Document: JAM-3-52-6

Spatial organization of communities: A unified null model methodology.

SANDERSON, J.G.*

Los Alamos National Laboratory, Los Alamos, NM ¹

Abstract:
Using a matrix null model, metric, and test that I have used previously to analyze island communities for species co-occurrence patterns, I reanalyzed a herpetofaunal assemblage collected along an elevational gradient from Mount Kupe, Cameroon, for unusual species co-occurrence patterns. The null space was generated using a recursive backtracking method and simultaneously conserved both row and column sums. Species' range spans were derived from the actual incidence matrix. Each of 1,000,000 unique uniform-random communities was generated beginning with a matrix of 0s. I compared the number of times each pair of confamilial species co-occurred in the null space with the same number derived from the actual incidence matrix. I also compared range gaps and overlaps for each pair of species with those in the null space. I found several examples of checkerboard patterns between congeneric species (patterns of mutual exclusivity) and other species' co-occurrence patterns that differed significantly from chance expectations (P < 0.01). My results differ from those obtained by others using a different null model approach in which the majority of species co-occurrences showed no significant pattern even among congeneric species suspected of exhibiting competitive interactions. This finding suggests that the null model methodology and associated metric developed to analyze island communities and described here is a powerful tool also applicable to the analysis of communities along gradients.

Keywords: null model, community analysis, species co-occurrences, elevational gradient

This abstract is being presented at: 8:45 AM in session:
Oral Session #56: Metapopulation Analysis.