Wheat-pea species mixtures as resource efficient and high-performance food cropping systems: Evaluation of contrasting wheat genotypes

Abstract

High fertiliser inputs for good yields and baking quality of wheat may cause nitrogen runoff. Cereal-legume mixtures have a range of advantages in terms of land and nitrogen use efficiency and protein content. We experimentally evaluated 15 contrasting wheat genotypes for their performance in mixture with pea and in monoculture without mineral N-fertilisation. All mixtures had land equivalent ratios above 1.0 and increased wheat protein. All but two wheat genotypes increased their quality either from fodder to baking or baking to top-baking quality, showing high relevance for practice. In contrast, wheat monocultures were constrained by a strong yield protein trade-off. The highest-ranking wheat genotypes are most often early cultivars from Hungary. Wheat genotype-based variation in performance could be harnessed for variety selection or breeding. Targeted strategies are needed to optimise threshing, separation and/or maximum allowance of pea residuals accepted in baking wheat.

Full text

Aspects of Applied Biology 146, 2021
Intercropping for sustainability: Research developments and their application
1
Wheat-pea species mixtures as resource ecient and
high-performance food cropping systems: Evaluation of
contrasting wheat genotypes
By TIMAEUS JOHANNES, WEEDON ODETTE and MARIA R FINCKH
Kassel University, Department of Ecological Plant Protection, Mönchebergstraße 5,
34127 Kassel, Germany
Summary
High fertiliser inputs for good yields and baking quality of wheat may cause nitrogen
run-o. Cereal-legume mixtures have a range of advantages in terms of land and nitrogen
use eciency and protein content. We experimentally evaluated 15 contrasting wheat
genotypes for their performance in mixture with pea and in monoculture without mineral
N-fertilisation. All mixtures had land equivalent ratios above 1.0 and increased wheat
protein. All but two wheat genotypes increased their quality either from fodder to baking
or baking to top-baking quality, showing high relevance for practice. In contrast, wheat
monocultures were constrained by a strong yield protein trade-o. The highest-ranking
wheat genotypes are most often early cultivars from Hungary. Wheat genotype-based
variation in performance could be harnessed for variety selection or breeding. Targeted
strategies are needed to optimise threshing, separation and/or maximum allowance of pea
residuals accepted in baking wheat.
Key words: Crop ecology, species mixtures, intercropping, baking quality, resource
eciency, trade-os, resource complementarity
Introduction
High-input fertilisation practices result in large amounts of nitrogen that are not consumed by
crops but leached into the environment (Häußermann et al., 2019). Wheat is - in terms of yield
and especially baking quality - a nutrient demanding crop. In addition, wheat traders assess baking
quality of wheat based on protein content to determine the wheat price payed to farmers. This creates
pressure to increase yield by early and protein content through late fertilisation potentially resulting
in nitrogen run-o. Ecological wheat cropping systems should aim simultaneously for (1) good
yields, (2) baking quality and (3) high nitrogen use eciency. Robust empirical evidence suggests
that wheat-pea species mixtures have the potential for resource ecient and high-performance food
cropping systems in terms of total yield, land use eciency, wheat protein content and reduced
weed pressure and lodging compared to sole crop legumes (Bedoussac et al., 2015). A recent study
suggests that intercropping cereals and legumes can decrease the required fertiliser globally by 26%
compared to sole crop legumes (Jensen et al., 2020). Ecological theory and empirical evidence
suggests fundamental trade-os with respect to ecological adaptations within each single species
(Tilman, 1990). From this point of view monocultural cropping systems focusing either on cereals
or legumes are limited by trade-os with respect to the specic resource use strategy for nitrogen

2
and underlying morphological and physiological adaptations. Specically, in monocultural wheat
cropping systems, empirical evidence supports a strong trade-o between wheat yield and protein
content (Simmonds, 1995). Possibly, monocultural cropping systems are caught in these within-
species trade-os both from a resource acquisition and performance perspective and intercropping
could represent an elegant way of circumventing these.
The goal of this study was to assess the performance of wheat-pea species mixtures with contrasting
wheat genotypes compared to the respective pure stands. Specic research questions were: (1) Do
positive mixture eects such as protein increase for wheat and land equivalent ratios (LER) above
unity hold true for a range of contrasting wheat genotypes allowing generalisation for wheat-pea
mixtures? (2) Is there wheat genotype-based variation in performance that can be harnessed for
variety selection or breeding indicated by: (a) genotype-based variation in pea yield and LER (one-
factorial evaluation) or (b) do wheat genotypes react dierently comparing a mixed and pure stand,
i.e. are the best performing genotypes in monoculture also the best performing in mixtures (system
genotype interaction)? Is there evidence suggesting that species mixtures transcend within species
trade-os, i.e. increasing total yield and protein content in mixtures compared to monocultures?
Results of the season 2018/2019 are reported here.
Materials and Methods
The experiment was set up at the Kassel University research station, Neu-Eichenberg, Germany
(51°22’24.7” N and 9°54’12.5” E, 247 m above sea level). The research elds are classied as
having ne loamy loess soil (Haplic Luvisol) with 76 soil points according to the German soil
classication system (0–100). Mean annual temperature from 2013–2019 was 9.6°C and mean
annual precipitation 560 mm. Sowing rate in mixtures was 70% wheat and 50% pea of the pure
stand sowing density. The experiment was set up as a split-plot design with four replicates, 15 wheat
genotypes each in pure and mixed stands with winter pea cultivar Fresnel. Plot size was 19.5m2 (13
× 1.5 m). Contrasting wheat genotype groups included western European/Hungarian genotypes,
composite cross populations (CCPs) and organically and conventionally bred line cultivars. No
fertiliser was applied for all treatments. Statistical analysis was carried out as mixed eect modeling
either with the lme4 (homogeneous variances) or nlme (heteroskedasticity) package for R (R. Core
Team, 2020). NIRS-based quality measurements (protein, gluten, sedimentation) were conducted
with a Foss Infratec 1241 Grain Analyser. NIRS results were used to categorise genotypes into
wheat quality groups based on the thresholds given for protein, gluten and sedimentation by BLE
(2011, see Table 1).
Table 1. Quality groups for wheat based on BLE (2011)
Quality groups Protein content (%) Gluten (%) Sedimentation (ml)
Baking wheat 1 > 11 > 26 > 35
Baking wheat 2 10–11 22–26 > 25–35
Fodder wheat <10 <22 <=25

3
Fig. 1. Pea grain yields in t ha-1. Shown are estimated marginal means from linear mixed models. Error bars
indicate standard errors. Letters indicate signicant dierences at (P<0.05) estimated from linear mixed
eects model with pairwise comparison and Holm correction.
Fig. 2. Wheat grain yields in t ha-1. Shown are estimated marginal means from linear mixed models. Error
bars indicate standard errors. Letters indicate signicant dierences at (P<0.05) estimated from linear mixed
eects model with pairwise comparison and Holm correction.
Results
Average emergence rate of peas was 90 seeds m2 in monoculture and in mixture 44 seeds m2,
which was within the expected range. Average pea grain yield was 2.0 t ha-1 in monoculture and
1.6 t ha-1 in mixture. Average relative pea yields across all variety combinations was 0.77 which is
considerably higher than expected from pea sowing density (50%). Pea yield varied signicantly
from 2.0–1.2 t ha-1 with wheat genotype (Fig. 1). The pea yields of the two highest yielding cultivar
combinations with the wheat varieties Karizma (2.0 t ha-1) and Elit CCP (2.0 t ha-1) did not dier
from the pea monoculture, but signicantly from the two lowest ranking cultivar combinations
CCP OQII (1.3 t ha-1) and Butaro (1.2 t ha-1) (Fig. 1).
Average emergence rate of wheat was 261 seeds m2 in monoculture and 181 in mixture and therefore
below absolute values expected (350 and 245 seeds m2 respectively), but the relative emergence
rate mix/mono of 70% was within the range of expectation. Emergence of Kolompos and Nemere

abcde cde abcde
e
ab
bcde
abcdeabcde abcd
bcde abcde
abc
abcde de abcde
a
0
0.5
1
1.5
2
2.5
Pea‐Achat
Pea‐Capo
Pea‐Butaro
Pea‐Wiwa
Pea‐Karizma
Pea‐Kolompos
Pea‐Nemere
Pea‐Toborzo
Pea‐Brandex
Pea‐BSFI
Pea‐BSFII
Pea‐Elitccp
Pea‐Liocharls
Pea‐OQII
Pea‐OYQII
Fresnel
conv.var. org.var. hung.var. CCP Monopea
Peagrainyield[t/ha]

**
********
*
****
0
1
2
3
4
5
6
Achat
Capo
Butaro
Wiwa
Karizma
Kolompos
Nemere
Toborzo
Brandex
BSFI
BSFII
EliteCCP
Liocharls
OQII
OYQII
mono
mix
conv.var. org.var. hung.var. CCP System
Mean
Wheatgrainyield[t/ha]
mix mono

4
Fig. 3. Land equivalent ratios (LER). Shown are estimated marginal means from linear mixed models. Error
bars indicate standard errors. Letters indicate signicant dierences at (P<0.05) estimated from linear mixed
eects model with pairwise comparison and Holm correction.
Fig. 4. Protein content of wheat kernels based on NIRS measurement. Shown are estimated marginal means
from linear mixed models. Error bars indicate standard errors. Letters indicate signicant dierences at
(P<0.05) estimated from linear mixed eects model with pairwise comparison and Holm correction.

abcde abcde
e
abcde
abc abcd ab
cde
abcd
bcde de
a
abcd bcde abcd
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
Achat
Capo
Butaro
Wiwa
Karizma
Kolompos
Nemere
Toborzo
Brandex
BSFI
BSFII
ElitCCP
Liocharls
OQII
OYQII
con.var. org.var. hung.var. CCP
LER

**
*
**
*
**
*
*
*****
*
0
2
4
6
8
10
12
14
16
Achat
Capo
Butaro
Wiwa
Karizma
Kolompos
Nemere
Toborzo
Brandex
ElitCCP
Liocharls
BSFI
BSFII
OQII
OYQII
Mixmean
Monomean
conv.var org.var. hung.var. CCP System
Proteincontent[%]
mix mono

5
Fig. 5. Wheat Quality groups based on thresholds for protein, gluten and sedimentation given by BLE (2011).
f = fodder wheat, b2 = baking wheat, b1 = high-quality baking wheat.
was above average in mixture (201 seeds m2 each). Average wheat grain yield in monoculture
was 4.2 t ha-1 and in mixture 3.1 t ha-1 with average relative yields of 0.74 being in the range of
expectation from 70% sowing density. All genotypes reacted with a signicant yield reduction in
mixture except cultivar Kolompos resulting in a system-genotype interaction. All over, genotype
wheat yield ranking between mixtures and pure stands changed only slightly.
Total yields and LER
Total yields including peas were on average 4.71 t ha-1 in mixture and 4.16 t ha-1 in monoculture,
the dierences were not statistically signicant. However, System-genotype interaction was
signicant and two genotypes, Elit CCP and Kolompos had statistically signicant higher total
yields in mixtures. Land equivalent ratio (LER) as indicator of resource eciency had an overall
mean of 1.5, ranging from 1.3–1.7 indicating a robust genotype independent increase above unity.
Signicant variation of LER was observed for wheat genotypes with the three top ranking genotypes
all being Hungarian early genotypes (Nemere>Karizma>Kolompos) (Fig. 3).
Protein content and wheat quality groups
Average wheat protein contents in mixture (10.4%) were signicantly higher by 1.6 percentage
points (13% increase) in mixtures compared to monocultures (12.0%) (Fig. 4). Protein increase was
robust for all variety combinations, ranging from 1.1–2.1 percentage points (10–20% increase) with
statistically signicant system-genotype interactions indicating some genotype specic variation.
Performance ranking of genotypes between mixture and pure stands did only slightly change.
Top-ranking in mixture were Toborzo>Nemere>Karizma>Wiwa>Elit CCP all being Hungarian
genotypes, except Wiwa. Protein increase was relatively large for Karizma (2.1%), Kolompos
(1.9%), Nemere (1.9%) and Elit CCP (1.8%) compared to BSFI (1.2%), Capo (1.1%), Liocharls
(1.2%), CCPs OQII (1.3) and OYQII (1.2%). All but two wheat genotypes improved their quality
groups in mixtures compared to pure stands from fodder to baking wheat or intermediate baking
to top baking quality, indicating a high relevance for practical farming (Fig. 5).

b1
b2
b1 b1
b2
b2 b2
b1 b1
b2 b2 b2 b2 b2 b2
fff
b2
b2
f
b2 b2
fff
ffff
Achat
Capo
Butaro
Wiwa
Karizma
Kolompos
Nemere
Toborzo
Brandex
ElitCCP
Liocharls
BSFI2
BSFII2
OQII2
OYQII2
conv.
var.
org.var hung.var. CCP
Wheatqualitygroups
mix mono

6
Discussion
Pea yields in monoculture reported here are relatively low (2 t ha-1) compared to up to 4 t ha-1
reported in the literature (Pelzer et al., 2012). Heavy lodging was observed in pea monocultures
possibly causing yield losses in pea monoculture resulting in high relative pea yields in mixtures.
This possibly explains a part of the high LER values. Wheat yields decreased in mixtures as expected
except for Kolompos. Part of that might be explained by a comparatively high emergence rate.
Generalisation of mixture eects across contrasting wheat genotypes
It appears that positive mixture eects of winter peas on wheat baking quality and LERs above
unity occur across contrasting wheat genotypes and can be generalised for wheat-pea mixtures.
The robust protein increase in wheat in mixtures conrms results from the literature (Bedoussac
et al., 2015). Ample evidence in the literature suggests this is caused by complementary resource
use strategies of cereals and legumes, and a change in competition for soil nitrogen for wheat in
mixtures compared to wheat monocultures. Cereals are highly ecient in soil nitrogen uptake
relative to legumes and the latter can draw on atmospheric nitrogen pools. If a part of a single species
wheat population is substituted by legumes, competition for wheat individuals for soil nitrogen
is reduced, increasing per plant available nitrogen. In contrast, for legumes competition for soil
nitrogen increases, facilitating increased rates of nitrogen xation. Pea cultivar Fresnel also was
mature earlier than the wheat cultivars possibly releasing nitrogen from senescent plants. These
explanations demonstrate the importance of conceptual models based on crop ecology to explain
crop community level performance, a point highlighted by Gliessmann (1987).
Genotype variation and breeding potential
For pea-wheat mixtures our data supports wheat-genotype based variation of pea yield and LER.
For wheat yield and protein content we found signicant system-genotype interactions pointing
to some potential to optimize variety selection and harness genotype-based variation for breeding.
Earliness and performance
All Hungarian genotypes were found to be early heading compared to western European genotypes
(Timaeus et al., 2021). The early Hungarian genotypes often occupy the highest performance
ranks with respect to pea yield, LER and protein content. This implies that traits contributing to
fast plant development contribute to an increase in these performance indicators. This possibly can
be explained by more ecient nitrogen uptake conferred by high early vigor (Liao et al., 2004).
The very early cultivar Toborzo had low yields and LER indicating a yield trade-o for extreme
earliness. However, the year 2018 was exceptionally dry also giving the Hungarian variety an
advantage. The expected trends with climate change will possibly continue favoring such genotypes
also in Germany.
Performance trade-os in species mixtures
Strong trade-os between wheat yield and protein content monocultural wheat cropping systems
hamper to increase both yields and protein content simultaneously (Simmonds, 1995). Our data as
well as the literature supports that both, total yields (including wheat and pea) and wheat protein
quality can increase in mixtures simultaneously compared to monocultures. Trad-os between total
yield and protein quality might still persist in mixtures but likely their performance limits shift
beyond the monocultural performance limits.
Strategies towards practical implementation
The improvement in quality groups shows that the mixture eect could be highly relevant
for practical farming and the food sector. A major obstacle for food grain mixtures is their use
downstream in the food supply chain. Industrial baking processes usually rely on high purity

7
ingredients combined in precise recipes. One strategy would be to accommodate these needs by
combining a high purity separation process with an optimised threshing process avoiding broken
peas that increase separation eorts. Another (perhaps complementary) strategy would be to accept
wheat and pea yields a certain amount of peas in baking wheat. Up to 10% peas in a baking mixture
still result in good baking and taste properties (Dabija et al., 2017). Allowing for a dened pea
residual could decrease separation eorts and deliver a product with dened composition without
interfering with baking quality. In the end, mixture benets will have to outweigh increased mixture
eorts but there seems to be a large unused potential to optimize the value chain for species mixtures.
Acknowledgements
The work described here was conducted within the EU project ReMIX. This project has recieved
funding from the European Union’s Horizon 2020 Research and Innovation Programme under
Grant Agreement N. 727217.
References
Bedoussac L, Journet E-P, Hauggaard-Nielsen H, Naudin C, Corre-Hellou G, Jensen E S,
Prieur L, Justes E. 2015. Ecological principles underlying the increase of productivity achieved by
cereal-grain legume intercrops in organic farming. A review. Agronomy for Sustainable Development
35:911–935.
Dabija A, Codina G G, Fradinho P. 2017. Eect of yellow pea our addition on wheat our
dough and bread quality. Romanian Biotechnological Letters 22:12888.
Gliessman S R. 1987. Species interactions and community ecology in low external-input
agriculture. American Journal of Alternative Agriculture 2:160–165. https://doi.org/10.1017/
S0889189300009279.
Häußermann U, Bach M, Klement L, Breuer L. 2019. Sticksto-Flächenbilanzen für Deutschland
mit Regionalgliederung Bundesländer und Kreise – Jahre 1995 bis 2017, pp. 167. In Methodik,
Ergebnisse und Minderungsmaßnahmen. Eds Uwe Häußermann, Martin Bach, Laura Klement and
Lutz Breuer. Dessau-Roßlau, Germany: Umweltbundesamt.
Jensen E S, Carlsson G, Hauggaard-Nielsen H. 2020. Intercropping of grain legumes and
cereals improves the use of soil N resources and reduces the requirement for synthetic fertiliser
N: A global-scale analysis. Agronomy for Sustainable Development 40(5). https://doi.org/10.1007/
s13593-020-0607.
Liao M, Fillery I R, Palta J A. 2004. Early vigorous growth is a major factor inuencing nitrogen
uptake in wheat. Functional Plant Biology 31:121–129.
Pelzer E, Bazot M, Makowski D, Corre-Hellou G, Naudin C, Al Rifaï M, Baranger E,
Bedoussac L, Biarnès V, Boucheny P. 2012. Pea–wheat intercrops in low-input conditions combine
high economic performances and low environmental impacts. European Journal of Agronomy
40:39–53.
R Core Team. 2020. A language and environment for statistical computing. Vienna, Austria: R
Foundation for Statistical Computing.
Simmonds N W. 1995. The relation between yield and protein in cereal grain. The Journal of the
Science of Food and Agriculture 67:309–315.
Tilman D. 1990. Constraints and tradeos: toward a predictive theory of competition and succession.
Oikos, pp. 3–15.
Timaeus J, Weedon O, Backes G, Finckh M R. 2021. Plant traits, plasticity and growth dynamics
in wheat-pea species mixtures: evaluation of contrasting wheat genotypes. Aspects of Applied
Biology 146, Intercropping for Sustainability 2021, pp. xx–xx.

8