Comparative efficiency of photographs and videos for individual identification of the Andean bear (Tremarctos ornatus) in camera trapping

Resumen

Identification of Andean bear (Tremarctos ornatus) specimens is essential for obtaining demographic estimates of their populations. Camera traps are a noninvasive tool that allows such identification. The efficiency of using photographic or video records for identifying specimens of this species in a wild population in Colombia was compared. A total of 18 camera traps were operated from November 2011 through November 2013; each sample station included a single camera at 0.6 m height, with a bait placed 2 m in front of it at 1.5 m height. Four key external morphological features were chosen for identifying the specimens: Presence, shape and colour of facial; presence, shape and colour of pectoral markings; estimated body size; and sex. For each recording event, a visual file (photograph or video) was scored as ¨good¨ if it showed at least three key identification features, thus allowing the correct identification of the specimen; or as ¨bad¨ if it showed fewer than three features, making identification impossible. Successful recording events were those that included at least one good visual file (photograph or video). A total of 4,588 visual files were obtained: 4,324 photographs in 325 recording events and 264 videos in 260 recording events. 5.25 % of the photographs and 53.03 % of the videos were scored as good files. 26.77 % of the photograph-based and 49.62 % of the video-based recording events were successful. There were statistically significant differences between the percentage of good photographs and good videos obtained every time a camera trap was activated in the presence of a bear (Mann-Whitney, P = 1.37E-11). The low percentage of successful recording events obtained with photographs (26.77 %) compared to that obtained with videos (50.38 %), is consistent with results previously reported for this same species in Ecuador using photographs (25 %). The higher percentage of good videos (53.03 %) compared to that of good photographs (5.25 %), is consistent with the statistically significant difference found between the percentage of good photos and good videos obtained every time a camera trap was activated in the presence of a bear (Mann-Whitney, P = 1.37E-11), and with results previously reported for the Asian black bear (Ursus thibetanus, 70 %) using sample stations including a single camera trap with video format. The use of video for recording Andean bear specimens allows the observation of individuals from different viewpoints and distances, even with the use of sample stations including a single camera trap, thus minimizing the effect of light reflection on the recognition of key identification features. Additionally, the video format allowed recognition of particular physical conditions, such as limp or rigid limbs in some specimens, which cannot be recognized in photographs. In this study case, information obtained with video records provides a greater ability to recognize individual marks in the specimens and to identify them.