Variation in forest richness, density, and size is explained by environmental gradients from the plot to landscape scale

A spatial gap exists between fine scale forest census plot dynamics and coarse scale landscape processes. Patterns observed at the plot scale do not necessarily continue outside plot boundaries and how such patterns scale across the landscape remains poorly understood. The sparse geographic extent of census plot data and the prohibitive cost of high resolution remote sensing remain the largest obstacles to extending the study of plot scale dynamics across the landscape, however this situation is changing rapidly and methods must be developed to integrate new data and close this spatial gap. The research presented in this dissertation utilizes plot level census data to study how plant richness, density and size vary across environmental gradients and presents a methodological framework for 1) predicting species richness where no plot data currently exists, 2) investigating guild level differences between plant growth forms in relation to resource gradients and 3) studying how forests vary across environmental gradients at the continental scale. The methodology utilizes spatial and non-spatial modeling to identify associations between plant richness, density and size at the plot scale. These associations are used to predict forest dynamics at the landscape scale and study forest structure at the continental scale. First, the findings provide evidence that increased tree species richness is associated with environmental heterogeneity in both the canopy and hydrologic environment. These associations explain nearly half of the variation in tree species richness and are used to make hectare scale predictions of tree species richness on Barro Colorado Island, Panama. Secondly, there is strong evidence of habitat filtering along resource gradients of light and water caused by guild level differences between plants. In particular, free-standing plant guilds are non-randomly arranged along the hydrologic gradient with short understory trees and shrubs clustered in wetter environments and midstory and canopy level trees clustered in drier environments. Finally, when compared to climatic predictors, topography and terrain slope in particular appear to be exerting strong controls on forest structure across Mesoamerica. Taller forests occur on steep slopes, high elevation, on well drained soils and these effects are insensitive to land cover, biome and spatial scale