Evaluation of fecal metabarcoding for studying the diets of sympatric dusky salamanders (desmognathus)

diet studies often rely on visual identification of prey obtained through forced regurgitation or dissection. These approaches are somewhat invasive and often lack taxonomic specificity, which can discourage diet studies involving at-risk species and limit fine-scale investigations of diet composition. Here, we employ and assess a non-invasive molecular technique to characterize the diets of three co-occurring stream-dwelling salamander species, Ocoee Salamanders (Desmognathus ocoee), Seal Salamanders (Desmognathus monticola), and Nantahala Black-bellied Salamanders (Desmognathus amphileucus) in northern Georgia, USA, and we investigate possible dietary partitioning within and among species. We used DNA metabarcoding to classify the arthropod prey communities from fecal samples of field-collected salamanders and investigated associations with predator species and snout-vent length (SVL). Of 200 salamanders captured and held for 24 h, 38 (19%) produced fecal samples containing arthropod DNA. We identified 53 prey taxa, of which we could classify 27 to species, 12 only to genus, 10 only to family, and four only to order. We found no evidence of dietary partitioning among species or by SVL. Individual fecal samples generally contained few taxa, and few taxa were shared among samples, suggesting that our sample size likely limited the power of our inference. Our results support the utility of fecal metabarcoding as a non-invasive and taxonomically precise alternative to traditional diet analysis techniques. Researchers should also consider the challenges associated with fecal metabarcoding (e.g., infrequent defecation by study organisms, inability to identify life stage of prey), however, before using it to complement more traditional methods.