期刊
FISHERIES OCEANOGRAPHY
卷 29, 期 4, 页码 349-365出版社
WILEY
DOI: 10.1111/fog.12476
关键词
American lobster; front; Northeast U; S; Shelf; random forest; secondary productivity; suitable habitat; temperature
American lobster (Homarus americanus) supports one of the most valuable regional fisheries in the United States, with its abundance and distribution profoundly influenced by environmental conditions. To explain how lobster distribution has changed over time and assess the role of environmental variables on these changes, we used random forest classification and regression tree models to estimate occupancy and biomass in two seasonal periods. The occupancy models were fit to static and dynamic variables, which yielded model fits with AUC scores of 0.80 and 0.78 for spring and fall, respectively. Biomass models were fit with the same data and resulted in models explaining 61% and 63% of the spring and fall biomass variance, respectively. Significant variables scored in the formation of the regression trees were secondary productivity (i.e., zooplankton), bathymetry characteristics, and temperature. American lobster suitable habitat has changed regionally; habitat has increased in the Gulf of Maine and declined in Southern New England. There is also evidence of declining habitat along the inshore margin of the Gulf of Maine, which has been accompanied by a shift in occupancy probability offshore. Habitat suitability results from the random forest models provide insights on the structure and function of lobster habitat and context to understand recent population trends.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据