A method to estimate pre‐exploitation population size

Abstract

Antarctic fur seals (Arctocephalus gazella) were commercially exploited on the subantarctic island of South Georgia for over 100 years and nearly driven to extinction. Since the cessation of harvesting, however, their populations have rebounded, and they are now often considered a nuisance species whose impact on the terrestrial landscape should be mitigated. Any evaluation of their current population requires the context provided by their historic, pre-exploitation abundance, lest ecologists fall prey to shifting baseline syndrome in which their perspective on current abundance is compared only with an altered state resulting from past anthropogenic disturbance. Estimating pre-exploitation abundance is critical to defining species recovery and setting recovery targets, both of which are needed for the International Union for the Conservation of Nature's recent efforts to develop a green list of recovering species. To address this issue, we reconstructed the South Georgia fur seal harvest from 1786 to 1908 from ship logbooks and other historical records and interpolated missing harvest data as necessary with a generalized linear model fit to the historical record. Using an approximate Bayesian computation framework, harvest data, and a stochastic age-structured population model, we estimated the pre-exploitation abundance of Antarctic fur seals on South Georgia was 2.5 million females (95% CI 1.5-3.5 million). This estimate is similar to recent abundance estimates, and suggests current populations, and the ecological consequences of so many fur seals on the island, may be similar to conditions prior to human harvest. Although the historic archive on the fur sealing era is unavoidably patchy, the use of archival records is essential for reconstructing the past and, correspondingly, to understanding the present. Article impact statement: Defining species recovery requires an understanding of baseline population state, which can be estimated through statistical methods.