Time‐varying effect of sea ice on the population dynamics of Steller sea lions ( Eumetopias jubatus) on Tuleny Island

• Published in: Ecological Research
• Main Authors: Mizuguchi, Daisuke, Kato, Keiko, Okamoto, Suguru, Hattori, Kaoru, Sakurai, Yasunori
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2019
• Subjects: Sea ice
• Online Access: http://dx.doi.org/10.1111/1440-1703.12043; https://onlinelibrary.wiley.com/doi/pdf/10.1111/1440-1703.12043; https://onlinelibrary.wiley.com/doi/full-xml/10.1111/1440-1703.12043

Abstract The population of Steller sea lions (SSLs) on Tuleny Island has been increasing continuously since the 1980s. However, the reasons for this increase have remained unknown. In this study, we investigated whether changes in sea ice distribution affected population dynamics on Tuleny Island by applying a Gompertz density‐dependent model in a state‐space framework with a time‐varying coefficient. We compared results from the time‐varying coefficient model with those from two other models that assumed time‐constant effects and no effect of sea ice. For both SSL pups and non‐pups, only the time‐varying coefficient model predicted a significant effect from the timing of sea ice breakup. The SSL population was larger when the sea ice broke up earlier. The sea ice effect was larger at the start of the study period and became insignificant in the late 1980s. Thus, our results suggest that the increase in SSL population on Tuleny Island during the first 20 years could be partly explained by earlier sea ice breakup, which promoted immigration from other rookeries. Because SSLs exhibit strong site‐fidelity as the habituation effect, individuals that visited once may have repeatedly returned to visit and breed on the island, regardless of the sea ice conditions in later years. This may explain why the sea ice effect was only significant in the first years of population increase. Negative density dependence was also only predicted by the time‐varying coefficient model. Adding the time‐varying coefficient to modeling analysis could simultaneously improve predictions of environmental effects and estimations of density dependence.