EMPIRICAL VALIDATION OF A METHOD FOR UMBRELLA SPECIES SELECTION

Empirical validation that putative umbrella species protect many co‐occurring species is rare. Using 10 sets of data, representing two taxonomic groups and three ecoregions, we tested the effectiveness of a recently developed index for selection of umbrella species. We also tested whether species identified with the index were more effective umbrellas than species selected at random, evaluated whether sample size and intensity affect selection of umbrella species, and examined whether the index could identify cross‐taxonomic umbrellas in a single ecoregion. Conserving all locations with at least one umbrella species would protect the vast majority of each assemblage. A more realistic scenario, conservation of subsets of locations with relatively high numbers of umbrella species, generally would protect ≥0.75 of each assemblage. Randomly selected sets of species often required that more locations be designated for protection than did sets selected using the umbrella index. The umbrella index tended to identify fewer locations that offered an equivalent level of species protection. Sampling intensity affected which species were identified as umbrellas, but not the proportion of species that would be protected. Umbrella species were no more effective than randomly selected species for cross‐taxonomic applications; nonetheless, neither group was significantly less effective than same‐taxon umbrellas. Particularly when selection of protected areas is not highly constrained, it may indeed be feasible to identify effective umbrella species. Unqualified utility of the umbrella index or umbrella species concept, however, is not supported by this study.