Evaluating the reliability of field identification and morphometric classifications for carnivore scats confirmed with genetic analysis

摘要: Scat surveys are commonly used to monitor carnivore populations. Scats of sympatric carnivores can be difficult differentiate and field-based identification misleading. We evaluated the success species for scats 2 carnivores—coyotes (Canis latrans) kit foxes (Vulpes macrotis). conducted scat in Great Basin desert Utah, USA, during winter summer 2013, we detected 1,680 scats. classified based on field identification, recorded morphometric measurements, collected fecal DNA samples molecular identification. subsequently classification predictive power nonparametric techniques—k-nearest neighbors trees—based measurements. Overall, 12.2% were misclassified by but misclassifications not equitable between species. Only 7.1% identified as coyote with misclassified, compared 22.9% fox. Results from both k-nearest neighbor classification-tree analyses suggest that measurements provided an objective alternative improved rarer Overall misclassification rates 11.7% 7.5%, respectively. Using trees, was reduced (8.5%) remained similar coyotes (7.2%), relative Although techniques provide unambiguous approaches may offer a cost-effective alternative. recommend monitoring programs employing utilize develop training data sets evaluate accuracy statistical approaches. © 2015 The Wildlife Society.