Many industries are required to monitor themselves in meeting regulatory policies intended to protect the environment. Self‐reporting of environmental performance can place the cost of monitoring on companies rather than taxpayers, but there are obvious risks of bias, often addressed through external audits or inspections. Surprisingly, there have been relatively few empirical analyses of bias in industry self‐reported data. Here, we test for bias in reporting of environmental compliance data using a unique data set from Canadian salmon farms, where companies monitor the number of parasitic sea lice on fish in open sea pens, in order to minimize impacts on wild fish in surrounding waters. We fit a hierarchical population‐dynamics model to these sea‐louse count data using a Bayesian approach. We found that the industry's monthly counts of two sea‐louse species, Caligus clemensi and Lepeophtheirus salmonis, increased by a factor of 1.95 (95% credible interval: 1.57, 2.42) and 1.18 (1.06, 1.31), respectively, in months when counts were audited by the federal fisheries department. Consequently, industry sea‐louse counts are less likely to trigger costly but mandated delousing treatments intended to avoid sea‐louse epidemics in wild juvenile salmon. These results highlight the potential for combining external audits of industry self‐reported data with analyses of their reporting to maintain compliance with regulations, achieve intended conservation goals, and build public confidence in the process.

中文翻译：

---

鲑鱼养殖业自我报告的寄生虫数据存在偏差。

许多行业需要自我监控，以达到旨在保护环境的法规政策。对环境绩效的自我报告可能会给公司而不是纳税人带来监督成本，但存在明显的偏差风险，通常通过外部审计或检查来解决。令人惊讶的是，行业自我报告数据中对偏差的经验分析相对较少。在这里，我们使用来自加拿大鲑鱼养殖场的独特数据集测试报告环境合规性数据时是否存在偏见，在该数据集中，公司在公海圈中监视鱼上寄生海虱的数量，以最大程度地减少对周围水域中野生鱼类的影响。我们使用贝叶斯方法将分层的人口动力学模型拟合到这些海虱计数数据。我们发现该行业在经过联邦渔业部门审核的月份中，Calgus clemensi和Lepeophtheirus鲑鱼分别增加了1.95倍（95％可信区间：1.57、2.42）和1.18倍（1.06、1.31）。因此，行业海虱计数不太可能触发代价高昂但强制性的去虱处理，目的是避免野生少年鲑鱼海虱流行。这些结果凸显了将行业自我报告数据的外部审计与报告分析相结合的潜力，以保持对法规的遵守，实现预期的保护目标并在此过程中建立公众信心。