SEARCH

SEARCH BY CITATION

Keywords:

  • camera-trap;
  • false positive error;
  • individual identification;
  • observer agreement;
  • red fox;
  • Vulpes vulpes

ABSTRACT

The individual identification of animals from photographs is increasingly used to obtain density estimates not only for animals with distinct natural markings but also for species with little or no distinct markings, such as coyotes (Canis latrans), pumas (Puma concolor) and tapirs (Tapirus terrestris). This lack of distinct natural markings may lead to large error rates in the assessment of photographs, and as a consequence result in poor abundance estimates. We conducted an experiment asking expert observers to identify individual red foxes (Vulpes vulpes) from a set of photographs taken by automatic camera-traps. Our objectives were to determine whether reliable individual identification of red foxes from photographs is possible and, if possible, to improve the identification process. Exact assessment of error rates in individual identification can only be achieved if photographs of known individuals are available. This is rarely the case; therefore, we used photographs of red foxes from different study sites and determined the lower limit of the proportion of false positive matches. Our analysis, based on 10 expert responses, suggested that individual identification of red foxes is not reliable. The number of individual foxes assessed by the observers varied between 4 and 23 individuals. The minimal proportion of false positive matches was very large (>50% of the photographs considered to be of the same individual were from 2 different study sites) and there was little agreement among experts on which photographs showed the same individuals. Hence, we caution against individual identification from photographs of red foxes and other animals with similar or less natural markings without further testing. © 2013 The Wildlife Society.