The influence of explanatory variables on local (sites within locations) and regional β-diversity as revealed by generalized linear models.

Models are based on variables standardized to run from 0 to 1. Local β-diversity model used a log-link function and Poisson error structure; regional β-diversity model used an identity link function and normal error structure (Exp  =  ratio of model to total mean square (or deviance), MS  =  mean square or deviance).