Linking camera‐trap data to taxonomy: Identifying photographs of morphologically similar chipmunks

Abstract

Remote cameras are a common method for surveying wildlife and recently have been promoted for implementing large-scale regional biodiversity monitoring programs.The use of camera-trap data depends on the correct identification of animals captured in the photographs, yet misidentification rates can be high, especially when morphologically similar species co-occur, and this can lead to faulty inferences and hinder conservation efforts. Correct identification is dependent on diagnosable taxonomic characters, photograph quality, and the experience and training of the observer. However, keys rooted in taxonomy are rarely used for the identification of camera-trap images and error rates are rarely assessed, even when morphologically similar species are present in the study area. We tested a method for ensuring high identification accuracy using two sympatric and morphologically similar chipmunk (Neotamias) species as a case study. We hypothesized that the identification accuracy would improve with use of the identification key and with observer training, resulting in higher levels of observer confidence and higher levels of agreement among observers. We developed an identification key and tested identification accuracy based on photographs of verified museum specimens. Our results supported predictions for each of these hypotheses. In addition, we validated the method in the field by comparing remote-camera data with live-trapping data. We recommend use of these methods to evaluate error rates and to exclude ambiguous records in camera-trap datasets. We urge that ensuring correct and scientifically defensible species identifications is incumbent on researchers and should be incorporated into the camera-trap workflow.

Towards the fully automated monitoring of ecological communities

“…The previous examples of fully automated monitoring of multidimensional data from multispecies systems pose the question of whether such frameworks could be developed for almost any other ecosystem. A first limitation is obviously the requirement for large and properly labelled training datasets when implementing accurate and reliable deep learning classifiers, preventing the monitoring of communities for which such data is not available (McKibben & Frey, 2021 ). Second, the environmental complexity of natural systems could hamper such designs, generating data with a very low signal‐to‐noise ratio in comparison with experimental systems.…”

Misinterpretation of Genomic Data Matters for Endangered Species Listing: The Sub-specific Status of the Peñasco Least Chipmunk (Neotamias minimus atristriatus)

“…The geographic origin of N. m. atristriatus, coupled with morphological or molecular characters, would allow for recognition of this subspecies. For instance, even photographs taken with remote cameras allow for accurate diagnosis of N. m. atristriatus when paired with information about location (McKibben and Frey, 2021).…”

To model or not to model: false positive detection error in camera surveys

“…While Miller et al (2012) found that training did little to reduce false positive errors in auditory surveys for anurans, training improves identification accuracy for other species and survey methods (Thornton et al 2019, McKibben and Frey 2021). We reiterate the recommendation made by McKibben and Frey (2021) that researchers using camera data should train observers, test the influence of training on identification accuracy for the study species, and test the identification accuracy of individual observers. While Miller et al (2012) encouraged trained observers to not record uncertain detections during auditory surveys to reduce false positive detections, observer‐reported confidence ranks can facilitate the inclusion or exclusion of ambiguous detections or could be used as covariates for modeling false positive error.…”