Factors affecting the (in)accuracy of mammalian mesocarnivore scat identification in South-western Europe

Research on terrestrial carnivore ecology frequently relies on scat identification and analysis. However, species assignment is commonly based on scat morphology. Potential errors in scat identification are rarely accounted for and might contribute to substantial bias of the final results. Using molecular methods, we evaluate the accuracy of species identification based on morphological characteristics of mammalian mesocarnivore scats collected in two areas in the Iberian Peninsula. Our results revealed that error rates in species assignment of scats based on morphology were highly variable, ranging from 14%, for putative red fox Vulpes vulpes samples, to 88%, for putative wildcats Felis silvestris. The developed models revealed that putative species, season, study area and target species abundance are among the factors involved in identification accuracy. However, the low variability explained suggests that unaccounted factors also had significant effects on accuracy rates. The error rates in scat species assignment constitute a potential source of bias in ecological studies, with serious consequences for the management of threatened species, as unrealistic estimates of status and distribution are prone to occur. Our results suggest that scat identification accuracy rates are circumstance-specific and therefore should not be transferred or extrapolated. We suggest that scat-based studies should implement measures (molecular or others) that allow researchers to determine their own circumstance-specific error rates in scat identification, which should be incorporated in subsequent analyses, ensuring reliable ecological inferences.
This work was partially supported by a PhD grant from the Fundação para a Ciência e a Tecnologia (FCT) to P.M. (SFRH/BD/37795/2007) and a research project from the Spanish National Plan (project ref: CGL2009-10741) funded by the Spanish Ministry of Science and Innovation and EU-FEDER funds. R.G. and T.S. are supported by postdoctoral and PhD grants, respectively (SFRH/BPD/36021/2007and SFRH/BD/73680/2010).