Results

Plant diversity-productivity relationships

To investigate the potential drivers behind plant diversity-productivity relationships, we compare the effects of food-web and resource-use scenarios (see Fig. 1) on productivity at both ends of the plant diversity gradient. In monocultures without animals, we find that a spatial overlap in plant resource access (‘spatial resource overlap’) has no effect on productivity (Fig. 2, plant species richness of one; Fig. S1A; green points). Instead, differences occur across the different food-web scenarios. Specifically, monocultures without animals are the most productive, closely followed by those embedded in spatially non-nested food webs (dark blue points). In spatially nested food webs, plant productivity of monocultures is the lowest on average but shows the largest variation with a weakly positive response to an increased spatial resource overlap (light blue points). We rarely found unviable monocultures. The few examples we recorded were spread across all resource-use scenarios and more common in spatially nested (93/1600) than in spatially non-nested food webs (30/1600), and never occurred in communities without animals. When focusing on monoculture productivity, an interaction with neighbouring plants through a spatial resource overlap therefore emerges as having little effect, rendering differences in food web architecture as the main driver.
Our analyses reveal some striking effects of having a spatial resource overlap on the diversity-productivity relationships in plant communities without animals (Fig. 2, green lines). Without a spatial resource overlap we find neutral relationships between productivity and species richness (Fig. 2A). However, as soon as plants are able to access resources of the neighbouring patches (i.e. with spatial resource overlap), we find positive effects of plant diversity on productivity that are similar across resource-use scenarios (Fig. 2B-E; Fig. S1B). Taken together, these results suggest that a positive response of productivity to species richness in plant communities without animals requires a spatial resource overlap, but already small amounts of spatial resource overlap (i.e. Fig. 2B) suffice to saturate these relationships.
In spatially non-nested food webs, plant communities show a strong decrease in productivity with increasing richness in most resource-use scenarios (Fig. 2A-D; Fig. S1B; dark blue lines). Diversity-productivity relationships are most negative when spatial resource overlap is smallest (Fig. 2B). Across the gradient of resource-use scenarios, plant monoculture productivity is constant (Fig. S1A), while it increases considerably at higher plant species richness (Fig. 2B-E; Fig. S1B). This culminates in neutral diversity-productivity relationships when spatial resource overlap is maximized (Fig. 2E). Thus, in communities with spatially non-nested food webs, a strong spatial resource overlap with neighbouring plants has a positive effect on plant diversity-productivity relationships.
In contrast, plant communities in spatially nested food webs display positive diversity-productivity relationships in the majority of cases (Fig. 2B-E, light blue line). We only find negative effects of plant diversity on productivity when there is no spatial resource overlap (Fig. 2A). However, productivity at both ends of the diversity gradient displays large amounts of variation. As soon as plants have access to resources of neighbouring patches (i.e. with spatial resource overlap), productivity increases with diversity, reaching values with little variation that are similar to those in plant communities without animals (Fig. 2B-E). Together with having the lowest average productivity in plant monocultures compared to all other food web scenarios (Fig. S1A), this makes plant communities in spatially nested food webs exhibit the most positive diversity-productivity relationships (Fig. 2B-E).

Plant community composition

Our prior results show that differences in the plant diversity-productivity relationships are mainly driven by varying productivity at the highest plant diversity levels (Fig. 2). To better understand these differences between food-web and resource-use scenarios, we investigated how plant community composition differs between scenarios at the highest plant diversity level of 16 species. Without a spatial resource overlap (i.e. spatial resource overlap at 0), realized species richness, realized plant density, and Shannon diversity display the highest values within each food web scenario considered (Fig. 3). In communities without animals, the values are at their absolute maximum (Fig. 3, green line). In spatially non-nested food webs, the plant communities show a tendency towards lower values of realized richness and density, and Shannon diversity is clearly lower, indicating an increased heterogeneity in the plant community (Fig. 3, light blue line). For spatially nested food webs, plant communities display a slightly reduced plant species richness and density and have the lowest Shannon diversity (Fig. 3, dark blue line). Thus, spatially nested food webs support the least diverse plant communities when there is no spatial resource overlap between neighbouring plants.
The compositional response of plant communities without animals to increasing the spatial resource overlap between neighbouring plants stands out as it displays a delayed but harsh drop for all three compositional variables (Fig. 3, green lines). This leads to plant communities that lose almost half of their plant individuals when spatial resource overlap is highest (Fig. 3B), and includes the extinction of slightly more than three species on average (Fig. 3A). Since Shannon diversity decreases more than species richness (Fig. 3A&C), an increased spatial resource overlap increases the heterogeneity in the plant communities without animals. Taken together, the effects of increasing the spatial resource overlap are most severe for plant communities without animals.
Plant communities in spatially non-nested food webs follow very similar patterns compared to communities without animals (Fig.3, dark blue lines). However, the negative effects of increasing the spatial resource overlap are less pronounced for plant richness and density, culminating in about a quarter of plants and only about two species lost when the spatial resource overlap was highest (Fig.3A&B, dark blue lines). Shannon diversity was generally lower than in communities without animals, reaching the lowest values at maximum spatial resource overlap compared to all other scenarios (Fig. 3C, dark blue line). When spatial resource overlap is high, spatially non-nested food webs are therefore enhancing differences between plant species more than any other scenario.
Compared to the other food web scenarios, plant community composition in spatially nested food webs show the weakest response to changes in spatial resource overlap. Especially realized plant species richness, which displays an average loss of only one species, was independent from spatial resource overlap (Fig.3A, light blue line). Similar to spatially non-nested food webs, only about a quarter of plants are lost when spatial resource overlap is highest (Fig.3B, light blue line). Shannon diversity again decreases with increasing spatial resource overlap but ends up stabilizing over the last two steps of the spatial resource overlap gradient (Fig.3C, light blue line). Overall, these findings suggest that spatial resource overlap between neighbouring plants matters the least in spatially nested food webs.