Tuesday, August 4, 2009 - 3:40 PM

OOS 20-7: Community assembly of an Amazonian forest: what does a phylogenetic perspective add?

Nathan Kraft1, Renato Valencia2, and David D. Ackerly1. (1) University of California, Berkeley, (2) Pontificia Universidad Católica del Ecuador

Background/Question/Methods

Phylogenetic community methods are a powerful set of tools for making inferences about the importance of ecological processes in structuring communities. As these methods use relatedness as a proxy for ecological similarity, they are most useful when direct measurement of the ecological strategy for each species is impractical. However the interpretation of phylogenetic community patterns is challenging in many situations as different ecological and evolutionary processes can produce similar patterns. In contrast, trait-based methods often present less equivocal results. Here we present a phylogenetic community analysis of one of the world's most diverse forests- the 25 ha. Yasuní Forest Dynamics plot in Ecuador, which contains over 1,100 tree species. We use a recent ecological strategy based-analyses (Kraft et al. 2008 Science) as a point of comparison. We then use an ecological simulation modeling approach (e.g. Kraft et al. 2007 Am. Nat.) to account for the effect of spatial scale on the statistical power of both sets of methods.

Results/Conclusions

We found evidence for habitat filtering from both trait and phylogenetic methods from small (25 m2) to intermediate (1,000 m2) spatial scales. Trait-based methods only detected even spacing of strategies, a pattern consistent with competition or enemy-mediated density dependence, at smaller spatial scales (25 m2 to 400 m2). Simulation modeling of different community assembly processes suggests that low power to detect even spacing of traits at larger spatial scales may contribute to the observed patterns. Trait and phylogenetic methods tended to identify the same areas of the forest as being subject to habitat filtering, but did not consistently report even spacing in the same areas. Phylogenetic community methods, which are far less data-intensive than trait based methods, captured much of the same filtering patterns detected by trait based methods, but often failed to detect processes that produce even spacing. Taken together, it appears that both habitat associations and niche differentiation shape species co-occurrence patterns in one of the most diverse forests in the world at a range of small and intermediate spatial scales.