Assessing changes in indicators of fish health measured between 1997 and 2019 relative to multiple natural and anthropogenic stressors in Canada’s oil sands region using spatio-temporal modeling

Abstract

Industrial development in Canada’s oil sands region influences the ambient environment. Some of these influences, such as the atmospheric deposition of emitted particles and gases are well-established using chemical indicators, but the effects of this process on bioindicators examined in field studies are less well-supported. This study used an extensive dataset available from 1997 to 2019, spatio-temporal modeling (Integrated Nested Laplace Approximation), and data on industrial and non-industrial covariates, including deposition patterns estimated using HYSPLIT (Hybrid Single Particle Lagrangian Integrated Trajectory) to determine if changes in sentinel fishes collected in streams from Canada’s Oil Sands Region were associated with oil sands industrial activity. While accounting for background variables (e.g., precipitation), estimated deposition of particles emitted from mine fleets (e.g., Aurora North), in situ stacks (e.g., Primrose and Cold Lake), mine stacks (e.g., Kearl), mine dust (e.g., Horizon), road dust (e.g., Muskeg River mine), land disturbance in hydrologically-connected areas, and wildfires were all associated with at least one fish endpoint. While many individual industrial stressors were identified, a specific example in this analysis parallels other work: the potential influence of emissions from both Suncor’s powerhouse and dust emitted from Suncor’s petroleum coke pile may both negatively affect fish health. Comparisons of fitted values from models with the estimated industrial effects and with deposition rates set to zero suggested some negative (and persistent) influences of atmospheric deposition at some locations, such as the gonadosomatic index (GSI) in the lower Muskeg and Steepbank rivers. While there is evidence of some large differences at individual locations the mean GSI and body condition estimates have improved throughout the region since the beginning of these collections in the late 1990s potentially highlighting improved environmental performance at the facilities, widespread enrichment effects, or interactions of stressors. However, mean liver-somatic indices have also slightly increased but remain low. These results, coupled with others suggest the utility of spatio-temporal approaches to detect the influence and effects of oil sands development at both local and regional scales.