Environmental effects on Neotropical liana species richness

Aim Lianas differ physiologically from trees, and therefore their species-richness patterns and potential climate-change responses might also differ. However, multivariate assessments of spatial patterns in liana species richness and their controls are lacking. Our aim in this paper is to identify the environmental factors that best explain the variation in liana species richness within tropical forests. Location Lowland and montane Neotropical forests. Methods We quantified the contributions of environmental variables and liana and tree-and-shrub abundance to the species richness of lianas, trees and shrubs >= 2.5 cm in diameter using a subset of 65 standardized (0.1 ha) plots from 57 Neotropical sites from a global dataset collected by the late Alwyn Gentry. We used both regression and structural equation modelling to account for the effects of environmental variables (climate, soil and disturbance) and liana density on liana species richness, and we compared the species-richness patterns of lianas with those of trees and shrubs. Results We found that, after accounting for liana density, dry-season length was the dominant predictor of liana species richness. In addition, liana species richness was also related to stand-level wood density (a proxy for disturbance) in lowland forests, a pattern that has not hitherto been shown across such a large study region. Liana species richness had a weak association with soil properties, but the effect of soil may be obscured by the strong correlation between soil properties and climate. The diversity patterns of lianas and of trees and shrubs were congruent: wetter forests had a greater species richness of all woody plants. Main conclusions The primary association of both liana and tree-and-shrub species richness with water availability suggests that, if parts of the Neotropics become drier as a result of climate change, substantial declines in the species richness of woody plants at the stand level may be anticipated.