A Multi-Point Identification Approach for the Recognition of Individual Leopards (Panthera pardus kotiya)

SIMPLE SUMMARY: All of the previous research on photography-based leopard identification was conducted based on the assumption that leopard spots and rosette formations do not change in shape or form. We observed 29 instances of changes to spot and rosette formations in continuously observed leopards at Yala National Park, Block 1. Since the previous approaches have flaws and errors, the same leopard may be misdiagnosed and counted numerous times, overestimating leopard populations if the spot and rosette formation of a leopard has changed. To address this issue, we developed the multi-point leopard identification method, which is a novel process for identifying Sri Lankan leopards. The minimum leopard population of Yala National Park, Block 1, on 31 March 2021, was established during the study. ABSTRACT: Visual leopard identifications performed with camera traps using the capture–recapture method only consider areas of the skin that are visible to the equipment. The method presented here considered the spot or rosette formations of either the two flanks or the face, and the captured images were then compared and matched with available photographs. Leopards were classified as new individuals if no matches were found in the existing set of photos. It was previously assumed that an individual leopard’s spot or rosette pattern would not change. We established that the spot and rosette patterns change over time and that these changes are the result of injuries in certain cases. When compared to the original patterns, the number of spots may be lost or reduced, and some spots or patterns may change in terms of their prominence, shape, and size. We called these changes “obliterate changes” and “rejig changes”, respectively. The implementation of an earlier method resulted in a duplication of leopard counts, achieving an error rate of more than 15% in the population at Yala National Park. The same leopard could be misidentified and counted multiple times, causing overestimated populations. To address this issue, we created a new two-step methodology for identifying Sri Lankan leopards. The multi-point identification method requires the evaluation of at least 9–10 spot areas before a leopard can be identified. Moreover, the minimum leopard population at the YNP 1 comprises at least 77 leopards and has a density of 0.5461 leopards per km(2).