Abstract
Over the last 70 years, seabird populations have declined by up to 70%, and represent the most endangered group of birds globally. When compared to other seabird species, there is little known about the Snow Petrel (Pagodroma nivea) and its marine habitat use-especially in East Antarctica. To better understand what drives Snow Petrel presence within this region, we modeled vessel-based observations of the Snow Petrel against remotely sensed environmental data using binomial generalized additive models (GAMs). Throughout the 16-year study period (1991-2006), Snow Petrel presence was associated with areas exhibiting shallower bathymetry, increasing sea-ice coverage, decreasing sea-surface height, and increasing wind speed. We then used a subset of the Snow Petrel data to generate a population density map and compare model outputs when data recording methods differ. Specifically, we tested how outputs change when inputs are binomial (presence/absence) versus when inputs include count and effort data. The outputs from both effort-quantified and presence/absence models identified similar environmental drivers of Snow Petrel presence. However, the effort-quantified models were more robust, yielding higher deviance explained values and narrower confidence intervals around the environmental variables associated with Snow Petrel presence. Snow Petrel interactions with the tested environmental variables may be driven by associated biological processes-specifically, foraging strategies that target niche areas of high biological productivity in the Southern Ocean. Our study provides an important baseline to compare the likely future distribution of Snow Petrels under different climate change scenarios.