Predicting and Correcting Helicopter Counts of Moose with Observations made from Fixed-Wing Aircraft in Southern Quebec

Abstract

(1) During winter moose (Alces alces L.) aerial surveys, a crew aboard a fixed-wing aircraft flew plots initially to delineate track networks visible in the snow; immediately following, a crew in a helicopter made counts of animals associated with each track network. (2) Two models that predict helicopter counts of moose, based on observations made only from fixed-wing aircraft, were computed by using eighty-five 60-km2 plots covered in 1980-84 by both types of aircraft. The crude model used the overall ratio aeroplane counts : helicopter counts (0.29) to predict helicopter counts. The elaborate model was composed of three linear regressions that used aeroplane counts and percentage area covered by track networks to predict helicopter counts; the regression differed when aeroplane counts equalled 0, 1-2, or more. (3) The elaborate model produced better predictions than the crude model: using cross-validation for seven recent aerial surveys, mean helicopter counts predicted by the elaborate model departed at worst by 23% from observed mean counts. (4) Visibility bias was estimated with the help of radio-collared moose; 88% (n = 59) of the animals occupied track networks visible from fixed-wing aircraft while 83% of moose within track networks were observed during helicopter searches. Therefore, 73% of the moose were seen during early winter helicopter counts over mixed forests of Quebec. (5) All recent surveys produced imprecise density estimates because of inadequate sample size and poor stratification. Double sampling with regression, using elaborate models, is proposed in order to improve precision of estimates without increasing costs.