Spatial modeling of butterfly species richness using tiger beetles (Cicindelidae) as a bioindicator taxon

General spatial patterns of species richness can be useful when determining conservation policy. Reliable species distribution data, however, are often rare or limited to a relatively few taxa in many parts of the world, and extensive species inventories tend to be expensive and time consuming. Consequently, the use of a few rigorously selected bioindicator taxa to represent broad-based inventories has been suggested as a viable alternative. Because spatial dependencies (spatial autocorrelations) are likely to exist in species richness data, common statistical techniques that assume independence are inappropriate for making cross-taxa comparisons of species ranges and distributions. We applied geostatistical methods that incorporate spatial dependencies to test the usefulness of a proposed bioindicator, tiger beetles, as a predictor of an unrelated taxon, butterflies, across North America and found a statistically significant relationship. We also showed how the application of statistical procedures that assume independence may be misleading. Finally, we showed how to make spatial predictions of species richness in intermediate areas where no sample species data are available.