Individual Identification of the Endangered Wyoming Toad Anaxyrus baxteri and Implications for Monitoring Species Recovery
Monitoring the fates of individuals after release in the wild is essential for building effective species recovery programs. Current conservation efforts for the endangered Wyoming Toad (Anaxyrus baxteri) are limited by the size and number of toads that can be individually marked using invasive tagging techniques. We evaluated the use of natural patterns of wart-like glands on the dorsum of Wyoming Toads as a potential identification technique. We photographed 194 known-identity individuals (822 total images, representing 1,554 true matching-image pairs of the same individuals) from two captive-breeding facilities in 2011 and 2012. Spot patterns provided stable markings from metamorphosis through adult life stages, and naïve observers correctly matched 100% of a subset of photo pairs “by eye.” In contrast, computer-assisted identification performed relatively poorly: the two software platforms tested (Wild-ID and Hotspotter) failed to match 47% and 64% of true matching-image pairs, respectively. The use of higher-quality cameras with faster automatic focusing speeds yielded the largest improvement in matching success of any variable tested when using identification software. Simulated capture–recapture data demonstrated that using software to identify individuals would bias abundance estimates high by up to 920%.Abstract
Example of photos used for wart pattern retention trials, showing the progression of a single toad over the course of the study, from newly metamorphosed toadlet (July 2011; ∼1 g) to fully grown adult (August 2012; ∼30 g).
Wild-ID similarity scores vs. weight difference (A) and time interval (B) between photo captures of the same individual. Weight change (A) does not include individuals measured and photographed multiple times on the same day.
Effect of misidentification on the bias of capture–recapture abundance estimates using different identification techniques. We assumed data were collected over three closed capture periods and modeled using the Huggins closed population model.
Image pair of the same adult toad taken August 2011 (left image) and August 2012 (right image) and matched with Wild-ID. The similarity score between these images was the highest (0.1316) among all photo pairs collected ≥50 d apart. Lines indicate individual SIFT features identified on matching photographs. Note that the large wart clusters identifiable to the human eye are not being fully characterized.
Contributor Notes
