Discussion

In our meta-food-web approach, we show that spatial processes related to plant resource exploitation and animal movement strongly affect plant diversity-productivity relationships. Positive relationships arise only when plant resource access overlaps spatially (‘spatial resource overlap’) at the cost of exploitative competition. Herbivore movement introduces apparent competition between plants, which can reduce plant productivity in diverse communities, yielding negative diversity-productivity relationships. However, a realistic body mass scaling of animal home range sizes moves apparent competition motifs up the food chain. The reduced plant competition together with the spatial integration of sub-food webs through the movement of top predators lead to the most positive effects of plant diversity on productivity, suggesting animal movement as a crucial driver of plant diversity-productivity relationships.

Plant-resource interactions

A spatial resource overlap between neighbouring plants has two important implications for plants. While it allows each plant to access a larger share of resources available in the ecosystem, it also forces them to engage in exploitative competition. In diverse plant communities, however, plants will differ in their resource requirements and thus in their competitive ability. Different resource requirements are usually accompanied by low competition (Tilman et al. 1997) and suggest a stoichiometric complementarity between neighbouring plants (Gonzálezet al. 2017) that is likely to have positive impacts on plant productivity. We find this mirrored in our results, where, in the absence of animals, even a weak spatial resource overlap is enough to maximize productivity in diverse plant communities of species with different resource requirements. Accessing small amounts of otherwise inaccessible resources can therefore already suffice to lift resource limitations, leading to positive diversity-productivity relationships due to a stoichiometric complementarity.
Differences in competitive abilities across pairs of plant species are rarely associated with such performance enhancements. Instead, they should lead to local extinctions of the weaker competitor (Tilman 1982), which we find mirrored in a loss of plants when the spatial resource overlap increases. Surprisingly, this does not come at the cost of a reduced productivity. Instead, plants with a competitive advantage, either due to a higher efficiency in resource acquisition (i.e. exploitative competition) or favourable multi-trophic interactions (i.e. apparent competition), can maximize their resource uptake and thereby increase their biomass (Wang & Brose 2018). This has the positive side effect of reducing energy requirements for metabolic processes relative to their mass (Enquist et al. 1998), contributing to a more energy efficient plant community. Our findings suggest that this is enough to counterbalance the loss of plants as well as the associated diversity loss.
When plants have a spatial resource overlap, diversified resource requirements create complementarity whereas selection (sensuLoreau 2000) due to competitive differences shift the community to be more energy efficient (Wang & Brose 2018), leading to an optimized resource uptake in both cases. Consistent with our hypothesis (H1), this increases productivity and leads to positive diversity-productivity relationships. The consistency of those relationships paired with the shifts in plant community composition additionally implies that the contribution of complementarity and selection processes to maximizing productivity varies depending on the strength of the competitive interaction between plants.

Plant-animal interactions 

When embedded in complex food webs, the response of plant productivity to varying plant biodiversity is rooted in food-web topology. As the number of plant species increases, there is an increasing number of apparent competition motifs in which two plant species are coupled by a shared herbivorous consumer population. In this motif, the plant species with a higher resource acquisition efficiency achieves a higher biomass density, leading to higher herbivore densities, which in turn has negative top-down effects on other plant species with lower resource acquisition efficiencies (Holt 1977). Accordingly, our simulations of spatially non-nested food webs have shown that as plant species richness increases, plant productivity decreases, which is, consistent with our hypothesis (H2), reflected in negative diversity-productivity relationships. However, when compared to scenarios without animals, the added apparent competition does not foster competitive exclusion. Instead, it seems to buffer some of the negative effects of an increased exploitative competition (i.e. increased spatial resource overlap) as more plant individuals and species are able to coexist when embedded in a food web (see also Brose 2008; Albert et al. 2022).
While the high levels of maintained plant individuals and species are similar between spatially nested and non-nested food webs, the effects of apparent competition on productivity are not. Specifically, our simulations of spatial non-nested food webs assume a well-mixed system without any differences in local biomass densities of animal species. Ignoring such differences results in herbivore populations that can feed simultaneously on different plants regardless of their location. In nature, however, animal communities have a complex spatial organization (e.g. Gonçalves-Souza et al. 2015). While almost all animal species move between resource patches, larger species travel longer distances and have larger habitats (Tucker et al. 2014; Tamburello et al. 2015; Hirt et al. 2021). As a result, meta-food webs have a structure in which smaller species from local food webs are spatially integrated within the home ranges of larger species. Apparent competition between plants in spatially nested food webs is therefore spatially constrained depending on the home range size of the herbivore. In addition, an increased amount of apparent competition motifs between sub-populations of herbivores reduces their top-down control on plants. Hence, instead of the negative plant diversity-productivity relationships found in spatially non-nested food webs, relationships in spatially nested food webs are the most positive, peaking at levels similar to plant communities without animals.
Apart from the positive effects of an altered spatial topology (i.e. effects of apparent competition) on diversity-productivity relationships in spatially nested food webs, the spatial integration of sub-food webs has additional dynamic benefits. In particular, biomass overshooting and unstable dynamics leading to local extinctions are buffered in spatially nested food webs by large top predators that stabilize biomass minima of populations in the local food webs away from critically low values (McCann et al. 2005). This is reflected in the relatively stable plant diversity of spatially nested food webs despite differences in the spatial resource overlap of plants. Consistent with our hypothesis (H3), we thus conclude that a spatial integration of sub-food webs associated with spatially nested food webs has positive effects on plant diversity-productivity relationships. The clear dynamic and topological differences between spatially nested and non-nested food webs, which may be negligible for biodiversity maintenance, can therefore have strong implications for plant productivity, leading to vastly different plant diversity-productivity relationships.

BEF: from multi-trophic to meta-food webs

BEF research has evolved from focusing on single functional groups (e.g. plant communities) to the complex multi-trophic structure of natural communities (e.g. Schuldt et al. 2019; Barnes et al. 2020; Albert et al. 2022). This development has shown that multi-trophic interactions can facilitate plant coexistence and thereby increase productivity. In our study, we extended this development by applying meta-ecosystem (i.e. plant-resource exploitation bridges between local habitats) and meta-food web approaches (i.e. spatially-explicit structure of the food webs). Some of our results on the effects of multi-trophic interactions differ significantly from previous conclusions. While prior studies reported generally positive effects of multi-trophic interactions on plant coexistence and diversity-productivity relationships (Thébault & Loreau 2003; Brose 2008; Albert et al. 2022), we found that under the assumption of spatially segregated plants (i.e. each plant inhabits its own local habitat) this is not necessarily the case. Spatially non-nested animal communities paired with spatially segregated plants instead result in negative relationships, which finds an explanation in the systematic increase in apparent competition motifs. In contrast, the spatially nested structure of animal communities yields strongly positive diversity-productivity relationships due to the positive effects of an apparent competition shift up the food chain (i.e. from between plants to between herbivores) and the spatial integration of sub-food webs by top predators.
By relaxing the classic assumption of well-mixed systems (e.g. Schneideret al. 2016; Albert et al. 2022), we gained accuracy in the description of the processes that drive ecosystems and their functioning. The assumption of a well-mixed system is also at the core of BEF research, as it usually compares the functioning of entire communities of varying diversity. While this helped to identify complementarity mechanisms as the main driver of positive BEF relationships, it remains difficult to identify their concrete causes (Barry et al. 2019), which may be related to focusing on the wrong spatial scale. Indeed, competition and the associated BEF processes (i.e. complementarity and selection; Loreau 2000) act between a few organisms and are thus spatially constrained. Our work demonstrates that a multi-trophic investigation of spatially-explicit plant-resource interactions additionally requires a spatially-explicit consideration of the entire food web. Moreover, our simulations show that the sign and strength of diversity-productivity relationships depends on the joint effects of animal movement and spatial resource overlap of plants. This renders the spatial organisation of multi-trophic communities, which can vary across landscapes, an important but often neglected aspect that can help to explain the variation observed in empirical BEF relationships (Cardinale et al. 2007). Overall, our findings on diversity-productivity relationships clearly demonstrate the importance of spatial community structure and animal movement in driving BEF relationships in meta-food webs.

Future directions

The development of accounting for spatial processes in BEF relationships can be progressed in multiple ways. We have advanced this field in one dimension by synthesizing spatially-explicit processes related to animal foraging movement with spatially-explicit plant-resource exploitation. Our model is flexible to also include other aspects of community structure across spatial scales, including (1) local factors and species traits influencing exploratory movement during foraging (Hirt et al. 2017), (2) neighbouring habitats coupled by lateral nutrient flows in meta-ecosystems (Loreau et al. 2003; Gounand et al.2018), (3) meso-scale landscape structures in community assembly models (Bannar‐Martin et al. 2018; Saravia et al. 2022), including plant and animal dispersal (Ryser et al. 2021), and (4) biogeographic differences between species pools (e.g. of plants; Sabatini et al. 2022). In this vein, merging our spatially-explicit meta-food web approach with food web assembly models (Bauer et al. 2022; Saravia et al. 2022) offer a particularly exciting avenue of future research as it allows to understand how local spatial processes scale to the diversity and ecosystem functioning patterns observed at larger spatial scales.

Conclusion

Despite its variability, the positive effects of diversity on productivity in plant communities are a widely recognized pattern that is consistent across ecosystems (Cardinale et al. 2007). To date, the most prominent among the proposed mechanisms driving these patterns and their variability is complementarity in the resource-use of plants (Barry et al. 2019), which has more recently been supplemented by multi-trophic complementarity (Poisot et al. 2013; Albertet al. 2022). To better understand their differences, we explicitly modelled the different spatial scales at which both mechanisms operate in a simulated biodiversity experiment. We could show that a spatial overlap in resource access between neighbouring plants is a fundamental requirement for positive plant diversity-productivity relationships, highlighting the tight association of exploitative competition with resource-use complementarity and plant compositional shifts due to selection. The realistic, spatially-explicit representation of meta-food webs that integrate nested local sub-food webs stabilizes plant coexistence and yields the strongest diversity-productivity relationships we observe. Our modelling framework can serve as a foundation to further enhance our mechanistic understanding of multi-trophic processes in driving plant diversity-productivity relationships. It provides a novel approach to managing biodiversity while explicitly accounting for the spatial processes that underpin the ecosystem functions that are the basis of our human society. Advancing in this direction is therefore crucial for guiding conservation efforts to maintain biodiversity and the functioning of ecosystems.