Support for and nuances underlying the stress-gradient hypothesis and the role of species interactions
We found strong support for our hypothesis that biotic interactions would become increasingly competitive in more productive environments (Table S4; Fig. 3). Although this pattern was observed for all four species (i.e., greater difference between blue and red lines in productive areas in Fig. 1A-D), there were also interesting nuances. For one species, growth rates increased with productivity when neighbors were present (Festuca : 𝛽1 = 0.167 ± 0.12; Table S2, red data in Fig. 1D), whereas for two, population growth rates decreased (Plantago : 𝛽1 = -0.10 ± 0.12, Bromus : 𝛽1 = -1.00 ± 0.18; Table S2, red data in Fig. 1B,C). As a result, each species’ fundamental and realized niches differed (see ‘Neighbor treatments’ term for each species in Table S3). Competition appeared to be absent for all species at the unproductive end of the gradient, with weak evidence of facilitation (i.e., λ without neighbors > λwith neighbors) only in a few specific cases. Plantago andFestuca patterns showed evidence of facilitation for seed production and early germinate survival (i.e., zero-inflated component of the model; Fig. S6B, D), while Bromus showed weak evidence of facilitation in the fecundity (i.e., conditional) component of λ ; Fig. S6 F).
Taken together, at a community level, productivity had opposing effects on the number of species that could persist depending on the presence or absence of neighbors: in the most productive locations (i.e., vegetation index of 2), persistence without neighbors was 1.58 times greater than persistence with neighbors (P < 0.001; Fig. 3; Table S5). Notably, these relationships were linear at the community level despite a diversity of non-linear relationships observed for individual species (Table S1; Fig. 2A-D).