Comparing hair-morphology and molecular methods to identify fecal samples from Neotropical felids

Abstract

To avoid certain problems encountered with more-traditional and invasive methods in behavioral-ecology studies of mammalian predators, such as felids, molecular approaches have been employed to identify feces found in the field. However, this method requires a complete molecular biology laboratory, and usually also requires very fresh fecal samples to avoid DNA degradation. Both conditions are normally absent in the field. To address these difficulties, identification based on morphological characters (length, color, banding, scales and medullar patterns) of hairs found in feces could be employed as an alternative. In this study we constructed a morphological identification key for guard hairs of eight Neotropical felids (jaguar, oncilla, Geoffroy's cat, margay, ocelot, Pampas cat, puma and jaguarundi) and compared its efficiency to that of a molecular identification method, using the ATP6 region as a marker. For this molecular approach, we simulated some field conditions by postponing sample-conservation procedures. A blind test of the identification key obtained a nearly 70% overall success rate, which we considered equivalent to or better than the results of some molecular methods (probably due to DNA degradation) found in other studies. The jaguar, puma and jaguarundi could be unequivocally discriminated from any other Neotropical felid. On a scale ranging from inadequate to excellent, the key proved poor only for the margay, with only 30% of its hairs successfully identified using this key; and have intermediate success rates for the remaining species, the oncilla, Geoffroy's cat, ocelot and Pampas cat, were intermediate. Complementary information about the known distributions of felid populations may be necessary to substantially improve the results obtained with the key. Our own molecular results were even better, since all blind-tested samples were correctly identified. Part of these identifications were made from samples kept in suboptimal conditions, with some samples remaining outdoors for up to seven days, simulating conditions in the field. It appears that both methods can be used, depending on the available laboratory facilities and on the expected results.