Variability in a crabeater seal population and the marine ecosystem near the Antarctic Peninsula

A population of crabeater seals, Lobodon carcinophagus, inhabiting the sea ice and waters surrounding the Antarctic Peninsula was studied to clarify the nature of variability in the population age structure. Understanding this variability is important for efforts to identify effects of climate change and marine resource exploitation in the Southern Ocean. The fundamental question is first addressed, of whether the variability is a genuine demographic phenomenon or an artifact of sampling or analysis. Then the evidence is examined for support of an interpretation of periodicity. Because age-estimation errors are known to reduce variability in estimates of cohort strength, and because crabeater seal ages are subject to estimation errors, the effects of those errors are evaluated. Features of the crabeater seal natural history were used to generate hypotheses about expected correlations between the cohort strengths and sea ice extent, surface air temperature, the Southern Oscillation, and an index of leopard seal (Hydrurga leptonyx) populations.;The primary data for this study are age estimates from 2,852 crabeater seals collected near the Antarctic Peninsula between 1964 and 1990. Assuming that the seals were collected at random with respect to age, these "catch-at-age" data were analyzed by a maximum likelihood technique, yielding a time series of relative cohort strengths for the 1945-1988 cohorts and a new life table for crabeater seals. Bootstrap and Monte Carlo techniques were used to assess the uncertainty of the estimates and the power of the sampling scheme to detect true fluctuations in cohort strengths. It was found that if the assumptions are correct, the relative cohort strengths are well determined by the data and therefore unlikely to be artifacts of sampling or analysis. Time series modeling of the data, however, indicated that there is little support for interpreting the fluctuations in cohort strength as periodic. Statistical modeling of the age estimation error process yielded a series of cohort strengths with more variability, as expected, but with features that may reflect invalid assumptions about the age estimates for a portion of the sample. Time series of sea ice extent, surface air temperature, a measure of the Southern Oscillation, and numbers of leopard seals sighted annually at Macquarie Island were unable to explain a significant fraction of the crabeater seal cohort variability. (Abstract shortened by UMI.).