Genetic testing of dung identification for antelope surveys in the Udzungwa Mountains, Tanzania

Dung counts are frequently employed to infer abundance of antelope species in African forests, but the accuracy of dung identification has rarely been tested. We used non-invasive genetic methods to test the accuracy of both field identification and morphometrics for identifying dung samples collected in the Udzungwa Mountains, Tanzania. Species identity was established by sequencing part of the mitochondrial control region from faecal DNA. Field identification was found to be correct in only 58-76% of cases depending on the observer. Discriminant analysis of dung pellet length correctly classified 80% of samples but a larger reference sample size is needed before using this method to classify dung of unknown origin. The results of this study illustrate the potential inaccuracy of dung counts as a monitoring tool for sympatric forest antelope species when the probability of correct identification is unknown. We recommend molecular testing of species identity during forest antelope surveys before conclusions are drawn on the basis of other identification methods.