Affordable Wildlife Monitoring. A New Approach to Line Transects Sampling From Vehicles

Monitoring is essential for evidence-based wildlife conservation and management. Conventional distance sampling (CDS) represents a methodology of election for population assessment of large herbivores. CDS requires that (1) animals' distribution is uniform around the transects and (2) transects must be randomly distributed over the study area. Monitoring costs are usually lower by using cars moving along dirty roads, instead of walking randomly located transects, but this choice may introduce biases in the estimate, as ungulates may avoid roads, which in their turn are not randomly distributed across the landscape. To address both problems, we used bivariate distance sampling (collecting both forward and perpendicular distances) to estimate detection probability, thus correcting for road avoidance. The resulting detection function is used as input for Density Surface Models to correct for non-random line placement. We demonstrate this methodology by considering a pilot survey of impala (Aepyceros melampus) and common duiker (Sylvicapra grimmia) in the Sandwe GMA (Zambia). Potentially, this approach can mitigate biases and increase the precision of estimates. We discuss the possibility of applying the proposed methodology for routine wildlife monitoring in underfunded areas, in Africa and elsewhere. To assist practitioners, we provide an easy-to-use R script which implements statistical procedures.