Linking ecological processes and animal movements to inform timing of long‐term surveys of a migratory game bird

Abstract

Managers typically estimate wildlife abundance using surveys within a timeframe that favors increased detectability; however, the ability to account for probabilities of inclusion, detection, and/or presence within a given sampling area is often limited. Cranes provide a good opportunity to research count accuracy because they are large, conspicuous, and often congregate during part of the year, typically on staging areas (i.e., fall and spring) or on wintering grounds. The objectives of this paper are twofold: (1) to evaluate how environmental factors influence crane movement in and out of crane survey areas to identify the best window of availability for annual survey counts; and (2) to evaluate environmental factors that influence overall crane survey counts from year to year. For Objective 1, a generalized linear mixed model was selected to model the probability of crane presence within survey areas using GPS transmitter data. A binary response variable representing crane movement within and outside formal survey areas was used as the dependent variable to investigate environmental covariates that influence movement into survey areas. Probability of crane presence was explained by seven covariates plus a quadratic term for Julian day number. Interactions between Palmer drought severity index (PDSI) and normalized difference vegetation index supported higher probability of crane presence in survey areas during times of drought. Probability of crane presence increased throughout the entire study period (May–October), suggesting that formal surveys in September could be augmented or replaced by surveys in October. For Objective 2, a negative binomial model with linear parameterization was selected to model crane counts using census data compiled from 1995 to 2019. Covariates were acquired at the watershed scale using Hydrologic Unit Code 6 boundaries. Of the 17 covariates investigated, we found that 18‐month precipitation (PPTgss), PDSI, and minimum temperature (Tempmin) explained most of the variability in crane census counts. High PPTgss (antecedent moisture), low PDSI (drought conditions), and low Tempmin (cold extremes) result in higher annual crane counts. The ability to link ecological processes to wildlife movement and population abundance both locally and at landscape scales has long‐ranging implications on resource projections, conservation, and the ability to deploy adaptive management.