Des facteurs environnementaux et le comportement des bélugas de l'estuaire du Saint-Laurent génèrent de l'hétérogénéité dans les biais de disponibilités associés aux inventaires aériens photographiques et visuels

Abstract

In absence of adequate data, abundance estimates for St. Lawrence Estuary (SLE) beluga obtained from visual surveys have been corrected for availability bias using factors developed for photographic surveys. Not accounting for the longer detection time associated with visual surveys will lead to an overestimation of beluga abundance relative to indices obtained from photographic surveys. This study offers a comprehensive analysis of the relative influence of multiple methodological, environmental and behavioural factors on availability bias estimates for both photographic and visual surveys using detailed dive profiles from 27 SLE beluga. As expected, availability estimates were systematically higher for visual surveys than for photographic surveys for which time-in-view is instantaneous. However, for photographic surveys the change in methodology for estimating availability from an approach based on group visibility to one where the detailed diving patterns of individuals were logged, led to a 26—42% decrease in mean availability estimates. Our results confirmed that dives are longer when animals are inside compared to outside areas of high density (AHD), consistent with the prediction that these areas are used for behaviour like foraging. They also indicate that while some of the behavioural or environmental factors such as latent processes associated with the zone used may have a notable effect on availability, survey design (photographic or visual), characteristics of survey platforms, and observer searching patterns may be the most influential factors on availability bias. We conclude that previous estimates of SLE beluga abundance from photographic surveys were likely biased downward by an overestimation of beluga availability, and by not considering the uneven distribution of beluga among different zones with specific but undefined underlying processes.