Context . Determining reliable estimates of carnivore population size and distributions are paramount for developing informed conservation and management plans. Traditionally, invasive sampling has been employed to monitor carnivores, but non-invasive sampling has the advantage of not needing to capture the animal and is generally less expensive. Faeces sampling is a common non-invasive sampling technique and future use is forecasted to increase due to the low costs and logistical ease of sampling, and more advanced techniques in landscape and conservation genetics. For many species, faeces sampling often occurs on or alongside roads. Despite the commonality of road-based faeces sampling, detectability issues are often not addressed.

Aim . We sought to test whether faeces detection probabilities varied by species – coyote (Canis latrans) versus kit fox (Vulpes macrotis) – and to test whether road characteristics influenced faeces detection probabilities.

Methods . We placed coyote and kit fox faeces along roads, quantified road characteristics, and then subsequently conducted ‘blind’ road-based faeces detection surveys in Utah during 2012 and 2013. Technicians that surveyed the faeces deposition transects had no knowledge of the locations of the placed faeces.

Key results . Faeces detection probabilities for kit foxes and coyotes were 45% and 74%, respectively; larger faeces originated from coyotes and were more readily detected. Misidentification of placed faeces was rare and did not differ by species. The width of survey roads and the composition of a road’s surface influenced detection probabilities.

Conclusion . We identified factors that can influence faeces detection probabilities. Not accounting for variable detection probabilities of different species or not accounting for or reducing road-based variables influencing faeces detection probabilities could hamper reliable counts of mammalian faeces, and could potentially reduce precision of population estimates derived from road-based faeces deposition surveys.

Implications . We recommend that wildlife researchers acknowledge and account for imperfect faeces detection probabilities during faecal sampling. Steps can be taken during study design to improve detection probabilities, and during the analysis phase to account for variable detection probabilities.