Sex-Specific Bias in Helicopter Surveys of Elk: Sightability and Dispersion Effects

Abstract

Sex-ratio data for adult elk (Cervus elaphus) collected from helicopters may be biased if males and females have different sighting probabilities. Sightability models can account for differential sightability if they are sex-specific or if sightability functions are the same for both sexes. I collected sightability data from a sample of radiocollared elk in Washington that included a large proportion of adult males to test 2 hypotheses about sightability: (1) male and female elk had equal sighting probabilities, and (2) the functional relationship between factors affecting sightability and the probability of sighting groups was the same for both sexes. I also tested the hypothesis that the spatial distribution of male and female elk was the same during experimental surveys. Using logistic regression modeling, 1 found that females were >9 times more likely to be seen than males during surveys. A multivariate logistic model containing the variables group size and overstory canopy closure explained sighting probabilities as well as models with more parameters. Sex was related (P < 0.001) to group sightability in univariate analyses, but the effect of sex was nonsignificant (P = 0.77) when group size and canopy closure were present in a multivariate model, indicating sex operated as a confounder but not an effect modifier. The spatial overlap of adult male and female elk in blocks of aerial sampling units was 67% during the sightability trials. The distribution of male and female elk among these blocks was related to sex (P≤0.05) and bivariate relocations from males and females were spatially distinguishable (P<0.001), indicating that males and females were spatially segregated. These results suggest that sightability models need not be sex-specific to reduce bias in sex-ratio data collected from helicopters. However, the spatial distribution of male and female elk during winter is a potential source of bias in helicopter surveys designed to yield accurate sex-ratio data if survey area boundaries are defined by the distribution of high-sightability groups, which are likely to be groups of females.