Published June 21, 2013

Crop Ecology & Physiology

Grass–Legume Proportions in Forage Seed Mixtures and Effects

on Herbage Yield and Weed Abundance
Matt A. Sanderson,* Geoffrey Brink, Robert Stout, and Leah Ruth

ABSTRACT
Formulating grass–legume mixtures requires knowledge of how the proportion of species in a seed mixture (i.e., species evenness)
affects productivity and weed abundance. We hypothesized that mixtures with more equal proportions of species in the seed
mixture (i.e., greater species evenness) would have greater productivity and fewer weeds than mixtures dominated by one or
two species or monocultures. Two experiments with 15 mixtures and monocultures of orchardgrass (Dactylis glomerata L.),
quackgrass (Elytrigia repens L.), alfalfa (Medicago sativa L.), and white clover (Trifolium repens L.) (Exp. 1) or 15 mixtures and
monocultures of meadow fescue [Schedonorus pratensis (Huds.) P. Beauv.], reed canarygrass (Phalaris arundinacea L.), red clover
(T. pratense L.), and kura clover (T. ambiguum L.) (Exp. 2) were sown in autumn 2008 at four locations in Pennsylvania and
Wisconsin. In each experiment, there were four monocultures, four mixtures dominated by one species, six mixtures dominated
by pairs of species, and one equal mixture. Mixtures and monocultures were harvested four to five times each year from 2009
to 2011. Mixtures often had more biomass than the average of legume or N-fertilized grass monocultures. Mixtures with more
equal proportions of species in the seed mixture, however, did not have more biomass or fewer weeds than other mixtures. Rather,
differences in yield were related to the dominant species in the mixture. Optimal legume percentages (30–40%) in the harvested
biomass were achieved with a wide range of grass and legume seed proportions, which suggested that farmers have wide flexibility
in formulating seed mixtures for pastures.

S tudies of grass–legume mixtures often focus on

binary combinations (e.g., Sheaffer et al., 1990; Sleugh et al.,
2000; Zemenchik et al., 2002). Few studies of mixtures have
cultural and mechanical means to control weeds in mixed stands
of grasses and legumes because some herbicides may injure or kill
legumes. Planting mixtures of forage species to create a highly
dealt with multiple forage grasses and legumes for agronomic competitive environment can be an effective measure to suppress
systems and varied the species proportion in mixtures (Kirwan weeds (Drenovsky and James, 2010). Research on temperate
et al., 2007). Small-plot (Skinner et al., 2004; Deak et al., 2007) grasslands of the United States (Picasso et al., 2008; Bonin and
and pasture-scale (Sanderson et al., 2005; Skinner et al., 2006) Tracy, 2012) and northwestern Europe (Kirwan et al., 2007;
studies have shown that complex forage mixtures (three to nine Frankow-Lindberg et al., 2009; Nyfeler et al., 2009) also demon-
species) yielded more herbage than grass–legume binary mix- strated that species-rich forage mixtures can suppress weeds.
tures or monocultures. An economic analysis showed that using From a farm management standpoint, knowledge of how
complex forage mixtures could be more profitable than using a the proportions of species in a seed mixture (i.e., species even-
grass monoculture (Sanderson et al., 2006; Deak et al., 2010). ness) affect production and weed abundance will be useful in
In Nova Scotia, Canada, three- and four-species mixtures of guiding the formulation of forage seed mixtures for farmers. In
grasses and legumes yielded more herbage than binary mixtures a multilocation (28 sites) study across Europe, herbage produc-
(Papadopoulos et al., 2012). Studies in Europe and England tion of grass–legume mixtures (two grasses and two legumes)
demonstrated that forage mixtures yielded more herbage than increased as the proportion of the species in mixture became
monocultures (Kirwan et al., 2007) or species-poor plant com- more equal (i.e., the seed mixtures had greater species evenness;
munities (Bullock et al., 2007). Kirwan et al., 2007). The effect of varying species proportions
Weeds in forage and pastureland can reduce forage yields and also depended on the dominant species in the plant community
nutritive value and compete for soil, water, and light resources (Mulder et al., 2004) and on the proportion of legume in the
(Masters and Mitchell, 2007). Farmers must often rely on mixture (Nyfeler et al., 2009).

M.A. Sanderson,* USDA-ARS Northern Great Plains Research Laboratory, Department of Agriculture (USDA) prohibits discrimination in all its
Mandan, ND 58554; G. Brink, USDA-ARS Dairy Forage Research Center, programs and activities on the basis of race, color, national origin, age,
Madison, WI 53706; R. Stout, and L. Ruth, USDA-ARS Pasture Systems and disability, and where applicable, sex, marital status, family status, parental
Watershed Management Research Unit, University Park, PA 16802. Received status, religion, sexual orientation, genetic information, political beliefs,
12 Mar. 2013. *Corresponding author ([email protected]). reprisal, or because all or part of an individual’s income is derived from any
public assistance program. (Not all prohibited bases apply to all programs.)
Published in Agron. J. 105:1289–1297 (2013) Persons with disabilities who require alternative means for communication
doi:10.2134/agronj2013.0131 of program information (Braille, large print, audiotape, etc.) should contact
Copyright © 2013 by the American Society of Agronomy, 5585 Guilford USDA’s TARGET Center at (202) 720-2600 (voice and TDD). To file
Road, Madison, WI 53711. All rights reserved. No part of this periodical may a complaint of discrimination, write to USDA, Director, Office of Civil
be reproduced or transmitted in any form or by any means, electronic or Rights, 1400 Independence Avenue, S.W., Washington, DC. 20250-9414,
mechanical, including photocopying, recording, or any information storage or call (800) 795-3272 (voice) or (202) 720-6382 (TDD). USDA is an equal
and retrieval system, without permission in writing from the publisher. opportunity provider and employer.

A g ro n o my J o u r n a l • Vo l u m e 10 5 , I s s u e 5 • 2 013 1289
Table 1. Description of the soil at the four locations used in an experiment to compare grass–legume mixtures. Soil analysis results
are for the top 15 cm of the soil profile.
Extractable Soil texture
Location Soil type pH P K Sand Silt Clay
————— mg kg–1 ————— ——————————— % ———————————
Rock Springs Hagerstown silt loam 6.8 35 79 19 47 34
Philipsburg Rayne 5.4 9 67 28 37 34
Lancaster Rozetta silt loam 7.0 43 93 15 62 23
Lancaster Palsgrove silt loam 7.3 20 121 17 53 27

Our objective was to test the hypothesis that mixtures with Research Station (42°51´0˝ N, 90°42´36˝ W; 340 m elevation).
more equal proportions of species in the seed mixture would The Palsgrove soil was susceptible to water-deficit stress and had
have greater productivity and fewer weeds than monocultures lower soil P than the Rozetta soil (Table 1). Both plot areas were
or mixtures dominated by one or two species. We conducted previously cropped to alfalfa in 2006 and 2007 that was killed with
two experiments each with different groups of grass and legume glyphosate [N-(phosphonomethyl) glycine] in June 2008. In mid-
species for 3 yr at four locations. In each experiment, we focused July, plot areas were tilled twice before seeding in early August.
on a limited number of grasses and legumes with contrasting Fifteen forage mixture and monoculture treatments were con-
morphology and establishment rate. Our previous research structed in each of two experiments at each location to include
showed that plant communities of about four forage species were grass and legume species that contrast in relative establishment
practical for management (Sanderson et al., 2005; Deak et al., rate (slow or rapid) and growth habit (bunch and rhizomatous
2007). Because of the importance of legumes in supplying N and grasses; erect and prostrate legumes). These specific traits often
influencing yields of forage mixtures (Nyfeler et al., 2009), we govern the persistence, production, and resulting species evenness
explored the relationship between legume proportion in the seed of mixtures (Kirwan et al., 2007). In Exp. 1, the mixtures and
mixture and the resulting abundance in the harvested biomass monocultures included orchardgrass (rapid establishment; bunch;
and relationships with weed abundance and herbage yield. Pre- cultivar Baridana), quackgrass (slow establishment; rhizomatous;
viously, we reported on weed abundance in monocultures and cultivar Everett), alfalfa (rapid establishment; erect; cultivar
mixtures at establishment of one of the experiments (Sanderson Winter), and white clover (slow establishment; prostrate; culti-
et al., 2012). Here we report weed abundances from two addi- var Will). In Exp. 2, the mixtures and monocultures included
tional years of that experiment. meadow fescue (rapid establishment; bunch; cultivar Pradel), reed
canarygrass (slow establishment; rhizomatous; cultivar Chieftan),
MATERIALS AND METHODS red clover (rapid establishment; erect; cultivar Marathon), and
Field studies were conducted at four locations that varied in kura clover (slow establishment; prostrate; cultivar Endura).
soil fertility and drainage in Pennsylvania and Wisconsin. The The 15 treatments in each experiment were planted in a sim-
location at Rock Springs, PA, was on a well-drained and fertile plex design (Cornell, 2002) that systematically varied relative
Hagerstown silt loam (fine, mixed, semiactive, mesic Typic abundance of each species with a fixed level of overall initial
Hapludalfs) soil in the Ridge and Valley physiographic province abundance (Ramseier et al., 2005; Kirwan et al., 2007; Sand-
(40°43´5˝ N, 77°56´28˝ W; 350 m elevation; Table 1). The plot erson et al., 2012; Table 2). Conducting mixture experiments
area was level with a cropping history of maize (Zea mays L.) in with several forage species can be difficult because of the large
2007 and winter wheat (Triticum aestivum L.) in 2007–2008. number of possible combinations. To pare the mixture combi-
The second location was near Philipsburg (40°1´55.1˝ N, nations to a manageable number, we adopted the simplex design
78°10´10.5˝ W; 600 m elevation) on a north-facing 6% slope in used in similar experiments in Europe (Kirwan et al., 2007;
the Allegheny Plateau physiographic province. The plot area was Nyfeler et al., 2009; Frankow-Lindberg et al., 2009; Frankow-
in a low-fertility naturalized pasture on a Rayne (fine-loamy, Lindberg, 2012; Sturludóttir et al., 2013). The simplex design
mixed, active, mesic Typic Hapludults) soil. The dominant spe- aids in efficiently comparing levels or combinations of several
cies in the pasture vegetation at Philipsburg were goldenrod factors in an experiment (Cornell, 2002). The simplex design
(Solidago canadensis L.), wild carrot (Daucus carota L.), sweet has been used to compare mixtures of plant species in other
vernalgrass (Anthoxanthum odoratum L.), orchardgrass, heal all ecological and agricultural contexts (e.g., Scheffé, 1963; Gibson
(Prunella vulgaris L.), yarrow (Achillea millefolium L.), oxeye et al., 1999; Bondari, 2005; Ramseier et al., 2005; Sheehan et
daisy (Leucanthamum vulgare Lam.), and narrow leaf plantain al., 2006; Suter et al., 2007). The treatments were planted at
(Plantago lanceolata L.). At both locations, the existing vegeta- two initial densities (500 and 1000 germinable seeds m–2) at the
tion was killed with glyphosate [N-(phosphomonomethyl) Pennsylvania locations and at 1000 germinable seeds m–2 at the
glycine] in August 2007. Winter wheat was planted in October Wisconsin locations. The 1000 seed m–2 density approximated
2007, wheat herbage was removed in June 2008, and the plot the recommended planting rate for the forage species (8–17 kg
areas were tilled twice in late June and late July before establish- seed ha–1; Penn State Agronomy Guide, 2011).The germination
ment of the grass–legume plots in early August. rate and seed mass of each species were measured so that the
The Wisconsin plots were established on a Rozetta silt loam soil appropriate mass of seeds could be determined to formulate the
(fine-silty, mixed, superactive, mesic Typic Hapludalfs) and on a seed mixtures to result in the initial seed densities. Legume seed
Palsgrove silt loam soil (fine-silty, mixed, superactive, mesic Typic were treated with the appropriate rhizobia species.
Hapludalfs) at the University of Wisconsin Lancaster Agricultural

1290 Agronomy Journal • Volume 105, Issue 5 • 2013

Table 2. Composition of the 15 monocultures and mixtures in seed bed on 18 Aug. 2008 at Rock Springs and 21 Aug. 2008 at
two experiments planted at four locations in Pennsylvania and
Wisconsin. Numbers are proportions of germinable seeds of Philipsburg. The plots were 10 rows wide with 18 cm between
each species in the seed mix. rows. Monocultures and mixtures were seeded at the Wiscon-
Grass 1† Grass 2 Legume 1 Legume 2 sin locations with a plot drill (Carter, Carter Manufacturing
Monocultures Company, Brookston, IN) in 1.7 by 3 m plots in a clean-tilled
1 0 0 0 seed bed on 13 Aug. 2008. Plots were 10 rows wide with 15 cm
0 1 0 0 between rows. Planting depth was about 0.5 cm.
0 0 1 0 Plots at Rock Springs, PA, were fertilized with 112 kg K ha–1 in
0 0 0 1 June 2009 and 49 kg P and 166 kg K ha–1 in October 2009; and
Mixtures dominated by one species 52 kg P and 391 kg K ha–1 in 2010 and 2011. Plots at Philipsburg,
0.7 0.1 0.1 0.1 PA, were fertilized with 1000 kg lime, 49 kg P, and 112 kg K ha–1
0.1 0.7 0.1 0.1
in autumn 2007. Lime and fertilizer were applied at Philipsburg
0.1 0.1 0.7 0.1
only to aid establishment of forages on the low fertility Rayne soil.
0.1 0.1 0.1 0.7
Mixtures dominated by two species
The Wisconsin plots were not fertilized before establishment.
0.4 0.4 0.1 0.1 Nitrogen fertilizer was applied to grass monocultures at 56 kg ha–1
0.4 0.1 0.4 0.1 in April and June at Rock Springs and the two Wisconsin loca-
0.4 0.1 0.1 0.4 tions and at 56 kg in April and 28 kg in June each year at Philips-
0.1 0.4 0.4 0.1 burg. Nitrogen was not applied to the grass–legume mixtures.
0.1 0.4 0.1 0.4 Plots were clipped each time the mean canopy height reached
0.1 0.1 0.4 0.4 30 cm to determine herbage production (Table 3). Biomass was
Equal mixture harvested by cutting a 1.52- by 4-m strip at a 7-cm stubble height
0.25 0.25 0.25 0.25
(Deak et al., 2007) through the center of each plot at Pennsylvania.
† Grass 1, orchardgrass (Exp. 1) or meadow fescue (Exp. 2); Grass 2, quackgrass At Wisconsin, biomass was harvested by cutting a 50-cm swath at
or reed canarygrass; Legume 1, alfalfa or red clover; Legume 2, white clover or
kura clover. a 7-cm stubble height through the center of each plot using a rotary
mower equipped with a catch basket. A 600- to 800-g subsample
The 15 treatments in each experiment included four monocul- was taken from each yield sample, dried at 65°C for 48 h, and
tures, four mixtures each of which was dominated by a different weighed to determine dry matter content. Botanical composition of
species (dominant species as 70% of the germinable seeds in the each plot was assessed in spring, summer, and autumn each year by
mixture, other species at 10%; low evenness), six mixtures domi- hand separating subsamples (about 300 g wet weight) of biomass to
nated by pairs of each species (dominant pairs at 40% each of seed the species level for both sown forages and unsown species (hereaf-
mixture, other species at 10%; moderate evenness), and one mix- ter “weeds”). In autumn 2008, plots were inspected visually about
ture with an equal proportion of all species (25% of each species 30 d after planting and dominant weed species present were noted.
in seed mixture; high evenness). The treatment structure and seed One group of plots in Exp. 1 and 2 was severely damaged by
mixture proportions were chosen to be consistent with those used snow and ice at Philipsburg during the winter of 2008–2009 and
in a 28-site experiment in Europe (Kirwan et al., 2007). was eliminated from the experiment. Kura clover did not estab-
The monocultures and species mixtures in each experiment lish at Rock Springs and Philipsburg; therefore the kura clover
were replicated twice in a randomized complete block design at monoculture treatment was omitted from statistical analysis of
Pennsylvania (monocultures and mixtures and seed density rep- Exp. 2. To satisfy the assumptions of normality and homosce-
licated twice, 60 plots per experiment) and Wisconsin (30 plots dasticity, weed percentage data were arcsin (ÖY) transformed for
per experiment). We compromised on the number of replicates analysis. Untransformed data are presented in the tables. Biomass
because of the scope of the research (two similar experiments (forage+weeds; sum of all harvests within a year) and weed per-
conducted at four locations for 3 yr). The monocultures and mix- centage data were analyzed with the mixed models procedure of
tures were planted with a plot drill (Hege1, Wintersteiger AG, SAS (Littell et al., 1996). The statistical model for the Wisconsin
Ried im Innkreis, Austria) in 1.9 by 5 m plots in a clean-tilled data included year, treatment (the 15 monocultures and mix-
tures), and block effects. Treatments were considered fixed effects
1 Mention of any product, brand, or trademark does not imply endorsement and blocks were considered random effects. Because of the loss
or recommendation of that product, brand, or trademark by USDA-ARS. of plots at the Phillipsburg location, we decided to analyze the

Table 3. Harvest dates for plots of forage monocultures and mixtures at each location. Plots on the two soil types at Wisconsin
were harvested on the same dates.
Pennsylvania
Rock Springs Philipsburg Wisconsin
2009 2010 2011 2009 2010 2011 2009 2010 2011
19 May 7 May 21 May 9 June 21 May 31 May 21 May 6 May 24 May
22 June 4 June 28 June 8 July 21 June 6 July 23 June 9 June 28 June
22 July 6 July 3 Aug. 18 Aug. 2 Aug. 17 Aug. 23 July 14 July 3 Aug.
12 Aug. 9 Aug. 16 Sept. 14 Oct. 13 Oct. 5 Oct. 25 Aug. 12 Aug. 7 Sept.
23 Sept. 20 Sept. 18 Sept. 29 Sept.

Agronomy Journal • Volume 105, Issue 5 • 2013 1291

Table 4. Harvested biomass in two experiments at four locations in Pennsylvania and Wisconsin averaged over 3 yr. Pennsylvania
data are least squares means of two seeding rates, two replicates, and 3 yr. Wisconsin data are least squares means of two repli-
cates and 3 yr.
Pennsylvania Wisconsin
Sowing proportions Exp. 1 Exp. 2 Exp. 1 Exp. 2
G1† G2 L1 L2 RS‡ PB RS PB RZ PG RZ PG
––––––––––––––––––––––––––––––––––––––––– kg dry matter ha–1 –––––––––––––––––––––––––––––––––––––––––
Monocultures
1 0 0 0 7760 4330 6720 4010 7470 6640 6580 7320
0 1 0 0 5920 3770 6960 4570 7330 5570 6510 6620
0 0 1 0 8750 3880 9310 3890 8800 6450 8200 5830
0 0 0 1 6570 3580 § § 6310 4820 5990 5370
Mean 7250 3890 7660 4160 7480 5870 6820 6290
Mixtures dominated by one species
0.7 0.1 0.1 0.1 7740 3570 8350 3370 6690 5480 6930 6420
0.1 0.7 0.1 0.1 8010 3520 8000 3290 7040 9270 6420 6170
0.1 0.1 0.7 0.1 8470 3830 10130 4170 9340 7360 8660 6660
0.1 0.1 0.1 0.7 8480 4070 7330 3420 8170 6370 8050 6780
Mean 8180 3750 8450 3560 7810 7120 7520 6510
Mixtures dominated by two species
0.4 0.4 0.1 0.1 6930 3330 8250 3170 6440 6310 7020 6290
0.4 0.1 0.4 0.1 8370 4310 8880 3840 7760 6350 7740 7390
0.4 0.1 0.1 0.4 8570 3900 8170 2920 7360 6980 7600 6660
0.1 0.4 0.4 0.1 8470 4120 9260 4220 8380 6660 8840 7080
0.1 0.4 0.1 0.4 8200 3730 7810 3180 7720 6390 7510 6400
0.1 0.1 0.4 0.4 8310 4450 8960 3570 8730 7960 8090 7660
Mean 8140 3970 8560 3480 7730 6780 7800 6920
Equal mixture
0.25 0.25 0.25 0.25 8080 4160 9370 3850 7550 6580 7950 6740

SE 315 325 438 261 454 423 409 588

Contrasts
Avg. of monocultures vs. avg. of mixtures ** ns¶ * ns ns ** ** ns
Equal mixture vs. avg. of mixtures dominated
ns ns ** ns ns ns ns ns
by one species
Mixtures dominated by one species vs. avg.
ns ns ns ns ns ns ns ns
of mixtures dominated by two species
Grass- vs. legume-dominated mixtures ** ** ** * ** ** ** *
G1 monocultures and mixtures vs. G2
* * ** * ns ns ns ns
monocultures and mixtures
L1 monocultures and mixtures vs. L2
** ns ** * ** ns
monocultures and mixtures
* P < 0.05.
** P < 0.01.
† G1, orchardgrass (Exp. 1) or meadow fescue (Exp. 2); G2, quackgrass or reed canarygrass; L1, alfalfa or red clover; L2, white clover or kura clover.
‡ RS, Rock Springs; PB, Philipsburg; RZ, Rozetta soil; PG, Palsgrove soil.
§ Kura clover did not establish at the Pennsylvania locations.
¶ ns, not significant.

Pennsylvania data by location. A preliminary analysis of the Penn- RESULTS AND DISCUSSION
sylvania data indicated that there was no effect of seeding density At Pennsylvania, seeding density did not affect annual yields
on biomass yield or weed abundance. Therefore, we treated the (data not shown); therefore all data reported are averaged for the
seeding densities as replicates for analysis (total of four replicates). two seeding densities. There were interactions of year and treat-
The statistical model for the Pennsylvania data included year, treat- ment for biomass and weed abundance (P < 0.05); however, these
ment, and block effects. In analyses across years, year was treated interactions were mainly due to shifts in magnitude of the differ-
as a repeated measure. In most instances the variance components ence between some means and not because of changes in ranks.
covariance structure best fit the data (lowest AIC value) compared Therefore, a combined analysis across years was conducted for each
with several other covariance structures (e.g, compound sym- experiment and location.
metry; Toeplitz; first-order autoregressive). Denominator degrees Averaged across 3 yr, harvested biomass (forage+weed) of
of freedom were calculated with the Kenward–Roger option. mixtures was greater than the average of monocultures at two
Planned contrasts were used to make specific comparisons when of four locations (Table 4). There were no differences in average
main effects were significant (P < 0.05). harvested biomass among mixtures dominated by one, two, or
The establishment year (2009) weed abundance information equal numbers of species, with the exception of the Rock Springs
from Exp. 1 was published by Sanderson et al. (2012). Those data location in Exp. 2. Legume-dominated mixtures consistently had
are not reported here. more biomass than grass-dominated mixtures in each experiment

1292 Agronomy Journal • Volume 105, Issue 5 • 2013

Fig. 1. Dry matter yield of grass–legume mixtures as a func-
tion of legume percentage in the harvested biomass in two
experiments in (top panel) Pennsylvania and (bottom panel)
Wisconsin. The regression relationship differed between the
Rock Springs and Philipsburg locations in Pennsylvania; how-
ever, a common regression line best described the Wisconsin
data. Data are means of 3 yr. RMSE, root mean square error.

at all locations. In most instances, there was a linear relationship

between harvested biomass and legume percentage at Pennsylvania
and Wisconsin (Fig. 1). At Pennsylvania in Exp. 1, orchardgrass
monocultures and mixtures had more biomass than quackgrass
monocultures and mixtures. Alfalfa monocultures and mixtures
generally had more biomass than white clover monocultures and Fig. 2. Relationship between legume abundance in the har-
mixtures in Exp. 1 at Pennsylvania and Wisconsin. vested biomass and the proportion of legume in the seed
Although overyielding (i.e., mixtures more productive than mix in Exp. 1. There were no significant relationships in 2011.
RMSE, root mean square error.
the average monoculture) occurred frequently in our experi-
ments, we did not observe any transgressive overyielding (i.e.,
mixture biomass yield greater than the most productive mono- most years indicating that the legume component substituted
culture; Schmid et al., 2008). Others have reported frequent for up to 112 kg N ha–1 (the amount applied to the grass mono-
instances of transgressive overyielding of grass legume mixtures cultures) in these environments. Research in Iowa demonstrated
compared with monocultures in studies with a similar experi- that orchardgrass–alfalfa mixtures yielded as much forage
mental design in several locations in Europe and Canada (Finn as orchardgrass monocultures fertilized with 269 kg N ha–1
et al., 2013; Sturludóttir et al., 2013). In the European studies, (Carter and Scholl, 1962). In the same study, orchardgrass–red
grass–legume mixtures and monocultures at many locations clover mixtures yielded as much forage as orchardgrass fertil-
received N, P, and K fertilizers, whereas in our study the grass ized with 120 kg N ha–1. Barnett and Posler (1983) reported
monocultures received N fertilizer but the grass–legume mix- average yield of 6500 kg dry matter ha–1 for grass monocul-
tures did not. This may have biased yields for grasses compared tures fertilized with 90 kg N ha–1 (average of four cool-season
with mixtures; however, we reasoned that in farm practice grass– grasses and 4 yr) vs. 8600 kg ha–1 for alfalfa–grass mixtures and
legume mixtures would not receive N fertilizer. 6100 kg ha–1 for red clover–grass mixtures.
Contrary to our hypothesis, we did not observe a pattern of As the proportion of legume in the seed mix increased, the
greater biomass for mixtures with more equal proportions of for- percentage of legume in the harvested biomass in 2009 (the year
age species (more evenness) in the seed mixture than with seed after seeding) increased linearly at most locations in both exper-
mixtures dominated by one species. Mixtures often had as much iments (Fig. 2 and 3). In subsequent years, however, there was
biomass as N-fertilized grass monocultures at most locations in generally no relationship between seeded legume proportion

Agronomy Journal • Volume 105, Issue 5 • 2013 1293

 Such instances require preventive management measures such
as providing dry hay before and during grazing, slowly acclimat-
ing grazing animals to grass–legume mixtures, or poloxalene
supplementation (Kahn et al., 2011).
Averaged across years, weed percentage in the harvested bio-
mass was consistently greater in monocultures than in mixtures
in both experiments at all locations (Table 5), which was consis-
tent with our previous results during the establishment phase of
Exp. 1 (Sanderson et al., 2012). At Rock Springs, the predomi-
nant weed species were annual forbs in 2009, whereas perennial
forbs and summer-annual grasses became more abundant in
2010 and 2011 (Table 6). At Phillipsburg, perennial forbs and
summer-annual grasses were the main weed species. At Wiscon-
sin, the dominant weed species in each experiment at both loca-
tions was dandelion (Taraxacum officinale F.H. Wigg.), followed
by the annual forbs lambsquarters (Chenopodium album L.),
red root pigweed (Amaranthus retroflexus L.), and Pennsylvania
smartweed (Polygonum pennsylvanicum L.; Table 6). There were
no differences in weed abundance among mixtures dominated
by one, two, or equal numbers of species. Orchardgrass mono-
cultures and mixtures consistently had lower weed abundance
than quackgrass monocultures and mixtures at all locations in
Exp. 1. Orchardgrass monocultures were more resistant to inva-
sion by dandelion than grass–legume mixtures because of strong
aboveground competition for light in Swedish research, in which
a similar experimental design was used (Frankow-Lindberg et
al., 2009). Perennial ryegrass (Lolium perenne L.) monocultures
were also more resistant to weed invasion than grass–legume
mixtures in a follow up experiment by Frankow-Lindberg
(2012). Mixtures of Kentucky bluegrass (Poa pratensis L.), timo-
thy (Phleum pratense L.), red clover, and white clover sown in the
same mixture combinations as in our experiment had a much
lower weed abundance (5% of harvested biomass) than monocul-
tures (10–60% of harvested biomass; Sturludóttir et al., 2013).
Kura clover established very poorly for unknown reasons at
the Pennsylvania locations. Establishment difficulties with kura
clover are well known (Sheaffer and Marten, 1991). Kura clover
monoculture plots were nearly all weeds at Wisconsin in Exp. 2
Fig. 3. Relationship between legume abundance in the har- (Table 5) and kura clover presence in mixtures was very low.
vested biomass and the proportion of legume in the seed mix Riday and Albrecht (2012) also reported very low proportions of
in Exp. 2. RMSE, root mean square error; ns, not significant. kura clover with red clover mixtures at Arlington, WI.
We hypothesized that weed abundance would be less in mix-
and percentage of legume in the harvested biomass. The lack of tures with more equal proportions of forage species (i.e., greater
a relationship between legume proportion in the seed and the species evenness) in the seed mixture than with seed mixtures
percentage of legume in the herbage indicates limited scope for dominated by one species. We based this hypothesis on previ-
manipulation or maintenance of desired legume proportions in ous research where we observed that weed abundance decreased
the herbage solely through the seed mixture. Other variables, as the evenness of forage species increased in clipped plots and
such as companion grass, stocking method, and landscape posi- on-farm surveys of plant species diversity in pastures (Tracy and
tion, often influence legume proportions in established pastures Sanderson, 2004; Tracy et al., 2004). The results reported here
(Harmoney et al., 2001). with established stands and our previous research during the
The percentage of legume in the harvested biomass frequently establishment phase (Sanderson et al., 2012) provided no evi-
was above 30% at all locations (Fig. 2 and 3). Legume dry matter dence that the evenness of grass–legume mixtures affected weed
components of 35 to 40% are considered optimum for sustain- abundance. Mixtures planted with an equal proportion of grasses
able herbage yields and forage quality of grass–legume pastures and legumes had similar weed abundance as mixtures with one
(Thomas, 1992). Excessively high legume abundance, however, or two dominant forage species in the seed mix. The species com-
increases the risk for bloat in grazing ruminants (Mouriño et position of the mixtures, however, had a strong effect on invasion.
al., 2003). The very high (>50% of harvested biomass) legume Orchardgrass monocultures had a strong suppressive effect on
component of mixtures at some locations in both Exp. 1 and 2 weeds (Table 5) and the legume abundance in the mixture had
could have presented a bloat hazard if such pastures were grazed. a strong negative effect on weed abundance in Pennsylvania but

1294 Agronomy Journal • Volume 105, Issue 5 • 2013

Table 5. Weed abundance (percentage of harvested biomass) in Exp. 1 and 2 at four locations in Pennsylvania and Wisconsin.
Pennsylvania data are least squares means of two seeding rates, two replicates, and 2 (2010 and 2011, Exp. 1) or 3 (Exp. 2) yr.
Wisconsin data are least squares means of two replicates and 2 (2010 and 2011, Exp. 1) or 3 (Exp. 2) yr.
Pennsylvania Wisconsin
Sowing proportions Experiment 1 Experiment 2 Experiment 1 Experiment 2
G1† G2 L1 L2 RS‡ PB RS PB RZ PG RZ PG
Monocultures
1 0 0 0 8 16 15 36 4 2 5 4
0 1 0 0 49 70 32 43 51 50 14 20
0 0 1 0 36 33 24 52 42 39 20 11
0 0 0 1 46 58 § § 42 31 89 86
Mean 35 44 24 44 35 31 32 30
Mixtures dominated by one species
0.7 0.1 0.1 0.1 2 13 11 23 1 4 4 8
0.1 0.7 0.1 0.1 4 11 10 20 12 10 7 6
0.1 0.1 0.7 0.1 5 8 6 21 6 1 5 7
0.1 0.1 0.1 0.7 4 14 14 23 5 2 7 14
Mean 4 12 10 22 6 4 6 9
Mixtures dominated by two species
0.4 0.4 0.1 0.1 9 13 8 22 3 3 2 4
0.4 0.1 0.4 0.1 3 9 5 14 6 2 9 5
0.4 0.1 0.1 0.4 3 6 10 23 4 2 6 4
0.1 0.4 0.4 0.1 5 11 8 20 7 5 5 4
0.1 0.4 0.1 0.4 4 20 11 17 10 5 8 7
0.1 0.1 0.4 0.4 4 6 9 25 5 6 12 4
Mean 5 11 8 20 6 4 7 5
Equal mixture
0.25 0.25 0.25 0.25 4 7 5 16 5 3 2 6

SE 3.5 3.8 1.8 4.6 4.1 3.6 3.2 3.1

Contrasts
Avg. of monocultures vs. avg. of mixtures ** ** * * ** ** ** **
Mixtures dominated by one species vs. avg.
ns¶ ns ns ns ns ns ns ns
of mixtures dominated by two species
Equal mixture vs. avg. of mixtures dominated
ns ns ns ns ns ns ns *
by one species
Grass- vs. legume-dominated mixtures ns ns ns ns ns ns ns ns
G1 monocultures and mixtures vs. G2
** ** ns ns ** ** ns ns
monocultures and mixtures
L1 monocultures and mixtures vs. L2
ns ns ns ns ns ns ** **
monocultures and mixtures
* P < 0.05.
** P < 0.01.
† G1, orchardgrass (Exp. 1) or meadow fescue (Exp. 2); G2, quackgrass or reed canarygrass; L1, alfalfa or red clover; L2, white clover of kura clover.
‡ RS, Rock Springs; PB, Philipsburg; RZ, Rozetta soil; PG, Palsgrove soil.
§ Kura clover did not establish at the Pennsylvania locations.
¶ ns, not significant.

less effect in Wisconsin (Fig. 4). This indicated that including the seed mixture would have greater productivity and fewer weeds
the appropriate species in mixtures to achieve a specific function than monocultures or mixtures dominated by one or two species.
was more important than the evenness of species in the mixture. The data did not support our hypothesis that seed mixtures of
Picasso et al. (2008) found a positive relationship between plant equal proportions of forage species would perform better than mix-
species richness of mixtures and their biomass yield. They also tures dominated by one or two species. Our results indicated that
noted that specific species (e.g., alfalfa and orchardgrass) had a productivity of the mixtures was related to the dominant species
greater weed suppressive effect than species richness. and the proportion of legume in the seed mixture. Overall, grass–
legume mixtures produced more biomass and had fewer weeds than
CONCLUSIONS the average of grass and legume monocultures. Mixtures also pro-
We conducted two experiments each with different groups duced as much biomass as N-fertilized grass monocultures. Opti-
of grass and legume species for 3 yr at four locations to test the mal legume components of 30 to 40% or greater were achieved with
hypothesis that mixtures with more equal proportions of species in a wide range of seed mixtures containing different grass and legume

Agronomy Journal • Volume 105, Issue 5 • 2013 1295

Table 6. Dominant weed species observed in plots at each location in descending order of abundance.†
2008 2009 2010 2011
Autumn Spring Summer Autumn Spring Summer Autumn Spring Summer Autumn
Rock Springs
VEPE3 VEPE3 CEFO2 CEOF BAVU TAOF PADI TAOF TAOF TAOF
CEOF2 CEFO2 BAVU BAVU TAOF PLMA2 PLMA2 PLLA PLMA2 PADI
SEPUP2 CABU2 OXST PADI CAAC COCA5 BAVU POPR DIIS PLMA
PADI BAVU COAR4 TAOF RUCR CAAC SEPUP2 SONE CYES DIIS
BAVU LECA5 CAAC OXST PLMA2 ERAN SONE POER81 CAAC POPR
Philipsburg
VEPE3 POPR AMAR2 AMAR2 RUAC3 DACA6 PLMA2 PACA6 PLMA2 DIIS
CEOF2 LEVU LEVU SEPUP2 LEVU PLCO3 COCA5 POER81 PLLA PLMA2
LAAM AMAR2 COAR DACA6 PLLA SONE PACA6 SOCA6 DACA6 PLLA
SEPUP2 RUAC3 POPR LEVU COCA5 ERAN PLLA ERAN PACA6 ERAN
PADI POER81 POER81 PLLA DACA6 PRVU SEPUP2 LEVU COCA5 POER81
Wisconsin Rozetta soil
TAOF TAOF TAOF TAOF TAOF TAOF TAOF TAOF TAOF TAOF
CHAL7 AMRE AMRE AMRE AMRE DIIS
LECA5 MEOF DIIS
Wisconsin Palsgrove soil
TAOF TAOF TAOF TAOF TAOF TAOF TAOF TAOF TAOF TAOF
CHAL7 CHAL7 CHAL7 CHAL7 CHAL7 POPE2
STME2 STME2 POPE2 POPE2
POPE2 MEOF
† Annual forbs: AMAR2, Ambrosia artemisiifolia L.; AMRE, Amaranthus retroflexus L.; BAVU, Barbarea vulgaris L.; CABU2, Capsella buras-pastoris L.; CEFO2, Cerastrium
fontanum Baumg.; CHAL7, Chenopodium album L.; COAR4, Convulvulus arvensis L.; COCA5, Conyza canadensis (L). Cronquist; ERAN, Erigeron annuus (L.) Pers.; LAAM,
Lamium amplexicaule L.; LECA5, Lepidium campestre (L.) W.T. Aiton; OXST, Oxalis stricta L.; RUAC3, Rumex acetosella L.; VEPE3, Veronica persica Poir. Perennial forbs:
DACA6, Daucus carota L.; PRVU, Prunella vulgaris L.; LEVU, Leucanthamum vulgare Lam.; PLLA, Plantago lanceolata L.; PLMA2, Plantago major L.; POER81, Potentilla erecta
(L.) Raeusch; POPE2, Polygonum pennsylvanicum L.; RUCR, Rumex crispus L.; SOCA6, Solidago canadensis L.; SONE, Solidago nemoralis Aiton; STME2, Stellaria media (L.)
Vill; TAOF, Taraxacum officinale F.H. Wigg. Biennial forbs: CAAC, Carduus acanthoides L; MEOF, Melilotus officinalis L. Annual grasses: DIIS, Digitaria ischemum L.; PACA6,
Panicum capillaire L.; PADI, Panicum dichotimiflorium Michx.; SEPUP2, Setaria pumila (Poir.) Roem. & Schult. Perennial grass: POPR, Poa pratensis L.; Perennial sedge: CYES,
Cyperus esculentus L.

proportions. This indicated that farmers have wide flexibility in

formulating seed mixtures for specific locations and to achieve spe-
cific functions in their forage operations.

REFERENCES
Barnett, F.L., and G.L. Posler. 1983. Performance of cool-season perennial
grasses in pure stands and in mixtures with legumes. Agron. J. 75:582–
586. doi:10.2134/agronj1983.00021962007500040004x
Bondari, K. 2005. Mixture experiments and their applications in agricultural
research. In: Proceedings SAS Users Group International (SUGI) Confer-
ence, Philadelphia, PA. 10–13 April. Paper no. 209-30. SAS Inst., Cary, NC.
Bonin, C.L., and B.F. Tracy. 2012. Diversity influences forage yield and stabil-
ity in perennial prairie plant mixtures. Agric. Ecosyst. Environ. 162:1–7.
doi:10.1016/j.agee.2012.08.005
Bullock, J.M., R.F. Pywell, and K.J. Walker. 2007. Long-term enhancement of
agricultural production by restoration of biodiversity. J. Appl. Ecol. 44:6–
12. doi:10.1111/j.1365-2664.2006.01252.x
Carter, L.P., and J.M. Scholl. 1962. Effectiveness of inorganic nitrogen as a
replacement for legumes grown in association with forage grasses. I. Dry
matter production and botanical composition. Agron. J. 54:161–163.
doi:10.2134/agronj1962.00021962005400020021x
Cornell, J.A. 2002. Experiments with mixtures: Designs, models, and the analy-
sis of mixture data. 3rd ed. John Wiley & Sons, New York.
Deak, A., M.H. Hall, M.A. Sanderson, and D.A. Archibald. 2007. Production
and nutritive value of grazed simple and complex forage mixtures. Agron.
J. 99:814–821. doi:10.2134/agronj2006.0166
Deak, A., M.H. Hall, M.A. Sanderson, A. Rotz, and M. Corson. 2010. Whole-
farm evaluation of Forage mixtures and grazing strategies. Agron. J.
102:1201–1209. doi:10.2134/agronj2009.0504
Drenovsky, R.E., and J.J. James. 2010. Designing invasion-resistant plant com-
munities. The role of plant functional traits. Rangelands 32:32–37.
Fig. 4. Weed abundance in grass–legume mixtures as a func- doi:10.2111/RANGELANDS-D-09-00002.1
tion of legume percentage in the harvested biomass in two Finn, J., L. Kirwan, J. Connolly, M.T. Sebastià, Á. Helgadóttir, O.H. Baad-
experiments in (top panel) Pennsylvania and (bottom panel) shaug et al. 2013. Ecosystem function enhanced by combining four func-
Wisconsin. Data are means of 2 (Exp. 1) or 3 (Exp. 2) yr. tional types of plant species in intensively managed grassland mixtures:a
RMSE, root mean square error. There was no significant rela- 3-year continental-scale experiment. J. Appl. Ecol. 50:365–375.
tionship at Wisconsin. doi:10.1111/1365-2664.12041.

1296 Agronomy Journal • Volume 105, Issue 5 • 2013

Frankow-Lindberg, B.E. 2012. Grassland plant species diversity decreases inva- Sanderson, M.A., G.E. Brink, R. Stout, and L. Ruth. 2012. Grass-legume mix-
sion by increasing resource use. Oecologia 169:793–802. doi:10.1007/ tures suppress weeds better than monocultures during establishment.
s00442-011-2230-7 Agron. J. 104:36–42. doi:10.2134/agronj2011.0130
Frankow-Lindberg, B.E., C. Brophy, R.P. Collins, and J. Connolly. 2009. Biodi- Sanderson, M.A., M.S. Corson, C.A. Rotz, and K.J. Soder. 2006. Economic
versity effects on yield and unsown species invasion in a temperate forage analysis of forage mixture productivity in pastures grazed by dairy cattle.
ecosystem. Ann. Bot. (Lond.) 103:913–921. doi:10.1093/aob/mcp008 Forage and Grazinglands. www.plantmanagmentnetwork.org/fg. doi:
Gibson, D., J. Connolly, D.C. Hartnett, and J.D. Weidenhamer. 1999. Designs 10.1094/FG-2006-0929-01-RS.
for greenhouse studies of interactions between plants. J. Ecol. 87:1–16. Sanderson, M.A., K.J. Soder, L.D. Muller, K.D. Klement, R.H. Skinner, and S.C.
doi:10.1046/j.1365-2745.1999.00321.x Goslee. 2005. Forage mixture productivity and botanical composition in
Harmoney, K.R., K.J. Moore, E.C. Brummer, C.L. Burras, and J.R. George. pastures grazed by dairy cattle. Agron. J. 97:1465–1471. doi:10.2134/
2001. Spatial legume composition and diversity across seeded landscapes. agronj2005.0032
Agron. J. 93:992–1000. doi:10.2134/agronj2001.935992x Scheffé, H. 1963. The simplex-centroid design for experiments with mixtures. J.
Kahn, C., S. Line, and S. Atello. 2011. Bloat (ruminal tympany). In: The Merck Royal Statistical Soc. Ser. B (Methodological) 25:235–263.
veterinary manual. 10th ed. Courier Kendallville, Inc., Kendallville, IN. Schmid, B., A. Hector, P. Saha, and M. Loreau. 2008. Biodiversity effects and
Kirwan, L., A. Lüscher, M.T. Sebastià, J.A. Finn, R.P. Collins, C. Porqueddu transgressive overyielding. J. Plant Ecol. 1:95–102. doi:10.1093/jpe/rtn011
et al. 2007. Evenness drives consistent diversity effects in inten- Sheaffer, C.C., and G.C. Marten. 1991. Kura clover forage yield, forage qual-
sive grassland systems across 28 European sites. J. Ecol. 95:530–539. ity, and stand dynamics. Can. J. Plant Sci. 71:1169–1172. doi:10.4141/
doi:10.1111/j.1365-2745.2007.01225.x cjps91-163
Littell, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS sys- Sheaffer, C.C., D.W. Miller, and G.C. Marten. 1990. Grass dominance and
tem for mixed models. SAS Institute, Cary, NC. mixture yield quality in perennial grass-alfalfa mixtures. J. Prod. Agric.
Masters, R.A., and R.B. Mitchell. 2007. Weed management. In: R.F Barnes et 3:480–485. doi:10.2134/jpa1990.0480
al., editors, Forages. Vol. II. The science of grassland agriculture. Blackwell Sheehan, C., L. Kirwan, J. Connolly, and T. Bolger. 2006. The effects of earth-
Publ., Ames, IA. p. 395–409. worm functional group diversity on nitrogen dynamics in soils. Soil Biol.
Mouriño, F., K.A. Albrecht, D.M. Schaefer, and P. Berghazi. 2003. Steer perfor- Biochem. 38:2629–2636. doi:10.1016/j.soilbio.2006.04.015
mance on kura clover-grass and red-clover grass mixed pastures. Agron. J. Skinner, R.H., D.L. Gustine, and M.A. Sanderson. 2004. Growth, water rela-
95:652–659. doi:10.2134/agronj2003.0652 tions, and nutritive value of pasture species mixtures under moisture stress.
Mulder, C.P.H., E. Brazely-White, P.G. Mitakopoulos, A. Hector, M. Crop Sci. 44:1361–1369. doi:10.2135/cropsci2004.1361
Scherer-Lorenzen, and B. Schmid. 2004. Species evenness and pro- Skinner, R.H., M.A. Sanderson, B.F. Tracy, and C. Dell. 2006. On-farm com-
ductivity in experimental plant communities. Oikos 107:50–63. parisons of CO2 uptake, aboveground productivity, root growth, and
doi:10.1111/j.0030-1299.2004.13110.x soil carbon dynamics in simple and complex pasture mixtures. Agron. J.
Nyfeler, D., O. Huguenin-Elie, M. Suter, E. Frossard, J. Connolly, and A. 98:320–326. doi:10.2134/agronj2005.0180a
Lüscher. 2009. Strong mixture effects among four species in fertilized agri- Sleugh, B., K.J. Moore, J.R. George, and E.C. Brummer. 2000. Binary legume-
cultural grassland led to persistent and consistent transgressive overyield- grass mixtures improve forage yield, quality, and seasonal distribution.
ing. J. Appl. Ecol. 46:683–691. doi:10.1111/j.1365-2664.2009.01653.x Agron. J. 92:24–29. doi:10.2134/agronj2000.92124x
Papadopoulos, Y.A., M.S. McElroy, S.A.E. Fillmore, K.B. McRae, J.L. Duyins- Sturludóttir, E., C. Brophy, G. Bélanger, A.-M. Gustavsson, M. Jørgensen, T.
veld, and A.H. Fredeen. 2012. Sward complexity and grass species compo- Lunnan, and Á. Helgadóttir. 2013. Benefits of mixing grasses and legumes
sition affect the performance of grass-white clover pasture mixtures. Can. for herbage yield and nutritive value in Northern Europe and Canada.
J. Plant Sci. 92:1199–1205. doi:10.4141/cjps2012-015 Grass Forage Sci. http://onlinelibrary.wiley.com/journal/10.1111/
Penn State Agronomy Guide. 2011. Publ. AGRS-026. The Pennsylvania State (ISSN)1365-2494 (accessed 7 June 2013). doi:10.1111/gfs.12037.
Univ., University Park, PA. http://extension.psu.edu/agronomy-guide/ Suter, M., D. Ramseier, S. Guesewell, and J. Connolly. 2007. Convergence pat-
cm/sec8 (accessed 15 Jan. 2013). terns and multiple species interactions in a designed plant mixture of five
Picasso, V.D., E.C. Brummer, M. Liebman, P.M. Dixon, and B.J. Wilsey. 2008. species. Oecologia 151:499–511. doi:10.1007/s00442-006-0594-x
Crop species diversity affects productivity and weed suppression in peren- Thomas, R.J. 1992. The role of the legume in the nitrogen cycle of pro-
nial polycultures under two management strategies. Crop Sci. 48:331–342. ductive and sustainable pastures. Grass Forage Sci. 47:133–142.
doi:10.2135/cropsci2007.04.0225 doi:10.1111/j.1365-2494.1992.tb02256.x
Ramseier, D., J. Connolly, and F.A. Bazzaz. 2005. Carbon dioxide regime, spe- Tracy, B.F., I. Renne, J. Gerrish, and M.A. Sanderson. 2004. Forage diversity and
cies identity and influence of species initial abundance as determinants weed abundance relationships in grazed pasture communities. Basic Appl.
of change in stand biomass composition in five-species communities: An Ecol. 5:543–550. doi:10.1016/j.baae.2004.08.007
investigation using a simplex design and RGRD analysis. J. Ecol. 93:502– Tracy, B.F., and M.A. Sanderson. 2004. Relationships between forage plant
511. doi:10.1111/j.1365-2745.2005.00999.x diversity and weed invasion in pasture communities. Agric. Ecosyst. Envi-
Riday, H., and K.A. Albrecht. 2012. Combining Kura clover with forage ron. 102:175–183. doi:10.1016/j.agee.2003.08.002
legumes and grasses to optimize pasture forage legume content. Agron. J. Zemenchik, R.A., K.A. Albrecht, and R.D. Shaver. 2002. Improved nutri-
104:353–362. doi:10.2134/agronj2011.0088 tive value of kura clover-and birdsfoot trefoil-grass mixtures com-
pared with grass monocultures. Agron. J. 94:1131–1138. doi:10.2134/
agronj2002.1131

Agronomy Journal • Volume 105, Issue 5 • 2013 1297