Can scat-based species identification be a misleading sign of presence? More evidences from northeastern Portugal

Abstract

Species identification of non-invasively collected samples using molecular genetics tools has become an important tool in ecological research. For decades, scat-based ecological studies were almost exclusively rooted in morphological identification of scats, within local context, in the field. However, this approach raised a controversial debate, due to species and context-specific probability of error and lack of validation. In this study, we aimed to test the accuracy of mesocarnivore scats identification, based on morphological criteria, using a carnivore guild in northeastern Portugal as a model and molecular identification as a standard for accuracy of morphological identifications, within local context. While using only expertbased identifications for comparison with molecular identification standard, we have also compared the identifications performed by observers with different levels of experience. We extracted DNA from 63 scats (NE Portugal), which was successfully amplified/sequenced from 83% (n= 52) of the extracts: 38 were molecularly assigned to red fox (Vulpes vulpes), eight to stone marten (Martes foina), two to pine marten (Martes martes) and domestic dog (Canis lupus familiaris) and one to European badger (Meles meles) and common genet (Genetta genetta). There was a tendency for better performance by more experienced researchers, with 67% of scats being correctly assigned, but differences among observers were not significant. Due to the small sample size, only for foxes and stone martens was possible to estimate the error rate in species assignment, based on morphological criteria. False positive rates (% of times a scat was misassigned to a given species) were 4% for fox samples and 62% for stone marten. False negative rates (i.e. the rate at which a scat of a given species was assigned to another species) reached 29% for fox (scats that were initially assigned to stone marten and domestic dog were in fact from fox) and 25% for stone marten (originally misassigned to weasel, Mustela nivalis), respectively. The results support the need to implement molecular methods in ecological studies based on scat identification, so researchers can determine the error rates associated with morphological discrimination to develop accurate monitoring studies.