A comparison of detection probabilities for the patch‐nosed salamander (Urspelerpes brucei) using six survey methods

Abstract

AbstractThe patch‐nosed salamander (Urspelerpes brucei), discovered in 2007, is a small amphibian endemic to northeast Georgia (GA) and northwest South Carolina (SC), USA. Given its recent discovery, little is known about the species’ conservation status. This is especially true in SC where this species is known from 4 stream localities in total, making monitoring and inventory a current priority. We applied 6 survey methods in 3 streams known to harbor the species (1 in GA; 2 in SC) using a randomized complete block design and conducted replicate surveys using each method 5 to 6 times between August 2021 and June 2022. We used occupancy models to estimate the probability of patch‐nosed salamanders using a randomly selected 5‐m stream segment while simultaneously investigating the effects of method and conditions on detectability. We recorded a total of 90 patch‐nosed salamander detections (60 from GA; 32 from SC) across 22 of the 39 stream segments in our study. Occupancy was similar in 2 streams (SCSouth: 0.94 [0.22 to −1.00 95% CI]; GARef: 0.70 [0.41–0.89 95% CI]) where it was also 3 to 4 times higher than in a third (SCNorth: 0.16 [0.04–0.47 95% CI]), though mechanisms driving the differences remain unclear. While we detected patch‐nosed salamanders with all 6 survey methods, aquatic and terrestrial versions of both leaf‐litter bags and area‐unconstrained surveys were more effective than randomized area‐constrained searches. Terrestrial applications of each method were biased toward adult detections while aquatic applications were biased toward larvae. Patch‐nosed salamander detectability was maximized during conditions (cloudless skies, >5 cm rainfall in the prior week, soil temperatures ≥20°C) that align with spring (March–June) in the species' range. A power analysis based on detectability estimates suggests that, under ideal sampling conditions, either 5 temporally replicated area‐unconstrained searches or a survey of 5 spatially replicated leaf‐litter bags (per 5‐m segment) are necessary to confirm presence‐absence of patch‐nosed salamanders with 90% confidence. Given that the 2 SC streams included in our study represent nearly half of the species’ range in that state, but collectively yielded only about one‐third of total detections and exhibited considerable disparity in occupancy, we recommend more work to clarify resource selection patterns, subpopulation stability, and potential threats to patch‐nosed salamanders in SC specifically.