Analysis
Spatial autoregression was used to investigate the relationship between bivalve diversity and environmental variables (R package spatialreg: Bivand et al. , 2013). For each model, logged diversity was compared with the eight environmental variables as predictors. This was repeated for each dataset and diversity metric. Finally, to identify the best predictor variables for each diversity metric and dataset, an optimal submodel was chosen using the function RegBest in the R package FactoMineR (LĂȘ et al. , 2008). This analysis was repeated using the taxonomic composition data to test for the underlying causes of any biogeographic gradient.
Results