Paper No. 12
Presentation Time: 11:30 AM
EXTENDING THE REACH OF PALEOENVIRONMENTAL INFERENCE MODELS USING A TRAIT-BASED MODEL OF COMMUNITY ASSEMBLY
MICHELSON, Andrew V.1, PARK, Lisa E.2 and KOHLMAN, Krystal2, (1)Program in Integrated Bioscience, University of Akron, Akron, OH 44325-4101, (2)Geology and Environmental Science, University of Akron, University of Akron, University of Akron, Akron, OH 44325-4101, avi1@zips.uakron.edu
A tendency for species with specific traits to be found in certain environments has been noticed for a long time. However, the central goal of community ecology today is
predicting the distribution and relative abundances of species across spatial, temporal, and environmental gradients, a capability that would have wide application in paleoenvironmental reconstruction. Shipley (et al., 2006; 2010) has proposed a model of community assembly whereby natural selection between individuals of different species determines the distribution of
traits among environments, allowing species’ abundances to be predicted providing the species pool is known. Natural selection among genotypes across species along environmental gradients results in a predictable distribution of traits, even though species cannot exchange genes. This project tests this trait-based model of community assembly using ostracode species in lakes on San Salvador Island, Bahamas. Understanding how species’ traits, in addition to species’ abundances, are distributed across environmental gradients will greatly enhance the usefulness of paleoenvironmental inference models since they will be able to be used with species not encountered in modern data and across wider geographic areas.
We sampled 11 species’ abundances across 32 lakes on San Salvador Island and measured 25 traits of each species that display greater interspecific than intraspecific variation. Using community-aggregated mean trait values at each site as constraints, we constructed a model to predict relative abundance of all 11 species at each of the 32 sites using maximum entropy techniques. Mantel tests and non-metric multidimensional scaling ordinations reveal that this model makes highly accurate predictions of species’ abundances at all sites (Mantel r=.9978, p<.001). While many traits were useful as predictors of species’ abundances, those traits related to valve size, degree of calcification, and valve shape were the best predictors of species’ abundances. Distribution of ostracodes species’ traits therefore follow predictable distributions across environments San Salvador Island, allowing for the creation of more widely applicable paleoenvironemntal inference models based on species’ traits rather than species’ abundances.
