Peer Reviewed A Comparison of Noninvasive Techniques to Survey Carnivore Communities in Northeastern North America MATTHEW E. GOMPPER, 1 Department of Fisheries and Wildlife Sciences, University of Missouri, Columbia, MO 65211-7240, USA ROLAND W. KAYS, New York State Museum, Albany, NY 12230, USA JUSTINA C. RAY, Wildlife Conservation Society Canada, Toronto, ON M5S 2T9, Canada SCOTT D. LAPOINT, State University of New York, College of Environmental Science and Forestry, Syracuse, NY 13210, USA DANIEL A. BOGAN, Department of Natural Resources, Cornell University, Ithaca, NY 14853, USA JASON R. CRYAN, New York State Museum, Albany, NY 12230, USA Abstract Carnivores are difficult to survey due, in large part, to their relative rarity across the landscape and wariness toward humans. Several noninvasive methods may aid in overcoming these difficulties, but there has been little discussion of the relative merits and biases of these techniques. We assess the value of 5 noninvasive techniques based on results from 2 multiyear studies of carnivores (including members of Carnivora and Didelphidae) in New York forests. Two metrics were particularly valuable in assessing the species-specific value of any particular survey technique: latency to initial detection (LTD) and probability of detection (POD). We found differences in the value of techniques in detecting different species. For midsized species (raccoon [Procyon lotor], fisher [Martes pennanti], opossum [Didelphis virginiana], and domestic cat [Felis catus]), camera traps and track- plates were approximately equivalent in detection efficiency, but the potential for wariness toward the survey apparatus resulted in higher LTD for track-plates than for cameras. On the other hand, track-plates detected small carnivores (marten [M. americana] and weasels [Mustela spp.]) more often than cameras and had higher PODs for small and midsized species than did cameras. Cameras were efficient mechanisms for surveying bears (Ursus americanus; low LTD, high POD) but functioned poorly for discerning presence of coyotes (Canis latrans; high LTD, low POD). Scat surveys and snowtracking were the best methods for coyotes, which avoided camera traps and artificial tracking surfaces. Our analysis of fecal DNA revealed that trail-based fecal surveys were inefficient at detecting species other than coyotes, with the possible exception of red foxes (Vulpes vulpes). Genetic analyses of feces and snowtracking revealed the presence of foxes at sites where other techniques failed to discern these species, suggesting that cameras and track-plates are inefficient for surveying small canids in this region. The LTD of coyotes by camera traps was not correlated with their abundance as indexed by scat counts, but for other species this metric may offer an opportunity to assess relative abundance across sites. Snowtracking surveys were particularly robust (high POD) for detecting species active in winter and may be more effective than both cameras and track-plates where conditions are suitable. We recommend that survey efforts targeting multiple members of the carnivore community use multiple independent techniques and incorporate mechanisms to truth their relative value. (WILDLIFE SOCIETY BULLETIN 34(4):1142–1151; 2006) Key words camera traps, carnivores, fecal DNA, latency to detection, New York, noninvasive survey, probability of detection, snowtracking, track-plates. Species in the mammalian order Carnivora receive a great deal of conservation attention, much of which is due to a charismatic and conflict-ridden image that draws attention from diverse segments of society, as well as their importance as furbearers (Gittleman et al. 2001, Ray et al. 2005). But there also is an interest in these species because of their potential to fundamentally influence communities and ecosystems in ways that are disproportional to their biomass in the system (Estes et al. 1998, Post et al. 1999, Terborgh et al. 1999, Ray et al. 2005). It is increasingly clear that even midsized and small predators may be fundamental drivers of ecosystem processes despite their relative rarity across landscapes (Je ˛drzejewska and Je ˛drzejewski 1998, Korpimaki and Norrdahl 1998, Gittleman and Gompper 2005). As a result an increasing number of studies are attempting to assess presence or absence, relative abundance, and interac- tion of carnivore species across the globe. Techniques to study the ecology of carnivores typically are invasive; that is, they require capture and handling of individual animals. This intensive work usually is impractical for studies addressing questions over larger geographic scales and also may be inappropriate because of local norms and regulations, costs and logistics, low potential to capture the target species, or risk to the target animal. Yet, because carnivores range over large areas and frequently interact with one another (Palomares and Caro 1999), biologists and wildlife managers are increasingly recognizing the need for large-scale studies of entire carnivore communities. Thus, there has been a push to develop noninvasive survey techniques that can be deployed over large areas and detect multiple species (e.g., Zielinski and Kucera 1995). Five of the most commonly used of these noninvasive techniques are camera traps, covered track-plates, scent stations, snowtrack- ing, and scat surveys. The latter technique is increasingly paired with DNA-based analyses to enhance accuracy. 1 E-mail: gompperm@missouri.edu 1142 Wildlife Society Bulletin  34(4)