Multi‐species occupancy modeling reveals methodological and environmental effects on eDNA detection of amphibians in temporary ponds

Aquatic environmental DNA is increasingly used for biodiversity monitoring, such as surveying threatened and invasive species. Mainstreaming these methods in practical applications, however, still requires significant standardisation and optimisation, namely regarding DNA capture methods. Here we evaluated how filter type (standard disc filters vs high-capacity capsules), number of sampling sites, volume of water filtered and environmental factors affected amphibian detection in Mediterranean temporary ponds. The study involved water filtering until clogging at one (capsules) and five (discs) sites from 16 small and shallow ponds, where three urodele and seven anuran species were recorded through sweep-netting and adult observations. Detection probabilities were estimated from site occupancy models based on replicate sampling and from an adaptation of time-to-detection models relating detection probability to volume of water filtered. Discs filtered relatively small volumes (15–1250 mL), with detection probabilities of the two abundant species (Pelobates cultripes, Hyla meridionalis) increasing rapidly with sample size and water volume, reaching almost perfect detection (0.95) at four and seven discs, and 420 mL and 1860 mL, respectively. However, reaching high detection probabilities for rare species (Pelodytes atlanticus, Pleurodeles waltl, Triturus pygmaeus) would require larger sampling effort than that used in our study. Despite filtering much larger volumes (600–5300 mL), filtering with capsules at a single site per pond provided lower detection probabilities for abundant species than filtering with discs at five sites. Rarer species showed no difference between methods, which may be due to small sample sizes and reduced statistical power for species with few detection. The effect of conductivity on species detectability was largely negative, while the influence of water clarity varied across species, and pH had no effects. Overall, our results suggest that eDNA amphibian surveys in Mediterranean temporary ponds need to consider filter clogging, heterogeneous DNA distribution, and highly conductive waters.