Exploring and validating observations of non-local species in eDNA samples

Abstract

The development of DNA-based methods in recent decades has opened the door to numerous new lines of research in the biological sciences. While the speed and accuracy of DNA methodologies are clearly beneficial, the sensitivity of these methods has the adverse effect of increased susceptibility to false positives resulting from contamination in field or lab. Here, we present findings from a metabarcoding study on the diet of and food availability for five insectivorous birds, in which multiple lepidopteran species not known to occur locally were discovered. After describing the pattern of occurrences of these non-local species in the samples, we discuss various potential origins of these sequences. First, we assessed that the taxonomic assignments appeared reliable, and local occurrences of many of the species could be plausibly ruled out. Then, we looked into the possibilities of natural environmental contamination, judging it to be unlikely, albeit impossible to fully falsify. Finally, while dissimilar combinations of non-local species' occurrences across the samples did not initially suggest lab contamination, we found overlap with taxa and sequences handled in the same lab, which was undoubtedly not coincidental. Even so, not all exact sequences were accounted for in these locally conducted studies, nor was it clear if these and other sequences could remain detectable years later. Although the full explanation for the observations of non-local species remains inconclusive, these findings highlight the importance of critical examination of metabarcoding results, and showcase how species-level taxonomic assignments utilizing comprehensive reference libraries may be a tool in detecting potential contamination events, and false positives in general.