Accounting for false positive detections in occupancy studies based on environmental DNA: A case study of a threatened freshwater fish (Galaxiella pusilla)

Abstract

Environmental DNA (eDNA) sampling is a promising method for surveying aquatic fauna. Recent eDNA studies have investigated the likelihood of false negative errors in the laboratory and in the field, but the likelihood of false positives remains poorly studied. We investigated the likelihood of both types of errors in eDNA surveys of an Australian threatened freshwater fish (Galaxiella pusilla) using laboratory experiments, field surveys, and recent advances in hierarchical site occupancy‐detection modeling. Laboratory experiments revealed high primer/probe specificity; absence of sample contamination in extraction and qPCR blanks; and rapid accumulation and deterioration of eDNA in aquaria. Hierarchical site occupancy‐detection models fitted to pilot data collected at 13 wetlands revealed that two water samples, each with two qPCRs, would be required to achieve a cumulative detection probability >.95. A more comprehensive survey, in which we simultaneously used dip netting and eDNA sampling at 29 wetlands, revealed similar mean detection probabilities of the two sampling methods (trapping: 0.74 vs. eDNA: 0.68), and low probabilities of false positive errors at the water sample level (0.0080) and at the qPCR level (0.0039) for eDNA sampling. Collectively, our results illustrate that eDNA sampling can be a sensitive and specific method for monitoring the occurrence of freshwater fauna. Detection probabilities of eDNA sampling were comparable to those of a traditional sampling method, and probabilities of laboratory‐induced false positives were low. Future studies employing eDNA sampling should estimate, and properly account for, false positive errors in addition to false negatives.

Optimising sampling and analysis protocols in environmental DNA studies

“…Additionally, the posterior conditional probability of species absence is high when only a small proportion of qPCR replicates amplify. This supports false positives being more likely to be represented by samples where only a few replicates amplify, and that it is important to consider them in the analysis 5 , 19 . Interpretations about presence-absence using eDNA have so far relied on arbitrary thresholds about the number of positive qPCR replicates required to indicate presence.…”

“…As stated by Tingley et al . 19 “Future eDNA studies should still estimate the likelihood of, and properly account for, false positive detections using modelling approaches”.…”

“…These conventional occupancy models only account for the probability of a false negative observation error at the two eDNA survey stages. However, it is known that false positive errors are non-negligible in eDNA surveys 5 , 7 , 19 . For example, contamination of samples, misclassification of DNA sequences or poor specificity within the PCR process may all result in false positive detections.…”

“…Royle and Link 27 ) the model is only locally identifiable with eDNA data alone 33 . However, this limitation can be overcome via calibration from unambiguous survey methods or calibration experiments to identify core modelling parameters, both non-trivial exercises, or introducing informative prior distributions 19 , 32 , 34 . In addition, Guillera-Arroita et al .…”

What do you mean by false positive?

“…Unfortunately, this fear is compounded by misunderstandings of what is meant by the term “false positive” and unrealistic expectations regarding eDNA detection error rates and sources of error. Estimating the probability of false-positive error is, of course, an important aim of eDNA research and has received considerable attention in the literature, leading to the development of sophisticated statistical approaches to error estimation (e.g., Griffin et al, 2020 ; Tingley et al, 2021 ). However, despite these advances, there still persist challenges to communicating the meaning of these errors when they exist and to establishing terminology that clarifies rather than confuses.…”