Understanding community assembly patterns and the underlying mechanistic processes has been a main focus of ecology. Progress in the field includes identifying neutral versus non-random assembly patterns and comparing patterns along commonly used environmental gradients. The stress-dominance hypothesis proposes that in stressed locations abiotic pressures (e.g., environmental filtering) dominate, creating trait convergence, while in more productive locations biotic pressures (e.g., competition) dominate, creating trait divergence, and moderately stressed locations display random trait dispersion due to neutral assembly. This study investigates to what extent will two different vegetation types (forest and wetland/grassland) display similar trait distribution patterns at two spatial scales (patch/relevé and local quadrat) and two species pools (regional and local) across a soil moisture gradient. Plant species were sampled along a soil moisture gradient with intraspecific trait variation being collected as well. Four common traits were measured: specific leaf area (SLA), leaf dry matter content (LDMC), vegetative height, and leaf area. Trait distributions were calculated using nine different dispersion measures to assess ecological selection. Two null models were created to compare the effect of the species pool (regional and local) on the observed communities’ trait distributions to random assembly by shuffling species traits independently of the fixed species abundances.
Results/Conclusions
Our results support the stress-dominance hypothesis in both the forest and the wetland/grassland (i.e., standardized effect sizes were positively correlated with soil moisture). Directional ecological selection was also seen in both the forest and the grassland, but was independent of the soil moisture gradient. The unweighted ranges of the traits increased with soil moisture in the two vegetation types, however, in the case of LDMC, examining trait range in the absence of trait maximum and minimum may lead to overlooked details. Standardized effect sizes based on the regional species pools were more strongly influenced by the soil moisture gradient than those based on the local species pools. The patch and quadrat spatial scales displayed equivalent results with only minor differences, indicating that in our systems, plant community assembly is not strongly influenced by sampling scale. Overall, our study supports the idea that different assembly patterns may occur at opposite ends of stress gradients and this context-dependency should be considered when attempting to predict community composition in a changing environment.