Bias in carnivore diet analysis resulting from misclassification of predator scats based on field identification

Abstract

ABSTRACTDiet studies are frequently used to improve understanding of predator ecology, potential effects of carnivores on prey populations, and competition among predators. However, field identification of carnivore scat typically relies on scat morphology, size, and contents resulting in possible subjective predator identification and potentially biased results. Advancements in noninvasive genetic sampling allow for molecular identification of predator scat, eliminating many issues associated with field identification methods. We collected scat samples once per month from June 2011 to May 2012 in western Virginia, USA, using morphological characteristics for field identification of the predator. We then used mitochondrial DNA to identify the predator species of each scat and identified prey remains visually. Using confusion matrices, we found a range of accuracy in field identification for the 3 target species: coyotes (Canis latrans; 54.0%), bobcats (Lynx rufus; 57.1%), and black bears (Ursus americanus; 95.2%), even though we only considered samples with high‐confidence field identification. We found a high coyote false‐positive rate (52.7%), indicating we often incorrectly identified scats as coyote (98% of misidentified bobcat scats and 75% of misidentified black bear scats were recorded as coyote in the field). This asymmetrical bias in predator identification resulted in inaccurate estimates of dietary niche breadth and overlap between competitors. Our results suggest that caution should be exercised when interpreting results from studies in which carnivore species are identified by scat morphology. Future studies should employ noninvasive genetic sampling for carnivore scat identification, especially in areas with sympatric predator species that have similar scat morphology. © 2016 The Wildlife Society.