Modeling the spatial structure of topical forests

The spatial structure of tropical forest stands under different management conditions was modeled as a series of different spatial point processes. Spatial patterns were first assessed by K-function analyses to help choose a point process appropriate for observed patterns. The homogenous Neyman–Scott process accurately described live tree distribution in clear cut areas, where tree patterns tended to be aggregated. Parameters were estimated by minimizing Diggle's modified least squares criterion, and goodness-of-fit was assessed by comparison to confidence envelopes constructed by Monte Carlo simulation. Parameter estimates can be interpreted to help understand the ecological processes influencing re-colonization of disturbed areas. The inhomogeneous Poisson process was investigated for simulating the spatial pattern of ingrowth trees in lower canopy strata. The intensity function of this process was inversely proportional to variables representing canopy density. As assessed by Monte Carlo generation of confidence envelopes, the inhomogeneous Poisson process successfully portrayed the influence of canopy structure on understory plant distribution in most stands. Tree mortality was modeled as a thinning process in which the probability of individual tree mortality was conditional on subject tree attributes and competitive environment. The thinning function took the form of a generalized linear model with a binomial error distribution and logit link function. In most stands, tree neighborhood variables were powerful predictors of mortality, but they were not important predictors in all plots. This suggests that the surrounding forest structure of a subject tree has considerable influence on its morality, but competition is not the sole cause of tree morality in tropical forests.