Long‐Term Studies of Fish Health before and after the Closure of a Bleached Kraft Pulp Mill in Northern Ontario, Canada

Abstract

Changes in ecosystems after the removal of stress provide a rich source of information for conservation science. We used a long-term regional data set on the performance of white sucker (Catostomus commersonii) collected before and after the closure of a pulp mill to explore recovery in fish. Physiological indicators, including liver enzymes and plasma steroids, showed some compelling changes after the closure of the mill consistent with reduced exposure to pulp mill effluent but did not unequivocally demonstrate recovery. However, persistent signals in these fish may indicate effects of impoundment or discharge of sewage. We also used quantile regression with environmental covariates and bootstrap iteration to determine if systematic variation remained in relative body weight, liver weight, and gonad weight. In fish formerly exposed to pulp mill effluent, we found evidence of improvements (male gonad weight and liver weight of males and females), degradation (gonad weight of females), and no change (body weight). Although the observed patterns may be associated with closure of the mill, some differences were also found at regional locations, suggesting roles of additional stressors and challenging the clear association of change at the Mattagami River exposure site with the closure of the mill. However, fish captured at this location show responses consistent with regional locations, suggesting no residual impacts and highlights the challenges of identifying changes in fish even after large and known interventions.

Potential Influence of Sewage Phosphorus and Wet and Dry Deposition Detected in Fish Collected in the Athabasca River North of Fort McMurray

“…Sample sizes are limited in some years, especially in males, but other challenges also require additional consideration in evaluating a potential relationship between precipitation and the health of fish. Given the challenges of including environmental covariates in a retrospective analysis where no new data can be collected [61], some of the variations we observed in fish could be misattributed to precipitation or are properly attributed to precipitation, but through a mechanism not yet discussed. For example, a weak but nonzero relationship between precipitation and discharge (Supplementary Figure S1) suggests a small effect of discharge downstream of the Fort McMurray gauge is embedded in Pr.…”

“…Similar to discharge, the temperature was used here as a natural stressor that can affect the performance of fish populations [43,86,87]. Characterizing the precise nature of the relationships between wild fish and variables, such as discharge and temperature in a retrospective analysis is, however, often challenging [61]. Similar to Pr and D, T is likely a complex descriptor of variation and may be influenced by industrial practices.…”

“…The specific procedure as implemented here is described in detail elsewhere [61], but briefly, the number of resamples (m) was set to the sample size of the sex, site, measurement, and test year being considered. During the first bootstrap, X* was generated by drawing m samples with replacement from the empirical distribution X.…”

“…The rule-set for ascribing the potential occurrence of a large difference was also modified to account for the small sample sizes. We used the occurrence of the observed test medians outside the 2.5 or 97.5 percentile proper from the expected ranges of the median residual to detect potentially relevant differences used in earlier work [61]. In contrast to the earlier approaches [61], we did not use the variability of the observed medians to ascribe relevance.…”

“…We used the occurrence of the observed test medians outside the 2.5 or 97.5 percentile proper from the expected ranges of the median residual to detect potentially relevant differences used in earlier work [61]. In contrast to the earlier approaches [61], we did not use the variability of the observed medians to ascribe relevance. Our approach here inflates the false positive rate but is suitable for our purposes of identifying potentially relevant differences, to identify patterns supporting the postulated linkages between local stressors and the health of fish in the Athabasca River, and for informing the methods for both the study design and the analysis of future monitoring data.…”