MEPS 263:275-285 (2003)  -  doi:10.3354/meps263275

ABSTRACT: Predictive models were constructed based on at-sea observations of 6 species of seabirds within the Prydz Bay region of East Antarctica. A spatial logistic model was used, and predictions were based on physical environmental data. The hypothesis that these models could predict observations of the same species in the disjunct Vincennes Bay region of East Antarctica was tested. The best results were obtained for the 2 endemic Antarctic breeding species examined (Antarctic and snow petrels). The accuracies of the predictions for Wilson¹s storm petrels and Cape petrels (which breed both in the Antarctic and on subantarctic islands) were low due to their broad distributions at sea. The predictions for 2 species that breed in subantarctic and temperate localities (white-chinned petrels and prion spp.) were poorest, reflecting the limited seasonal use of Antarctic waters for foraging by these species. The approach is applicable for estimating the probability of observing seabird species in unsurveyed areas of the Southern Ocean, but is likely to be limited to those species breeding regionally, and not for species foraging in the area and breeding elsewhere.

KEY WORDS: Seabirds at sea · Southern Ocean · Spatial model · Predictive · Environmental determinants