A species distribution model of the Antarctic minke whale (Balaenoptera bonaerensis)

• Published in: Theriologia Ukrainica
• Main Author: Volodymyr Tytar
• Format: Article in Journal/Newspaper
• Language: English; Ukrainian
• Published: National Academy of Sciences of Ukraine. National Museum of Natural History 2022
• Subjects: balaenoptera bonaerensis; southern ocean; species distribution modelling; bayesian additive regression trees; biological indicator; climate change; Zoology; QL1-991; Antarctic; Southern Ocean; The Antarctic; Austral; Antarc*; Antarctic Minke whale; Balaenoptera bonaerensis; minke whale; Sea ice
• Online Access: https://doi.org/10.15407/TU2309; https://doaj.org/article/c0c766cf9a4b4a2ba00cbaca19670ece

The Antarctic minke whale (Balaenoptera bonaerensis) is regarded a Southern Hemisphere endemic found throughout the Southern Hemisphere, generally south of 60°S in austral summer. Here they have been routinely observed in highest densities adjacent to and inside the sea ice edge, and where they feed predominantly on krill. Detecting abundance trends regarding this species by employing visual monitoring is problematic. Partly this is because the whales are frequently sighted within sea ice where navigational safety concerns prevent ships from surveying. In this respect species-habitat models are increasingly recognized as valuable tools to predict the probability of cetacean presence, relative abundance or density throughout an area of interest and to gain insight into the ecological processes affecting these patterns. The objective of this study was to provide this background information for the above research needs and in a broader context use species distribution models (SDMs) to establish a current habitat suitability description for the species and to identify the main environmental covariates related to its distribution. We used filtered 464 occurrences to generate the SDMs. We selected eight predictor variables with reduced collinearity for constructing the models: mean annuals of the surface temperature (°C), salinity (PSS), current velocity (m/s), sea ice concentration (fraction, %), chlorophyll-a concentration (mg/m3), primary productivity (g/m3/day), cloud cover (%), and bathymetry (m). Six modelling algorithms were tested and the Bayesian additive regression trees (BART) model demonstrated the best performance. Based on variable importance, those that best explained the environmental requirements of the species were sea ice concentration, chlorophyll-a concentration and topography of the sea floor (bathymetry), explaining in sum around 62% of the variance. Using the BART model, habitat preferences have been interpreted from patterns in partial dependence plots. Areas where the AMW have particularly ...