A key challenge in ecology is to understand how changes in primary productivity alter spatial variation in community composition (β-diversity). Yet β-diversity often shows variable responses to productivity. Such variability could reflect shifts in the relative importance of assembly processes—dispersal, ecological drift, or niche-selection. However, shifts in these processes across productivity gradients are generally inferred from species β-diversity, where functional differences among species are not explicitly considered. Here, we tested how the relative importance of dispersal, ecological drift, and niche-selection for structuring herbaceous plant communities changes with productivity in a temperate forest understory. First, we used null models to determine the extent to which patterns of species and functional β-diversity reflect shifts in the importance of ecological drift and non-random assembly mechanisms. Second, we used spatially-explicit variation partitioning to determine whether changes in the non-random patterns with productivity reflect dispersal limitation or differences in the strength of species-environment and trait-environment relationships (light & soil resource availability).
In contrast to the hypothesis that productivity increases stochastic community assembly, we show that niche-selection acting on species traits becomes increasingly important in more productive environments. Observed patterns of species and functional β-diversity across the productivity gradient were correlated—β-diversity was highest in low-productivity sites and declined with increasing productivity. However, null-model analyses revealed that higher species β-diversity in low-productivity sites was driven by small community size and high functional similarity; two conditions that may increase the relative importance of ecological drift. After accounting for differences in community size and species pools, β-diversity increased with productivity, suggesting a stronger influence of non-random processes in more productive environments. The niche-selection mechanisms underpinning these non-random β-diversity patterns emerged primarily when considering trait sorting along environmental resource gradients. This corroborates other studies that suggest species-centric approaches may inflate the importance of stochastic processes. However, the importance of functional differences among species appears to increase with community size. Therefore, predictions for how changes in primary productivity will alter biodiversity may be contingent upon the scale-dependent factors that alter local community size and the functional similarity of species.