Canadian Journal of Fisheries and Aquatic Sciences, Ahead of Print.
Moving toward ecosystem-based fisheries management requires integration of biotic and abiotic factors into our understanding of population dynamics. Using blue crab (Callinectes sapidus) in the Chesapeake Bay as a model system, we applied Gaussian Graphical Models (GGMs) to understand the influence of climatic, water quality, and biotic variables on estimates of key indices of blue crab recruitment for 1990–2017. Variables included the North Atlantic Oscillation (NAO), Susquehanna River discharge, wind forcing, hypoxic volume, submerged aquatic vegetation, and the catch per unit effort of striped bass (Morone saxatilis). Direct effects of age‐1+ crabs and summer salinity on recruitment were significant. Phase of the NAO in summer and spring, summer and winter discharge, and hypoxic volume indirectly affected the recruitment. A simulation study showed that GGM model selection achieved nominal coverage and outperformed structural equation modeling (SEM) and Multivariate Adaptive Regression Splines (MARS). GGMs have the potential to improve ecosystem-based management of blue crabs in Chesapeake Bay. Specifically, the approach can be used to examine ecosystem impacts on blue crab productivity and to improve forecasts of blue crab recruitment.

中文翻译：

---

使用经验高斯图形模型对切萨皮克湾蓝蟹（Callinectes sapidus）的生态系统影响进行评估

《加拿大渔业和水生科学杂志》，印刷前。
向基于生态系统的渔业管理迈进需要将生物和非生物因素整合到我们对种群动态的理解中。以切萨皮克湾的蓝蟹（Callinectes sapidus）为模型系统，我们运用高斯图形模型（GGM）来了解气候，水质和生物变量对1990-2017年蓝蟹招募关键指标的估计的影响。变量包括北大西洋涛动（NAO），萨斯奎哈纳河流量，强迫风，低氧量，淹没的水生植物以及条带鲈（Morone saxatilis）的单位捕获量。1岁以上的螃蟹和夏季盐度对募集的直接影响是显着的。夏季和春季的NAO阶段，夏季和冬季的分泌物以及缺氧量间接影响了募集。仿真研究表明，GMG模型的选择可实现名义覆盖率，并且性能优于结构方程模型（SEM）和多元自适应回归样条（MARS）。GGMs有潜力改善切萨皮克湾蓝蟹的基于生态系统的管理。具体而言，该方法可用于检查生态系统对蓝蟹生产力的影响并改善对蓝蟹招募的预测。