Testing community assembly predictions for nominal and continuous plant traits in species-rich grasslands

We investigated patterns of trait-based community assembly in grasslands sampled along a moisture gradient (216 species) using three continuous species-level traits (maximum species height, minimum species height, seed mass), as well as seven nominal traits (flowering phenology, fruit type, pollen vector, clonal growth organs – CGOs, branching type, leaf distribution, lateral spread), some of which are unusual in that a species may possess more than one state per trait simultaneously (e.g. CGO). Additionally, this study tests whether patterns of assembly vary with plot biomass and moisture using both presence-absence and abundance-weighted analyses (two plot sizes: 25 × 25 cm, 75 × 75 cm). Plant species in these grasslands were randomly organized with respect to both maximum and minimum species height; however, coexisting plant species had a significantly smaller mean seed mass than expected by chance, and were significantly convergent in seed mass, consistent with observations from one previous study, and with theory related to environmental or competitive filtering. These patterns were consistent across plot sizes, and were similar whether analyses were abundance-weighted or not, although partitioned analyses show that these patterns are most pronounced in wet grasslands. Grasslands were non-randomly assembled with respect to five of our ten traits including seed mass, fruit type, CGO, branching type and leaf distribution. Among these, only patterns of branching type clearly conformed to the predictions from Limiting Similarity Theory, suggesting that variation in this trait may contribute to species coexistence in these grasslands. In two cases (fruit type, branching type), results differed in significance depending on whether analysis used presence-absence or abundance data; incorporating abundance may be more relevant, however, cover-based abundance measures in small plots can bias trait representation in favour of size over ramet number. In general, patterns were consistent across the two plot sizes. For four traits (seed mass, flowering phenology, leaf distribution and lateral spread) plot-level effect sizes of our tests varied significantly with plot-level biomass production; the slope of these relationships was positive for seed mass, but negative for flowering phenology, leaf distribution and lateral spread. For seven of ten traits, plot-level effect sizes varied significantly with plot-level soil moisture, with positive regression slopes for some traits (flowering phenology, leaf distribution), and negative slopes for others (minimum height, seed mass, fruit type, pollen vector, CGO). These results indicate that community assembly rules related to different functional traits can be responsive to variation in either biomass or soil moisture, or both.