Temporal dynamics of biodiversity effects and light‐use‐related traits in two intercropping systems

Funding information Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung, Grant/Award Number: PP00P3_170645 Abstract Introduction: Intercropping systems can be more productive than their respective monocultures and this positive net biodiversity effect is caused by complementarity and selection effects. While the complementarity effect is caused through resource partitioning or facilitation, the selection effect operates via the greater probability that a more diverse community contains a dominant and high‐yielding species which will account for the majority of productivity in that community. Here, we investigated how light‐use‐related traits contribute to the net biodiversity effect via complementarity or selection effects and how these qrelationships change throughout an annual growing season. Materials and Methods: We conducted weekly destructive harvests to examine temporal dynamics of biodiversity effects in two crop mixtures (oat–lupin and oat–camelina) and their respective monocultures. We linked the biodiversity effects to traits related to light use (i.e., light interception, plant height, photosynthetic efficiency and photosynthetic capacity) and investigated how these relationships changed over time. Results: We found that the net biodiversity and selection effect increased over time in both mixtures, while complementarity effects increased only in the oat–lupin mixture. More intercepted light and taller plants in mixtures compared to monocultures positively contributed to biodiversity effects in both mixtures. Strategies for shade tolerance differed between the mixtures, that is, increased photosynthetic capacity and increased photosynthetic efficiency contributed to a positive net biodiversity effect in the oat–lupin and oat–camelina mixture, respectively. Conclusion: By linking the temporal dynamics of the net biodiversity effect and its two additive components to light‐use‐related traits in two different crop mixtures, this study demonstrates that complementary light use contributes to overyielding in intercropping systems. Such understanding is important for the design of effective intercropping systems and developing new crop cultivars suited to these environments.


| INTRODUCTION
In modern agriculture, high productivity often comes at the price of sustainability. 1 A key strategy to implement sustainable agriculture is to restore on-farm biodiversity through diversified farming systems. 2 One route to increase biodiversity in agricultural systems is intercropping, where at least two crop species are cultivated on the same field at the same time. Intercropping aims to sustainably increase yields through benefits such as improved resource capture and lower artificial inputs. 3 Resource partitioning is considered a driving force for positive biodiversity effects in diverse plant communities. 4 This resource partitioning can occur above-and belowground and minimizes the niche overlap between species and thus enables an increased resource capture in the crop mixture compared to the monoculture. While many studies have observed partitioning of belowground resources, 5,6 evidence that these processes contribute to positive biodiversity effects remains limited. 7 This suggests that complementary use of light might be an important, but to-date, overlooked mechanism driving increased productivity in diverse communities. 8,9 Crop mixtures are known to be more efficient at intercepting light compared to monocultures, which is due to complementary use of aboveground space when intercropped species differ in their aerial architecture and thus create more complex canopies that can intercept more light. 10,11 The increased light interception in crop mixtures comes at the cost of shading, where shorter crops suffer shading from taller crops. 12 As shading is omnipresent in nature, plants have adapted to tolerate shade and have developed different strategies to optimize carbon gain even under low light conditions. These adaptions encompass-among others-increased photosynthetic capacity 13 or photosynthetic efficiency. 14 Photosynthetic capacity describes the maximum rate at which a leaf is able to fix C and has been tightly associated with leaf N content. 15 For instance, as a response to lower light conditions in the mixed cropping system, leaf N content in watermelon increased when cultivated in mixture compared to when cultivated in monoculture. 13 Photosynthetic efficiency describes the efficiency by which captured light is converted into biomass. 16 Photosynthetic processes are known to be highly sensitive to shading and plants can adapt their photosynthetic characteristics to various light environments, 17 as shown in a recent study where increased efficiency of photosystem II (PS II ) in proso millet was observed in response to being grown in a mixture. 14 Thanks to these adaptions, mixed cropping systems can enable positive light-driven biodiversity effects. In this study, we relate positive biodiversity effects mainly to the positive effect of increased plant species richness on plant primary productivity. However, the relative extent to which complementary use of light contributes to positive biodiversity effects in intercropping is poorly understood. 18 Positive biodiversity effects are measured through the net biodiversity effect (NE), which describes the productivity in mixtures compared to the average of the monocultures and-when positiveindicates overyielding of the mixture. The NE can be partitioned into the complementarity effect (CE; individual species contributing more to productivity than predicted from monoculture) and the selection effect (SE; covariance of monoculture and mixture productivity, describing the greater probability of more diverse communities including highly productive species which account for the majority of productivity). 19 Distinguishing whether positive biodiversity effects are driven by CEs or SEs is elementary to optimize farm management practices as well as breeding programs.
It is important to note that earlier work has shown that the contribution of CE and SE to the NE can change over time. 20 Lately, studies examining temporal dynamics of plant interactions have gained popularity, as they have unraveled important processes that would have gone by unnoticed if not detected through a series of destructive harvests 21 and have generally contributed to a better understanding of dynamic processes in diverse plant systems. [22][23][24] While earlier studies have shown that the amount of intercepted light increases during the growing season, 25 to the best of our knowledge, there are no studies that examined temporal changes of light-use associated traits in annual intercropping systems, important information when considering which traits might be the target for-for example-future breeding programs.
In summary, while differences in light use have been detected when crops are grown in mixtures compared to monocultures, there is little knowledge available on how differences in light use between mixtures and monocultures contribute to biodiversity effects and how the partitioning of light among co-occurring crops changes over time. Applying the additive partitioning method in combination with the study of light-use associated plant traits to mixed cropping systems can help to identify mechanisms that lead to yield advantages and can help identify target traits for breeding programs for crop species in mixtures. Therefore, the objectives of this study were (1) to quantify how NE and its two additive components, CE and SE, change over time and (2) how the differences of light-use associated traits in mixtures compared to monocultures contribute to biodiversity effects and how this changes over time in two different intercropping systems. To assess changes in light use over time, we analyzed two traits related to light acquisition (intercepted light and plant height) and two traits related to light conversion (photosynthetic efficiency and capacity) and measured these on a weekly basis. To quantify biodiversity effects, we measured aboveground biomass during weekly destructive harvests and-once available-quantified biodiversity effects based on final seed yields during the later stages of the growing season. As CEs were expected to be particularly strong in mixtures with crops from differing functional groups, we combined oat (Avena sativa) with either a legume (lupin, Lupinus angustifolius) or a Brassicaceae (camelina, Camelina sativa).

| Site description
The site and experimental design are identical to the one used in Engbersen et al. 24 The study was carried out at the field site Aprisco de las Corchuelas, near Torrejón el Rubio, Cáceres, Spain. The site is located at 290 m a.s.l. (39°48′47.9" N 6°00′00.9" W). Total precipitation between February and June 2019 was 77.4 mm, daily average hours of sunshine during the growing season were 10.5 h and daily mean temperatures ranged between 9.6°C and 21.9°C, averaging 16°C. All climatic data are from the national meteorological service (www.aemet.es).
The experimental garden covered 120 m 2 , divided into 480 square plots of 0.25 m 2 which were arranged in 12 beds of 10 × 1 m, with two rows of 20 plots, resulting in 40 plots per bed. The beds containing the plots were raised by 40 cm above the soil surface. A penetrable fleece was placed on the soil surface, allowing for root growth beyond 40 cm depth. Each bed on top of the fleece was filled by hand with 40 cm homogenized standard, unenriched, local agricultural soil. The soil consisted of 78% sand, 20% silt, 2% clay and contained 0.05% total nitrogen, 0.5% total carbon and 254 mg total P/kg with a mean pH of 6.3.
The experimental garden was irrigated throughout the growing season and all plots received the same amount of irrigation water. The automated irrigation system was configured for a dry threshold of soil moisture at 17% of field capacity and with a target value of 25% of field capacity. When the dry thresholds were reached, irrigation started automatically and irrigated until reaching the target value. Soil moisture was measured in six randomly selected plots at 10 cm below the soil surface with PlantCare soil moisture sensors (PlantCare Ltd.) and the average soil moisture of these six plots defined the soil moisture used for irrigation control.

| Experimental design
A complete randomized block design with three different crop species and two different diversity levels was used. The crop species were oat (Avena sativa, cv. Canyon), lupin (Lupinus angustifolius, cv. Boregine) and camelina (Camelina sativa, cv. unknown) and the two diversity levels were monocultures and 2-species mixtures. One block consisted of five plots: one plot of monoculture of each of the three species, one plot with an oat-lupin mixture and one plot with an oat-camelina mixture. A monoculture plot consisted of four identical rows of the respective crop species and a mixture plot consisted of two alternating rows of each crop species, following a speciesA|speciesB|speciesA|speciesB pattern ( Figure 1). The sowing densities and sowing depths were: 400 seeds/m 2 , 2 cm for oat, 160 seeds/m 2 , 5 cm for lupin and 592 seeds/m 2 , 0.5 cm for camelina and were based on current cultivation practice. 26 A monoculture plot consisted of four rows of 25 seeds of oat, 10 seeds of lupin and 37 seeds of camelina. For mixtures, we followed a substitutive design, where 50% of the seeds for the monocultures was used per species in the mixtures, to sum up to 100% sowing density per plot. Each block was repeated 54 times to allow for 18 destructive harvests with three replicates at each harvest. Sowing was done by hand on 2 and 3 February 2019.

| Light measurements
Photosynthetically active radiation (PAR) was measured with an LI-1500 (LI-COR Biosciences GmbH) every week just before the destructive harvest. In each plot, three PAR measurements were taken around noon by placing the sensor on the soil surface in the center of each of the three in-between rows. Light measurements beneath the canopy were put into context through simultaneous PAR measurements of a calibration sensor,

| Data analyses
To explain differences in community-level yield between mixtures and monocultures, we quantified the NE and its two additive components, CE and SE, according to Loreau and Hector 19 : where N is the number of species in the plot. ΔRY is the deviation from the expected relative yield of the species in the mixture in the respective plot, which is calculated as the ratio of observed relative yield of the species in the mixture to the yield of the species in monoculture. M is the yield of the species in monoculture. The first ) is the SE. Yield refers to total aboveground biomass for the harvest weeks (HWs) where no total grain yields were available (i.e.,  and to total grain yield when grain yields were available (i.e., .  were excluded from analyses, as they were not representative for total biomass anymore due to lupin leaves starting to wilt and fall and not yet representative for total grain yield, as the crop species had not yet produced mature grains.
As the NE and its additive components express the difference in productivity between monocultures and mixtures, we aimed to explain this difference through differences in light-use associated plant traits between mixtures and monocultures. We used a Δ to indicate differences between mixtures and monocultures. Δ trait values were calculated as the difference between community-weighted means of the respective trait value in mixture and monoculture. For example, Δheight was calculated as: where height mix is the average of all three measurements of height per mixture plot and height mono the average of all three measurements of height of the respective monoculture plot. Weights for communityweighted means were the total biomass of each species. For FPAR, we used mean values instead of community-weighted means.
All statistical analyses were performed in R version 3.6.0. 29 We  19 We tested for correlation among the light-use associated traits using Pearson's correlation coefficient.
If traits were correlated (i.e., Pearson's correlation coefficient > 0.45), we removed one of the two, keeping the one trait which would lead to the best model fit based on the Akaike Information Criterion (AIC).

| Biodiversity effects
Biodiversity effects were based on total aboveground biomass during the vegetative period (i.e., HWs 1-14) and on total grain yields during the reproductive period (i.e.,

| Light-use-associated traits and biodiversity effects
Collinearity among the light-use associated plant traits occurred between Δdiameter and Δheight (Table S1). Model comparison based on AIC indicated that the model fit improved after removing Δdiameter as explanatory variable from the model.

| ΔFPAR
Increases in ΔFPAR significantly increased with NE, CE and SE (Table 1 and Figures 3a, 4a and 5a) indicating that higher light interception in mixtures compared to monocultures was positively related to all three biodiversity effects. This effect did not differ significantly between mixture compositions or during the growing season (interactions ΔFPAR × mix and ΔFPAR × HW in Table 1).
Although insignificant, the strength of the positive relationship between ΔFPAR and all three biodiversity effects tended to increase with time in the oat-lupin mixture.

| ΔHeight
Overall, all three biodiversity effects increased with Δheight (Table 1 and Figures 3b, 4b and 5b). In the oat-lupin mixture, the SE de-

| ΔEfficiency of PS II
The interaction ΔΦPS II × mix × HW (Table 1) Table 1 and Figure 3D). ΔLeaf N was negatively correlated to SE in the oat-camelina mixture and positively in the oat-lupin mixture (interaction ΔLeaf N × mix. in Table 1 and Figure 5d). In the oat-lupin mixture, the relationship between Δleaf N and SE was negative during the early growing season but positive afterwards (interaction Δleaf N × mix. × HW in Table 1 and Figure 5d).
No effect of Δleaf N was observed on CE (Table 1).

| DISCUSSION
Understanding the underlying mechanisms of positive biodiversity effects in intercropping systems is essential when developing intercrops as a tool for sustainable agriculture. To address these needs, we investigated how light-use-related traits contribute to the NE via CE or SE and how these relationships change throughout an annual growing season.
We found increasing NE and SE in two different crop mixtures over time during the vegetative period. CEs were found to increase only in the mixture containing a legume. While the NE and CE also increased during the reproductive period, no increase was observed for the SE. This could suggest a discrepancy between the effects of biodiversity on biomass and seed yield.
The net biodiversity effect (a, d), complementarity effect (b, e) and selection effect (c, f) based on total biomass for the vegetative period (a-c) and based on total grain yields for the reproductive period (d-f) shown for oat-camelina (red) and oat-lupin (blue) mixtures. Lines in (a-c) show the marginal effect associated with the full model presented in Table 1. Data in (d-f) are mean and 95% confidence interval and significance analyses are based on linear models presented in Table S2 We found that higher light interception in mixtures compared  Table 1 decades. 30 Table 1 Alternatively, it could be that the higher biomass of the highly productive species causing most of the SE did not translate into an equally high seed yield. A discrepancy between the effects of diversity on biomass and seed yield has been observed before and is possibly due to currently commercially available crops having a higher harvest index in monocultures than in mixtures. 33 Increasing CEs in the oat-lupin mixture but the absence of a similar increase in the oat-camelina mixture suggests that the presence of the legume potentially contributed strongly to the CE and that cereal-legume mixtures are not without reason considered a successful combination for intercropping. 34 Most complementarity effects in cereal-legume mixtures are attributed to the legume meeting most of its N demand by fixing atmospheric N 2 , thus leaving most soil N for the neighboring cereal, which has been observed before for oat-lupin mixtures. 24 However, the present study could also show that specifically for the oat-lupin mixture, complementarity in light use due to the differences in canopy architecture between the intercropped species could further contribute to complementarity in this mixture.  Table 1 4.2 | Biodiversity effects and light-associated traits  41 This could support the assumption of increased photosynthetic capacity in the oat-lupin mixture, as we observed that higher leaf N in mixtures compared to monocultures contributed to the NE in the oat-lupin mixture. Higher leaf N in oat and lupin, when grown in mixture compared to when grown in monoculture, is in line with earlier observations in this mixed cropping system 24 and are due to the lupin meeting its N-demand by symbiotic N 2 -fixation, leaving more soil N for the intercropped oat. However, examining leaf N on a mass basis comes with certain caveats: (1) it does not account for the possibility that nitrogen is likely allocated to different light-harvesting compounds while total N of the leaf remains the same. For instance, total nitrogen to chlorophyll ratios have been shown to increase in deeper shade among individuals 42 ; (2) leaf N also depends on nutrient availability and competitive ability of the crop in the mixture. We, therefore, highlight the need for more detailed studies investigating the relative contributions of N 2 -fixation and increased photosynthetic capacity and their interdependence, for increasing biodiversity effects in cereal-legume mixtures over time.

| CONCLUSION
This study provided evidence that the NE and SE increased over time systems has the potential not only to improve yields but can also be a key strategy to increase sustainability in modern agriculture.
Laura Stefan collected the data; Nadine Engbersen analyzed samples in the laboratory; Nadine Engbersen assembled and analyzed the data with the help of Christian Schöb; Nadine Engbersen wrote the first draft of the paper. All authors discussed data analyses and results and revised the manuscript.