Attributing variability in fish demographic processes to environmental conditions is helpful when assessing population status and forecasting changes in ecosystem function. Empirical orthogonal function (EOF) analysis has long been used to explore variability in physical processes, but has been only recently employed to study variability in biological processes. EOF analysis estimates dominant modes of variability (indices) and produces maps representing the spatial response for the dependent variable to each of these indices. In the eastern Bering Sea (EBS), research has linked demographic processes to the spatial extent of bottom temperatures less than or equal to 2 °C (the “cold-pool extent” or “CPE”), but has generally not compared effects among different demographic processes. We applied EOF analysis to four types of data measuring the outcome of demographic processes for EBS walleye pollock (Gadus chalcogrammus) over the period 1982–2019: numerical density (outcome of movement), morphometric condition (outcome of bioenergetics), length-at-age (outcome of growth), and prey-biomass-per-predator-mass (a proxy for stomach contents; outcome of consumption). We first designed exploratory factor analysis (EFA) models that did not include a CPE effect. We then applied confirmatory factor analysis (CFA), which differed from EFA by attributing observed patterns to a spatially varying response of demographic processes to CPE. We inferred that CPE was a proxy for demographic variability when there was a strong correlation between (1) the first or second mode of variability in the EFA and CPE or (2) the spatial map associated with the positive phase of the first or second mode of variability from the EFA model and the spatially varying response of CPE from the CFA model. Results showed that prey-biomass-per-predator-mass had the strongest correlation with CPE, numerical density and morphometric condition were also strongly correlated with CPE, and length-at-age was moderately correlated with CPE. The models also identified several anomalous years: 1999 and 2010, which were characterized by a very large CPE and high indices for variables related to demographic processes; and 2016–2019, which were characterized by a small CPE and low indices for variables related to demographic processes. We conclude that demographic processes for EBS walleye pollock show the finger-print of bottom-up environmental variation. Future research can employ CPE projections to forecast spatio-temporal changes in variables related to demographic processes, thereby informing estimates such as weight-at-age that are used in stock assessment models.

在评估种群状况和预测生态系统功能的变化时，将鱼类人口统计过程的可变性归因于环境条件是有帮助的。经验正交函数（EOF）分析长期以来一直用于探索物理过程中的变异性，但是直到最近才被用于研究生物过程中的变异性。EOF分析估计变异的主要模式（指数），并生成表示因变量对这些指数中每个指数的空间响应的图。在白令海东部（EBS），研究已将人口统计过程与底部温度小于或等于2°C的空间范围（“冷池范围”或“ CPE”）相关联，但通常没有将这些影响进行比较不同的人口统计过程。甘草）在1982年至2019年期间：数字密度（运动结果），形态条件（生物能学结果），年龄长度（生长结果）和每个生物的捕食生物量（代表胃）内容；消费结果）。我们首先设计了探索性因素分析（EFA）模型，该模型不包含CPE效应。然后，我们应用了确认因素分析（CFA），该分析与EFA有所不同，其原因在于将观察到的模式归因于人口统计学过程对CPE的空间变化响应。我们推断，当（1）EFA中的第一或第二种模式的变异性与CPE或（2）与第一或第二种模式的正相相关的空间图之间存在很强的相关性时，CPE是人口统计学变异性的代理。 EFA模型的可变性以及CFA模型的CPE的空间变化响应。结果表明，每个生物的捕食生物量与CPE的相关性最强，数字密度和形态条件也与CPE的相关性很强，年龄长度与CPE的相关性中等。这些模型还确定了几个异常年份：1999年和2010年，其特征是CPE很大，并且与人口统计过程有关的变量的指数很高；和2016–2019年，其特点是CPE较小，与人口统计过程相关的变量的索引较低。我们得出的结论是，EBS角膜白斑鳕的人口统计过程显示出自下而上的环境变化的指纹。未来的研究可以使用CPE预测来预测与人口统计过程相关的变量的时空变化，从而提供诸如存货评估模型中使用的年龄权重之类的估计值。