Probabilistic Forecasts of Fish Abundances With Spatio‐Temporal Models to Support Fisheries Management

Details

• Journal Title:
  Fish and Fisheries
• Personal Author:
  Grüss, Arnaud ; Allyn, Andrew J. ; Thorson, James T. ; Mills, Katherine E. ; Yasumiishi, Ellen ; Pinkerton, Matthew H. ; Parsons, Darren M. ; Sibanda, Nokuthaba
• NOAA Program & Office:
  NMFS (National Marine Fisheries Service) ; AFSC (Alaska Fisheries Science Center)
• Description:
  Spatio-temporal species distribution models (SDMs) are valuable tools to support fisheries management, as they account for long-term and time-varying unmeasured variation (spatial and spatio-temporal variation), thereby providing more accurate and statistically efficient estimates than simpler SDMs. However, the application of spatio-temporal SDMs for probabilistic forecasts of fish abundances has been slowed by two main challenges. First, guidance surrounding the decisions needed to generate forecasts with a spatio-temporal SDM is lacking. Second, there is limited functionality to efficiently produce forecasts with spatio-temporal SDMs while also propagating uncertainty about initial conditions and process errors. We developed new approaches to forecasting with spatio-temporal SDMs, which allow for efficient predictions with spatio-temporal SDMs and quantifying the influence of different model components on prediction uncertainty. We illustrate our approaches with two contrasting applications: west coast New Zealand snapper (Chrysophrys auratus), using fisheries data and examining retrospective forecasts; and Bering Sea capelin (Mallotus villosus), employing fisheries-independent data and generating forecasts to 2100. The applications showed that spatio-temporal variation should be included in spatio-temporal SDMs to produce forecasts, to explain a much larger fraction of the variability in the data, thereby providing more accurate reconstructions of population trends and better characterising uncertainty around forecasts. Our results also highlight that forecasts with spatio-temporal SDMs work best in data-rich situations, particularly if the time series of fish data is relatively long. Our approaches can help unlock the use of spatio-temporal SDMs to make both near-term and long-term forecasts, providing better information to fisheries managers and informing future data collection.
• Source:
  Fish and Fisheries, 26(6), 1180-1197
• DOI:
  https://doi.org/10.1111/faf.70025
• ISSN:
  1467-2960 ; 1467-2979
• Format:
  pdf
• Publisher:
  Wiley
• Document Type:
  Journal Article
• License:
  https://creativecommons.org/licenses/by/4.0/
• Rights Information:
  CC BY
• Compliance:
  Submitted
• Main Document Checksum:
  urn:sha-512:91f3224c1d64aafcb016b3d43cc2e7214d95f743e86f49a7a07c127cbb4760762a89b63c03355e8ba4e85bf3c2c949c1a6fe48716b448f68c999c8f623596a53
• Download URL:
  https://repository.library.noaa.gov/view/noaa/72358/noaa_72358_DS1.pdf
• File Type:
  [PDF - 2.27 MB ]