Print PageResearchers have used static distributional data to infer the role of functional traits in assembly of tree communities. However, process-oriented models that use dynamic community data to reveal the role of traits may improve our ability to disentangle mechanisms. We present a hierarchical Bayesian spatial model of trait and size-based tree growth and mortality. We included in the model the roles of local filtering of traits and of competition between species of similar traits. We used growth and mortality data from two censuses (in 2004 and 2009) of all trees with DBH ≥ 1 cm on a 25 ha plot of wet subtropical forest in Fushan, Taiwan. Data on four functional traits were collected for 87% of species. We compared estimates of trait assembly parameters to estimates from data simulated under a null model where traits play no role in assembly. We compared variation in the strength of trait filtering to underlying environmental gradients.
Results/Conclusions
We found that parameters measuring the effects of traits, estimated from the community dynamic data, are significantly greater than trait parameters estimated from data simulated under the null. We revealed patterns of spatial variation in the strength of trait filtering and the optimal trait value in filtering. Trait filtering variation is correlated to variation in some topographic and soil conditions. Our model estimated separately the action of trait-based filtering and competition in a novel way that may improve our ability to build predictive models of assembly based on environmental conditions and species traits. Finally, we suggest extensions to the model to include additional aspects of assembly.
