Polychlorinated biphenyl (PCB) contamination has historically posed constraints on recreational and commercial fishing industry in the Great Lakes. Empirical evidence suggests that PCB contamination represents a greater health risk from fish consumption than other legacy contaminants. The present study attempts a rigorous assessment of the spatio-temporal PCB trends in multiple species across the Canadian waters of the Great Lakes. We applied a Bayesian modelling framework, whereby we initially used dynamic linear models to delineate PCB levels and rates of change, while accounting for the role of fish length and lipid content as covariates. We then implemented Bayesian hierarchical modelling to evaluate the temporal PCB trends during the dreissenid pre- and post-invasion periods, as well as the variability among and within the water bodies of the Great Lakes system. Our analysis indicates that Lake Ontario is characterized by the highest PCB levels among nearly all of the fish species examined. Historically contaminated local areas, designated as Areas of Concern, and embayments receiving riverine inputs displayed higher concentrations within each of the water bodies examined. The general temporal trend across the Great Lakes was that the high PCB concentrations during the early 1970s followed a declining trajectory throughout the late 1980s/early 1990s, likely as a result of the reductions in industrial emissions and other management actions. Nonetheless, after the late 1990s/early 2000s, our analysis provided evidence of a decline in the rate at which PCB concentrations in fish were dropping, accompanied by a gradual establishment of species-specific, steady-state concentrations, around which there is considerable year-to-year variability. The overall trends indicate that reduced contaminant emissions have brought about distinct beneficial changes in fish PCB concentrations, but past historical contamination along with other external or internal stressors (e.g., invasive species, climate change) continue to modulate the current levels, thereby posing potential risks to humans through fish consumption.

从历史上看，多氯联苯（PCB）的污染对大湖区的休闲和商业捕鱼业构成了限制。经验证据表明，多氯联苯污染比其他传统污染物对鱼类的健康危害更大。本研究试图对加拿大大湖水域中多种物种的时空多氯联苯趋势进行严格评估。我们应用了贝叶斯建模框架，该框架最初使用动态线性模型来描述PCB水平和变化率，同时考虑了鱼的长度和脂质含量作为协变量的作用。然后，我们实施了贝叶斯（Bayesian）分层模型，以评估在入侵前和入侵后Dreissenid的时间PCB趋势，以及大湖系统水体之间和内部的可变性。我们的分析表明，安大略湖的特征是几乎所有所检查鱼类中的PCB含量最高。历史上受污染的本地区域，称为“关注区域”，接受河流输入的隔离物在每个被检查的水体中显示出较高的浓度。大湖区的总体时间趋势是，在1970年代初，PCB浓度很高，在整个1980年代后期/ 1990年代初期，轨迹呈下降趋势，这可能是由于工业排放量减少和其他管理措施的结果。但是，在1990年代后期/ 2000年代初之后，我们的分析提供了证据，表明鱼类中PCB浓度下降的速率有所下降，伴随着物种特定的稳态浓度的逐步建立，其周围存在相当大的逐年变化。总体趋势表明，减少的污染物排放已导致鱼类多氯联苯浓度发生明显的有益变化，但过去的历史性污染以及其他外部或内部压力源（例如，入侵物种，气候变化）继续影响当前水平，从而带来潜在风险通过食用鱼类给人类。