Fire is increasingly driving loss and degradation of tropical habitats, but factors influencing biodiversity responses to fire are inadequately understood. We conduct a pan-tropical analysis of systematically collated data – 5257 observations of 1705 plant species (trees and shrubs, forbs, graminoids and climbers) in burnt and unburnt plots from 28 studies. We use model averaging of mixed effect models assessing how plant species richness and turnover (comparing burnt and unburnt communities) vary with time since fire, fire type, protected area status and biome type. More long-term studies are needed, but our analyses highlight three key findings. First, prescribed and non-prescribed burns have contrasting impacts on plant communities, the direction of which depends on focal life form and biome. Forb richness, for example, increases following non-prescribed (but not prescribed) burns in savannahs and flooded grasslands, but in moist broadleaved forest forb richness increases strongly following prescribed (but not non-prescribed) burns. Second, protected areas mitigate fire impacts on plant communities. Species richness of trees/shrubs increased (by ~50%) following fires in non-protected sites but tended to remain similar in protected sites. Similarly, ten years after a fire event graminoid community composition had recovered fully to resemble non-burnt communities in protected areas, but remained highly divergent in unprotected sites. Finally, this persistence in divergence of community composition following fire events occurs across a number of life forms. Composition of tree/shrub communities remained divergent from unburnt communities ten years after a fire, and composition of forb communities only returned to those of unburnt sites after ten years. Fire intervals are already less than ten years in some tropical locations, and future climate and land use change are predicted to further shorten these intervals. Plant communities across much of the tropics are thus likely to change substantially in response increased exposure to fire.

Selective logging of tropical forests substantially alters the composition and spatial arrangement of plant communities. Previous studies examining logged-forest tree assemblages have focused primarily on adult communities, leaving major knowledge gaps regarding the diversity patterns of earlier life stages. A key question is to elucidate the temporal dynamics of community assembly in human-modified forests. Sampling 8,664 sapling, juvenile, and adult trees from a heavily logged forest in Sabah, Malaysian Borneo, we tested whether compositional variation and the relative importance of different environmental and spatial factors explaining patterns of ß-diversity differed between life stages, and whether dissimilarity was driven by species turnover or nestedness. We found positive ß-deviations in all communities, consistent with a strong influence of assembly processes that result in aggregated spatial distributions of individual species. Across life stages, ß-diversity was largely explained by spatial distance, rather than measures of environmental heterogeneity. Dissimilarity was driven by species turnover not nestedness, with compositional variation in early life stages strongly correlated with turnover in adult communities. Collectively, our findings indicate that despite increased spatial heterogeneity in forest structure, liana infestation, and canopy openness post-logging, these factors do not sufficiently explain fine-scale patterns of tree composition. Alternatively, diversity patterns of earlier life stages more closely reflect potential assembly processes related to aggregated adult distributions and associated dispersal limitations resulting from spatial variation in logging activity.