Intraspecific Trait Variability At Multiple Spatial Scales And Community Assembly In A Subalpine Meadow

Trait-based approaches have been increasingly used to study community assembly and provided the evidence for the existence of habitat filtering and limiting similarity. However, species are commonly described by traits averaged at species level without considering intraspecific variability, although traits are known to vary both between and within species. We investigated the role of intraspecific variability in three functional traits:Height, LDMC(leaf dry matter content), and SLA(specific leaf area) along soil water content gradient in a subalpine meadow. A null model approach was used to test for two niche-based processes:habitat filtering and limiting similarity. Three biodiversity index(Simpson, Pielou and Margalef) were used to explore the trends of biodiversity along slope aspect gradient. We found that there was a large amount of within-species variation, roughly equal to interspecific trait variability, for all the three traits. Moreover, trends in community mean trait values along the soil water content gradient were mainly due to intraspecific trait variability, while interspecific trait variability accounted for a relatively negligible part. The null model approach revealed significant patterns of habitat filtering and limiting similarity for all three traits. Accounting for intraspecific variability increased detection of habitat filtering using mean trait values but decreased detection of habitat filtering using the range of trait values and limiting similarity using CV_ND(coefficient of variation of neighbor distance). Biodiversity showed quadratic relationship with slope aspect, which meant that biodiversity was relatively lower at two ends of the gradient than at the middle. This trends could be explained by about three fators:restricted resources(e.g, soil water content), intensity of competition and limiting similarity. We conclude that intraspecific trait variability, rather than species turnover, was the primary driver of community-level trait-gradient relationships in a subalpine meadow, but its role in generating nonrandom community assembly patterns depended on the traits and metrics selected.