Rapid transgenerational adaptation in response to intercropping reduces competition

  1. Laura Stefan  Is a corresponding author
  2. Nadine Engbersen
  3. Christian Schöb
  1. Institute of Agricultural Sciences, ETH Zurich, Switzerland
  2. Plant Production Systems, Agroscope, Switzerland
  3. Área de Biodiversidad y Conservación, Universidad Rey Juan Carlos, Spain

Abstract

By capitalising on positive biodiversityproductivity relationships, intercropping provides opportunities to improve agricultural sustainability. Intercropping is generally implemented using commercial seeds that were bred for maximal productivity in monocultures, thereby ignoring the ability of plants to adapt over generations to the surrounding neighbourhood, notably through increased complementarity, that is reduced competition or increased facilitation. This is why using monoculture-adapted seeds for intercropping might limit the benefits of crop diversity on yield. However, the adaptation potential of crops and the corresponding changes in complementarity have not been explored in annual crop systems. Here we show that plantplant interactions among annual crops shifted towards reduced competition and/or increased facilitation when the plants were growing in the same community type as their parents did in the previous two generations. Total yield did not respond to this common coexistence history, but in fertilized conditions, we observed increased overyielding in mixtures with a common coexistence history. Surprisingly, we observed character convergence between species sharing the same coexistence history for two generations, in monocultures but also in mixtures: the six crop species tested converged towards taller phenotypes with lower leaf dry matter content. This study provides the first empirical evidence for the potential of parental diversity affecting plantplant interactions, species complementarity and therefore potentially ecosystem functioning of the following generations in annual cropping systems. Although further studies are required to assess the contextdependence of these results, our findings may still have important implications for diversified agriculture as they illustrate the potential of targeted cultivars to increase complementarity of species in intercropping, which could be achieved through specific breeding for mixtures.

Editor's evaluation

This study reports that interactions between crop species grown over three generations in mixture instead of monoculture become less competitive/more facilitative, suggesting a way to breed for increased mixture yield. This fundamental finding is of high interest to the fields of ecology and agriculture, as well as society seeking new solutions to satisfy increasing food demands. The methodological approach is compelling, yet distinguishing between reduced competition and increased facilitation remains challenging.

https://doi.org/10.7554/eLife.77577.sa0

eLife digest

Plants have two ways of interacting with each other: they can compete with each other if they use the same resources; or they can ‘help’ each other in what is known as facilitation, for example, when a larger plant protects a smaller plant in harsh environments. These interactions can vary over several generations in response to changes in the environment or the surrounding plant community. For instance, in plant communities formed by many different species, like in most natural systems, competition usually decreases over time as the plants ‘learn’ to grow together.

In agriculture, intercropping – defined as growing at least two species of crop at the same time on the same field – takes advantage from a reduction in competition. The idea is that planting two species that grow differently together will lead to less competition than having a single crop because the two species will use slightly different resources, or use them at different times. However, intercropping has traditionally overlooked changes in the interactions between plants as a result of the crop species evolving after being grown together for generations. Indeed, farmers that practice intercropping generally use standard seeds that have been bred to produce high yields when planted on their own, in what is known as monoculture. If plants can adapt and become less competitive when they are grown together over several generations, then using these standard seeds might limit the success of intercropping.

Stefan, Engbersen and Schöb wanted to know whether crop species adapt to the levels of plant diversity surrounding them over generations, and if so, how they do it. To find this out, they investigated how competition and facilitation changed when six crop species (wheat, oat, lentil, coriander, flax and camelina) that grow annually were grown together in different combinations over several generations. Stefan, Engbersen and Schöb started off with seeds normally used for growing these crops on their own, and planted them either on their own, or in different combinations of two or four species. They then repeated the experiment over the course of three years, each year using seeds from the previous year, recording both crop yields and changes in how the plants interacted with each other.

The experiments showed that interactions among these annual crops shifted towards reduced competition and/or increased facilitation when the plants were growing alongside the same crops as their parents did in the previous two generations.

Improving and promoting the development of intercropping is essential for agricultural sustainability, as it could offer alternatives to intensive monocultures (crops grown on their own that require increased resources). Stefan, Engbersen and Schöb’s findings are relevant for programmes aimed at developing seeds for intercropping, as they highlight the importance of including diversity when developing these seeds. However, before these results can be used in the field, longer experiments (of more than three generations) in different environments should be carried out to confirm the findings. Another question that remains open is what the mechanisms underlying adaptations to intercropping are: more in-depth research will be needed to determine whether the changes observed have a genetic basis.

Introduction

Following decades of studies demonstrating the positive relationship between species diversity and plant primary productivity in natural systems (Spehn et al., 2005; Tilman et al., 2001), intercropping, that is growing more than two species in the same field during the same period, has been increasingly considered as a promising option to increase agricultural sustainability (Gurr et al., 2016; Brooker et al., 2015; Vandermeer, 1992). The productivity benefits of increasing species diversity rely on two main mechanisms, namely selection effects and complementarity effects, the latter encompassing both facilitation and niche differentiation (Loreau and Hector, 2001; Hooper et al., 2005). In perennial natural grasslands, complementarity effects have been shown to increase over time due to evolutionary processes (Zuppinger-Dingley et al., 2014; van Moorsel et al., 2019; van Moorsel et al., 2019). Notably, greater species complementarity can result from evolutionary changes (Anderson et al., 2011) that is changes in gene frequency or from heritable epigenetic changes (Verhoeven et al., 2016) affecting species traits in response to surrounding plant diversity, which either increases niche differentiation (i.e. reduces competition) or increases facilitation (Schöb et al., 2018; Meilhac et al., 2020).

The mechanisms of selection for plant facilitation remain very poorly understood (Bronstein, 2009; Brooker, 2008). Whereas some facilitative traits characterizing facilitator species are well known in some systems for example legume species fostering soil nitrogen enrichment (Wright et al., 2017), microclimate amelioration through shading by large canopy (Aguirre et al., 2021), nectar reward to attract pollinators (Losapio et al., 2021) traits of facilitated species remain much more obscure (Bronstein, 2009). These ‘facilitated’ traits, which allow organisms to benefit from their neighbours, may be a target of evolutionary selection in natural systems (Bronstein, 2009), and it is therefore reasonable to think that they might depend on neighbour identity (Schöb et al., 2018). The neighbour-dependent evolution of facilitation in grassland plant communities was demonstrated by Schöb et al., 2018, who showed that selection for net facilitative interactions was favoured in plant mixtures.

The evolutionary potential of plantplant interactions in diverse communities has tremendous implications for the diversification of agricultural systems (Isbell et al., 2017). This is of particular relevance for mixed cropping systems, where the use of commercial seeds domesticated and bred for maximum yield in monoculture is the norm (Thrall, 2012). These commercial varieties have been selected to express a particular phenotype or traits that would lead to the best yield in monoculture. Yet the optimal monoculture phenotype might not necessarily be the most adequate to promote positive diversity effects in mixtures, and may actually compromise the diversity benefits (Thrall, 2012; Chen et al., 2021; Chacón-Labella et al., 2019; Wuest et al., 2021; Annicchiarico, 2019). Despite the paramount importance of this question, the yield potential of mixture-adapted varieties is, to our knowledge, unknown, as are the trait differences of monoculture- compared to mixture-adapted crops.

Furthermore, the evolution and occurrence of plantplant interactions is notably context-dependent (Bertness and Callaway, 1994). This environmental dependence of the direction and strength of plant-plant interactions has been conceptualised as the stressgradient hypothesis (SGH), which suggests that competition between plants is stronger and more important in benign environments where resources are abundant while facilitation is more likely to occur in harsher environments where resources are scarce (Soliveres et al., 2015; Maestre and Cortina, 2004; Maestre et al., 2009). In the context of intercropping, this means that competition would be the dominant interaction in highly productive systems, and therefore, the benefits of increasing niche differentiation thereby reducing competition would be higher (Li et al., 2020; Stefan et al., 2021a). On the other side of the stress gradient, enhanced facilitation in resource-poor, low-productive systems may also increase diversity effects (Soliveres et al., 2015). Therefore, the effect of environmental severity on plantplant interactions and biodiversity effects in intercropped systems is unclear.

In this project, we determined whether, how, and under which soil fertility conditions crop species adapt over three generations to the level of plant diversity that they are surrounded by. We investigated how plantplant interactions, that is competition and facilitation, and plant traits changed within different coexistence histories over time, and whether these changes translated into yield benefits. To that end, we conducted an intercropping experiment in Switzerland with six different crop species belonging to four functionally different phylogenetic groups, namely wheat, oat, lentil, flax, camelina, and coriander. These species are commonly cultivated in Europe as monocrops and some of them for example oat, lentil, camelina are also partly cultivated in intercrops (Neumann et al., 2007; Kraska et al., 2004). We used commercially available seeds commonly used for monoculture practices and selected, whenever possible, open-pollinated varieties as seed source to provide the genetic variability needed for evolutionary processes to occur. The mesocosms square plots of 0.25 m2 included monocultures, 13 different 2-species mixtures, four different 4-species mixtures, and isolated single plants, and was replicated in two different fertilizing conditions. To assess potential transgenerational changes, we repeated the experiment over the course of three years with seeds from plants grown from either monocultures, mixtures, or single individual plants of the previous year (Figure 1, Figure 1—figure supplement 1). In the third year, we assessed plantplant interactions within each community using the Relative Interaction Index (Armas et al., 2004), a symmetrical and standardized index that compares the performance in terms of grain yield of a plant growing in a community to its performance when growing in isolation (Michalet et al., 2014; Schöb et al., 2014) (see Methods). This index takes a plant’s eye view by quantifying plantplant interaction intensity experienced by crop species in monocultures and mixtures. We therefore directly quantify species complementarity underlying classical biodiversity effects (Armas et al., 2004; Schöb et al., 2018). Yield was also used to derive the metrics of biodiversity effects following the method of Loreau and Hector, 2001. Finally, we measured standard above-ground plant traits that is plant height, plant width, Specific Leaf Area (SLA), Leaf Dry Matter Content (LDMC), and mass per seed and investigated whether there was a change in mean and variability at the species and community levels in response to the coexistence history of the community.

Figure 1 with 1 supplement see all
Experimental design.

(a) Six crop species were used to sow single plant individuals (Loreau and Hector, 2001), monocultures (Loreau and Hector, 2001), 2-species mixtures (Schöb et al., 2018) and 4-species mixtures (Brooker et al., 2015) in 2018 (Year 1) (see Supplementary file 1s for the list of species mixtures); seeds were collected at the end of the growing season and resown in 2019 (Year 2) in the same diversity setting as their previous generation. Seeds were collected again and resown in 2020 (Year 3), this time either in the same community their seeds were collected from [same coexistence history], or in a community different to the one of their parents [different coexistence history] (n=468 plots). This process was replicated in two different fertilizing conditions. We expected that crops growing in the same community as their parents would have adapted over the two generations, and therefore would exhibit less competition and have higher productivity than crops growing in a community different to the one of their parents. Av: Avena sativa; Ca: Camelina sativa; Co: Coriandrum sativum; Le: Lens culinaris; Li: Linum usitatissimum; Tri: Triticum aestivum (b) Left: part of the experimental garden, showing the plots within beds, and planted with single individuals. Right: a plot is outlined in red, showing a 2-species mixture, with oat alternated with camelina.

We hypothesized that crop mixtures composed of offspring of plants that had been grown for two generations in the same community type (as their offspring do now) would show increased niche differentiation (i.e. less competition) and/or increased facilitation compared to communities of offspring of plants that had been grown in a different community type (as their offspring do now). We also expected that these changes in plantplant interactions would lead to changes in complementarity effects in crop mixtures, that is communities with the same coexistence history would show higher complementarity effects than the communities with a different history. We hypothesised that the increased niche differentiation would be due to enhanced character displacement. Finally, following the stress gradient hypothesis (Bertness and Callaway, 1994), we expected more facilitation and/or less competition in conditions of reduced soil fertility.

Results and discussion

Results from the third year showed that plantplant interactions shifted towards increased complementarity, that is weaker competition and/or stronger facilitation distinguishing between these two mechanisms was not possible in this study when the plants were growing in the same community types as their two previous generations (Figure 2, Figure 2—figure supplement 1, Supplementary file 1a). More precisely, the net Relative Interaction Index, which compares the performance of focal plants growing in communities to the performance of single plants growing alone focal plants and single plants having the same coexistence history treatment was significantly higher (+54% [F=30.4; p-value < 0.001; n=276]) when the offspring was grown in the same community type as their parents did than when they were growing in a community type different to the one of their parents (Figure 2). Pairwise comparisons further showed that this effect of coexistence history was particularly true in mixtures and only a trend in monocultures, for both fertilizing conditions (Supplementary file 1b). This notably demonstrates that in mixtures, mixture-adapted communities (i.e. with the same coexistence history) exhibited less competition and/or more facilitation than monoculture-adapted communities or single-adapted communities (i.e. with a different coexistence history). Furthermore, when looking at the effect of fertilization, we observed that competition was weaker and/or facilitation was stronger in unfertilized plots (Figure 2, Figure 2—figure supplement 1;+64% F=44.5; p-value < 0.001; n=276), which is in accordance with the stressgradient hypothesis.

Figure 2 with 1 supplement see all
Relative Interaction Index in response to coexistence history and fertilization.

Net relative interaction index of monocultures, 2- and 4-species mixtures in response to coexistence history, for fertilized and unfertilized conditions. n=276. Dots represent the mean values across plots; lines represent the standard error. Stars placed above or next to the results represent the significance of the coexistence history effect. The net Relative Interaction Index (RII) compares the performance of plants growing in communities to the performance of single plants growing alone, with the same coexistence history treatment as the focal plant (see Methods). Negative RII indicates competition within a community, positive RII indicates facilitation. The closer this index gets to 1, respectively –1, the stronger the facilitation, respectively competition. ‘Same coexistence history’ indicates that crops were grown in the same community type as their parents (i.e. monocultures with seeds coming from monocultures, 2-species mixtures with seeds coming from the same 2-species mixtures [e.g. oat-lentil with seeds coming from oat-lentil], 4-species mixtures with seeds coming from the same 4-species mixtures [e.g. oat-lentil-coriander-flax with seeds coming from oat-lentil-coriander-flax]). “Different coexistence history” refers to crops grown in a community type different to the one of their parents (i.e. monocultures with seeds coming from singles, monocultures with seeds coming from mixtures, mixtures with seeds coming from singles, mixtures with seeds coming from monocultures). See Supplementary file 1a for the complete statistical analysis, and Figure 2—figure supplement 1 for the corresponding boxplots.

This shift in plantplant interactions was accompanied by a similar shift in net biodiversity effect (NE) in fertilized plots (Figure 3a, Figure 3—figure supplement 1a). Net biodiversity effect or overyielding represents the deviation from the expected yield in the mixture, based on the yield of the corresponding monocultures with the same coexistence history as the focal mixture (Loreau and Hector, 2001). The interaction between fertilization and coexistence history had a significant effect on NE (Supplementary file 1c, [F=9.60, p-value = 0.0023, n=204]). Posthoc pairwise comparisons further showed that under fertilized conditions, across all species combinations, NE was on average 58% higher with the same coexistence history than with a different coexistence history (Figure 3a, Supplementary file 1d p-value of the pairwise comparison: 0.0587). This indicates that in fertilized plots, overyielding of crop mixtures tended to be higher with mixture-adapted individuals compared to monoculture-adapted and single-adapted individuals. In unfertilized plots we did not observe the same result, which suggests that even though the shifts in plantplant interactions were consistent across fertilizing conditions, overyielding was not. When looking at the partitioning of net effects into complementarity and selection effects (Loreau and Hector, 2001), we observed a significant interaction effect between fertilizer, coexistence history and planted diversity on selection effects (Figure 3—figure supplement 2b, Supplementary file 1c, [F=4.09, p-value = 0.045, n=204]). More precisely, for 4-species mixtures under fertilized conditions, SEs were higher in plant communities composed of offspring of plants that had been grown in the same community type (as their offspring do now) than plant communities of offspring of plants that had been grown in a different community type (+109%, Figure 3—figure supplement 2b, Supplementary file 1c, [p-value of the pairwise comparison: 0.0286]). Coexistence history did not affect complementarity effects (Figure 3—figure supplement 2a, Supplementary file 1c, [F=1.57, p-value = 0.21, n=204]), nor total yield (Figure 3b, Figure 3—figure supplement 3, Figure 3—figure supplement 4, Supplementary file 1f, [F<1, p-value > 0.5, n=276]).

Figure 3 with 4 supplements see all
Effects of coexistence history on net biodiversity effects (a) and total yield per plot (b).

Effects of coexistence history and crop species number on (a) net biodiversity effect – reflecting the yield advantage of mixtures compared to monocultures – and (b) total yield per plot in fertilized and unfertilized plots. (a) n=204; (b) n=276. Dots represent the mean values across plots; lines represent the standard error. Stars or dots placed above or next to the legend represent the significance of the coexistence history effect. ‘Same coexistence history’ indicates that crops were grown in the community their seeds were collected from. ‘Different coexistence history’ refers to crops grown in a community different to the one of their parents. See SI Supplementary file 1c-f for the complete statistical analysis, Figure 3—figure supplement 1 for complementarity and selection effects, and Figure 3—figure supplements 2 and 3 for the corresponding boxplots.

To investigate the ecological mechanisms behind the shift in plantplant interactions with coexistence history, we assessed the response of standard above-ground plant traits and compared the average values and coefficients of variation at the species and community levels of single-, monoculture- and mixture-adapted varieties. Results pointed towards a reduction in trait variation at the community level, notably of height and leaf dry matter content (Figure 4): the coefficient of variation of height was lower in plant communities composed of offspring of plants that had been grown in the same community type (as their offspring do now) compared to plant communities of offspring of plants that had been grown in a different community type (–9%, Figure 4d, Supplementary file 1l, [F=3.93, p-value = 0.049, n=271]), and for leaf dry matter content it was 15% lower with the same history compared to a different history (Figure 4e, Supplementary file 1o, [F=4.18, p-value = 0.042, n=271]). Furthermore, the coefficient of variation of mass per seed was also lower under the same history compared to a different history, but this effect was only significant in monocultures (–33%, Figure 4f, Supplementary file 1p, [F=5.48, p-value = 0.020, n=271]). The community-weighted means of plant traits (CWM, calculated at the community level) further suggest that when growing in the same community type as their parents, plants seemed to converge towards taller individuals with lower leaf dry matter content. Indeed, the community-weighted mean of leaf dry matter content was significantly lower in plant communities composed of offspring of plants that had been grown in the same community type compared to plant communities of offspring of plants that had been grown in a different community type (–3%, Figure 4b, Supplementary file 1o, [F=4.33, P-value = 0.039, n=271]); height community-weighted mean was not significantly different between coexistence histories (Figure 4a, Supplementary file 1l, [F<1, p-value = 0.48, n=271]), but at the species level we did observe a consistent increase in plant height under the same coexistence history compared to a different history (Figure 4—figure supplement 1, Figure 4—figure supplement 2, Supplementary file 1g, [F=4.29, p-value = 0.040, n=1,726]). We observed similar consistent responses of leaf dry matter content at the species level (Figure 4—figure supplement 2, Supplementary file 1j).

Figure 4 with 2 supplements see all
Community-level trait responses to coexistence history.

Effects of coexistence history and crop species number on community-weighted mean (CWM) of height (in cm) (a), Leaf Dry Matter Content (LDMC) (b), and mass per seed (in g) (c), and on coefficient of variation at the community level of height (d), LDMC (e), and mass per seed (f). n=271. Dots represent the mean values across plots; lines represent the standard error. Stars placed above represent the significance of the coexistence history effect. See Supplementary file 1l-p for the complete statistical analyses, and Fig. S6-7 as well as Supplementary file 1g-k for responses at the species level.

Our research demonstrates that, after only two generations, annual crop plant communities composed of offspring of plants that had been grown in the same community type (as their offspring do now) showed reduced competition and/or increased facilitation compared to plant communities of offspring of plants that had been grown in a different community type (as their offspring do now). In fertilized conditions, common coexistence history also increased overyielding, but this was not the case in unfertilized conditions. Furthermore, common coexistence history had no effect on total yield per plot. We further investigated whether character displacement was responsible for this evolution of plantplant interactions; contrary to our hypothesis, results did not show evidence for character displacement, but rather for character convergence in plant aboveground traits.

The observed shift in plantplant interactions towards reduced competition and/or increased facilitation is consistent with a grassland study investigating the effects of community evolution on plantplant interactions (Schöb et al., 2018). However, the lack of response of total yield and biodiversity effects across fertilizing conditions does not agree with several grassland studies examining the effects of common evolution on community productivity and niche differentiation, where it was found that common rapid evolution in plant communities can lead to increases in ecosystem functioning (van Moorsel et al., 2018; Zuppinger-Dingley et al., 2014; van Moorsel et al., 2019; van Moorsel et al., 2021; Meyer et al., 2016; Allan et al., 2011). Only in fertilized plots did we observe a positive effect of common coexistence history on net biodiversity effects (i.e. overyielding), which means that the yield benefit of mixtures compared to monocultures was higher when the plants had been adapted to growing in mixtures (Figure 3). Yet we did not observe a significant increase in complementarity effects in response to common coexistence history (Figure 3—figure supplement 2a, Supplementary file 1c). Surprisingly, selection effects also increased in 4-species mixtures in response to coexistence history (Figure 3—figure supplement 2b). This is unexpected, as selection effects have to our knowledge not been shown to increase over time (Cardinale et al., 2007). However, it might be that this short common coexistence history has favoured a specific species or a specific trait that was particularly plastic or strongly linked to productivity (Colom and Baucom, 2021; Turcotte and Levine, 2016).

The apparent discrepancy between the response of plantplant interactions and the response of net biodiversity effects to coexistence history can stem from various reasons. First, net biodiversity effects are driven both by complementarity and selection effects Loreau and Hector, 2001; therefore, a reduction in competition does not necessarily lead to an increase in net biodiversity effects, as this can be compensated by concurrent changes in selection effects. Changes in RII should however correlate with complementarity effects, which they do in our study (Figure 5, p-value = 0.033), indicating that reduced competition and/or increased facilitation correlates with higher complementarity effects. Most importantly though, our RII calculations and net biodiversity effects use different reference levels, that is the single plant vs the monoculture. Indeed, the biodiversity effect calculations ignore the intensity of intra-specific competition and only assess changes in plantplant interactions from monoculture to mixture, while RII calculations quantify plantplant interaction intensity in monocultures and mixtures and therefore also allow to assess effects of coexistence history on intra-specific interactions. This can explain why the effect of coexistence history on plant interactions between individuals (quantified through RII) might diverge from the effects of coexistence history on the diversification of a monospecific community (quantified through the net biodiversity effect, complementarity effect or selection effect). Finally, we also think that the limited timeframe of this study two generations might be a reason for the lack of more significant changes in total yield and emphasize the need for longer-term research to confirm the trend identified at the individual level.

Figure 5 with 2 supplements see all
Correlation plot between Net RII index and complementarity effect across all plots.

There is a significant positive correlation (F=4.62, p-value = 0.033, n=204).

Further investigation is also needed to understand the contextdependence of the effects of common coexistence history, notably when reflecting on the important role of fertilization in our results. Our findings are consistent with several recent studies demonstrating that biodiversity effects are higher in high-inputs systems (Chen et al., 2021; Li et al., 2020; Stefan et al., 2021b), and emphasize the role of fertilization in driving yield benefits in diverse crop communities. Indeed, by promoting crop growth and, consequently, higher competition between plants, fertilization may foster higher benefits of niche differentiation that is reduced competition in mixtures (Bertness and Callaway, 1994; Stefan et al., 2021a; Goldberg and Novoplansky, 1997).

Overall, increases in biodiversity effects are associated with changes in species traits in response to surrounding plant diversity (Zuppinger-Dingley et al., 2014; Schöb et al., 2018; Abakumova et al., 2016). Traditional hypotheses of trait and niche theory indeed predict that when several species co-occur closely together, selection over generations would favour character displacement that would reduce resource overlap and consequently increase niche differentiation (Pfennig and Pfennig, 2009; Losos, 2000). Surprisingly, here we found the reverse and observed that a common coexistence history led to a reduction in trait variation, which would suggest a decrease in niche differentiation. Furthermore, functional diversity calculated as the volume occupied in the space of the traits considered in this study (Petchey and Gaston, 2002) did not respond to common coexistence history (Figure 6—figure supplement 1, Supplementary file 1q). While surprising, this result is not unheard of Colom and Baucom, 2021; Fox and Vasseur, 2000; Grant, 1972; Weedon and Finckh, 2021; notably, because competition for light is asymmetrical, plant height generally converges towards increased plant height in species-rich communities, as a response to a denser and taller canopy (Lipowsky et al., 1972; Falster and Westoby, 2003). Leaf traits in constrast are usually diverging Lipowsky et al., 1972; this was not the case in our study, where we found convergence towards taller plants with lower leaf dry matter content in response to common coexistence history, that is soft leaves associated with rapid biomass production (Diaz et al., 2016), and consequently less resource-conservative strategies (Reich and Cornelissen, 2014). Lower leaf dry matter content has recently been associated with lower parental or ambient competition (Puy et al., 2021b), which is consistent with our results of plantplant interaction intensities. The traits examined here did not allow to understand the mechanisms behind the observed reduction in competition; we suggest that other traits or processes not measured in this experiment might have responded to the coexistence history treatment. Notably, there could be a shift in belowground traits, such as root-associated traits (Puy et al., 2021b), or temporal differentiation of resource capture (Engbersen et al., 2021), such as light. We indeed observed a significant increase in light capture ability in communities with a common coexistence history compared to the same communities but with a different coexistence history (Figure 6, Supplementary file 1r), which indicates that plants used to growing in the same community during several generations might capture the resources more fully than plants coming from a different community. This suggests increased niche differentiation for light use with a common coexistence history. However, here we only rely on our light interception measurements and suggest more longer term studies to understand changes in the use of other resources, such as nutrients or water, and how this is associated to plant traits. Finally, the limited duration of the study as well as the lack of evolutionary potential of some of the chosen crops might also explain why we did not observe clearer signs of increased niche differentiation with common coexistence history.

Figure 6 with 1 supplement see all
Response of absorbed photosynthetically active radiation to coexistence history.

n=271. Fraction of PAR absorbed (in %) according to the day of year, for plants with the same or different coexistence history. The lines represent local polynomial regression fittings, with the grey area representing the 0.95 confidence interval. Stars placed next to the legend represent the significance of the result. n=2484. See Supplementary file 1r for the complete statistical analysis.

Furthermore, the scope of this study did not allow us to investigate the transgenerational mechanisms behind these changes in plantplant interactions and traits in response to coexistence history. The adaptation responses might be genetically based and due to natural selection (van Moorsel et al., 2019), as we specifically selected, whenever possible, open-pollinated varieties in order to ensure a maximum amount of genetic variability. This was notably the case for crops that are not standardly used in western European rotations, such as camelina and coriander. Potentially evolutionary mechanisms include sorting out from standing variation, recombination, mutation (Prentis et al., 2008), or heritable epigenetic processes (Rapp and Wendel, 2005; Sentis et al., 2018; Sobral and Sampedro, 2022). Because our study only accounted for two generations, recombination and mutation are unlikely, as these are long-term processes (Hendry et al., 2007). Rapid adaptation from standing genetic variation is a more plausible mechanism, especially in non-standard species with higher initial variation, such as coriander or camelina (Prentis et al., 2008; Herman and Sultan, 2011). Particularly, outcrossing could have occurred in the first year of this experiment, as we had a similar experiment running in the same experimental garden with Spanish varieties from the same species (Chen et al., 2021; Stefan et al., 2021a). However, considering the short timeframe of this study and the low rate of outcrossing in most of our species, we suggest that epigenetic changes that is stable heritable changes in cytosine methylation might also have played an important role as potential evolutionary mechanisms (Verhoeven et al., 2016; Sentis et al., 2018; Herman and Sultan, 2011; Cortijo et al., 2014; Van der Graaf et al., 2015; Saze et al., 2003; Springer, 2013; Puy et al., 2021a). Non-genetic mechanisms can also potentially underpin the observed transgenerational adaptive plasticity; these include seed provisioning, which refers to the carbohydrate, lipid, protein, and mineral nutrient reserves allocated by the maternal plant to the developing seed (Steets and Ashman, 2010; Donohue, 2009; Moles and Westoby, 2006), or changes in maternally derived proteins, mRNAs, or in the relative concentrations of hormones (Herman and Sultan, 2011; Sultan, 1996; Dyer et al., 2010). Finally, a recent study demonstrated the transgenerational role of the seed mycobiome that is fungal seedendophytes for improved resilience and adaptive phenotypes in several generations of wheat Vujanovic et al., 2019; thus, heritable transmission of a specific seed mycobiome and/or microbiome might also be a possible non-genetic mechanism (Vivas et al., 2015).

For the first time, our study provides empirical evidence for rapid transgenerational adaptation in response to common coexistence history in annual crop communities. Notably, we demonstrated that when plants were growing in the same diversity setting as their parents did for two generations, plantplant interactions shifted towards reduced competition and/or increased facilitation. This history effect was particularly true for mixtures and was associated with increased overyielding under fertilized conditions. However, there was no significant increase in total yield and no yield benefits in unfertilized conditions. Common coexistence history did surprisingly not lead to character displacement in mixtures, but we instead observed character convergence towards taller plants with lower leaf dry matter content. While further research is needed to assess the validity of our findings in other environmental conditions and for other species, this research emphasizes the importance of considering transgenerational effects of diversity for crop mixtures. This is particularly relevant for breeding programs and highlights the need of including diversity when breeding for crop mixtures, in order to design varieties that could be specifically adapted for intercropping.

Methods

Study sites

The Crop Diversity Experiment took place in 2018, 2019, and 2020 in an outdoor experimental garden located at the Irchel campus of the University of Zurich, Switzerland (47.3961 N, 8.5510 E, 508 m a.s.l). Zurich is characterized by a temperate climate (Stefan et al., 2021a). The experimental garden was irrigated during the growing season with the aim of maintaining a sufficient amount of water for optimal plant growth. The dry threshold of soil moisture was set at 50% of field capacity, with a target soil moisture of 90% of field capacity. Whenever dry thresholds were reached measured through PlantCare soil moisture sensors (PlantCare Ltd., Switzerland), irrigation was initiated, and water added until reaching the target value.

Each experimental garden consisted of square plots of 0.25 m2. The uppermost 30 cm of the square plots were filled with standard, not enriched, agricultural soil coming from the local region. This soil consisted of 45% sand, 45% silt, and 10% clay, and initially contained 0.19% nitrogen (N), 3.39% carbon (C), and 332 mg total phosphorous (P)/kg, with a mean pH of 7.25. Beneath that, there was local soil of uncharacterized properties that allowed unlimited root growth. The plots were embedded into larger beds of 7x1 m, each bed containing 28 plots. Inside a bed, plots were separated from each other by metal frames. The metal frames reached 10 cm aboveground and until 30 cm belowground. While the relatively small plot sizes allowed us to undertake a large experiment under environmentally highly controlled but realistic outdoor conditions, some variables can suffer edge effects and interferences with neighbouring plots. However, such effects would probably increase residual variation more than between-treatment variation, because randomization was used to prevent confounding of between-plot interactions with treatments. In the only relevant study of which we are aware, the biodiversityproductivity relationship in herbaceous communities was not affected by plot size (Roscher et al., 2005) while a recent theoretical study showed that, if anything, biodiversity effects should increase with plot size (Isbell et al., 2018).

We therefore assume that effect size in our experiment, if anything, is probably rather conservatively estimated compared with that in studies using larger plot sizes.

Every year, we fertilized half of the beds with N, P and potassium (K) at the concentration of 120 kg/ha N, 205 kg/ha P, and 120 kg/ha K. Fertilizers were applied three times per year, namely once just before sowing (50 kg/ha N, 85 kg/ha P, 50 kg/ha K), once when wheat was at the tillering stage (50 kg/ha N, 85 kg/ha P, 50 kg/ha K), and once when wheat was flowering (20 kg/ha N, 34 kg/ha P, 20 kg/ha K). The other half of the beds was left unfertilized. In 2018, we randomly allocated individual beds to a fertilized or non-fertilized treatment. In the following years, we kept the initial fertilization treatment allocation.

Crop species

Experimental communities were constructed with six annual crop species of agricultural interest. We selected only seed crops with similar growth requirements in terms of climate and length of growing season, and with similar plant sizes to fit at least 40 individuals in the rather small plots. The six species belong to four different phylogenetic groups with varying functional characteristics: we first separated monocots [Triticum aestivum (wheat, C3 grass, Poaceae) and Avena sativa (oat, C3 grass, Poaceae)] and dicots. Among the dicots, we differentiated between suparasterids [Coriandrum sativum (coriander, herb, Apiaceae)] and superrosids. Among the superrosids, we separated legumes [Lens culinaris (lentil, legume, Fabaceae)] from non-legumes [Linum usitatissimum (flax, herb, Linaceae) and Camelina sativa (false flax, herb, Brassicaceae)]. Furthermore, we chose crop varieties that were locally adapted and commercially available in Switzerland (Table 1).

Table 1
List of crop species ecotypes and their suppliers.

Avena sativa (oat) is mainly self-pollinating, with outcrossing rates of around 1% (Shorter et al., 1978). The variety Canyon was acquired in 2014 through conventional selection processes.

SpeciesSwitzerland
EcotypeSupplier
Avena sativaCanyonSativa Rheinau
Triticum aestivumFiorinaDSP, Delley
Coriandrum sativumIndianZollinger Samen, Les Evouettes
Lens culinarisAniciaAgroscope, Reckenholz
Camelina sativan.a.Zollinger Samen, Les Evouettes
Linum usitatissimumLirinaSativa Rheinau

Triticum aestivum (wheat) is principally self-pollinating, with outcrossing rates generally between 1 and 4% (Hai et al., 2005; Loureiro et al., 2012), although some cultivars have been shown to have outcrossing rates up to 8% (Lawrie et al., 2006). Fiorina is an accession originating from Switzerland, acquired in 2015, specifically for organic agriculture.

Coriandrum sativum (coriander) has a generally high genetic variability, with studies showing up to 70.46% polymorphism, indicating the presence of high degree of molecular variation in the studied coriander varieties (Choudhary et al., 2019; Singh et al., 2012). The variety that we used originally came from an Indian market and was not a fixed variety, which ensured a minimum of genetic variability. The flowers of coriander are self-incompatible but plants are self-compatible. Geitonogamy is therefore common. Cross-pollination is facultative but can reach up to 20% (Diederichsen, 1996).

Lens culinaris (lentil) is mainly self-pollinating; depending on the cultivar, outcrossing rates reach between 1 and 5% (Horneburg and Weber, 2006).

Camelina sativa (camelina) is mainly self-pollinating, with outcrossing rates of less than 1% (Walsh et al., 2006; Walsh et al., 2015). In the study, we used a local landrace that was not a fixed variety.

Linum usitatissimum (flax) is mainly self-pollinating but outcrossing does occur, at a rate of 1–5% (Jhala et al., 2011). Lirina, the variety of Linum that we used has been defined by ProSpecieRara as a rare or ancient variety. ProSpecieRara ensures the preservation of rare traditional varieties (Begemann, 2002). Furthermore, studies have shown that linseed varieties have higher genetic variability than fiber flax and should therefore be considered as valuable genetic resources (Vromans et al., 2006; Hoque et al., 2020).

Experimental crop communities

Experimental communities consisted of single plots with one individual, monocultures, 2- and 4-species mixtures (Figure 1, Figure 1—figure supplement 1). We planted every possible combination of 2-species mixtures with two species from different phylogenetic groups and every possible 4-species mixture with a species from each of the four different phylogenetic groups present (Supplementary file 1). We replicated the experiment two times with the exact same species composition, except for single individuals which were replicated 4 times. Single plants were allocated to separate beds in order to minimize interference among neighbouring plots (Figure 1), and randomized within each fertilized treatment. Monoculture and mixture plots were randomized among plots and beds within each fertilizer treatment. Each monoculture and mixture community consisted of one, two or four species planted in four rows. Two species mixtures were organized following a speciesA|speciesB|speciesA|speciesB pattern. The order of the species was chosen randomly. Four species mixtures were organized following a speciesA|speciesB|speciesC|speciesD pattern. The order of the species was also randomized for each 4-species mixtures to avoid having the same order of species for all the replicates of a same mixture. Density of sowing differed among species groups and was based on current cultivation practices: 160 seeds/m2 for legumes, 240 seeds/m2 for superasterids, 400 seeds/m2 for cereals, and 592 seeds/m2 for superrosids. These correspond to the densities in monocultures; in mixtures, we kept these densities for each species (e.g. for legume, we planted 10 individuals per line in the monocultures and also 10 individuals per line in the mixtures). Each year, seeds were sown by hand in early April.

Adaptation treatment

In 2019, we used the seeds collected in 2018 to add a coexistence history treatment: we repeated the experiment with seeds coming from single individuals, monocultures, and mixtures, respectively. This means that each plot described above was repeated three times: once with seeds coming from single plants, once with seeds coming from monoculture plants, and once with seeds coming from mixture plants. We respected the fertilizing treatment, that is there was a history treatment for each fertilizing condition. When planting the mixtures with a mixture history, we specifically used seeds coming from the same species combination. When planting the monocultures and singles with a mixture history, we used seeds coming from a common pool combining all 4-species mixtures. Plots were fully randomly re-allocated each year to avoid soil legacy effects.

In 2020, we repeated this process and selected seeds from 2019 to sow the single and community plots. We only selected seeds that had a ‘pure’ history, that is that were always grown in the same coexistence history (for instance, for single history seeds in 2020 we selected only seeds that were grown as singles also in 2018 and 2019).

Data collection

Photosynthetically active radiation (PAR)

Interception of PAR by the plant canopy was measured weekly with a LI-1500 (LI-COR Biosciences GmbH, Germany). 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 compared to ambient radiation through simultaneous PAR measurements of a calibration sensor, which was mounted on a vertical post at 2 m above ground in the middle of the experimental garden. FPAR (%) indicates the percentage of PAR that was intercepted by the crop canopy.

Traits measurements

At the time of flowering, three individuals per crop species per plot were randomly marked. We measured the height of each individual with a ruler from the soil surface to the highest photosynthetically active tissue. We then measured plant width with a ruler by taking the largest horizontal distance between two photosynthetically active tissues. We sampled one healthy leaf from each marked individual and immediately wrapped this leaf in moist cotton; this was stored overnight at room temperature in open plastic bags. The following day, we removed any excess surface water on the leaf and weighed it to obtain its water saturated weight (Cornelissen et al., 2003). Then this leaf was scanned with a flatbed scanner (CanoScan LiDE 120, Canon), oven-dried in a paper envelope at 80 °C for 72 hr, and subsequently reweighed to obtain its dry weight. We calculated Leaf Dry Matter Content (LDMC) as the ratio of leaf dry mass (g) to water saturated leaf mass (g). Using the leaf scans, we measured leaf area with the image processing software ImageJ (Schneider et al., 2012). Specific Leaf Area (SLA) was then calculated as the ratio of leaf area (cm2) to dry mass (g).

Plot grain yield and biomass

Grain yield and aboveground biomass of each crop species was determined per plot at maturity. This corresponded to July/August. As time of maturity slightly varied among the different crop species, we conducted harvest species by species. We clipped plants right above the soil surface and separated seeds from the vegetative parts. Seeds were sun-dried for 5 days and weighed. Biomass was oven-dried at 80 °C until constant weight and weighed.

Individual yield and biomass

We harvested the three marked individuals for the trait measurements separately; we separated seeds from aboveground biomass and they were both dried and weighed as previously mentioned. Furthermore, for each marked individual we weighed ten randomly selected seeds to obtain the mass per seed.

Data analyses

All analyses were performed using R version 4.1.0 (R Development Core Team, 2019).

Plant Interaction Index

Plant interaction intensity in the plots was calculated for each marked individual by means of the relative intensity index (RII) defined as such (Diaz Sierra et al., 2017):

(1) RII= yieldcomm-yieldsingleyieldcomm+yieldsingle

, where yieldsingle is the grain yield (in grams) of a single plant grown in isolation, and yieldcomm is the grain yield (in grams) of an individual of the same species when grown in a community. yieldsingle was calculated for each species, fertilizing conditions and coexistence history by taking the average of the four corresponding replicates. RII is a standardized index with commutative symmetry commonly used to measure plantplant interactions (Armas et al., 2004). A positive RII means that the individual is benefiting in terms of productivity, that is yield from being in a community compared to growing alone, and therefore indicates facilitation. On the contrary, a negative RII means that the individual is suffering from being in a community compared to growing alone, and therefore indicates competition. RII values of all species (a,b,c,d) composing the community (i.e. species a in case of a monoculture and species a to d in case of a mixture of four species) were averaged and subsequently weighted by their relative abundance ri= 1number of species to calculate the mean net interaction in the community (RIInet):

(2) RIInet= i=ad(RIIiri)

This net index thus indicates whether on the community level, plants are experiencing facilitation or competition. The closer this index gets to 1, respectively –1, the stronger the facilitation, respectively competition. To check the applicability of this net index, we looked at the correlation between this index and the complementarity effect from Hector & Loreau (Vandermeer, 1992) (see below for the calculations) and indeed we found a positive correlation across all plots (Figure 5, [F=4.62, p-value = 0.033, n=204]). This shows that a higher net index that is decreased competition indeed correlates with higher complementarity effects and therefore, we are confident that net RII reasonably describes plant interactions.

The reference values for RII per species were computed per fertilizer and coexistence history, which means that each coexistence history has a different reference value. We chose this way of calculating these metrics as this allows to explicitly distinguish the effects of coexistence history on the interactions, independently of the baseline effect on plant performance. This follows the classic framework of plantplant interaction and facilitation work (Michalet et al., 2014). To further investigate potential changes in reference plant performance, we calculated RII coexistence for each community (i.e. for single plants, for monocultures, and for mixtures) using the following calculations.

RIIcoexistence=yieldsinglewithcommunityhistoryyieldsinglewithsinglehistoryyieldsinglewithcommunityhistory+yieldsinglewithsinglehistory

for single plants and

RIIcoexistence=yieldmonoculturewithmixorsinglehistoryyieldmonoculturewithmonohistoryyieldmonoculturewithmixorsinglehistory+yieldmonoculturewithmonohistory

for monocultures (Figure 5—figure supplements 1 and 2).

Net biodiversity effect

For all mixture communities we quantified the net biodiversity effect (NE) defined as the overyielding relative to the expected yield based on monocrop values.

NE=Y=Yo-YE=Yo-i=1s(riMi)

where Yo is the observed yield of the mixture, YE is the expected yield measured as the sum of the monocrop yield of each species (Mi) weighted by the species proportion in the mixture. The monocrop yield was calculated for each species, fertilizing conditions and coexistence history by taking the average of the two corresponding replicates.

We partitioned net biodiversity effect into its two components, the complementarity and selection effects according to Loreau and Hector, 2001.

(5) NE=NΔRY¯M¯+ Ncov(ΔRY,M)

where N is the number of species in the plot, ΔRY is the deviation from expected relative yield of the species in mixture in the respective plot, which is calculated as the ratio of observed relative yield of the species in mixture to the yield of the species in monoculture, and M is the yield of the species in monoculture. The first component of the biodiversity effect equation (NΔRY¯M¯) is the complementarity effect (CE) and represents how much individual species contribute more to productivity than predicted from monoculture. The second component (Ncov(RY,M)) is the selection effect (SE) and describes the greater probability of more diverse communities including highly productive species which account for the majority of productivity.

Total crop yield

To assess crop performance, we calculated total crop yield per plot as the sum of total seed mass per species.

Trait analyses

Traits were analysed both at the species-level and at the plot-level. At the species level, we calculated the mean and coefficient of variation (CV) per species for each trait per plot. At the plot-level, we calculated Community-Weighted-Means (CMW, which is defined as the average of trait values for each species weighted by the species relative biomass Miller et al., 2004), and coefficient of variation per plot for each trait.

Functional richness (FRic) was calculated in each plot using the function dbFD from the package FD (Laliberté and Legendre, 2010), by measuring the convex hull volume occupied by the individuals of a plot in the space of the considered traits.

To analyze the effects of the experimental treatments on RIInet, NE, CE, SE, total crop yield, FRic, and CWM and CV per plot, we used generalized linear mixed models using the function lmer. Fixed factors included fertilizing condition (yes or no), coexistence history (considered as ‘same’ or ‘different’), crop species number (2 vs 4) nested in monoculture vs mixture, as well as the interactions between them. Species composition, bed and column were set as random factors.

Responsevariables(e.g.yield) fertilizationcoexistencehistory(monovsmix+cropspeciesnumber)+(1|comb)+(1|bed)+(1|column)

Effect sizes were calculated from marginal means obtained using the function emmeans, and pairwise comparisons were calculated using Tukey tests from the emmeans function (Lenth, 2021). To analyze the effects of the experimental treatments on the mean and coefficient of variation of the different traits per species (height, width, SLA, LDMC, mass per seed, respectively), we used generalized linear mixed models using lmer with the same fixed factors as previously described. Species, species composition, bed and columns were set as random factors. The response variables were log-transformed or square-root-transformed where needed. To analyse the response of FPAR, we calculated the average of the three measurements per plot for each week and analysed its response by using similar linear mixed models as described above on all the dates, with day of year as a random factor. For all models, we tested for normality of the residuals using a ShapiroWilk test and homogeneity of the variance using a Levene test.

Data availability

The data that support the findings of this study are available on Zenodo: https://doi.org/10.5281/zenodo.5223410.

The following data sets were generated
    1. Stefan L
    (2021) Zenodo
    Rapid adaptation in Intercropped Systems.
    https://doi.org/10.5281/zenodo.5223410

References

    1. Begemann F
    (2002)
    Sorten-und rassenvielfalt im supermarkt–eine kooperation zwischen der schweizer erhaltungsinitiative pro species RARA (PSR) und der supermarktkette coop
    Vielfalt Auf Dem Markt, Schriften Zu Genet. Ressourcen 17:143–152.
    1. Choudhary S
    2. Sharma R
    3. Nehru J
    4. Vishwavidyalaya K
    5. Vishal MK
    (2019)
    Molecular diversity in coriander (coriandrum sativum) using RAPD and ISSR markers
    Indian Journal of Agricultural Sciences 89:193–198.
  1. Conference
    1. Diederichsen A
    (1996)
    Coriander (coriandrum sativum L.)
    promoting the conservation and use of underutilized and neglected crops.
    1. Donohue K
    (2009) Completing the cycle: maternal effects as the missing link in plant life histories
    Philosophical Transactions of the Royal Society of London. Series B, Biological Sciences 364:1059–1074.
    https://doi.org/10.1098/rstb.2008.0291
  2. Software
    1. R Development Core Team
    (2019) A language and environment for statistical computing
    R Foundation for statistical computing, Veinna, Austria.
    1. Singh SK
    2. Kakani RK
    3. Meena RS
    4. Pancholy A
    5. Pathak R
    6. Raturi A
    (2012)
    Studies on genetic divergence among indian varieties of a spice herb, coriandrum sativum
    Journal of Environmental Biology 33:781–789.
  3. Book
    1. Vromans J
    2. Pieper B
    3. Stam P
    (2006)
    Molecular Genetic Studies in Flax
    Wageningen University.

Decision letter

  1. Bernhard Schmid
    Reviewing Editor; University of Zurich, Switzerland
  2. Meredith C Schuman
    Senior Editor; University of Zurich, Switzerland
  3. Piter Bijma
    Reviewer; Wageningen University, Netherlands

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Rapid transgenerational adaptation in response to intercropping reduces competition" for consideration by eLife. Your article has been reviewed by 4 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Meredith Schuman as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: Piter Bijma (Reviewer #3).

The reviewers agree that the idea behind the paper is a good one and the experiment is appropriate and well analyzed.

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

In your revision, please respond point-by-point to these essential revisions, which all reviewers agreed upon. In addition, the individual reviews, which are provided below, may contain some further points of interest.

1) The major result as suggested in the title seems to be reduced competition of offspring with a co-occurrence history. However, reviewers feel that results about biodiversity and yield effects should be the focus: greater weight is given to results on RII, whereas results on NBE/yield are only marginally considered. The emphasis should rather be put on NBE and yield. They are the variables of interest to characterize the consequences of plant-plant interaction at the group level and they are the ones classically used in the BEF field (see Zuppinger-Dingley et al., Nature, 2014).

2) However, the significance of the effect of coexistence history on NBE in the fertilized treatment is not apparent in the manuscript. The analysis of fertilized vs. unfertilized treatments should be improved to better show if indeed there is an indication of greater net biodiversity effects under fertilized conditions. The analysis of traits should also be improved by separating between- from within-species variation and referring to work that shows different traits respond differently (height generally converging and leaf traits generally diverging, work by Roscher et al.).

3) The discrepancy between 1) and 2) is very hard to understand and must be discussed by the authors. The way the single-plant reference is used for RII may affect the interpretation: is the apparent reduction in competition due to the mixture coexistence history, or due to the single reference plant? Because competition is always measured in terms of yield effects, the RII and total yield results would seem to rely on the same data. Hence, how can there be a true decrease in competition without an increase in yield? If this does not become clear, then the paper lacks a clear message.

It is quite difficult to interpret RII and NBE as we do not know how the reference values were computed and if the same reference values were used across treatments. Because these indices are relative to single plants or monocultures, changes in index values can be caused by changes in the reference values (i.e. single plant or monoculture, respectively), not only by changes in the value measured in the mixtures.

4) Compared with the SGH, there is as much evidence for the opposite response of increased biodiversity effects with improved environmental conditions. The authors should discuss their absence of evidence for SGH.

5) The authors should tone done facilitation and focus on reduced competition (as they do in the title and abstract). They should probably omit the partitioning of the competition index into competition and facilitation components. For example, when RII goes from -0.5 to -0.3 using the same single isolated plant yield as a reference, it means that single plant yield has increased in the mixtures. This can be caused either by higher facilitation or lower competition. How can you tell the two mechanisms apart?

6) Results for the second year might be included in the SI.

7) Potential mechanisms underpinning transgenerational effects and reduced competition should be better explained. Thus, for the first it should be mentioned that it could occur by evolution via sorting out from standing variation (for highly selfing species if at least there was variation between initially sown genotypes), recombination (probably minor contribution in only 2 generations) or mutation (even less likely) or by epigenetic/physiological carry-over processes.

8) The methods should be described in more detail.

Reviewer #1 (Recommendations for the authors):

I think the manuscript can be improved by removing the first index (RII) and focusing on plot-level analysis. This RII is not commonly used and not very explicit for the reader compared to NBE (or NE). Overall, the plot level data and trait data point to the same direction, which is no increase in complementarity between species over a generation of coexistence. I would focus on this message and discuss the limits of the design to detect a potential increase in complementarity if they were to occur as described in grasslands (inbred lines with low evolutionary potential, short evolutionary time, alternate rows which limit interactions, etc).

I would avoid commenting on "trends" and non-significant results in the "Results and Discussion" section. I would also add some significance measures in this section (either p-value or stars on the Figures, or ANOVA Tables).

Line by line comment:

l. 15: Why using seeds selected in monoculture could compromise yield benefit in mixtures? This is explained in the following sentence, but we lack the connection with this sentence.

l. 37: ref Vandermeer, J.H. (1992) The Ecology of Intercropping, Cambridge University Press.

l. 44-47: ref Meilhac, J. et al. (2020) Both selection and plasticity drive niche differentiation in experimental grasslands. Nat. Plants 6, 28-33.

l. 44: The first sentence is very general and does not convey any information. In the second sentence, it needs to be explained why the use of commercial seeds bred for monocultures might not be optimal to promote positive diversity effects.

l. 52: "changed and evolved" is redundant.

l. 54: I would provide more information on the species (species or functional groups) and say if they are commonly grown as intercrops.

l. 56: mesocosms is only used here and not defined.

l. 57-8: "We selected open-pollinated varieties", This does not seem to be the case for oat, wheat, and lentil in the Methods section.

l. 62: It is not clear what is the difference between this Relative Interaction Index, and the classical Relative Yield Total used in the partitioning of Loreau and Hector at this stage. Why use both? What is the information added by Relative Interaction Index?

l. 69 to 79: it should be stated that this hypothesis applies to mixture communities only.

l. 77-8: The SGH arrives a bit "out of the blue" here. It needs to be defined and developed within the context of intercropping before.

l. 81-82: Figure 2 and Table S1 suggest that the "history" treatment effect is primarily driven by facilitation, not by competition. Put differently, the difference between "same" and "different" evolutionary history is much bigger for the positive interaction index than for the negative ones. The sentences here suggest the opposite, i.e. greater effect on competition than on facilitation.

l. 83 and throughout the manuscript: "SI" can be removed before supplementary material references.

l. 91: "less competition": I think it should be "more facilitation" instead (cf previous comment).

l.99 to 105: this whole part only comments on non-significant results.

l. 112: I could not find to which data and test the p-value and F value reported here refer.

l. 142-144: "height community-weighted mean was not significantly different between coexistence histories".

Paragraph 120-145: I would add the results on Functional hypervolume and PAR here. I think it is especially interesting to discuss the differences in PAR interception between communities with different coexistence histories (you do have significant evidence of complementary effects increased in mixtures with a mixture history here).

l. 172-173: "In fertilized conditions, this shift in plant-plant interactions was associated with an increase in overyielding": I could not find the statistics supporting that statement. In Table S4, the Fertilizer x history effect is significant, but we do not have the tests of the history effect within each fertilization treatment. Also, the p-value used to support this in the result section (l. 103) is higher than 0.05.

l. 194-200: this discussion should be confronted with the Stress-Gradient Hypothesis, as presented in the Introduction. Indeed, the SGH predicts the opposite pattern as the one observed in the study, i.e., stronger biodiversity effects in low-input systems.

l. 205-206: "a reduction in trait variation favoured increased yield benefits in mixtures": I cannot find any result supporting that statement.

l. 209: "plants might have adapted to express the phenotype that would maximise their fitness": again, I do not see any relationship between traits and fitness in the results.

l. 218-222: this result should be presented in the main text.

l. 227-228: "we specifically selected open-pollinated varieties in order to ensure a minimum amount of genetic variability". There are two problems here. First, it was not open-pollinated varieties for all species. Second open-pollinated varieties have more genetic variability than other varietal types such as inbred lines or hybrids.

l. 262: 30 cm of what?

l. 268: were the metal frames only belowground or also aboveground?

l. 286 and the whole sub-section: It is not easy to assess the extent of genetic variation within each species in this section. It seems that there was no within-species variation for wheat, oat, and lentil, and maybe some variation for the 3 others. I think having this kind of information would be important to interpret the results phenotypic changes can occur both through selection and plasticity, depending on the amount of standing variation in the species.

l. 335 to 337: I first did not understand that single plant plots were grouped together in separate beds. In fact, I only got it later with the picture in Figure S10. I would reword this part.

l. 340-342: We have no information on how the different species were arranged in 4-species mixtures.

l. 342-343: Are these densities the monoculture densities? If yes, how were these densities adjusted in mixtures?

l. 400: "yield": is it grain or biomass yield? Also, how was "Yield single" computed? Did you consider only a single plant from the same year, same fertilization, with a pure single plant history? Were the different single plants from the same species averaged, corrected for design effects, etc?

l. 410: "proportional": amend to "relative".

l. 418-420: Why not use directly the Loreau and Hector partitioning then?

l. 431 to 436: Since RIINet and NInC are highly correlated, it is not surprising that you get the same results: you just repeated your analysis with a redundant index. I would either remove or just mentioned that RIInet and NIntC were strongly correlated and thus provided identical results.

l. 440: as for RII and single plant yield, how were monoculture yields computed here? Was it averaged across replicates of the same species, only considering replicates with a "pure" monoculture history?

l. 451-452: Community-Weighted Mean needs to be defined and referenced.

l. 456 and 457: "NIntC" should be amended to "RII" and "LER" should be amended to "SE"

Table S17: Which PAR values were used in this analysis of variance? Was it a single date, or averaged across dates?

Figure S10: I would move this Figure to the main text or provide an illustration of the experimental set-up there to help the readers visualize what are beds, plots, and how are the different species arranged, notably in mixture plots.

Reviewer #2 (Recommendations for the authors):

1. Please give a more balanced discussion of the proposed mechanisms for how co-occurrence can lead to more facilitation. I see how co-occurrence could lead to trait displacement and less niche overlap, so less competition. But what is the facilitation part of this? Please include something in the intro that covers this.

2. Your RII_fac and RII_comp appear to be superficially separated based on whether the net effects of neighbors were positive or negative, is this correct? What would happen if one species in a mixture plot had net positive effects of neighbors and another species had net negative effects of neighbors? Would you calculate the whole plot as just the sum of that? Or how is it calculated in that case?

3. Also, how do you separate intraspecific competition and interspecific competition from these metrics? For example, if intraspecific comp > interspecific comp, you would still see a less negative RII_comp at higher diversity. But this isn't necessarily because competition is actually weaker at higher diversity (but instead a shift from one type of competition to another). Similarly, if intraspecific comp > interspecific comp, you could also see these shifts in RII_fac. A stronger RII_fac value could be completely driven by just alleviation of intraspecific competition at higher diversity.

4. I'm thus consistently tripped up by the many phrases in the MS where you say things like: "shifted towards weaker competition, and in some cases, stronger facilitation" (lines 81-82). All of your RII values are net effects, so how can you conclude that what drives the changes you see is due to competition or facilitation? Why not: "shifted towards weaker competition and/or stronger facilitation (though teasing out the differences is not possible in our given dataset)".

5. Overall trends I see in RII: most things do worse when growing next to same species neighbors (vs. alone), but the same or better if they grow next to heterospecific neighbors. This gets stronger if they have coexisted for multiple generations. And the positive aspects of it are most strongly affected by coexistence. Your data indicate that this ISN'T due to trait divergence which is super interesting! You also show a strong correlation with selection effects which indicates that a couple of species are driving the patterns. Did you take a look at how selection effects were correlated with RII? Do you know which species are doing that? Have you split into plots with and without legumes? Do legume effects on neighbors get stronger with coexistence history? The only other mechanism I could imagine is more specialization of enemies over time leading to stronger dilution effects? Please discuss these possible mechanisms in some way.

6. The loss of variability in each species when they have coexisted for multiple generations also seems to point towards some kind of restricted gene flow. How much interbreeding was there between plots? Is there something about the environment that you could imagine selecting for these very tall plants? Also, the fact that they are getting more like each other overall, but still having more positive effects on each other is very strange. Need more context about how this might be explainable.

Reviewer #3 (Recommendations for the authors):

Background on main comments / further general comments

The presence of clear effects of the coexistence history on competition together with absence of effects on yield is worrying. Particularly because effects on competition relative to single plants (Figure 2) are statistically very significant while those on Net Effect relative to monocrop are borderline (P = 0.0715). At the same time, the plot of total yield excludes the results for single plants (these should be included).

The decrease in competition due to the same coexistence history shown in Figure 2 is relative to single plants. Hence, the comparison is relative. Do we really see a decrease in competition here, or is the single plant getting worse due to the coexistence history? Importantly: It is unclear in Figure 2a whether the two points that make up a pair of points for the same x-axis value are relative to a single plant with the same coexistence history. For example, for x = 1 (monoculture) the two y-axis values are about -0.3 and -0.2. Is the reference single plant the same for these two points the same, so that we can also infer that the difference between both points is 0.1? Or is the reference single plant for the first point (-0.3) a single plant with also a Different co-existence history, and the reference single plant for the second point (-0.2) a single plant with also a Same co-existence history. If the latter is the case, then both points have a different reference, and the apparent difference of 0.1 may very well originate from having two different single plant references, rather than an effect on competition in the monocrop.

Related to the previous comment: I cannot reconcile the results for No Fertilizer in Figures 3a and 3b. NE is a measure of overyielding relative to monocrops. Figure 3a NoFertilizer shows that overyielding is greater with Different. However, 3b shows greater total yield with Same. This strongly suggests that the two points making up a pair in Figure 3a do not have the same monocrop as the reference. Hence, then we don't know whether the effect is due to the coexistence history of the 2-mixture or due to the co-existence history of the monocrop used as reference. Things get very confusing because the reference point seems to be shifting all the time (to the best of my understanding).

The Introduction refers to "genetic variability needed", but it is not so clear whether (natural) selection happens (or is the focus), or whether interest is in non-genetic transgenerational effects. Whether or not seed selection for the next generation was affected by natural selection should be discussed.

Related to the previous comment: natural selection for individual performance in mixture populations is expected to lead to an increase in competition (e.g. Griffing 1967 and later similar work on IGE). This seems to agree with the observation of taller plants containing more water. This could be discussed.

In the statistical analysis at the species level (L464-466) it seems a species effect is included as a random effect? That is surprising. Why?

Detailed comments

L21: My impression is that, in Figure 3b Fertilizer, x-axis values 2 and 4, these differences are not significant. IF so, is this statement warranted?

L24: Taller with more water seems to be indicative of more competition. This could be discussed (in Discussion).

Introduction: please partition the Introduction into several paragraphs. At present, it is hard to read. For example at L44, L50, L61, L69.

Can something be said about the selection history of the source material? Have all these varieties been selected for monocrop performance?

L81: It seems Figure 2b shows differences for facilitation, not (or sometimes) for competition, which contradicts this sentence. Maybe y-axis labels in Figure 2b and c have accidentally been swapped? Anyway, as suggested above, I suggest dropping the distinction between less competition and more facilitation. Hence, I suggest dropping panels b and c.

For mixtures, the meaning of "same coexistence history" is not fully clear. Is this 2 with a 2 history, or also e.g. 2 with a 4 history, etc. And how often?

Figure 2: I propose to explicitly state in the text that this (RII) is a measure relative to single plants.

L100: In the context of the current manuscript, the "corresponding monoculture" is not fully clear. Does it also mean: "with the same coexistence history" as the focal mixture?

L105-107: Is this sentence warranted given the absence of any indication of statistical significance (P = 0.67)?

Figure 3b: For easier interpretation: Could the y-axis be presented on the original (not square root transformed) scale? So can the estimates be back-transformed even though the P-values come from the transformed data?

Figure 4 panel e,f, monocrop (x = 1) seems to suggest that plants become more competitive when grown in the mixture (in the previous 2 generations), which would agree with the theory (Griffing 1967).

L456-470 The description of the statistical analysis is very verbal. A model presented as a mathematical equation would be much easier to read, and particularly to find back.

The Net Biodiversity Effect (L440) could be explained much easier, by first stating that it measures overyielding relative to monocrops, giving a simple equation for that, and only after that splitting this up into a main effect (the CE) and an interaction (the SE). Then the interpretation of the CE and SE should be clarified. (This is very similar to general and specific combining ability, an analogy that may help the reader).

L457: I may have missed it, but what is LER?

References

Griffing, B. (1967). Selection in reference to biological groups I. Individual and group selection applied to populations of unordered groups. Australian Journal of Biological Sciences, 20(1), 127-140.

https://doi.org/10.7554/eLife.77577.sa1

Author response

Essential revisions:

In your revision, please respond point-by-point to these essential revisions, which all reviewers agreed upon. In addition, the individual reviews, which are provided below, may contain some further points of interest.

1) The major result as suggested in the title seems to be reduced competition of offspring with a co-occurrence history. However, reviewers feel that results about biodiversity and yield effects should be the focus: greater weight is given to results on RII, whereas results on NBE/yield are only marginally considered. The emphasis should rather be put on NBE and yield. They are the variables of interest to characterize the consequences of plant-plant interaction at the group level and they are the ones classically used in the BEF field (see Zuppinger-Dingley et al., Nature, 2014).

While we agree that NBE and total yield are the main measures classically used in the BEF and intercropping fields, we strongly believe that the added value of this project is precisely that we use a plant-plant interaction perspective (i.e. plant’s eye view) to improve our understanding in the BEF field. Rather than BEF relationships, we believe that plant-plant interactions are the core of this study, and this manuscript proposes to apply plant-plant interaction methods and metrics in an ecological approach to intercropping, also to understand mechanisms underlying potential NBE.

RII is a metric that has been extensively used in the field of plant-plant interactions and facilitation, such as Schöb Nat. Ecol. Evol. (2018), Diaz-Sierra Methods Ecol. Evol (2017), Schöb New Phyt (2014), Michalet Fun Ecol. (2013).

The classic method used in BEF studies, i.e. the additive partitioning of Loreau and Hector, compares the performance of mixtures vs monocultures. Because the reference is a monoculture, this method does not allow to understand the processes happening within monocultures, where plants can already suffer more or less from intraspecific competition. Since the core of this project was to study plant-plant interactions, we focused on the Relative Interaction Index because it does give additional information regarding the behaviour of plants both in monocultures and mixtures. For this metric, the reference is the individual plant growing in isolation, i.e. with zero interactions. This allows to compute and compare plant-plant interactions between different monocultures (in our case, fertilized vs unfertilized, or with different community histories). Furthermore, this index gives more insight into the mechanisms driving the classic net effects sensu L and H: indeed, if a plant grows more in a community compared to a single plant, this means that the net effect has at least a facilitative component. In contrast, if a plant grows less in a community in comparison to single plants, this means that the net effect has a competitive component.

Thus, we feel that it would be regretful to diminish the weight of RII and plant-plant interactions in the manuscript. However, we think it is reasonable to give an equal weight to community-level variables, such as NBE and yield, and we therefore developed these results more extensively in the discussion.

2) However, the significance of the effect of coexistence history on NBE in the fertilized treatment is not apparent in the manuscript. The analysis of fertilized vs. unfertilized treatments should be improved to better show if indeed there is an indication of greater net biodiversity effects under fertilized conditions. The analysis of traits should also be improved by separating between- from within-species variation and referring to work that shows different traits respond differently (height generally converging and leaf traits generally diverging, work by Roscher et al.).

The posthoc table of the significant interaction between fertilization and coexistence history was now added to the Supplement (Supplementary file 1d). The posthoc test does reveal that there is a marginally significant difference between coexistence history treatments in fertilized conditions, with a p-value of 0.0587.

The analysis of between and within species variation for traits has already been performed (see SI): we did the analyses at the plot level [i.e. between species] and at the species level [i.e. within species]. The very relevant work of Roscher at al has now been included and discussed in more detail. Thank you for drawing attention to it.

3) The discrepancy between 1) and 2) is very hard to understand and must be discussed by the authors. The way the single-plant reference is used for RII may affect the interpretation: is the apparent reduction in competition due to the mixture coexistence history, or due to the single reference plant? Because competition is always measured in terms of yield effects, the RII and total yield results would seem to rely on the same data. Hence, how can there be a true decrease in competition without an increase in yield? If this does not become clear, then the paper lacks a clear message.

It is quite difficult to interpret RII and NBE as we do not know how the reference values were computed and if the same reference values were used across treatments. Because these indices are relative to single plants or monocultures, changes in index values can be caused by changes in the reference values (i.e. single plant or monoculture, respectively), not only by changes in the value measured in the mixtures.

RII is a measure of competition/facilitation and it indeed correlates well with CE from Loreau and Hector (in both fertilizing conditions) -> see Figure 5. Therefore, this is consistent, in the sense that higher RII (=reduced competition and/or increased facilitation) correlates with higher complementarity effects.

In theory, there is no reason why RII has to correlate with NE, as NE is driven both by CE and SE. In our case, RII actually correlates negatively with SE. Therefore, the positive correlation with CE is compensated by the negative one with SE, and in the end, we observe no correlation with NE. An SE can easily be explained by compensatory changes in RII of two species composing the mixture. If the high yielding species shows a 20% higher RII in mixture than in monoculture and the low yielding species shows a 20% lower RII in mixture than in monoculture, then we get a net RII change of 0, but a positive NE due to a positive SE.

Another reason that might explain the discrepancy between RII and NBE is the different levels of reference plants: indeed, RII uses a single plant as a reference, while NBE uses a monoculture. To further investigate this, we calculated RII in the mixtures using the individual in monoculture as reference hereafter called RII_monoculture:

RII_monoculture = (yield of individual of species A in the mix – yield of individual of species A in mono)/(yield of individual of species A in the mix + yield of individual of species A in mono). RII_monoculture significantly correlates with the complementarity effects from Hector and Loreau (see Author response image 1, p-value < 0.001). Furthermore, when running the linear model on this new index, we did not find any significant effect of coexistence history on RII_monoculture (see Author response table 1). This is in line with the lack of a significant history effect on CE and emphasizes the role of using single individual plants as a reference to better understand changes in plant-plant interactions with coexistence history that underly NBE.

Author response table 1
Anova table for RII_monoculture.
NumDFDenDFF valuePr(>F)
Fertilizer17.8831.3060.286645
History1180.4590.98980.321121
Diversity115.0980.14880.705034
Fertilizer x history1180.3677.85310.005628 **
Fertilizer x diversity1175.7030.2170.641907
History x diversity1175.840.97120.325747
Fertilizer x history x diversity1175.6410.24650.620166
Author response image 1
Complementarity effects.
Author response table 2
Posthoc test for the significant interaction on fertilizer x history.
estimateSEdft.ratiop.value
no diff – yes diff0.1290.065113.21.9810.2436
no diff – no same0.09640.0562180.71.7150.3187
no diff – yes same0.0590.072419.60.8150.8466
yes diff – no same-0.03270.072819.7-0.4490.9691
yes diff – yes same-0.070.0546177-1.2840.5745
no same – yes same-0.03740.079427.1-0.4710.9649

Finally, the reference values of the single individuals per species were computed per fertilizer and coexistence history. And these reference values do indeed change with coexistence history (Figure 5 —figure supplement 1 and 2).

We chose this way of calculating these metrics as this allows to explicitly distinguish the effects of coexistence history on the interactions, independently of the baseline effect on plant performance overall. This follows the classic framework of plant-plant interaction work (see Michalet et al. 2014 Fun Ecol for instance). If we had kept the same reference plants for all coexistence histories (say single with single history, for instance), then we would not have been able to determine if the observed effect in the communities was due to the neighbour effect or because coexistence history was affecting plant performance in general (i.e. both single plants and plants in communities).

To investigate further the potential changes in reference plant performance, we calculated “RII coexistence” for each community (as they did in Michalet et al. 2014 for the effect of altitude on plant-plant interactions). This allows to see whether, within each diversity level (i.e. single, monocultures, mixtures), coexistence had an effect per se on plant performance in general.

This additional index allows to see that the effect of coexistence history on the performance of single plants varies across species: in the case of oat, flax and wheat, coexistence history does not change the performance of single individuals. In the case of camelina, coriander and lentil, the single plants grew worse when they had a community history than when they had a single history, which corroborates our results regarding adaptation in response to coexistence history. Indeed, the fact that single plants coming from single plants perform better than single plants coming from communities already suggests that plants have adapted to their surrounding community. Therefore, we believe that changes in these reference plants are already an effect of coexistence history, and that they should not be ignored by taking the same reference for all coexistence history treatments.

When calculating changes in plant performance in response to coexistence history within monocultures, we see again that this “RII monoculture” index is mostly negative, which means that the performance of plants with a mixture or single history is lower than plants with a monoculture history. This corroborates our adaptation results: in monocultures, plants coming from monocultures do better than plants coming from singles or mixtures. It also shows that coexistence history has an effect of plant performance per se and justifies the choice to take this effect into account when calculating the effect of neighbour on plant-plant interactions (i.e. by taking the appropriate reference).

This choice of index can explain why the effect of coexistence history on relative metrics (such as RII) does not appear in absolute metrics (such as total yield). We also suggest that the limited timeframe of this study – two generations – might be the reason for the lack of more significant changes in total yield. However, looking at the trend, we believe that there are clues indicating increased yield with a common coexistence history. We discuss this now extensively in the main manuscript (L211-226).

These additional results have now been added in the supplement (Figure 5 —figure supplement 1 and 2).

4) Compared with the SGH, there is as much evidence for the opposite response of increased biodiversity effects with improved environmental conditions. The authors should discuss their absence of evidence for SGH.

We believe that we actually see evidence for the SGH, according to our initial hypotheses: “we expected more facilitation and/or less competition in conditions of reduced soil fertility”. This is what we observed: in fertilized plots, competition was stronger and/or facilitation weaker than in unfertilized plots. Furthermore, in unfertilized plots only do we see “true” facilitation in mixtures (i.e. positive RII in mixtures and negative RII in monocultures). Furthermore, we suggest that increased biodiversity effects with improved environmental conditions are also in agreement with the SGH: if competition is more important or more intense in fertile conditions due to higher plant growth, then the benefits of niche differentiation coming from increasing diversity should also be higher. This has been visible in several intercropping studies, where authors have found increased biodiversity effects in highly-productive systems (see Li et al. 2020 Nat Plants, and Chen et al. 2021 Nat Plants). (5) The authors should tone done facilitation and focus on reduced competition (as they do in the title and abstract). They should probably omit the partitioning of the competition index into competition and facilitation components. For example, when RII goes from -0.5 to -0.3 using the same single isolated plant yield as a reference, it means that single plant yield has increased in the mixtures. This can be caused either by higher facilitation or lower competition. How can you tell the two mechanisms apart?

We removed the partitioning into competition and facilitation components as indeed a reduction in competition could also mean an increase in facilitation, and vice-versa. We also tried to rephrase and be more careful in our statements, such as “reduced competition and/or increased facilitation” to be more precise as it is indeed impossible to disentangle the two mechanisms when RII is negative in both the monoculture and the mixture (but less negative in the mixture). (6) Results for the second year might be included in the SI.

We collected only partial data after year one, as this was considered as an intermediate stage, where adaptation was less likely to give significant results. Notably, we put fewer efforts into collecting data at the individual-level and reduced the number of traits measured, which prevented us from having a full picture of the response of plant-plant interactions as well as of the trait space. Therefore, we decided to not present these partial pieces of data in the study.

7) Potential mechanisms underpinning transgenerational effects and reduced competition should be better explained. Thus, for the first it should be mentioned that it could occur by evolution via sorting out from standing variation (for highly selfing species if at least there was variation between initially sown genotypes), recombination (probably minor contribution in only 2 generations) or mutation (even less likely) or by epigenetic/physiological carry-over processes.

This was more extensively discussed in the discussion based on these helpful recommendations (L271-89).

8) The methods should be described in more detail.

This was improved according to the detailed comments below.

Reviewer #1 (Recommendations for the authors):

I think the manuscript can be improved by removing the first index (RII) and focusing on plot-level analysis. This RII is not commonly used and not very explicit for the reader compared to NBE (or NE). Overall, the plot level data and trait data point to the same direction, which is no increase in complementarity between species over a generation of coexistence. I would focus on this message and discuss the limits of the design to detect a potential increase in complementarity if they were to occur as described in grasslands (inbred lines with low evolutionary potential, short evolutionary time, alternate rows which limit interactions, etc).

As indicated in the response to the review summary provided by the review editor (i.e. point (1) above), we feel that the plant’s eye view represented by the RII metric was the core of the study and of this manuscript; therefore, we think that it would be regretful to remove this element, which brings additional information regarding plant-plant interactions and how they relate to community-level metrics such as NE. However, we reconcile now extensively in the discussion the individual-level results and plot-level results.

I would avoid commenting on "trends" and non-significant results in the "Results and Discussion" section. I would also add some significance measures in this section (either p-value or stars on the Figures, or ANOVA Tables).

We added stars to the figures and removed all the comments regarding trends.

Line by line comment:

l. 15: Why using seeds selected in monoculture could compromise yield benefit in mixtures? This is explained in the following sentence, but we lack the connection with this sentence.

This was rephrased.

l. 37: ref Vandermeer, J.H. (1992) The Ecology of Intercropping, Cambridge University Press.

This was added.

l. 44-47: ref Meilhac, J. et al. (2020) Both selection and plasticity drive niche differentiation in experimental grasslands. Nat. Plants 6, 28-33.

This was added, thank you.

l. 44: The first sentence is very general and does not convey any information. In the second sentence, it needs to be explained why the use of commercial seeds bred for monocultures might not be optimal to promote positive diversity effects.

This was made clearer.

l. 52: "changed and evolved" is redundant.

We kept only “changed”.

l. 54: I would provide more information on the species (species or functional groups) and say if they are commonly grown as intercrops.

This was added.

l. 56: mesocosms is only used here and not defined.

This was defined.

l. 57-8: "We selected open-pollinated varieties", This does not seem to be the case for oat, wheat, and lentil in the Methods section.

As we could not know the initial amount of standing variation, we clarified that we selected, whenever possible, open-pollinated varieties.

l. 62: It is not clear what is the difference between this Relative Interaction Index, and the classical Relative Yield Total used in the partitioning of Loreau and Hector at this stage. Why use both? What is the information added by Relative Interaction Index?

The classic method of Loreau and Hector, and also the Relative Yield Total or Land Equivalent Ratio, compares the performance of mixtures vs monocultures. Because the reference is a monoculture, this method does not allow to understand the processes happening within monocultures, where plants can already suffer more or less from intraspecific competition. Since the core of this project was to study plant-plant interactions, we focused on the Relative Interaction Index because it does give additional information regarding the behaviour of plants both in monocultures and mixtures. For this metric, the reference is the individual plant growing in isolation, i.e. with zero interaction. This allows to compute and compare plant-plant interactions between different monocultures (in our case, fertilizer vs unfertilized, or with different community history). Furthermore, this index gives more insight into the mechanisms driving the classic net effects sensu L and H: indeed, if a plant grows more in a community compared to a single plant, this means that the net effect has at least a facilitative component. On the contrary, if a plant grows less in a community in comparison to single plants, this means that the net effect has a competitive component.

l. 69 to 79: it should be stated that this hypothesis applies to mixture communities only.

This was specified.

l. 77-8: The SGH arrives a bit "out of the blue" here. It needs to be defined and developed within the context of intercropping before.

This was added in the intro (L68-78).

l. 81-82: Figure 2 and Table S1 suggest that the "history" treatment effect is primarily driven by facilitation, not by competition. Put differently, the difference between "same" and "different" evolutionary history is much bigger for the positive interaction index than for the negative ones. The sentences here suggest the opposite, i.e. greater effect on competition than on facilitation.

This was changed as indeed, the distinction between increased facilitation/reduced competition is blurry. We therefore tried to give equal weights to facilitation and competition.

l. 83 and throughout the manuscript: "SI" can be removed before supplementary material references.

This was removed.

l. 91: "less competition": I think it should be "more facilitation" instead (cf previous comment).

We changed to less competition and/or more facilitation to be consistent with the general comments.

l.99 to 105: this whole part only comments on non-significant results.

The interaction effect between fertilization and coexistence history on net biodiversity effect is significant. We now added the full table of the corresponding posthoc test to the supplement and referred to it in the main text. While the p-values of this posthoc test are indeed higher than 0.05 (p-value = 0.0587 for the history effect in fertilised conditions), this is nonetheless a marginally significant result and is worth reporting, especially considering the natural variation due to the external conditions and inherent environmental variability. We also looked into fertilized and unfertilized plots separately. We did find a significant effect of coexistence history in the fertilized plots (p-value of 0.02364, see Author response table 3); however, separating all the analyses would be confusing for the reader. We therefore suggest to keep the original model and the marginally significant effect revealed by the posthoc test (Supplementary file 1d).

Author response table 3
Author response table.
Sum Sq Mean Sq NumDFDenDF FvaluePr(>F)
history2801.42 2801.42 181.8145.31750.02364*
diversity580.76 580.76 114.8041.10240.31059
history:diversity41.19 41.19 180.2460.07820.78048

l. 112: I could not find to which data and test the p-value and F value reported here refer.

This was clarified and the corresponding tables for the ANOVA and the posthoc test were referenced and fully added to the Supplement. l. 142-144: "height community-weighted mean was not significantly different between coexistence histories".

This was changed. We also added precisions regarding the significant effect of coexistence history on height but at the species-level (not at the community level). Paragraph 120-145: I would add the results on Functional hypervolume and PAR here. I think it is especially interesting to discuss the differences in PAR interception between communities with different coexistence histories (you do have significant evidence of complementary effects increased in mixtures with a mixture history here).

We moved the PAR results to the main text but decided to keep the functional diversity as a supplement, as there was no response of history for this measure.

l. 172-173: "In fertilized conditions, this shift in plant-plant interactions was associated with an increase in overyielding": I could not find the statistics supporting that statement. In Table S4, the Fertilizer x history effect is significant, but we do not have the tests of the history effect within each fertilization treatment. Also, the p-value used to support this in the result section (l. 103) is higher than 0.05.

The posthoc test was now added in the supplement (Supplementary file 1d), showing the marginally significant effect of coexistence history in fertilised plots (p-value = 0.0587). As mentioned above, we also looked into fertilized and unfertilized plots separately. We did find a significant effect of coexistence history in the fertilized plots (p-value of 0.02364, see Author response table 4); however, we feel like separating now all the analyses would be confusing for the reader and suggest to keep the original model and the marginally significant effect- shown through the posthoc test.

Author response table 4
Author response table.
Sum Sq Mean Sq NumDFDenDF FvaluePr(>F)
history2801.42 2801.42 181.8145.31750.02364*
diversity580.76 580.76 114.8041.10240.31059
history:diversity41.19 41.19 180.2460.07820.78048

l. 194-200: this discussion should be confronted with the Stress-Gradient Hypothesis, as presented in the Introduction. Indeed, the SGH predicts the opposite pattern as the one observed in the study, i.e., stronger biodiversity effects in low-input systems.

The SGH predicts stronger facilitation in low-input systems, but not necessarily stronger biodiversity effects. Indeed, if competition is lower in low-input systems, the benefits of niche differentiation (so reduced competition) would also be lower. In highly productive systems where plants strongly compete, the benefits of niche differentiation would then be bigger. Of course, facilitation would be lower in highly-productive systems, but here in our Swiss agroecosystem, we suggest that competition is the dominant interaction, even in unfertilized conditions, as the soils were initially quite fertile. This result was found in other intercropping studies, including a large meta-analysis (see Li et al. 2020 in Nat Plants), where they found that biodiversity effects were higher in highly-productive systems.

l. 205-206: "a reduction in trait variation favoured increased yield benefits in mixtures": I cannot find any result supporting that statement.

This was rephrased to focus on the effect of coexistence history on traits.

l. 209: "plants might have adapted to express the phenotype that would maximise their fitness": again, I do not see any relationship between traits and fitness in the results.

This was rephrased to focus on the effect of coexistence history on traits.

l. 218-222: this result should be presented in the main text.

This was moved to the main text.

l. 227-228: "we specifically selected open-pollinated varieties in order to ensure a minimum amount of genetic variability". There are two problems here. First, it was not open-pollinated varieties for all species. Second open-pollinated varieties have more genetic variability than other varietal types such as inbred lines or hybrids.

This was clarified and we stated that open-pollinated varieties included camelina and coriander.

l. 262: 30 cm of what?

This was clarified.

l. 268: were the metal frames only belowground or also aboveground?

They were only belowground (30 cm belowground). This was specified.

l. 286 and the whole sub-section: It is not easy to assess the extent of genetic variation within each species in this section. It seems that there was no within-species variation for wheat, oat, and lentil, and maybe some variation for the 3 others. I think having this kind of information would be important to interpret the results phenotypic changes can occur both through selection and plasticity, depending on the amount of standing variation in the species.

We did not assess the initial standing variation within our crop populations, as investigating the underlying mechanisms was not the main point of this study and the genetic analyses would have required much more time and funding. We described the seed sources in the Method part as precisely as possible considering the information available, but regarding the mechanisms, we can only speculate. It is however true that the most common species (i.e. wheat and oat) probably had the lowest standing variation, as these two species have been selected and controlled for homogeneity for decades. This was added as an element of discussion.

l. 335 to 337: I first did not understand that single plant plots were grouped together in separate beds. In fact, I only got it later with the picture in Figure S10. I would reword this part.

This was rephrased, and we also moved Figure S10 to Figure 1 to give readers a better grasp.

l. 340-342: We have no information on how the different species were arranged in 4-species mixtures.

This was specified.

l. 342-343: Are these densities the monoculture densities? If yes, how were these densities adjusted in mixtures?

Yes indeed, these are the monoculture densities. The densities were not adjusted in the mixtures (i.e. if we had 25 individuals per vertical line in the monocultures, we kept 25 individuals per line in the mixtures for this species) to avoid concomitant effects of density changes on productivity. This was specified in the text.

l. 400: "yield": is it grain or biomass yield?

Yield is defined as grain yield.

Also, how was "Yield single" computed? Did you consider only a single plant from the same year, same fertilization, with a pure single plant history? Were the different single plants from the same species averaged, corrected for design effects, etc?

This was specified in the Methods and discussed in the response to the general comments. We always used the average value over the four replicates from the same year, with the same fertilization and coexistence history combination, e.g. a single plant with a monoculture history in unfertilised plots as a reference for monoculture history treatments in unfertilised plots.

l. 410: "proportional": amend to "relative".

This was changed.

l. 418-420: Why not use directly the Loreau and Hector partitioning then?

As mentioned earlier, the RII metric allows to have a finer picture of plant-plant interactions, both at the individual and also at the community level. This allows to compare different monocultures but also to disentangle, in some cases, net facilitative from competitive effects. Since the core of this study was on changes in plant-plant interactions underlying mixture benefits, we feel that this is the right metric. Furthermore, it has already been used in several facilitation and competition studies, such as Schöb et al. Nat. Ecol. Evol. (2018), Diaz-Sierra et al. Methods Ecol. Evol (2017), Schöb et al. New Phyt (2014), Michalet et al. Fun Ecol. (2013).

l. 431 to 436: Since RIINet and NInC are highly correlated, it is not surprising that you get the same results: you just repeated your analysis with a redundant index. I would either remove or just mentioned that RIInet and NIntC were strongly correlated and thus provided identical results.

We removed this extra index as the information is indeed redundant.

l. 440: as for RII and single plant yield, how were monoculture yields computed here? Was it averaged across replicates of the same species, only considering replicates with a "pure" monoculture history?

This was specified in the methods; it was averaged across replicates of the same species, fertilizing condition and coexistence history.

l. 451-452: Community-Weighted Mean needs to be defined and referenced.

This was added.

l. 456 and 457: "NIntC" should be amended to "RII" and "LER" should be amended to "SE"

Thank you, this was corrected.

Table S17: Which PAR values were used in this analysis of variance? Was it a single date, or averaged across dates?

We took all the dates (indicated in L541) but added day of year as a random factor in the model, to have the response “across date”.

Figure S10: I would move this Figure to the main text or provide an illustration of the experimental set-up there to help the readers visualize what are beds, plots, and how are the different species arranged, notably in mixture plots.

This was moved to the main text along with Figure 1.

Reviewer #2 (Recommendations for the authors):

1. Please give a more balanced discussion of the proposed mechanisms for how co-occurrence can lead to more facilitation. I see how co-occurrence could lead to trait displacement and less niche overlap, so less competition. But what is the facilitation part of this? Please include something in the intro that covers this.

We included a paragraph covering the evolution of facilitation in the introduction (L49-57).

2. Your RII_fac and RII_comp appear to be superficially separated based on whether the net effects of neighbors were positive or negative, is this correct? What would happen if one species in a mixture plot had net positive effects of neighbors and another species had net negative effects of neighbors? Would you calculate the whole plot as just the sum of that? Or how is it calculated in that case?

The distinction between facilitation and competition was dropped as indeed it was redundant and did not bring any more insight than the net RII.

3. Also, how do you separate intraspecific competition and interspecific competition from these metrics? For example, if intraspecific comp > interspecific comp, you would still see a less negative RII_comp at higher diversity. But this isn't necessarily because competition is actually weaker at higher diversity (but instead a shift from one type of competition to another). Similarly, if intraspecific comp > interspecific comp, you could also see these shifts in RII_fac. A stronger RII_fac value could be completely driven by just alleviation of intraspecific competition at higher diversity.

We think that stronger intraspecific vs interspecific competition is an inherent mechanism underlying positive BEF effects. This was actually one of the reasons why we focused on the plant’s eye view in BEF studies and quantified plant-plant interaction metrics here. And we agree with your assessment of changes in RII_fac from monoculture to mixture being potentially due to alleviation of intraspecific competition – but this only if there is also facilitation going on. We would not get RII_fac (i.e. individuals growing bigger in a community than as single plant) when there is no facilitation but only alleviation of intraspecific competition. In any case, we decided to remove the distinction between RII_fac and RII_comp, as outlined above.

4. I'm thus consistently tripped up by the many phrases in the manuscript where you say things like: "shifted towards weaker competition, and in some cases, stronger facilitation" (lines 81-82). All of your RII values are net effects, so how can you conclude that what drives the changes you see is due to competition or facilitation? Why not: "shifted towards weaker competition and/or stronger facilitation (though teasing out the differences is not possible in our given dataset)".

This was changed throughout the manuscript, and we tried to be more precise and careful in our statements.

5. Overall trends I see in RII: most things do worse when growing next to same species neighbors (vs. alone), but the same or better if they grow next to heterospecific neighbors. This gets stronger if they have coexisted for multiple generations. And the positive aspects of it are most strongly affected by coexistence. Your data indicate that this ISN'T due to trait divergence which is super interesting! You also show a strong correlation with selection effects which indicates that a couple of species are driving the patterns. Did you take a look at how selection effects were correlated with RII? Do you know which species are doing that? Have you split into plots with and without legumes? Do legume effects on neighbors get stronger with coexistence history? The only other mechanism I could imagine is more specialization of enemies over time leading to stronger dilution effects? Please discuss these possible mechanisms in some way.

Since all our 4-species mixtures necessarily included a legume, the plots without legumes were only monocultures and 2-species mixtures, and therefore it did not give a representative picture of the experimental design. We can only compare 2-species mixtures if we want to compare with and without legumes, therefore we lose many plots and, as we saw in the other results, planted diversity is often significant. This is notably the case for RII and CE, which are both higher in 4-species mixtures compared to 2-species mixtures and where we expected the most important effects of coexistence history. When only considering the 2-species mixtures and separating with and without legumes, we for instance see a positive effect of coexistence history on RII in fertilized plots with legumes, but when looking at CE, we see a positive effect of coexistence history in fertilized plots without legumes. Disentangling the role of legumes would require having 4-species mixtures without legumes and much more in depth analyses that would fall out of the scope of this paper, which aimed at investigating general patterns across species. Furthermore, since the plots were reshuffled each year to avoid soil legacy effect, dilution effects of pathogens are, in our opinion, unlikely. However, this is highly speculative; further research and analyses are certainly needed to investigate the mechanisms behind these results.

6. The loss of variability in each species when they have coexisted for multiple generations also seems to point towards some kind of restricted gene flow. How much interbreeding was there between plots? Is there something about the environment that you could imagine selecting for these very tall plants? Also, the fact that they are getting more like each other overall, but still having more positive effects on each other is very strange. Need more context about how this might be explainable.

We could not measure precisely how much interbreeding there was between plots; however, since the plots were very close to each other, we believe that there was no limitation to interbreeding between plots. The selection for tall plants has been referenced and observed previously, notably because competition for light is asymmetric (see Lipowsky 2015, Falster 2003), meaning that a taller plant gets a much larger amount of the resource than a shorter one. This puts a high selection pressure on species to evolve plasticity for increased plant height in response to lower light. We were also surprised by the evidence pointing towards character convergence; however, we do see an increase in light interception with a common coexistence history, which points towards improved resource use and increased niche differentiation, at least for light. We therefore believe that the measured traits did not allow us to see the changes in resource use; light interception can be influenced by leaf architecture or vertical leaf distribution, which we did not measure here.

Reviewer #3 (Recommendations for the authors):

Background on main comments / further general comments

The presence of clear effects of the coexistence history on competition together with absence of effects on yield is worrying. Particularly because effects on competition relative to single plants (Figure 2) are statistically very significant while those on Net Effect relative to monocrop are borderline (P = 0.0715). At the same time, the plot of total yield excludes the results for single plants (these should be included).

The results for single plants were included in Figure 3b.

The decrease in competition due to the same coexistence history shown in Figure 2 is relative to single plants. Hence, the comparison is relative. Do we really see a decrease in competition here, or is the single plant getting worse due to the coexistence history? Importantly: It is unclear in Figure 2a whether the two points that make up a pair of points for the same x-axis value are relative to a single plant with the same coexistence history. For example, for x = 1 (monoculture) the two y-axis values are about -0.3 and -0.2. Is the reference single plant the same for these two points the same, so that we can also infer that the difference between both points is 0.1? Or is the reference single plant for the first point (-0.3) a single plant with also a Different co-existence history, and the reference single plant for the second point (-0.2) a single plant with also a Same co-existence history. If the latter is the case, then both points have a different reference, and the apparent difference of 0.1 may very well originate from having two different single plant references, rather than an effect on competition in the monocrop.

See response to main comments, in particular (3).

Related to the previous comment: I cannot reconcile the results for No Fertilizer in Figures 3a and 3b. NE is a measure of overyielding relative to monocrops. Figure 3a NoFertilizer shows that overyielding is greater with Different. However, 3b shows greater total yield with Same. This strongly suggests that the two points making up a pair in Figure 3a do not have the same monocrop as the reference. Hence, then we don't know whether the effect is due to the coexistence history of the 2-mixture or due to the co-existence history of the monocrop used as reference. Things get very confusing because the reference point seems to be shifting all the time (to the best of my understanding).

See response to main comments, in particular (3).

The Introduction refers to "genetic variability needed", but it is not so clear whether (natural) selection happens (or is the focus), or whether interest is in non-genetic transgenerational effects. Whether or not seed selection for the next generation was affected by natural selection should be discussed.

We recognize that the underlying mechanisms remain unclear, as investigating these mechanisms was not the goal of the study. Therefore, we can only speculate regarding the potential mechanisms, and we have now added an extensive paragraph discussing the possibilities in the discussion (L271-289).

Related to the previous comment: natural selection for individual performance in mixture populations is expected to lead to an increase in competition (e.g. Griffing 1967 and later similar work on IGE). This seems to agree with the observation of taller plants containing more water. This could be discussed.

Height indeed generally converges towards taller plants in denser and more diverse communities; this was now more discussed in the discussion. Previous work shows the same trend, due to the asymmetrical character of competition for light (see Lipowsky et al. 2015). However, lower leaf dry matter content indicates less resource-conservative strategies and it has recently been associated with lower parental or ambient competition, so this trait goes in line with our interaction results (see Reich et al. 2014, Puy et al. 2020).

In the statistical analysis at the species level (L464-466) it seems a species effect is included as a random effect? That is surprising. Why?

For the analyses at the species level only, we added species as a random effect, because we wanted to see whether there was a response across all species and were not interested in the response of species per se.

Detailed comments

L21: My impression is that, in Figure 3b Fertilizer, x-axis values 2 and 4, these differences are not significant. IF so, is this statement warranted?

Indeed, for total yield the difference is not significant, but for yield benefits (=overyielding, or net effects), it is in fertilized plots. We clarified this.

L24: Taller with more water seems to be indicative of more competition. This could be discussed (in Discussion).

See response above.

Introduction: please partition the Introduction into several paragraphs. At present, it is hard to read. For example at L44, L50, L61, L69.

This was done.

Can something be said about the selection history of the source material? Have all these varieties been selected for monocrop performance?

These varieties come from standard seed providers used for agricultural purposes, hence they are usually used as monocultures. However, some less “usual” crops, such as camelina or coriander, surely have a shorter breeding history as wheat or oat.

L81: It seems Figure 2b shows differences for facilitation, not (or sometimes) for competition, which contradicts this sentence. Maybe y-axis labels in Figure 2b and c have accidentally been swapped? Anyway, as suggested above, I suggest dropping the distinction between less competition and more facilitation. Hence, I suggest dropping panels b and c.

We removed the partitioning into competition and facilitation components.

For mixtures, the meaning of "same coexistence history" is not fully clear. Is this 2 with a 2 history, or also e.g. 2 with a 4 history, etc. And how often?

This was clarified in the Figure 2 legend.

Figure 2: I propose to explicitly state in the text that this (RII) is a measure relative to single plants.

This was made clearer in Figure 2.

L100: In the context of the current manuscript, the "corresponding monoculture" is not fully clear. Does it also mean: "with the same coexistence history" as the focal mixture?

Yes, this is what it means. This was clarified.

L105-107: Is this sentence warranted given the absence of any indication of statistical significance (P = 0.67)?

This sentence was actually referring to the response of net effect (overyielding), but we moved the sentence earlier to be clearer and removed the commentary on the nonsignificant response of yield.

Figure 3b: For easier interpretation: Could the y-axis be presented on the original (not square root transformed) scale? So can the estimates be back-transformed even though the P-values come from the transformed data?

This was changed.

Figure 4 panel e,f, monocrop (x = 1) seems to suggest that plants become more competitive when grown in the mixture (in the previous 2 generations), which would agree with the theory (Griffing 1967).

Taller plants would indeed suggest more competitive plants. However, the plant interaction index does not indicate whether a plant is competitive or not, but rather whether they experience competition (i.e. whether their growth is more or less hindered by other plants). In that sense, lower LDMC indicates softer leaves, and consequently less resource-conservative strategy. On the contrary, a higher LDMC would increase a plant’s ability to cope with stress. We therefore believe that plants in the same coexistence history might be less “stressed” because they might experience less competition (i.e. this does not mean that they are not more competitive, but that at the community-level they experience less competition from their neighbours).

L456-470 The description of the statistical analysis is very verbal. A model presented as a mathematical equation would be much easier to read, and particularly to find back.

We added the general equation in the method for more clarity.

The Net Biodiversity Effect (L440) could be explained much easier, by first stating that it measures overyielding relative to monocrops, giving a simple equation for that, and only after that splitting this up into a main effect (the CE) and an interaction (the SE).

This was done.

Then the interpretation of the CE and SE should be clarified. (This is very similar to general and specific combining ability, an analogy that may help the reader).

The interpretation was added and clarified in the methods (L509-513).

L457: I may have missed it, but what is LER?

This was a mistake, thank you for pointing this out.

https://doi.org/10.7554/eLife.77577.sa2

Article and author information

Author details

  1. Laura Stefan

    1. Institute of Agricultural Sciences, ETH Zurich, Zurich, Switzerland
    2. Plant Production Systems, Agroscope, Nyon, Switzerland
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Writing - original draft, Writing - review and editing
    For correspondence
    laura.stefan@m4x.org
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0798-9782
  2. Nadine Engbersen

    Institute of Agricultural Sciences, ETH Zurich, Zurich, Switzerland
    Contribution
    Data curation, Validation, Investigation
    Competing interests
    No competing interests declared
  3. Christian Schöb

    1. Institute of Agricultural Sciences, ETH Zurich, Zurich, Switzerland
    2. Área de Biodiversidad y Conservación, Universidad Rey Juan Carlos, Móstoles, Spain
    Contribution
    Conceptualization, Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Investigation, Methodology, Project administration
    Competing interests
    No competing interests declared

Funding

Swiss National Science Foundation (PP00P3_170645)

  • Christian Schöb

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Elisa Pizarro Carbonell, Carlos Barriga Cabanillas, Anja Schmutz, Sandra Gonzalez Sanchez, Lukas Meile, Carlos Federico Ingala, Roman Hüppi, Simon Baumgartner, Benjamin Wilde, Manon Longepierre, Marijn Van de Broek, Leonhard Späth, Inea Lehner, Anna Bugmann, Jianguo Chen, Nicola Haggenmacher and Zita Sartori for their help in the field, and Johan Six for comments on the experimental design. We also thank the Aprisco de Las Corchuelas Field Station and the University of Zurich for the use of their facilities. The study was funded by the Swiss National Science Foundation (PP00P3_170645).

Senior Editor

  1. Meredith C Schuman, University of Zurich, Switzerland

Reviewing Editor

  1. Bernhard Schmid, University of Zurich, Switzerland

Reviewer

  1. Piter Bijma, Wageningen University, Netherlands

Version history

  1. Preprint posted: January 17, 2022 (view preprint)
  2. Received: February 3, 2022
  3. Accepted: August 22, 2022
  4. Accepted Manuscript published: September 13, 2022 (version 1)
  5. Version of Record published: October 11, 2022 (version 2)

Copyright

© 2022, Stefan et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,704
    Page views
  • 507
    Downloads
  • 5
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Laura Stefan
  2. Nadine Engbersen
  3. Christian Schöb
(2022)
Rapid transgenerational adaptation in response to intercropping reduces competition
eLife 11:e77577.
https://doi.org/10.7554/eLife.77577

Further reading

    1. Ecology
    Jingxuan Li, Chunlan Yang ... Zhong Wei
    Research Article Updated

    While bacterial diversity is beneficial for the functioning of rhizosphere microbiomes, multi-species bioinoculants often fail to promote plant growth. One potential reason for this is that competition between different species of inoculated consortia members creates conflicts for their survival and functioning. To circumvent this, we used transposon insertion mutagenesis to increase the functional diversity within Bacillus amyloliquefaciens bacterial species and tested if we could improve plant growth promotion by assembling consortia of highly clonal but phenotypically dissimilar mutants. While most insertion mutations were harmful, some significantly improved B. amyloliquefaciens plant growth promotion traits relative to the wild-type strain. Eight phenotypically distinct mutants were selected to test if their functioning could be improved by applying them as multifunctional consortia. We found that B. amyloliquefaciens consortium richness correlated positively with plant root colonization and protection from Ralstonia solanacearum phytopathogenic bacterium. Crucially, 8-mutant consortium consisting of phenotypically dissimilar mutants performed better than randomly assembled 8-mutant consortia, suggesting that improvements were likely driven by consortia multifunctionality instead of consortia richness. Together, our results suggest that increasing intra-species phenotypic diversity could be an effective way to improve probiotic consortium functioning and plant growth promotion in agricultural systems.

    1. Computational and Systems Biology
    2. Ecology
    Vanessa Rossetto Marcelino
    Insight

    High proportions of gut bacteria that produce their own food can be an indicator for poor gut health.