Effects of spring wheat / faba bean mixtures on early crop development


 
 Intercropping cereals and grain legumes has the potential to increase grain yield in comparison to the respective sole crops, but little is known about mixture effects at the early crop developmental stage. In cereal legume mixtures, the cereal is usually the dominating partner. We aimed to find out when domination starts, which factors may enhance early domination, and if there is a legacy effect of early domination on later growth stages.
 
 
 We set up field trials at a low input conventional site in 2020 and an organic site in 2020 and 2021. Treatments included all possible monocultures and 1:1 mixtures of twelve spring wheat (SW) entries, and two faba bean (FB) cultivars. All combinations were each sown in two sowing densities. To measure the effect of the mixture on early crop development, we counted crop emergence (plant m-2) at ~ 23 days after sowing (DAS) and crop biomass dry matter at ~ 52 and ~ 82 DAS.
 
 
 We found positive mixture effects on SW emergence at the conventional site and on SW and FB emergence at the organic site in 2021. Spring wheat was the dominating partner in all three environments; SW domination, without suppressing FB, was already noticed at emergence at the conventional site. There, a small head start of SW at emergence favored dominance at later growth stages and lead to superiority over FB in terms of plant biomass.
 
 
 Understanding early dominancy as observed here may help managing competition in mixture to enhance complementarity and improve productivity.



Introduction
Intercropping, i.e. the mixed cultivation of crop species, is recognized as a pathway towards more sustainable agriculture (Li et  Crop emergence, i.e. the emergence of the shoot from a germinated seed through soil, is the rst essential stage in the life cycle of plants because this process often affects components of plant tness (Verdú & Traveset, 2005) and determines the future crop performance both at the individual (plant) and the population (crop stand) level. Presumably the nal yield bene t of cereal legume crop mixture compared to their monocrops could not appear without earlier below-and aboveground competition and/or facilitation processes. In barley and pea intercropping, for example, the reduction in yield of pea has been found likely to be induced in early growth stages (To nga et al., 1993). However, the exact proportion and direction of the mixture effects on that early growing stage are currently largely unknown.  (Orrock & Christopher, 2010;Schiffers & Tielborger, 2006) between seedlings. Indeed, facilitation was observed to enhance seedling survival and growth, especially in extreme weather events such as harsh winters (Batllori et al., 2009). By accelerating or reducing the emergence rate, crop mixtures might have a decisive role on plant biomass in the early growth stage, and thereby also on the balance of competition among partners at later growth stages.
Crop mixtures of cereals and grain legumes have a high potential of increasing the system productivity compared to the respective sole crops (Xiao et al., 2018). In general, in such crop mixtures, cereals have been shown to gain a higher relative yield than grain legumes (Xiao et al., 2018;Yu et al., 2016). However, estimating the nal yield as an overall growth dynamic is obviously insu cient for drawing conclusions about lifetime dominancy over the whole growing stage of the cereal over the legume. It is currently unknown when the domination of the cereal over the legume starts, in particular, whether this domination is evident already at an early growing stage. Further, it is not known if there is a legacy of early dominance (if any) on later performance of the crops. It is important to understand the competitive balance between the two intercropping partners at an early growing stage because this may play a major role in determining productivity in mixtures. To understand the complexity of crop mixtures and predict their productivity and the relative performance of the partners, it is vital to study the entire life cycle of the interacting plants from crop emergence.
We therefore conducted an experiment to test the mixture effect of spring wheat/ faba bean crop mixtures of emerging seedlings in three different environments. In particular, we tested the following four hypotheses: (H1) There is a positive mixture effect of spring wheat/faba bean crop mixture at the early growing stage (emergence); (H2) The early mixture effect depends on factors such as SW cultivar, FB cultivar, environment, and sowing density; (H3) Domination of SW in the mixture can be observed already at crop emergence; here, by domination we mean that the relative proportion of dominating partner in the mixture is higher compared to the other partner; (H4) There is a legacy of early mixture effects on later growth stages as measured by crop biomass, i.e. the strength of domination of one partner over the other at crop emergence is carried over to later stages.

Experimental Site
The eld experiment was carried out in three environments, namely in one year (2020) at Campus Klein-Altendorf (CKA) and in two years (2020 and 2021) at Wiesengut (WG), both research stations of the University of Bonn, Germany. CKA is a conventionally managed research station located at 50°36'North, 6°59'East, and at an altitude of 186 m above sea level (a.s.l.) in Rheinbach, about 40 km south of Cologne. The site is characterized by a fertile Haplic Luvisol with a loamy silt texture with a eld evaluation index ("Ackerzahl") between 85 and 90. The German "Ackerzahl" evaluates the soil's quality ranging from 1 (very poor quality) to 100 (highest quality). The small variance in the values of the CKA eld evaluation index demonstrates relatively homogeneous soil conditions at that site. WG, an organically managed research station, is located at 50°47'North, 7°15'East at an altitude of 65 m a.s.l. in the lowland of the river Sieg near Hennef. The soil type is a Fluvisol with a silt loam texture on gravel layers, a eld evaluation index between 20 to 70 and uctuating groundwater level. This wide range of values for the eld evaluation index shows the heterogeneous soil conditions across different sites at WG. Within WG farm, however, the eld trials in 2020 and 2021 were placed to minimize within-trial heterogeneity of the soil. Compared to CKA, total organic carbon (C t ) and total N (N t ) concentrations in the soil were higher at WG. Mineral N (N min ) concentrations in 2020 were higher at CKA than at WG, and the latter site, higher in 2021 than in 2020 (Table 1). The average yearly temperature and annual rainfall at the experimental sites was 10.3°C and 669 mm at CKA, and 10.7°C and 732.8 mm at WG between 1991 and 2020, respectively. In April and May, the most relevant months for early crop development of the spring sown trial crops, precipitation was relatively low in 2020, but higher in 2021 with an average temperature mean at both sites (Fig. S1). Soil temperature at 15cm soil depth during April and May was higher in 2020 (15.52°C at WG and 12.28°C in CKA) compared to 2021 (10.14°C in WG).

Experimental Design
The trial was replicated in three environments; at the organic site (WG) in 2020 and 2021 and the conventional site (CKA) in 2020, below referred to as WG2020, WG2021 and CKA2020. Each of the three eld experiments was performed as a randomized complete block design with four replicates and four experimental factors. Factor A compared the mixture of spring wheat (SW) and faba bean (FB) with the respective monocultures. In the mixtures, both crop species were mixed within the row during sowing (see section 2.4 for sowing technique). Factor B was the FB cultivar with two levels (cv. Mallory and cv. Fanfare). Factor C had 12 levels of different SW entries, using ten spring SW cultivars ( Table 2) Table 2). The required amounts of seeds for each cultivar within the group were calculated separately according to their seed weight (thousand kernel weight, TKW), germination percentage (GP) (Table S1), plot area, and sowing density (equation S3). Finally seeds of the ve cultivars were mixed homogeneously before sowing. Factor D varied the sowing density and had two levels, 120% and 80% of the recommended sole crop densities (%RD) of 400 seed m − 2 for SW and 45 seed m − 2 for FB (Table 3). In total, the experiment included 76 treatments per block and both crops, SW and FB. The monocultures of FB were doubly replicated to stabilize comparisons with the numerous mixtures. This resulted in 320 plots for each of the three eld trials.

Management Practices
After harvesting the pre-crop (Table 4), the soil was ploughed once (30 cm depth, with a moldboard plough) in WG and twice (7 and 10 cm depth, with a chisel plough) in CKA, and the seedbed was prepared with a rotary harrow at both sites. The single plot size was 1.5 m × 10 m with 6 rows and 21 cm row to row distance. FB was sown rst, at 6 cm depth; to maintain the appropriate FB sowing density, a direct seeding machine type Hege 95 B was used and optimized by rotation setting according to sowing density (Table S2). Thereafter, SW was sown directly over FB with a Hege 80 seeder at 3 cm soil depth. No seed treatment was performed on FB or SW before sowing. The trials were run with no use of chemical fertilizers, herbicides, or pesticides. In both years, mechanical weeding was carried out twice (3 and 5 weeks after sowing) using a hoe, and once four weeks after sowing with a harrow.

Measurements
In order to monitor the emergence and early growth of the crop, crop emergence (plant m − 2 ) was measured by counting plants on de ne areas per plot about 23 days after sowing (DAS), and biomass measurements (dry matter, DM, in t ha − 1 ) were performed twice about 52 and 82 DAS, respectively; the exact times in days after sowing (DAS) are reported in Table 4. For all measurements, 1m long sections from the 3rd and 4th rows of the plots were chosen after discarding 1m for biomass and 3m from the 3rd rows and 4m from the 4th rows for crop emergence from the edge to avoid potential edge effects.
Counting of emerged seedlings and the rst biomass cut was performed from one side of the plot whereas a second biomass cut was performed from another side of the plot (Fig. 1).
During the rst biomass cut, only 12 treatments (all FB monocultures, FB mixtures with SW cultivars Lennox and SU Ahab, and SW monocultures with cultivars Lennox and SU Ahab; which makes 48 plots in total out of 320 plots per site) were taken. During both biomass cuts, both crop and weed aboveground biomasses were taken at the base. Fresh biomass samples from the sole crops were manually separated into weeds and crops; samples from the intercrops were separated into weeds, SW, and FB before further processing. Biomass dry weight (t ha − 1 ) of the samples was obtained by oven-drying at 105°C until weight constancy.

Data Processing and Calculations
To quantify the effect of the mixture in comparison to the monocultures, land equivalent ratio (LER), partial LER (PLER), and competitive ratio (CR) were calculated for the crop biomass. The LER of a mixture measures the relative land area that is required for the crop monocultures to produce the same biomass as observed in the mixture; it was calculated as the sum of the PLERs of the two species (Equations 1 and 2) in the mixture by using Eq. (3) (6) To measure the competitive ability of intercrops, to identify the dominant partner producing relative higher proportion of biomass in the mixture than the other partner and to indicate the degree of dominance, a simple competitive ratio (CR) is considered. CR refers to the ratio of the PLERs of the two partners in the intercropping (Willey & Rao, 1980), and, by analogy, to the ratio of the PEERs.

Statistical Analyses
We examined the mixture effects at the early growth stage by using a two-sided Wilcox test against PEER above 0.5. Dominance of one partner over the other at emergence as well as at later biomass was assessed by comparing the two partners using a two-sided Wilcox test of PLER and PEER. PLER and PLER were non-normally distributed according to the Shapiro-Wilk test and the variances between the groups were not homogeneous according to Fisher's F test which was complemented by graphical assessments. In cases when data was strongly non-normal and could not be normalized by transformations, non-parametric Kruskal-Wallis test was used followed by Dunn test with Bonferroni correction for multiple comparisons.
Analysis of variance (ANOVA) was carried out to test the effects of the mixture on cultivar spring wheat, cultivar faba bean, environment, and sowing density. In the rst step, a model was prepared considering all possible interactions between those four factors. Non-signi cant factors or interactions were iteratively removed from the model to improve model performance according to Akaike's Information Criterion (AIC) (Burnham et al., 2011). The nal model included two independent variables spring wheat cultivar and environment, and their interaction. An alternative ANOVA model based on the logtransformed data was also carried out; its results are provided as supplementary information.
We assumed a legacy of early mixture effects on later biomass. To verify this, we tested for a correlation (Pearson) and linear regression model between the competitive ratio (CR) at emergence and CR at later crop biomass. All statistical analyses were conducted in RStudio version 1.4.1106 (R Core Team., 2020).

Results
Mixture effects at the early growth stage Although the expected crop emergence was the same in the three environments, the observed emergence was highly different between the three trials (Table 5). At CKA2020, the observed SW emergence was much lower for both densities compared to the other two environments (WG2020 and WG2021, Table 5 and Fig. 2). Compared to the expected crop emergence, the observed SW emergence at CKA2020 in monoculture was 46.1% at low density and 46.2% at high density. In mixture, the observed SW emergence at CKA compared to the expected SW emergence was comparatively higher than in monoculture. We observed a high variation of SW emergence at CKA2020 compared to the other two environments (WG2020 and WG2021; Fig. 2A). At the environment WG2021, a relatively low FB emergence was noticed compared to the expected emergence (in monoculture 83.3% and 81.4% for high density (HD) and low density (LD), respectively; and in mixture 86.2% and 94.7% in LD and HD, respectively; Table 5 and Fig. 2F).
Depending on the trial, we marked positive mixture effects already at crop emergence (PEER > 0.5) about 23 days after sowing (DAS) ( Table 6). In particular, the mixture effect was consistently positive for SW at both densities at the environment CKA2020 where the observed SW emergence was low (Table 5). At WG2021, the mixture effect was positive for both species (FB and SW) and both densities (LD and HD) where the observed FB emergence was lower compared to the expected FB emergence. Neither species nor densities showed any positive effect at WG2020 (Table 6).  (Table S3).
Step by step all non-signi cant factors and interactions between those factors were removed from the model to improve model performance according to Akaike's Information Criterion (AIC). The nal model included two independent variables SW cultivar and environment, and their interaction (Table 7). An alternative minimal model based on the logtransformed data was similar to the one presented in Table 7, but additionally contained a signi cant interaction of FB cultivar and density (Table S4).

Spring wheat dominance at the early stage
At the environment CKA2020, domination of SW over FB was already observed at crop emergence (Fig. 4A), as the mean partial emergence equivalent ratio (PEER) of SW was 0.11 higher than FB mean PEER (Wilcox test; P < 0.001). This level of domination of SW over FB at CKA2020 increased over time (Fig. 4B and 4C). The PLER mean difference between SW and FB at CKA2020 increased to 0.4 (Wilcox test; P < 0.001) at about 52 DAS and to 0.50 (Wilcox test; P < 0.001) at about 82 DAS.
Although no signi cant PEER mean differences between SW and FB were observed during the early developmental stage at WG2020 and WG2021, the SW domination over FB changed over time for these environments (Fig. 4). At WG2021, a signi cant mean difference between SW and FB was rst observed about 52 DAS (PLER mean difference 0.11; P < 0.05) which increased by about 82 DAS to a PLER mean difference of 0.31 P < 0.001; Fig. 4B and 4C). On the other hand, at WG2020, a highly signi cant mean difference between SW and FB PLER was only observed after about 82 DAS (PLER mean difference of 0.25; P < 0.001; Fig. 4C).

Legacy of early mixture effects on later biomass
In cereal legume crop mixtures, the cereal is known to be a dominant partner by achieving a higher relative yield than grain legumes. However, it is not known if the degree of early cereal domination affects its domination at later biomass production, i.e. if there is any legacy of early mixture effects on later biomass. To estimate the legacy effect, SW competitive ratio (CR SW ) was calculated by Eq. (7) and the relationship between CR SW at emergence vs. later CR SW was tested with linear regression. There was a strong linear relationship between the two CR SW values at CKA2020 for both densities (P < 0.001, Fig. 5 and Table 11), i.e. the higher the proportion of SW early on in SW-FB mixtures, the higher was its dominance later in the season. In addition, dominance, as measured by the CR SW , increased over time. In the two other environments (WG2020 and WG2021) and two densities (high and low), only a weak signi cant correlation was observed, which at WG2020 for high density was signi cant at p < 0.05 (Table 11). Signi cance codes: ***= P < 0.001, **= P < 0.01, * = P < 0.05 and ns = not signi cant.

Discussion
Mixture effects at the early development stage Because of the small size of the young plants and the short time of coexistence for the two species it is not unreasonable to expect that at the early crop developmental stage there is a lack of interference and therefore a neutral mixture effect. In contrast to this expectation, however, this study showed clear evidence of positive mixture effects at this stage in some of the trial environments. In two out of three environments, we observed higher crop emergence in spring wheat/faba bean mixtures compared to their respective monocultures. The strongest mixtures effects were found in those environment-crop combinations where absolute crop emergence was low ( Fig. 2A; Tables 5, 6), but we were unable to identify the exact reasons for spring wheat emergence being low at CKA2020. Although in this environment SW emergence was low (Table 5), its PEER was high (Table 6), i.e. higher relative SW emergence in mixtures compared to monocultures. It may be assumed that the higher SW emergence in mixtures was due to the temporal complementarity effect in resource use between SW and FB (Li et al., 2016;Xiao et al., 2018) in the very early stage of plant development (Elsalahy et al., 2021). In our eld experiment, we observed that FB was slower to germinate than SW, potentially because of the (necessarily) deeper seed depth of FB (6 cm) than of SW (3 cm). In the mixture, this asynchronous germination between SW and FB may allow SW to temporarily access more resources and induce higher germination than in monoculture (Elsalahy et al., 2021).
Results also showed the same pattern of high mixture effects and low absolute emergence in the case of WG2021 where a lower FB emergence at high density and a positive mixture effect for both species was observed. These phenomena may be explainable by compensation effects in the mixture (Elsalahy et al., 2021). The mechanism of compensation means that if one of the partners in a mixture fails, the other partner takes its place; it has been shown to lead to high-yield stability in crop mixtures in response to different environmental stresses (Creissen et al., 2013). More speci cally, with regard to emergence, a locally acting mortality factor that only affects one partner will reduce competition for the remaining partner and nally may induce higher emergence or survival in the mixture. A positive mixture effect was also observed at CKA2020 for FB emergence at low density.
Our results provide evidence of compensation by one species; this may be particularly important for the low emergence of species in case of climate extremes and increasing environmental stresses. However, further experiments with different combinations of other species are required to con rm the generality of our results.
Dependence of mixture effects on cultivar identity, environment and density The mixture effect is an outcome from a complex combination of different factors and the interactions between them. In this study, the SW cultivars, environments and their interactions turned out to be the factors in uencing the mixture effects on crop emergence. Higher differences among SW emergence were observed at the low input conventional site (CKA2020) than at the organic sites (WG2020 and WG2021). At CKA2020, the SW cultivars Lennox and Anabel and at WG2021, the SW cultivar SU Ahab showed signi cantly higher crop emergence in mixture than in monoculture. While the existence of mixing potential in some SW cultivars may been indicated by this nding, the identity of cultivars and the direction of their differences was not consistent across environments. Furthermore, the effect of mixture completely disappeared for all SW cultivars in WG2020.
Although a wide range of SW cultivars are currently available for sole cropping, there is lack of cultivars suitable for intercropping (Bančič et 4A). On the other hand, dominance does not necessarily mean suppression of the other partner. The relative emergence of SW was 16.4% higher than FB (Table 6) at that stage, indicating dominance, but the PEER value of FB (0.51 at LD and 0.50 at HD; Table 6) showed no reduction of FB emergence in the mixture compared to FB monoculture. The higher emergence of SW in the mixture than in the SW monoculture was observed presumably due to the complementarity effect (Xiao et al., 2018), and could be linked to asynchronous germination between the two species (Elsalahy et al., 2021). At the same time, the observed dominance of SW did not lead to the suppression of FB emergence in the mixture possibly due to the small size of the cereal, and the short time of co-growth between the two species.
At WG2021, the domination of SW started about 52 DAS with a slight suppression of FB biomass (PLER of 0.47; Fig. 4B) where we also noticed low FB emergence in both the mixture and the monoculture, especially at high density (Fig. 2F). One mechanism to explain this result could be that crop failure of FB may have relaxed competition and thereby enhanced the domination of SW in the mixture. In line with this, a study with grassland species, in which Asteraceae were sown together with different neighboring species, reported the dominance of these companion species to be particularly strong after a severe drought disturbance (Fenesi et al., 2020), which was linked to the failure of the weaker species. Finally, the SW dominated the mixture about 82 DAS in all three environments by suppressing FB biomass by 38%, 16%, and 18% at CKA2020, WG2020, and WG2021, respectively, compared to monoculture (Fig. 4C).
Characteristics that allowed rapid access of SW cultivars to environmental resources are likely to be the determining factors leading to the dominance at a later stage. A rapid growth because of deeper root

Legacy of early mixture effects on later biomass
Despite the fact that the SW domination was detected at all three environments at about 82 DAS, the degree of domination was different among those three environments (Fig. 4C). The higher degree of SW domination (PLER 62% higher than monoculture; Fig. 4C) and a higher degree of FB suppression (PLER 38% lower than monoculture; Fig. 4C) was observed at CKA2020 where we also recorded an early dominating effect of SW in the mixture about 23 DAS. That means the larger the proportion of SW seedlings early on in the SW-FB mixture, the higher SW domination and FB suppression later in the season. The domination intensity of SW for both densities exhibited a consistent expanding trend through the overall growth period at CKA2020 (Fig. 5A). It has been reported that earlier emerging seedlings of dicotyledonous sand dune annual plants tended to become larger adults (Turkington et al., 2005). In line with this, also other studies found that the strength of superiority that started at a very early developmental stage expanded as plants grew (Benincasa et al., 2012;Mangla et al., 2011;Schiffers & Tielborger, 2006). The increasing intensity of SW domination over FB was also observed at WG2020 for low density (Fig. 5B).
While some experiments were conducted on grassland species regarding the direction and strength of

Conclusions
Our study highlights that positive mixture effects may be detected at the early developmental stage but these mixture effects depend on the environmental conditions and selected SW cultivars. The results also suggest that SW and FB crop mixtures may also improve the ability of crop emergence to buffer environmental stresses, especially during the partial crop failure of one partner. Spring wheat domination in the mixture was noticed in all the environments but the domination was initiated and intensi ed through low crop emergence of SW cultivars.
Cereals-grain legumes crop mixtures have a high potential to bring stability in productivity compared to the respective sole crops. However, due to the lack of cultivars suitable for mixed cropping, this experiment depended on the cultivars currently available in the market which have been bred for sole cropping. Therefore, it is recommended to expand breeding SW cultivars targeted at mixed cropping systems. On this way, our insights may also be used to develop an effective indirect selection tool for crop breeding purposes. More generally, a better understanding of the mechanisms of early domination and its intensity changing over time may help to improve the development and management of diversi ed cropping systems towards sustainable agriculture.
Declarations Figure 1 Schematic illustration of a single plot showing the locations where the individual measurements were taken.

Figure 2
Page 24/27 Crop emergence (plant m -2 , each symbol representing one plot) of spring wheat (SW, panels A, C, and E, green symbols) and faba bean (FB, panels B, D, and F, blue symbols) in monocultures (mono) and mixtures (mix) for high ( lled circles) and low (open triangles) sowing densities in the environments CKA2020 (A, B), WG2020 (C, D) and WG2021 (E, F). Points above the black diagonal (PEER = 0.5) indicate higher crop emergence in the mixture than in monoculture. Sown high and low densities are represented by solid and dashed lines, respectively.

Figure 3
Mean of emergence equivalent ratio (EER) of 12 levels of different spring wheat entries (ten cultivars and two 5-component mixtures of these wheat cultivars) in three different environments. Within each environment, SW cultivars with different letters are signi cantly different according to Tukey's HSD test.    (Table 11). SW competitive ratio (CR SW ) was calculated by Eq. (7) and Eq.