Species' traits do not converge on optimum values in preferred habitats.

Plant trait expression is shaped by filters, which can alter trait means and variances, theoretically driving species toward an "optimum" trait value for a set of environmental conditions. Recent research has highlighted the ubiquity of intraspecific variation in functional traits, which can cause plants to diverge from a hypothesized "optimum". We examined whether species occurring in "core" habitats (where they occur frequently, abundantly, and consistently) express traits that are nearer to "optimum", as captured by the community-weighted mean (CWM). We also asked whether trait variance showed signs of environmental filtering. We used cluster analysis to group plots based on environmental factors along a wet-to-dry ecotone. We used indicator species analysis to identify species with strong associations within each cluster. Trait means and variances were compared, and evidence of variance filtering was tested using a null-model approach. Trait means and trait variances respond to local-scale environmental filtering and species in core habitats were not necessarily nearer to the CWM than in other habitats. Intraspecific trait variability shows a strong signal of filtering, as variability was reduced for nearly all species and all traits compared to estimates of variability generated in the absence of environmental filtering. Our results provide strong evidence that species traits are not necessarily near "optimum" trait values in core habitats, and that trait distributions within species are strongly shaped by the environment. Future analyses should account for this divergence when calculating metrics of functional diversity, and extrapolating to ecosystem function.