Putting space into trait ecology: Trait, environment and biodiversity relationships at multiple spatial scales.

Ecological processes such as environmental filtering and biotic interactions that shape species' traits and community diversity often vary with geographic distance, potentially generating spatial structures in trait variation, covariation and biodiversity data. Understanding spatial structures of trait, environment and biodiversity, or the spatial link between those factors, is fundamental to identifying spatially explicit assembly processes or biodiversity distributions in spatially heterogeneous landscapes but remains unclear.To address the issue, we gathered individual‐level leaf and diameter traits data paired with environmental data from a 4.8 ha subtropical Chinese forest and divided the forest into 25, 100, 400 and 1936 m2 grids representing contrasting spatial grains. Using Moran's correlograms, we quantified the spatial structures of trait variation and covariation, environmental conditions and biodiversity. We assessed the links between those variables using path analyses.Most variables were spatially positively autocorrelated. However, trait mean was more autocorrelated than trait variation or covariation, and intraspecific trait variation was more autocorrelated than interspecific variation. Autocorrelations in those community properties were generally weak at the large grain.Path analyses indicated positive associations between interspecific trait variation and species diversity at a very small to medium scale and a positive association between intraspecific variation and small‐scale functional diversity. Trait covariation constrained biodiversity, and multi‐trait means were negatively linked to very small‐ to medium‐scale species diversity but positively to medium‐ and large‐scale functional diversity. Patterns regarding multi‐trait community structure–environment–biodiversity associations were generally held for individual traits. However, depending on the trait, spatial scale and plant ontogenic stage, the pattern's strength changed, or occasionally, their sign reversed.We attribute spatial patterns in multi‐trait mean and covariation to scale‐dependent variation in environmental heterogeneity and trait variation to scale‐dependent competition.Synthesis. Our study provides novel insights into spatial and scale‐dependent variability in functional community structure, environment and biodiversity, and their relationships. Our results demonstrate the usefulness of spatial trait analyses in identifying scale‐dependent assembly processes or finding the importance of processes to biodiversity distributions in spatially heterogeneous landscapes. A spatially explicit perspective is thus helpful for the progress of trait ecology. [ABSTRACT FROM AUTHOR]