Modeling moose sightability in South-Central British Columbia

We developed a model to correct sightability bias in aerial surveys of moose in the southern interior of British Columbia. Sightability trials were conducted by searching sample blocks where radio-collared moose were known to occur. Relevant attributes that were believed to affect the sightability of moose were recorded and used as independent variables for a logistic regression model of sightability. Univariate analysis revealed that percent vegetative cover, percent snow cover, and daily temperature all significantly influenced the probability of detecting a moose. However, multivariate analysis retained only vegetative cover as the significant variable to predict sightability probability. This mirrors the results of analysis of similar trials done in western Wyoming that used the same independent variables as well as 6 additional ones. Two logistic regression models were developed for moose sightability; one based on 5 classes of vegetative cover and a second, less accurate, based on 3 classes of vegetative cover.