Understanding mesoscale patterns of floristic diversity are of direct relevance to operational management and conservation, particularly in tropics where natural vegetation persists as fragments. Data on forest stand structure and composition, gathered across large scale environmental and/or degradation gradients, can elucidate species-habitat relationships, forest succession dynamics, and allow hypothesis tests about the ecological processes that maintain species diversity. We examined the influence of environmental (rainfall, soil, length of dry season and altitude) and surrogate disturbance variables (forest vs. non-forest area proportion, proximity to villages, and density of human habitation) on tropical woody plant species diversity, based on data from 96, 1-ha plots (macroplots) spread across 21,970 km² in the Central Western Ghats (CWG) of Karnataka, India. In the macroplots, all trees and lianas above 10 cm girth at breast height (gbh) were recorded and within them, all woody plants below 10 cm gbh but above 1 m height were enumerated in 3 microplots of 10 x 10 m.

Results/Conclusions

A total of 61,906 individual trees comprising 400 species under 254 genera and 75 families were recorded in macroplots. From microplots, 14,863 individuals comprising 342 species under 235 genera and 80 families were recorded.A symmetric co-correspondence (COCA) analysis of tree species from macroplots and understory plant species from microplots showed similar mesoscale patterns of distribution, with a progressive change in the mixture of evergreen plots (negative axis) to deciduous plots (positive axis) along COCA axes gradients. We coupled the floristic tables (macroplots and microplots) to the explanatory variables through non-symmetrical correspondence analysis on instrumental variables. The statistical significance of the portion of initial floristic variance captured by the approximated tables was tested through Monte-Carlo method with 999 iterations. The environmental and disturbance variables significantly explained 52.2% of the total variance observed in macroplots and 47.3% of total variance in microplots. Our results suggest that at CWG regional scale, alpha diversity is related most strongly to the rainfall gradient, followed by soil and length of dry season. For beta diversity, the influence of seasonality may be limited to the wet evergreen forests (Viz. high-rainfall zone), while soil types also influence them. Increase in the proportion of deciduous species in potential evergreen zone (2000–5000 mm rainfall/yr) indicated secondary forest stands associated with higher disturbance. This study disentangles environmental and disturbance regimes that drive patterns of tropical plant species diversity in a biodiversity hotspot.