Climate impacts on fished populations. Part 2: Effects of climate and environmental variability on fishery stock assessment accuracy

Fish stocks are influenced through direct physiological effects from changes in their marine habitat, such as temperature, dissolved oxygen, and acidity. At the same time, indirect bottom-up and top-down effects of these changing environmental factors, including impacts on food resources and predation, may also determine productivity changes. In combination, these direct and indirect influences may result in productivity changes in local fish stocks, particularly if the latter are unable to shift spatially to avoid environmental changes. In turn, these changes in productivity interact with fisheries, which also affect stock productivity through plastic, density-dependent effects, including increased growth of fish at reduced densities.

This study used a model of individual eco-physiological response to environmental and climate factors to derive population level outcomes of fish stocks. It was also used to investigate how fisheries stock assessments are influenced by climate and bottom-up variability in production parameters.

The outcome from this investigation showed that, on average, the assessments provided unbiased estimates of stock status even though there were annual and decadal fluctuations in all production-related parameters. Nevertheless, a considerable spread in assessed stock status was not readily explained by production changes or the mis-specification of production parameters. We hypothesise that this unexplained variation in assessment error is predominantly related to information content of stock-specific data as opposed to environmentally-determined parameter variability per se.

Beyond this basic variability, there were persistent and predictable biases introduced by directional (e.g., climate change related) changes in temperature and bottom-up constraints. Nevertheless, the assessment error in the presence of interannual and decadal variability only (i.e., without directional trends) in production was considerably greater than variability induced by climate scenarios considered here. On the basis of these scenarios, it is suggested that assessments focus on structural and model uncertainties such as the exploration and evaluation of time-varying parameters to obtain more realistic estimates of model uncertainty, and more relevant management quantities. These improvements in assessment methodologies and practice could inform and facilitate more responsive management in variable environments, and mitigate risks from over-confidence in model estimates, yet provide opportunity by closely tracking environmental variability.