Modeling animal habitats based on cover types: a critical review.

The simplest type of model describing animal habitats is a "cover-type model," in which a species is assumed to be present in certain vegetation types and absent in others. Ecologists and managers use these models to predict animal distributions for gap analysis and conservation planning. Critics, however, have suggested that the models are overly simplistic and inaccurate. We reviewed the use of cover-type models including assessing their error rates, diagnosing the problems with these models, and determining how they should best be used by managers. To determine models' accuracy rates, we conducted a meta-analysis of 35 studies in which cover-type models were tested against data on animal occurrences. Models had a mean accuracy rate of 0.71 +/- 0.18 (SD). Rates of commission error averaged 0.20 +/- 0.16, and omission errors averaged 0.09 +/- 0.11. A review of the effects of errors in conservation planning suggests that the observed error rates were high enough to call into question any management decisions based on these models. Reasons for the high error rates of cover-type models include the fallibility of expert opinion, the fact that the models oversimplify how animals actually use habitats, and the dynamic nature of animal populations. Given the high rate of errors in cover-type models, any conclusions based on them should be taken with extreme caution. We suggest that these models are best used as coarse filters to identify locations for further study in the field.