Cluster analysis of spatial patterns in Malaysian tree species.

Tree species in tropical rain forests exhibit a rich panoply of spatial patterns that beg ecological explanation. The analysis of tropical census data typically relies on spatial statistics, which quantify the average aggregation tendency of a species. In this article we develop a cluster-based approach that complements traditional spatial statistics in the exploration and analysis of ecological hypotheses for spatial pattern. We apply this technique to six study species within a fully mapped 50-ha forest census in peninsular Malaysia. For each species we identify the scale(s) of spatial aggregation and the corresponding tree clusters. We study the correlation between cluster locations and abiotic variables such as topography. We find that the distribution of cluster sizes exhibits equilibrium and nonequilibrium behavior depending on species life history. The distribution of tree diameters within clusters also varies according to species life history. At different spatial scales, we find evidence for both niche-based and dispersal-limited processes producing spatial pattern. Our methodology for identifying scales of aggregation and clusters is general; we discuss the method's applicability to spatial problems outside of tropical plant ecology.