Evaluating scat surveys as a tool for population and community assessments

Abstract

Context: Reliable, cost-effective monitoring methods are essential for managing wildlife populations. Scat-and-sign surveys (i.e. monitoring defecation, animal scratching, footprints, food remains) are a rapid, low-cost, non-invasive monitoring approach, but unquantified biases and uncertainties associated with these methods have caused researchers to doubt their reliability. Aims: We sought to quantify how richness, occupancy and activity estimates derived from a long-term camera-trap study differed from those of scat surveys in the same locations, to determine scat-survey reliability and model bias corrections. Methods: We used transect-based scat surveys at 110 sites in the temperate forests of southern Tasmania (Australia), to estimate occupancy, activity and community richness for common, ground-dwelling vertebrates. These results were compared with estimates derived from a long-term passive camera-trap study at the same sites. In addition, time-lapse imagery taken with the camera traps was used to monitor the persistence of rufous-belled pademelon (Thylogale billardierii) and Tasmanian devil (Sarcophilus harrisii) scats in relation to environmental correlates. Key results: Scat persistence differed between these two species. The half-life of S. harrisii scats was 113 days, compared with 63 days for T. billardierii. Generalised linear modelling showed that scat surveys were most efficacious at sites with little disturbance and homogenous substrates. Overall, scat surveys consistently underestimated site occupancy and richness relative to the camera traps (μ = 2.7:1), but this bias was inconsistent, with the ratio exceeding 15 for the arboreal brushtail possum (Trichosurus vulpecula). Scats were most reliably detected for large, trail-using mammals such as S. harrisii, T. billardierii, and common wombat (Vombatus ursinus). Scat surveys were less useful for the surveillance of low-density and arboreal species. Scats were uncommon for the two bird species examined, but alternative superb lyrebird (Menura novaehollandiae) signs were detected reliably. Conclusions: Scat surveys reliably detected large, trail-using mammals. However, estimates of activity were poorly correlated between camera traps and scat surveys. Implications: When used appropriately, scat surveys can provide an effective and cheap 'snapshot' index for wildlife monitoring, especially if the species-specific biases have been calibrated for the vertebrate community and environment under monitoring.