I find linear models are not appropriate for my comparisons of cross-taxa species richness, and want to find more information on the defects of linear models in ecology. A review is preferable.
Ben Bolker's entire text is a good primer on non-gaussian models in ecology.
http://ms.mcmaster.ca/~bolker/emdbook/
But for a shorter version, he also has a nice review on generalized mixed models in ecology for non-normal data and for considering random effects, both common in ecological data.