VEECH, J.A.; Colorado State University: Using structural equation modeling to compare hypotheses of local and regional species diversity

Structural equation modeling (SEM) was used to evaluate several hypotheses previously proposed to explain local and regional species diversity. The North American Breeding Bird Survey provided estimates of species richness within communities (local diversity) and ecoregions (regional diversity). SEM is a powerful technique for testing hypotheses that involve latent variables as explanatory factors. A latent variable is a theoretical construct that cannot be directly measured; instead, latent variables are assessed by indicator variables. I tested the effect of six latent variables on local and regional bird diversity: amount of available environmental energy, landscape compositional complexity, configurational complexity, geographic size of ecoregion, geographic position, and sampling intensity. The average amount of environmental energy in an ecoregion had a negative effect on local and regional diversity, opposite of the prediction of the �energy-availability� hypothesis. The compositional complexity of the landscape within an ecoregion had a positive effect on local and regional diversity, which thus supported the �habitat-heterogeneity� hypothesis. As expected, the size of an ecoregion had a positive effect on regional diversity. Likewise, sampling intensity also had a positive effect on regional diversity. Overall, this study demonstrates the application and advantages of SEM. Although SEM has been widely used by social scientists, it is not as well known to biologists. Thus, another goal of this study is to promote the use of SEM in the study of species diversity and other biological topics.