analysis agroforestry trees, Kenya, 111, 121
cotton 30 fodder shrubs, Kenya, 125, 136, 140–141
credit 71, 151, 153, 176 hedgerow intercropping, Kenya, 90,
crop residues 45 105–107
Crotalaria spp. 7, 84, 169–170 improved fallows, Kenya 84, 156
improved fallows, Zambia 41, 55–57, 59,
65
dairy cows 6, 24, 26, 32, 89, 94, 104, 106, policies 157–162
125–141, 154, 156, 171 role in assessing adoption potential of
dairy meal concentrate 25, 128, 137, 152, agroforestry 12, 13, 19, 20, 21,
155, 164, 182 22, 30, 31, 32, 33, 159–162
dambos 39
decision trees and decision tree modelling
fallow, natural 40, 54, 65, 66, 68, 70, 71,
6, 23, 28, 93, 104–105
79, 171
Desmodium intortum 140
farm models 23, 25, 43, 52
development projects, organizations and
farm size 39, 68, 71, 93–94, 105,
practitioners 2, 4–5 112–114, 126, 132, 134, 153
see also extension staff and programmes farmer-designed trials see type 3 trials
direct seeding 40, 46, 53, 58, 78, 80, farmer groups and organizations 3, 6, 8,
81–82, 83, 133, 146, 158, 178 19, 20, 21, 30, 34, 151, 160, 180
discount rate 50, 51, 64, 70, 102, 131, agroforestry tree trials, western Kenya
137, 150, 153, 175 111
dissemination see extension staff and fodder shrubs, central Kenya 140, 141,
programmes 147–148
donors 3 hedgerow intercropping, western Kenya
drought 44, 114–116, 119, 120 92
 Index 191

improved fallows, western Kenya 69, feed 1, 25, 104, 105, 113, 128, 137, 152,
119 155, 164, 169
improved fallows, Zambia 55, 56, 57, 58, feedback from farmers 12, 13, 14, 15, 16,
175 21, 22, 27, 34
farmer managed trials see type 2 and type 3 agroforestry trees, western Kenya 121
trials improved fallows, western Kenya 83–84
farmer-to-farmer diffusion 6, 27, 32, 139 fodder shrubs, central Kenya 126, 136
farmer-to-farmer visits and training 16, hedgerow intercropping, western Kenya
56, 59, 93, 136, 147, 159, 160, 179 106–107
farmer workshops 28 improved fallow, Zambia 55–57, 59
farmers’ assessments of agroforestry practices see also farmers’ assessments; farmers’
4–6, 177 innovations
of agroforestry trees, western Kenya Ferrasols 68, 91, 112
112–113, 115–119, 121 fertilizer 1–2, 25, 27, 90, 99–100, 103,
of fodder shrubs, central Kenya 130 112, 120, 153, 164, 173, 175, 182
of hedgerow intercropping, western in comparison with improved fallows,
Kenya 92–93, 98, 107 western Kenya 66, 68, 71
of improved fallows, western Kenya 81 in comparison with improved fallows,
of improved fallows, Zambia 40, 53–55, Zambia 38, 40–41, 43, 46, 49, 51,
59 52, 62–64
methods 14, 15, 16, 19, 21–22, 23, 27, financial analysis of agroforestry practices
28, 31, 33 2–6, 9, 13, 23–27, 31, 43, 152, 153,
farmers’ evaluations see farmers’ assessments 155, 176–179
farmers’ experiments see type 3 trials of alternative seed supply approaches
farmers’ innovations 4–5, 13, 14, 17, 148–151
29–30, 32, 177–179 fodder shrubs 125, 126, 130, 131, 140,
of fodder shrubs, central Kenya 137–39
136–137, 141 hedgerow intercropping 90, 91, 92–93,
of improved fallows, western Kenya 83 98–103, 106
of improved fallows, Zambia 44, 53–54 improved fallows, western Kenya 65,
farmers’ problems 12, 14, 23 66, 69, 70–71, 74–79, 80–82,
agroforestry trees, western Kenya 119 87–88
fodder shrubs, central Kenya 136 improved fallows, Zambia 38, 41, 43,
hedgerow intercropping, western Kenya 47–53, 62–64
90–91, 98 methods for conducting 12, 13, 16, 17,
improved fallows, western Kenya 66 23, 24–26, 28, 29, 31, 33
improved fallows, Zambia 40, 43, 46, returns to capital 70
59 returns to labour 49, 50, 52, 53, 70,
farming systems and farming systems 75–80, 87, 89, 93, 98
approach 4–5, 11, 12, 24, 112 returns to land 49, 50, 53, 74–78, 80,
feasibility of agroforestry practices 3, 6, 87, 89, 93, 98, 102, 106
13, 15, 17, 28 fires 44, 95, 163
agroforestry trees, western Kenya fodder 24, 71, 90, 115, 117–118, 120,
114–115 125–141, 147–148, 153
fodder shrubs, central Kenya 132–135 fodder shrubs, central Kenya 125–142,
hedgerow intercropping, western Kenya 146, 164, 165, 169, 171–173,
96–98 177–178, 181
improved fallows, western Kenya 81 fodder trees 6–7, 23, 156, 157, 171
improved fallows, Zambia 38, 44–47 food security 2, 8
192 Index

forests 1, 8, 145, 146, 153, 155, 156–157, hedges 69, 128

174 herbaceous legumes 58, 126
French beans 68 Heteropsylla cubana (psyllid) 96, 116, 120
fuelwood 1, 5, 7, 47, 53, 153–155, 170
in agroforestry tree trials, western Kenya
112, 115, 117–119 International Centre for Research in
in fodder shrubs, central Kenya 135, Agroforestry (ICRAF) 12, 20, 59,
138 128, 139, 140, 161, 162, 173
in hedgerow intercropping, western impact assessment 4, 28, 59
Kenya 91, 99–101, 102, 105 see also farmers’ assessments; financial
in improved fallows, western Kenya 66, analysis
71–78, 80 improved fallow 2, 6–7, 22, 151, 156, 158
in improved fallows, Zambia 47, 48 western Kenya 18, 23, 24, 31, 32,
65–88, 146, 153, 154, 155, 165,
177, 179, 181
gapping up Zambia 18, 19, 22, 23, 24, 25, 26, 27,
in improved fallows, western Kenya 69 28, 29, 30, 37–64, 146, 148, 154,
in improved fallows, Zambia 44, 45, 155, 157, 164, 165, 169–170,
46, 54 172–181
gender issues and women farmers 19, 20, incentives, extension staff 57
26, 29, 156–157, 160, 164 incentives, providing to farmers 2–4, 8,
agroforestry trees, western Kenya 111, 21–22, 90, 176
113, 114, 115, 119 India 80
fodder shrubs, central Kenya, 126, 130, Indonesia 80
132
induced innovation 82
hedgerow intercropping, western Kenya
inflation 48, 50
93, 97, 105–106
input use and supply 151–154
improved fallows, western Kenya 68, 72
see also tree management, fertilizer
improved fallows, Zambia 43, 46, 54,
institutional support for on-farm research
58
and extension 19–22, 29–31, 32–34,
genetic diversity 147, 151
157–162, 164–166, 175–182
geographic information systems 34, 182
germination 44, 136 fodder shrubs 140–141
see also survival rates improved fallows, western Kenya 86–87
germplasm production and distribution see improved fallows Zambia 55–57
seed production and distribution intercropping trees and crops 18, 30, 153,
Ghana 163 178
Gliricidia sepium 42, 58, 91, 92 in fodder trees central Kenya 134
goats 39, 127, 129, 132, 135, 140, 147 in hedgerow intercropping, western
grazing communal 45, 59, 83, 157, 163, Kenya 101, 103
164, 176 in improved fallows, western Kenya 69,
Grevillea robusta 7, 11, 148, 155, 170, 70, 73
173, 174 in improved fallows, Zambia 40–42,
groundnuts 39, 54, 95 43, 45, 46, 47, 48, 53–54, 58, 59,
157, 179
International Livestock Research Institute
hedgerow intercropping 2, 6–7, 17, 18, (ILRI) 128, 139
22, 23, 26, 28, 31, 89–108, 113, 146,
153, 156, 157, 158, 159, 165,
169–170, 172 Java 80
 Index 193

kale 82 in association with improved fallows,

Kenya Agricultural Research Institute Zambia 38–64, 152
(KARI) 136, 140 Makoka Research Station 58
Kenya Forestry Research Institute (KEFRI) Malawi 106, 170, 173
112, 161 manure
Kenya 6, 19, 20, 23, 24, 29, 32, 34, use in hedgerow intercropping, western
65–144, 146, 147–148, 157, 162, Kenya 96, 99, 100, 103, 107
169–170 use in improved fallows, western Kenya
Kenya Agricultural Research Institute 66, 71
(KARI) 128, 139, 161 marigold 136
Kisumu District 67, 112 markets and market access 4, 5–6, 13, 30,
41, 50, 59, 82, 141, 146, 147, 163,
165, 172
Maseno Agroforestry Research Centre 95
labour availability and use 5, 14, 25, 26, Maseno, Kenya 67, 80, 81, 94
31, 170–171, 173 matrix ranking 6, 23, 28, 32
costs see financial analysis medicine 1
estimation of 14, 16, 17 mesoplatys beetle 46, 58
fodder shrubs, central Kenya 141 milk 155, 164
hedgerow intercropping, western Kenya minimum returns analysis 27, 93, 102
89, 92, 96–103, 106, 146 models, economic 23, 25, 43, 52, 82
improved fallows, western Kenya 68, monitoring 24, 25, 27, 57, 90, 93, 103,
70–71, 74, 76, 79, 80 179
improved fallows, Zambia 41, 46, 47, see also surveys
48, 50, 52, 58 Msekera Research Station 55
land and soil degradation 1, 8, 37, 80, 81, Mucuna pruriens 81
82, 158, 169 mulch 117
land transactions 71 Morus alba (mulberry) 129, 140
Leucaena diversifolia 111–121
Leucaena leucocephala 7, 89, 91, 92, 95,
105 111–121, 128, 169–170 napier grass 127, 128
Leucaena trichandra 129, 140 National Agroforestry Research Project
livestock 45, 114–115, 118, 120 (NAFRP) 128, 132
livestock feed 1, 104, 105, 113, 169 National Dairy Development Project
Lufisols 39, 44, 46 (NDDP) 130, 132
natural resource management and associated
practices 1, 4, 13
neem 136
Madoqua kirki (antelopes) 72, 83, 95 nematodes 58, 173
maize 154, 155, 156 networks, adaptive research and extension
in association with trees in agroforestry networks 4, 8, 21, 32, 55, 57, 59,
tree trials 112 121, 161, 180–181
in association with fodder shrubs 127, niches where trees are planted see tree
128, 132 management
in association with hedgerow Nigeria 106
intercropping, western Kenya 92, Nitisols 68, 91, 112, 126
96–108 nitrogen
in association with improved fallows, deficiency 7, 40, 68, 170
western Kenya 65, 68, 69–71, 73, fixation 1, 90
74, 75, 76 see also fertilizer
194 Index

non-governmental organization 3, 8, 20, Organic Matter Management Network

21, 55–56, 121, 140, 147, 151, 160, 161
162, 180, 182 ornamental benefits of trees 115, 119, 138
nurseries 147–151, 160, 161, 164, oxen 14, 19, 39, 44, 54, 98
175–176 Oxford Forestry Institute (OFI) 140
agroforestry trees, western Kenya 113
hedgerow intercropping, western Kenya
92, 100, 105 parastatal 155
fodder shrubs, central Kenya 129, partial budget 23–24, 32, 130–131, 137
132–134, 136, 140, 146 participatory on-farm trials see on-farm trials
improved fallows, Zambia 45, 46, 47, participatory rapid appraisals see surveys,
57, 58, 59, 62–64 informal or semi-structured
participatory technology development see
technology design and development
off-farm enterprises and income 68, 71,
payback period 48, 153, 170
80, 82, 94, 105, 114
Pennisetum purpureum (napier grass) 127,
on-farm trials 4, 5–6, 8
128
on-farm experiments, involvement of
extension 57 pests and diseases 24, 115, 135, 136–137,
on-farm trials 12–20, 159–162, 176–182 140
farmer and technology selection 19 Philippines 106–107
handling complexity 18–19 phosphorus 24, 68, 83, 90, 95, 96, 106,
monitoring 20 117, 121, 153
objectives of 13–14 see also fertilizer
suitability of trial types for different pigeon pea 38, 42, 44– 45, 53, 173
objectives 16–19 poles 7, 111–113, 115, 118–120, 148,
supervision of 20 154, 164
type 1 trials (designed and managed by policy 3, 8, 13, 22, 145–166, 176
researchers) 15 constraints 146
improved fallows, western Kenya 65, input supply 151–154
66, 83, 160 land and forest tenure and regulatory
improved fallows, Zambia 54, 160 systems 156–157
type 2 trials (designed by researchers but makers 4, 12, 157, 162, 163, 181
managed by farmers) 15, 16–17, product markets 154–156
29, 33, 120, 130, 160, 176 recommendations 165
improved fallows, western Kenya 98, tree germplasm supply 146–151
160 population density 1, 6
improved fallows, Zambia 40–42,
central Kenya 126
43–44, 46, 49
eastern Zambia 39, 169
type 3 trials (designed and managed by
western Kenya 66, 68, 70, 72, 81, 83,
farmers) 5, 16, 17–18, 29, 30, 33,
91, 106, 112
177
agroforestry tree trials, western Kenya poultry 132, 135
111–121 private sector, Zambia 59
fodder shrubs, central Kenya 128, profitability see financial analysis
141 pruning 18, 22, 73, 113, 128–130,
improved fallows Zambia, 40–41, 134–135, 139, 157
43–44, 46, 49, 53 in hedgerow intercropping, western
improved fallows, western Kenya 65, Kenya 89, 90, 95–97, 100,
160 105–106, 113
 Index 195

rabbits 132, 135 sensitivity analyses see financial analysis

rainfall Sesbania sesban 7, 31, 128, 148, 169–170,
central Kenya 126, 140 179
eastern Zambia 39, 44, 45, 46, 58 improved fallows, western Kenya 65–84
western Kenya 67–68, 94, 95, 107, 113 improved fallows, Zambia 37–64
recommendation domains 12, 170–171, management of indigenous stands
173 71–72
Regional Land Management Unit (RELMA) shade 73, 115
140 shading, reduction 73
regression 29, 30, 96, 114, 118 sheep 132, 135
relay cropping 81–82, 158 Siaya district 67, 91, 112
rental rates, land 93 slope 89, 90, 91, 92, 96, 105, 106
research stations 2, 14, 55, 56, 58, 90, 95, social analysis 3, 8, 92–93
112, 120, 179 see also farmer assessment or acceptability
research–development continuum 21, social time preference rate 70
161–162, 180–181 sociology 6
resource budgets 23, 26 soil
rhizobia 40, 92 conservation 90, 115, 128, 135
risk 5, 23, 27, 153, 155, 172–173, 175 erosion 1, 7–8, 95, 104, 105, 106, 107,
risk, hedgerow intercropping relative to 120, 138, 154, 169–170
other enterprises, western Kenya 91, fertility 2, 6, 169
93, 102, 106 agroforestry trees, western Kenya
risk, of improved fallow relative to fertilized 111, 113, 115, 117–19, 120
maize, Zambia 50–51 hedgerow intercropping, western
rock phosphate 83, 153 Kenya 89, 90, 96, 98, 105
root system 45, 46, 54 improved fallows, western Kenya
Rwanda 140, 170–171 66–68, 71–72, 152
improved fallows, Zambia 37–38, 40
organic matter 38, 51
sample size see surveys; trial design structure 38, 51
Sapium ellipticum 129 sorghum 96, 100
seed production and distribution 30, 140, southeast Asia 11
146–151, 164, 175, 182 subsidies 147, 154, 164, 176
agroforestry trees, western Kenya 120 sugarcane 68
fodder shrubs, central Kenya 130, sunflower 39, 54
135–136, 140 surveys
hedgerow intercropping, western Kenya diagnostic 14, 59, 160
105 formal 27, 28, 31,
improved fallows, western Kenya 66, in assessing agroforestry trees, western
69, 72, 76 Kenya 113
improved fallows, Zambia 57, 59 in assessing fodder shrubs, central
seeding, direct 40, 46, 53, 58, 78, 80, Kenya 129–130, 130–137
81–82, 83, 133, 146, 158 in assessing hedgerow intercropping,
seedlings, bare-rooted 40, 42, 44, 45, western Kenya 92–93
53–54, 55, 98, 102, 105, 129, 146, in assessing improved fallows, western
178 Kenya 68, 71–72
seedlings potted 40, 42, 53–54, 76, 80, in assessing improved fallows, Zambia
81–82, 92, 98, 129, 132–133, 178 41, 43
196 Index

surveys, cont hedgerow intercropping, western Kenya

informal or semi-structured 27, 28, 30, 89, 94–95, 103, 104, 105, 106
33, improved fallows, western Kenya 72, 83
in assessing fodder shrubs, central tree management 30, 146–148, 153, 158
Kenya 129 agroforestry trees, western Kenya
in assessing agroforestry trees, western 114–115, 120
Kenya 113 fodder shrubs, central Kenya 128–29,
in assessing hedgerow intercropping, 132–137, 141
western Kenya 91, 92, 93 hedgerow intercropping, western Kenya
in assessing improved fallows, western 89, 91, 94–95, 103, 104, 105, 106
Kenya 68, 72 improved fallows, western Kenya 69,
in assessing improved fallows, Zambia 72, 73, 83
43 improved fallows, Zambia 44, 46, 47
sustainable agriculture 13 tree survival rate
sweet potato 73 agroforestry trees, western Kenya 111,
Systemwide Livestock Programme 139 113, 114, 116, 120
fodder shrubs, central Kenya 132
hedgerow intercropping, western Kenya
Tanzania 83, 140, 163, 170, 173, 174 95
tea 68, 105, 132, 135 improved fallows, western Kenya 72
technology design and development 4, 6, improved fallows, Zambia 44, 46, 53,
9, 11, 12, 13, 121, 146, 157–162, 164, 54–55, 57
176–182 type 1, 2, and 3 trials see on-farm trials
see also on-farm trials tree species and provenances 2, 5, 83, 95,
technology testing 5 111, 120
see also on-farm trials trial design 14–22
technology testing, village approach to agroforestry tree trials, western Kenya
19–20 113
tenancy 71 hedgerow intercropping, western Kenya
tenure systems and regulations 30, 91–92, 178
103–104, 126, 156–157, 165, 176 improved fallows, western Kenya 69
Tephrosia vogelii 7, 136, 169–170, 179 improved fallows, Zambia 40–41
western Kenya 84, 173
Zambia 38, 42, 44, 45, 46, 47, 48, 53,
54, 58, 173 Uganda 140, 170–171, 173, 174
termites 95, 115, 116, 118, 119, 120 United States Agency for International
terraces 96, 114, 128 Development (USAID) 57
threshing 47 upper-storey trees 7, 111, 159 128, 139
timber 1, 7, 111, 113, 115, 117, 119, 146, uses of trees see tree management
148, 154, 164, 182
Tithonia diversifolia 156
tobacco 136, 154 Vihiga district 67, 91, 112
tomatoes 156 village approaches 30
transplanting 58
tree germplasm supply policy 146–151
tree growth wealth status 14, 20, 29, 93, 94, 105, 114,
agroforestry trees, western Kenya 134
114–17 of farmers planting improved fallows,
fodder shrubs, central Kenya 137, 141 Zambia 43, 44, 55, 57
 Index 197

weeds and weeding, wood 5, 73

agroforestry trees, western Kenya 117, see also fuelwood or timber
120 World Vision International 57
hedgerow intercropping, western Kenya
87, 98
improved fallows, western Kenya 66, Yala 67
72, 73, 82
improved fallows, Zambia 44, 46, 47,
54, 55 Zambia 6, 16, 19, 20, 21, 22, 23–24, 26,
western Kenya 28, 31, 32, 169–170 27, 31, 32, 34, 37–64, 80, 146–148,
Winam 67 152, 155, 157, 158, 160, 169–171
windbreaks 1, 120 Zimbabwe 140, 170–171, 173, 174