Scat happens and why wildlife managers need to be aware: Quantifying observer uncertainty in predator dietary analysis
Effective wildlife management requires the collection of accurate data by repeatable and precise methods. Without such data, management actions may be misguided. Collecting informative field data is complicated by cryptic or rare species for which data collection typically relies on indirect methods such as track counts, hair collection, or scat surveys. While these methods provide valuable information, they are prone to observer uncertainty which may result in misidentification of sign and contribute bias to the outcomes of analyses. We use a case study of dietary overlap between dingoes (a native apex predator) and feral cats (an invasive mesopredator) in Western Australia to explore the consequences of uncertainty in scat identification when used to inform predator management decisions. Control of either or both feral cats and dingoes has been proposed to protect declining native mammals in the ‘critical weight range’ (35 to 5500g). Over two years, 708 dingo and feral cat scats were collected and analysed for prey content. Scats were identified in the field and again in the lab, with each observer ranking their confidence in identification as positive, probable, or uncertain. Seventy-two scats had conflicting identifications between observers and a subset (12) were analysed for identification by DNA.
The extent of dietary overlap between dingoes and feral cats varied markedly depending on observer. Field-identified scats showed a 17.2% overlap in diet whereas lab- identified scats showed a dietary overlap of 36.7%. A multivariate analysis of scat contents revealed that 77.9% of the variability in diets could be explained along two axes yet considerable variability in contents occurred between identification classes for each observer. We attempted to resolve uncertainty in scat identifications using a combination of observer-ranked confidence estimates and DNA test results. In doing so, the diets of dingoes and feral cats could be tightly characterised, and little overlap between the two species was found. Based on these results we conclude that dingoes pose little threat to critical weight range mammals in our study area, and we recommend that predator control efforts focus on feral cats whose impacts are more severe, particularly for small mammals. Our study provides an effective, repeatable approach for dealing with measurement uncertainty in scat collection studies. Given the widespread use of scat identification and dietary analysis for wildlife management, we recommend that practitioners and researchers account for uncertainties such as these to avoid potentially inappropriate and poorly targeted management actions.