Semi-automated identification of individual big brown bats via collagen–elastin patterns in the wing membrane

Collagen-elastin (CE) bundle patterns in the wing membrane have been used to identify individual bats; however, this method has not been widely adopted, likely owing to the laborious nature of manually comparing wing images through visual inspection. We tested the effectiveness of using an accessible, feature-based, pattern-recognition software-HotSpotter-to identify individuals using patterns of CE bundles in the bat wing. We collected photos from 24 adult (<i>n </i>= 192 photos) big brown bats (<i>Eptesicus fuscus</i>) and their direct offspring (<i>n </i>= 34 pups; <i>n </i>= 136 photos) by illuminating the ventral surface of the wing with ultraviolet light. Upon running a query match comparison on a selected reference image, HotSpotter ranks every other photo in the database based on an assigned similarity score. We found that HotSpotter correctly presented the top-ranked image as another image of the same individual at higher<i>-</i>than<i>-</i>chance performance. The software also performed better than chance when considering matches to images with the same age (adult/juvenile), sex (male/female), wing side (left/right), and known-relatedness (mother-offspring or twin) to the bat in the queried image. The proportion of correct matches increased with the number of top-ranked images included in the initial query. These results are encouraging because they suggest that pattern-recognition software has the potential to automate recognition of bats based on CE bundle patterns in photos of bat wings. With further refinements in the technology, we think it may be possible to achieve nearly 100% accuracy of individual identification.