Antarctic sea ice: measuring habitat complexity, and seasonal and regional variability in habitat use for minke whales
M.I. Garcia, D. Thiele, E.T. Chester, K. Asmus
Minke whales, Balaenoptera bonaerensis, are found in association with sea ice in the Antarctic year round, often occurring hundreds of kilometres into the pack. The sea ice is a dynamic and complex region of the Antarctic marine ecosystem in both physical and biological terms that provides habitat for many species. Cetacean surveys are often conducted on Antarctic vessels that record sea ice data, but few of these have incorporated standardised sea ice data collection protocols for simultaneous collection. None have attempted to determine the extent to which sea ice can be categorized in an ecologically meaningful way for cetacean species, particularly how the patchiness of their distribution in ice relates to the heterogeneity of the ice landscape. Sea ice physicists use a standard shipboard data collection system around the Antarctic that measures complexity in sea ice structure (ASPeCt sea ice program). Here, we have coupled the ASPeCT system with cetacean surveys to investigate the potential for classification of minke whale habitat types in sea ice and prediction of minke whale distribution in sea ice. Over three years (2004/05/06) cetacean sighting surveys were conducted during autumn, spring and summer with sampling occurring across three ‘regions’ (East Antarctica, Weddell and Ross Sea). Sea ice data were collected simultaneously every 10 minutes: high resolution digital still images of sea ice and ASPeCt sea ice data fields. Sea ice complexity was tested across regions and seasons, with summer showing the greatest complexity and variability. We investigated the relationship between minke whale distribution and sea ice characteristics by fuzzy coding the sea ice classifications for a subset of the data, and then using multiple correspondence and discriminant analysis to explain variability at a high temporal sampling resolution. Marginal frequencies for fuzzy-coded ice variable modalities were examined to characterize ice in each data set: marginal frequencies for records where minkes were sighted indicated potential habitat. Ordination through multiple correspondence analysis explained variability among the fuzzy-coded records according to the influence of the original ice variables: complexity was high with samples at a ten minute temporal resolution, and relatively many ordination axes were required to explain > 50% of variance in the data. Using canonical functions from axis scores to discriminate ice records associated with minke whale presence, significant potential for classifying minke habitat was found in summer, but not in other seasons. In summer, the sea ice is more complex and variable than in other seasons, due to melting processes. Our results show that it is possible to predict minke whale distribution in summer, as whales are significantly associated with particular combinations of sea ice ‘habitat’ at that time. During autumn and spring, the sea ice is more homogenous and can not be used as a predictor for minke whale distribution. These patterns are likely due to prey association with summer ice melt processes that do not occur in other seasons at a scale we can measure. Fine scale surveys may provide further insight into linkages with prey aggregation and movements and potentially provide further insight into habitat use across all seasons.