Discussion
Most biodiversity-ecosystem-functioning studies address the effect of diversity within a trophic level, such as plants, on functions such as primary productivity (e.g., Isbell et al. 2015). We have introduced an integrated model of producer species richness and resource-use complementarity that yields positive diversity-productivity relationships consistent with patterns found in experimental (Isbell et al. 2015) and natural communities (Duffy et al. 2017; van der Plas 2019). When introducing resource-use complementarity by creating dissimilarities in how producer species access resource-pool compartments, monocultures generally become less productive than species mixtures as they utilize a smaller proportion of the total resource pool (Tilman et al. 1997; Loreau 2001). By increasing this resource-use dissimilarity across primary producers, we could enhance such net diversity effects through bottom-up mechanisms. However, trophic interactions can affect diversity-productivity relationships in producer communities in similar ways. We find that adding animal communities embedded in food-webs of multi-trophic interactions strengthened the net diversity effect on primary production. This similarly results from a reduction in monoculture productivity. Such top-down reductions due to herbivory can be compensated in more species-rich producer communities (e.g., Jactel et al. 2021), where trophic interactions shape the composition and interactions among producer species (e.g., Naeem et al. 1994; Thébault & Loreau 2003; Brose 2008; Zhao et al. 2019). By addressing the interplay of resource-use complementarity and multi-trophic interactions, our study synthesizes bottom-up and top-down drivers of BEF relationships. While both create complementarity to create positive net diversity effects, our model suggests that diversity across trophic levels can additionally change selection mechanisms and thereby producer community composition. Specifically, a dominance of highly productive monoculture species (positive selection effects) at low animal richness shifts to a community that also includes the less productive monoculture species (negative selection effects) as animal richness increases. An increased complementarity among producer species at high animal richness therefore allows more species to coexist. In consequence, complementarity effects increase with animal richness and overcompensate the negative selection effects. To which degree multi-trophic mechanisms can increase net diversity effects is determined by resource-use dissimilarity. At high levels of resource-use dissimilarity, multi-trophic interactions show only weak effects, whereas lower levels of resource-use dissimilarity allow top-down mechanisms to enhance net diversity effects more. Hence, our results suggest that multi-trophic interactions and resource-use complementarity among producers shape diversity-productivity relationships interactively. This finding implies that processes across trophic levels are strongly interwoven, which renders the integration of multi-trophic mechanisms in the analysis of diversity effects in complex communities highly important.
In simple communities without animals, we tested for the consequences of resource-use dissimilarities between producer species. It promotes coexistence, creates complementarity and consequently positive net diversity effects, thereby confirming findings of earlier theoretical studies (Vandermeer 1981; Tilman 1982; Loreau 2004). Further, resource-use dissimilarity can create a range of different shapes of diversity-productivity relationships (e.g., exponential, sigmoidal, or saturating on a log2-scale of producer richness), as found in experimental and natural studies (Balvanera et al. 2006; Duffy et al. 2017). Our simulated producer communities show how at low levels of resource-use dissimilarity (i.e., substantial overlap in the resource compartments used by different producer species), saturating diversity-productivity relationships emerge where only a few species are necessary to maximize primary production. On the contrary, at high levels of resource-use dissimilarity (i.e., producer species mostly exploit different compartments of the total resource pool), the majority of producer species is necessary to maximize productivity. This highlights how an increasing resource-use dissimilarity not only increases complementarity between species but also reduces their functional redundancy in resource-use (Loreau 2004). When producer species are lost, communities with a low functional redundancy are more prone to become less productive and thus show weaker net diversity effects. Resource-use dissimilarity that enhances complementarity and thus drives net diversity effects in producer communities can therefore also be responsible for weakening such effects as species are lost.
In ecosystems with animal species, our results confirm that multi-trophic interactions can create positive net diversity effects even without any resource-use dissimilarity amongst producers (Thébault & Loreau 2003). As long as producer species are not limited to access distinct resource compartments, multi-trophic interactions consistently enhance net diversity effects. Whether herbivores are predominantly specialists or generalists determines if such effects are strong or negligible, respectively (Thébault & Loreau 2003; Jactel et al. 2021). In our simulations, generalism is constraint by predator-prey body-mass ratios known from aquatic and belowground ecosystems (Schneider et al. 2016). Regardless, they are sufficient to reproduce the decreasing influence of herbivores on primary production as producer diversity increases that is common to forests, grasslands, and agroecosystems alike (Barnes et al. 2020; Wan et al. 2020; Jactel et al. 2021). We find that animals largely influence net diversity effects by reducing primary production in monocultures (Barry et al. 2018). In mixtures, reductions in productivity can be compensated by producers that access the same resource compartments (i.e., functional redundancy in resource-use; Naeem 1998). The potential of compensatory effects therefore scales with resource-use dissimilarity and producer species richness. When animals are present, a lack of compensation inevitably leads to a reduced primary production in mixture. Even though this may weaken the positive impact of multi-trophic interactions on net diversity effects, our results suggest that it is rarely the case.
Our findings show that enhanced net diversity effects in multi-trophic ecosystems can largely be attributed to complementarity mechanisms (Thébault & Loreau 2003; Barry et al. 2018), which reduce interspecific competition among producer species. Apart from competing for resources, animals can shift the competitive interaction to being additionally mediated by consumers and their trophic interactions (Holt 1977; Loreau 2010). For example, multi-trophic interactions reduce competition between producer species by limiting productivity and thereby inhibiting the dominance of single species (Brose 2008). As a result, producer species can coexist even if their resource-niches overlap entirely (Brose 2008; Loreau 2010). Similar to an increased vertical diversity (Wang & Brose 2018), we found that an increased animal richness can enable more producer species to coexist, which is indicative for the higher complementarity among them. In addition, a complementarity in herbivorous feeding links sorts producer species into different trophic groups common to our simulated and natural food-webs alike (Gauzens et al. 2015; Schneider et al. 2016). This top-down aspect of trophic complementarity can enhance net diversity effects similar to the bottom-up complementarity of resource-use (Thébault & Loreau 2003; Poisot et al. 2013). Despite the increased complementarity, a limited resource availability caps primary productivity in multi-trophic ecosystems to not exceed primary production in ecosystems without animals.
While multi-trophic interactions determine net diversity effects in producer communities largely through complementarity mechanisms, selection effects draw a less conclusive picture. Differences in the functional expression of producer species in monoculture are a fundamental requirement for non-neutral selection effects (Loreau & Hector 2001). In our simulations, the maximum productivity of all producer species is largely determined by their access to resource compartments, which is the same for all co-occurring species. Their functional expression in monoculture (i.e., primary production without competition) is therefore equal in the absence of animal consumers. Hence, we do not find selection effects in simple producer communities. This model simplification disregards processes such as associations between competitive ability and productivity that can determine selection in producer communities (Tilman et al. 1997), but it enables us to isolate multi-trophic selection mechanisms. Specifically, we find that the productivity of large producer species is less susceptible to herbivory. The low mass-specific metabolic rates of large species may play an important role in minimizing losses to herbivory (Brown et al. 2004; Schneider et al. 2016). The ability of large species to better cope with herbivory also increases their chances to persist in mixtures (Schneider et al. 2016; Wang & Brose 2018), which should lead to positive selection effects. However, we find negative selection effects at high animal richness and resource-use dissimilarity, which both tend to enhance complementarity. Because complementarity mechanisms reduce interspecific competition, the small producer species that are excluded when complementarity is low can persist in mixtures. Once in the mixture, a disproportionally strong effect of herbivores on strong competitors (Brose 2008) elevates the productivity of the otherwise excluded, small and less productive producer species. Even though this does not imply their dominance, it alters community composition enough to turn selection effects negative in response to an increasing complementarity. This interdependence of complementarity and selection effects in multi-trophic ecosystems becomes apparent in our findings of their inverse relationship to realized producer richness. However, not all complementarity mechanisms have to be linked to selection mechanisms that influence net diversity effects (e.g., resource-use dissimilarity as defined in this study). Identifying causes of selection can therefore serve as an important tool to disentangle drivers of diversity effects.
Despite the evidence that multi-trophic interactions (Thébault & Loreau 2003) and resource-use complementarity (Tilman et al. 1997) can create positive net diversity effects on primary production independently, how they interact has remained speculative (Tilman et al. 2014; Barry et al. 2018). We find that both mechanisms create positive net diversity effects by lowering primary production in monoculture. Hence, an already low monoculture primary production at high resource-use dissimilarity, which leads to high net diversity effects on its own, cannot be reduced much further by animals before driving the single producer species and thus the entire food-web extinct. A high resource-use dissimilarity therefore limits the ability of multi-trophic interactions to enhance net diversity effects. Similarly, it promotes the coexistence of producer species by reducing competition but simultaneously limits the ability of multi-trophic mechanisms to alter competition in favor of a more diverse producer community (Brose 2008). In both cases, bottom-up forces fundamentally limit the strength of top-down mechanisms to improve either net diversity effects or species coexistence. Plastic responses in resource-use to changes in producer diversity (Von Felten et al. 2009; Mueller et al. 2013), consumer diversity, or vertical diversity (Zhao et al. 2019) might affect such limitations but should not change our conclusion that multi-trophic interactions become especially important when the resource-use dissimilarity of producers is low.
The interactive effect of resource-use complementarity and multi-trophic interactions creates positive net diversity effects that generally exceed their independent effects. Both mechanisms jointly support diverse communities of complementary producer species. However, multi-trophic interactions determine the community composition of the producer species depending on the animal diversity of the multi-trophic ecosystem, whereas selection in simple producer communities is solely driven by resource competition. Hence, different mechanisms can similarly create complementary communities, but the associated selection mechanisms may differ. To identify drivers of positive diversity effects common to natural ecosystems, instead of focusing on identifying causes of complementarity it may therefore be more expedient to identify causes of selection and understand how they relate to complementarity mechanisms. In bridging the gap between food-web and BEF theory, our novel simulation-framework can guide such efforts as it integrates effects of diversity within and across trophic levels on functions of complex, multi-trophic ecosystems. Its results highlight the interplay between bottom-up and top-down forces in these ecosystems, emphasizing the need to adopt a multi-trophic view on BEF relationships.