Classifying carnivore tracks using dimensions that control for snow conditions

Snow-tracking is important for elucidating patterns of carnivore behavior, but misclassifying tracks reduces the accuracy of snow-tracking studies. Quantitative methods improve accuracy by distinguishing between similar tracks left by different carnivores. American marten (Martes americana) and fisher (Pekania pennanti) tracks are difficult to distinguish. We studied martens and fishers in northern Wisconsin, USA, during winter 2008–2010, to determine whether dimensions of tracks left in snow differed by snow conditions, and if marten and fisher tracks could be accurately classified by analyzing track dimensions that controlled for snow conditions. Snow depth, snow compaction, and crust depth correlated strongly with fisher step depth. Classification trees accurately classified marten and fisher tracks, and were 5–14% more accurate when track dimensions controlled for snow conditions. Species-only classifications were 91–96% accurate. Trees that classified sex and species were 75–89% accurate, indicating that snow-tracking can be used to estimate sex-specific marten and fisher habitat selection, distribution, and abundance. Controlling for snow conditions improves track classification accuracy for martens and fishers, and would likely improve classification accuracy for other carnivores. © 2017 The Wildlife Society.