Category Archives: Fisheries Think Tank
Using Machine Learning with Fisheries Data
Using Machine Learning with Fisheries Data
Caitlin Allen Akselrud; Alaska Fisheries Science Center, NOAA
ABSTRACT TBD
When do environmental drivers improve advice from stock assessment models?
ABSTRACT TBD
Modelling Selectivity in Fisheries Stock Assessment
ABSTRACT TBD
Dynamic structural equation models and tinyVAST: expressive models for time-series and spatio-temporal synthesis
ABSTRACT TBD
Adapting monitoring to a changing seascape: efficiency, flexibility and continuity for bottom trawl surveys
Fisheries research surveys provide important information on the state of marine populations and ecosystems over time. Survey data are critical inputs to stock assessment, ecosystem-based fishery management initiatives, and applied ecological research. However, environmental changes may affect system stationarity, species distribution, and sampling effectiveness, so it can impact the consistency and trustfulness of abundance estimates time series computed from survey data. Therefore, it is essential to design flexible surveys that can adapt to climatic variability and budget limitations while keeping high-quality time series to effectively manage marine resources. To address these issues, we investigated multiple sampling survey designs in the Eastern Bering Sea because of the ongoing and rapid environmental change that is causing species distribution shifts to deeper and northern regions in the system. We used the eastern Bering Sea groundfish bottom trawl survey which is a systematic fishery-independent survey with a fixed station design and carried out since 1982. We conducted a multispecies survey design optimization using a spatiotemporal operating model and a genetic algorithm that optimizes the minimal optimal allocation of samples without compromising the precision of abundance estimates. Results revealed that reducing and redistributing samples and expanding the sampling range can provide accurate abundance estimates of marine populations under changing environmental conditions and budget limitations. This simulation study provided a framework that may help to increase the efficiency of sampling marine resources to maintain the quality of such time series of abundance estimates for the management of marine populations.
Modelling the impacts of Ocean Acidification on North Pacific crabs – integrating experiments and biology (or ten years of integrating experiments, population ecology and economics)
Management strategy evaluation of sardine harvest control rules under climate change
Climate-driven changes in ocean temperatures, currents, or plankton dynamics may disrupt pelagic forage fish recruitment. Being responsive to such impacts enables fisheries management to ensure continued sustainable harvest of forage species. We conducted a management strategy evaluation to assess the robustness of current and alternative Pacific sardine harvest control rules under a variety of recruitment scenarios representing potential projections of future climate conditions in the California Current. The current environmentally-informed control rule modifies the harvest rate for the northern sardine subpopulation based on average sea surface temperatures measured during California Cooperative Oceanic Fisheries Investigations (CalCOFI) field cruises. This rule prioritizes catch at intermediate biomass levels but may increase variability in catch and closure frequency compared to alternative control rules, especially if recruitment is unrelated to ocean temperatures. Fishing at maximum sustainable yield and using dynamically estimated reference points reduced the frequency of biomass falling below 150,000 mt by up to 17%, while using survey index-based biomass estimates resulted in a 14% higher risk of delayed fishery closure during stock declines than when using assessment-based estimates.
Hands-on demonstration for phylogenetic comparative methods using R-package phylosem: improving stock assessment (natural mortality) and ecosystem modelling (consumption over biomass)
stock assessment (natural mortality) and ecosystem modelling (consumption over biomass)
Comparative study across species are foundational to many topics in fisheries science, including to justify biological reference point proxies, predict natural mortality rates, or parameterize ecosystem models. Surprisingly, however, fisheries science has largely missed the research interest in phylogenetic comparative methods (PCM), which were pioneered by Joe Felsenstein at University of Washington in
the 1970s and remain a well-published topic in ecology and evolutionary biology. This is unfortunate, because PCM offers many improvements over conventional nested-taxonomic methods used in fisheries science.
In this workshop, I provide a hands-on introduction to the R-package phylosem available on CRAN. Attendees are encouraged to bring a laptop with package-install privileges, or pre-install phylosem. I will walk through vignettes which show how phylosem generalizes existing packages for phylogenetic linear models (phylolm), structural equation models (sem), phylogenetic trait imputation (Rphylopars), and phylogenetic path analysis (phylopath).
I will then provide a quick overview of results using two fisheries case studies. The first applies phylosem to the Then et al. natural mortality database, where phylosem reduces variance for out-of-sample prediction of natural mortality rates. The second replicates Palomeres and Pauly (1998), and confirms that consumption over biomass (a parameter in Ecopath with Ecosim models) is elevated for small-bodied with a high caudal-fin aspect ratio in warmer waters. I hope to demonstrate that PCM is broadly applicable for routine use throughout fisheries science.