Evaluating facial recognition for photographic-mark-recapture of four species of northern ungulates

• Author / Creator

  Ness, Isobel, FG

• Estimating animal abundance is a key component of wildlife management and mark-recapture surveys are one of the most commonly used methods of obtaining population estimates. Photographic identification has recently been explored as a method of ‘marking’ individuals for mark-recapture surveys. It is most often used on species with spots, stripes of other unique markings however, in a few cases, species that do not have such obvious marks have been successfully identified using morphological measurements. There is a need for the development of non-invasive, affordable and accurate methods of censusing wide-ranging and elusive northern populations of ungulates. In this study, I tested a likelihood-based photographic identification method on 4 species of ungulates: muskox (Ovibos moschatos), Dall sheep (Ovis dalli), mountain goats (Oreamnos americanos) and mule deer (Odocoileus hemionus). False-rejection (FRR) and false-acceptance (FAR) error rates were identified for each species. These measures varied widely among species (i.e., FRR = 0–13%, FAR = 2–22%). Matching success was also determined for each species and ranged from 48% to 96%. Muskox (FRR=0% and FAR =3%) and sheep (FRR=4% and FAR=2%) had the lowest misidentification rates and the highest matching success rates (96% and 88%, respectively). Moderate results were obtained for goats (FRR=11% and FAR=6%) and deer (FRR=11% and FAR=0%) with matching success rates of 81% and 80%, respectively. An automated matching success rate was calculated based on the top-ranked photograph for each potential match and was compared to the observer matching success rate. The observer matching success rate was significantly higher for all analyses (t9=7.2, p<0.05), indicating that the final subjective user choice step of the method was important. An observer bias test was conducted for deer and sheep and significant observer bias was found for deer, and affected matching success rates for both species pointing to the importance of observer training and/or experience in the species of interest and the use of the program. This study provides a proof-of-concept for the use of photogrammetric identification on sheep and muskox and lays the groundwork for future capture-mark-recapture studies on wild populations by establishing misidentification rates and determining the effect of observer bias. Future work should address the stability of horn measurements over time and the feasibility of capturing useable photographs from remote cameras.

• Subjects / Keywords

  • photographic identification
  • ungulates
  • photographic mark-recapture
  • facial recognition

• Graduation date

  Fall 2019

• Type of Item

  Thesis

• Degree

  Master of Science

• DOI

  https://doi.org/10.7939/r3-z1vq-gz55

• License

  Permission is hereby granted to the University of Alberta Libraries to reproduce single copies of this thesis and to lend or sell such copies for private, scholarly or scientific research purposes only. Where the thesis is converted to, or otherwise made available in digital form, the University of Alberta will advise potential users of the thesis of these terms. The author reserves all other publication and other rights in association with the copyright in the thesis and, except as herein before provided, neither the thesis nor any substantial portion thereof may be printed or otherwise reproduced in any material form whatsoever without the author's prior written permission.