Improving the reliability of Individual animal identification based on Camera-Trapping Data

A Novel Method to Improve Individual Animal Identification Based on Camera-Trapping Data

EDUARDO MENDOZA, PIERRE R. MARTINEAU, ELLIOTT BRENNER, RODOLFO DIRZO.

The Journal of Wildlife Management 75(4):973–979; 2011; DOI: 10.1002/jwmg.120

Picture-based individual identification of animals is largely based on ad hoc protocols. This in turn tends to strongly rely on human visual inspection of pictures.  Mis-identification can produce significant biases in population estimates. In the paper referred to above the researchers present a novel method to improve individual identification of animals based on camera-trapping data. Application of their method to bobcat (Lynx rufus) resulted in a marked increase (300%) in agreement between classifiers.

An initial ARI value of 0.28 increased to a final value of 0.84 (1 ¼ maximum agreement). In contrast, comparisons involving random picture groupings consistently rendered low ARI values (<0.05). The numbers of individuals named by the 2 users decreased from initial values of 46 and 43 to final values of 25 and 29, respectively.

The researchers sign off saying “tool presented here will help researchers and wildlife managers to identify individual mammals and monitor populations.”

I thank, Dr Eduardo Mendoza, for graciously sending me a copy of the paper