Wildlife density estimation by distance sampling: A novel technique with movement compensation.

Abstract

Estimates of population density are fundamental to wildlife conservation and management. Distance sampling from line transects is a widely used sample count method and is most often analysed using Distance software. However, this method has limited capabilities with mobile populations (e.g., birds), which tend to encounter an observer more often than immobile ones. This paper presents a novel distance sampling method based on a different set of models and assumptions, named WildlifeDensity after its associated software. It is based on mechanistic modelling of visual detections of individuals or groups according to radial distance from the observer or perpendicular distance from the transect line. It also compensates for population-observer relative movement to avoid the detection overestimates associated with highly mobile populations. The models are introduced in detail and then tested in three ways: 1) WildlifeDensity is applied to several 'benchmark' populations of known density and no-to-low mobility, 2) the movement compensation model is tested on two highly mobile songbird populations, and 3) a fairly difficult case is analysed: a low-density, highly mobile bird population in a forest habitat. The results show that 1) using either radial or perpendicular distance data from surveys of immobile populations, WildlifeDensity provides similar estimates (and errors) to Distance, with radial WildlifeDensity analysis appearing to be slightly better for surveys of low-mobility populations (kangaroos), 2) the movement compensation model effectively removes the correlation between observer speed and detection numbers, and 3) WildlifeDensity provides acceptable estimates where conventional Distance analysis overestimates density due to high movement. In summary, WildlifeDensity extends the capabilities of distance sampling by 1) compensating for movement, 2) not requiring complete detectability on the transect line and 3) supporting the use of radial distances, which simplifies fieldwork and increases measurement accuracy.