Incorporating environmental covariates into a Bayesian stock production model for the endangered Cumberland Sound beluga population

Abstract

The Cumberland Sound (CS) beluga (Delphinapterus leucas) population inhabits CS on eastern Baffin Island, Nunavut, Canada, and is listed as threatened under the Canadian Species at Risk Act. The population dynamics of CS beluga whales have been modelled in the past, but the effect of environmental covariates on these models has not previously been considered. An existing Bayesian population model fitted to CS beluga whale aerial survey data from 1990 to 2017 and harvest data from 1960 to 2017 was modified to include sea ice concentration (ICE) and sea surface temperature (SST). ICE and SST were extracted for all years from the CS study area in March and August, respectively, and incorporated into the state process component of the state-space model. The model resulted in a 2018 population estimate of 1245 (95% credible interval [CI] 564-2715) whales and an initial population estimate of 5147 (95% CI 1667-8779). Determining sustainable harvest by calculating the probability of population decline estimated 30% probability of decline after 10 yr with a harvest of ~15 whales annually. Compared to the previous model without environmental covariates, which followed a relatively linear trajectory, our model had more noticeable peaks and troughs in the population trend and wider CIs. The model suggested harvest levels be reduced by ~7 individuals for a management goal with a low risk of decline. The novelty of this approach for beluga whales provides an opportunity for further model development via the addition of various other abiotic and biotic variables related to beluga whale ecology.