The miniaturization of video cameras and associated hardware has allowed ecologists to remotely observe the private life of wild animals. New technologies such as camera traps (i.e., fixed cameras activated via movement sensors or time-lapse recorders) or video-collars (i.e., cameras combined to telemetry collars) open up new possibilities for the study of animal behavior across various spatial and temporal scales. However, the large amount of footage collected using these technologies creates a new analytical challenge; it is not unusual to collect hundreds of thousands of photos or seconds of film during a single field season. In this poster, we share experiences acquired during two research projects focused on predator-prey interactions between black bear Ursus americanus and woodland caribou Rangifer tarandus caribou. One study used camera traps to collect data on space use and local density of predators and prey in a boreal forest; the other used video-collars to quantify the predatory behavior of bears towards caribou calves in a subarctic ecosystem. The first study allowed us to study space use and local densities at the broad scale of a whole study area, whereas the second study focused on much finer metrics such as predation rates and prey handling times.
Results/Conclusions
Both studies yielded considerable amounts of data. In the first study, we detected more than 100 black bears on camera traps out of 850 photos of other large mammals and many more thousands of empty photos collected during two summers (2017-2018). In the second study, we collected an average of 11.5 hours of footage per collared bear (n = 2) during a period of only 70 days (end of summer 2018). The latter study, which now plans to include data from 22 more collared individuals, highlighted variations in the predatory behavior of black bears at the northern edge of their geographic distribution. Indeed, one individual was largely carnivorous (64% of its time spent eating) and predominantly consumed caribou, especially calves, whereas the other was more focused on vegetation (95%). Both individuals were males of relatively similar size. Advances in automatized image recognition through machine learning or other artificial intelligence algorithms would have greatly enhanced our capacity to filter raw data in both studies. New ideas and collaborations will be sought to improve our performance in future endeavours using these emergent technologies.