Scholarlink

Background <p>Bycatch of Chinook salmon (<i>Oncorhynchus tshawytscha</i>) in the Gulf of Alaska walleye pollock (<i>Gadus chalcogrammus</i>) fishery poses both conservation and management challenges. While fleet communication and excluder devices have improved mitigation efforts, effective reduction ultimately depends on anticipating overlap—especially near the seafloor where pollock are targeted. To support context-specific bycatch mitigation strategies, we developed a probabilistic deep learning model that predicts Chinook salmon depth occupancy using fisheries-independent telemetry data and environmental, temporal, and spatial covariates.</p> Results <p>The best-performing model incorporated season, diel period, and salinity as key predictors of Chinook salmon depth occupancy in the Gulf of Alaska. Seasonal and diel effects were observed to represent a repeating annual cycle, with strong diel patterns appearing in late summer and fall. Salinity was also predictive in some months with shallower depth occupancy corresponding to lower surface salinity. Using the model, spatial and temporal patterns in the likelihood of Chinook salmon occupying depths targeted by trawl gear were mapped, revealing seasonal and diel hotspots of overlap.</p> Conclusions <p>The model reveals a dynamic landscape of depth occupancy for Chinook salmon in the Gulf of Alaska. The predictions so derived provide a foundation for bycatch mitigation strategies that account for both behavioral rhythms and environmental variability, underscoring the role such models could have in advancing dynamic ocean management.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A depth model for informing Chinook salmon bycatch risk in the Gulf of Alaska

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月5日

查看全文
  • Marcel Gietzmann-Sanders,
  • Michael Courtney,
  • Andrew Seitz,
  • Curry J. Cunningham

摘要

Background

Bycatch of Chinook salmon (Oncorhynchus tshawytscha) in the Gulf of Alaska walleye pollock (Gadus chalcogrammus) fishery poses both conservation and management challenges. While fleet communication and excluder devices have improved mitigation efforts, effective reduction ultimately depends on anticipating overlap—especially near the seafloor where pollock are targeted. To support context-specific bycatch mitigation strategies, we developed a probabilistic deep learning model that predicts Chinook salmon depth occupancy using fisheries-independent telemetry data and environmental, temporal, and spatial covariates.

Results

The best-performing model incorporated season, diel period, and salinity as key predictors of Chinook salmon depth occupancy in the Gulf of Alaska. Seasonal and diel effects were observed to represent a repeating annual cycle, with strong diel patterns appearing in late summer and fall. Salinity was also predictive in some months with shallower depth occupancy corresponding to lower surface salinity. Using the model, spatial and temporal patterns in the likelihood of Chinook salmon occupying depths targeted by trawl gear were mapped, revealing seasonal and diel hotspots of overlap.

Conclusions

The model reveals a dynamic landscape of depth occupancy for Chinook salmon in the Gulf of Alaska. The predictions so derived provide a foundation for bycatch mitigation strategies that account for both behavioral rhythms and environmental variability, underscoring the role such models could have in advancing dynamic ocean management.