We constructed a stochastic model of Northwest Atlantic grey seal population dynamics and fit it to available pup production data from 1977-2010 divided into three breeding regions: Sable Island, Gulf of St. Lawrence, and Coastal Nova Scotia (including Hay Island and other small colonies along the coast of Nova Scotia). The model and fitting methods are the same as those used in a previous population assessment (Thomas et al. 2007), updated with 2010 pup production data and revised estimates of historical harvest. The model assumes that fecundity rates are age-dependent but are constant over time, that adult survival rates are constant, and that pup survival is density dependent. Females are assumed to be able to move to a new region to breed if pup survival is higher there, but once they start breeding they do not move. We used a Bayesian computer-intensive method (particle filtering) to fit the model, with informative priors on model parameters.
The posterior estimates for some parameters were close to their priors, indicating little information about these parameters in the pup production data and highlighting the importance of carefully choosing the priors. Other parameters were far from the prior: in particular the posterior estimates of carrying capacity were far higher than the prior values, indicating little evidence of density dependent population regulation at current levels of pup production. The total estimated population size at the end of the 2010 breeding season (i.e., including pups) was 348,900 (95% CI 291,300-414,900). This is 4% higher than the equivalent estimate for 2009 of 335,200 (95%CI 292,000-395,100) and 975% higher than the estimate for 1977 of 35,800 (95%CI 24,700-53,100). Average annual rates of population increase are estimated to be 6% in the 1980s, 9% in the 1990s and 6% in the 2000s. These estimates should be treated with some caution because: (1) the biological model showed clear lack of fit, particularly to the Gulf data where extending the model to account for ice and weather conditions would be useful; (2) sensitivity of the results to the priors used has not been assessed; and (3) the fitting algorithm may have caused some (small) biases.
View complete PDF document
(27 pages; 309K)
This document is available in PDF format. If the following document is not accessible to you, please contact the Secretariat to obtain another appropriate format, such as regular print, large print, Braille or audio version.
This report uses scientific and technical terms and is published in the official language of the working group or scientific expert that produced the document. If this document is not accessible to you in the official language of your choice, please contact the Secretariat.