Using ecological-niche modeling as a conservation tool for freshwater species: live-bearing fishes in central Mexico.

Ecological-niche modeling is an important tool for conservation assessment of terrestrial species; however, its applicability has been poorly explored in the aquatic realm. Goodeines are a monophyletic group of viviparous freshwater fishes that are well known in central Mexico, with 41 species in 19 genera. Given the number of threats to biodiversity in the region, goodeines represent an excellent model with which to test novel conservation approaches. We assessed the conservation status of the goodeines (37 species), based on their potential distributions predicted by ecological-niche models generated with the genetic algorithm for rule-set prediction (GARP). Predictions of species' distributions performed well in six out of eight species for which sufficient information was available to perform estimations of the area under the curve (AUC) in receiver operating characteristic plots. Extensive field surveys conducted in recent years in most cases confirm the models' predictions. Species richness exhibited a nested pattern, in which the number of species increased toward the center of the distribution of the group. At the basin level, the Río Ameca Basin had the highest number of species (11), chiefly because of the high number of microendemic species (6). Human activities within water bodies (e.g., extensive aquaculture) and drainages (e.g., agriculture, ranching, industrial activities) have affected most goodeines severely, given the deleterious effects of pollution and introductions of exotic species, such as carp (Cyprinus carpio, Ctenopharingodon idella) and tilapia (Oreochromis spp.). Our results paint a pessimistic picture for the long-term survival of many goodeines in their natural environment, and realistic conservation measures are complex and would require immediate protection of specific areas that we have identified. Ecological-niche modeling is a suitable tool for conservation assessment of freshwater species, but availability of environmental information on aquatic systems (e.g., temperature, water speed, pH, oxygen concentration) would improve distributional predictions.