Geolocation of a demersal fish (Pacific cod) in a high-latitude island chain (Aleutian Islands, Alaska)

Abstract

Abstract Background Fish geolocation methods are most effective when they are customized to account for species behavior and study area characteristics. Here, we provide an example of customizing a hidden Markov model (HMM) for reconstructing movement pathways of a high-latitude demersal fish species in a remote island chain using Pop-up Satellite Archival Tag (PSAT) data. Adult Pacific cod were tagged with PSATs while occupying winter spawning grounds in the Aleutian Islands in February 2019. We adapted a demersal fish application of the HMM to (1) add light-based longitude to the data likelihood model, (2) account for possible off-bottom behavior of demersal fishes in the maximum daily depth likelihood, and (3) modify the model framework to accommodate convoluted island topography in the study area. A simulation study was conducted to explore the two primary modifications to the model framework, reflecting boundary for the movement kernel and the Viterbi method of pathway reconstruction, under known conditions. Results Geolocation was performed on satellite-transmitted and detailed archival data sets from 6 adult Pacific cod at liberty for 21–277 days. Migration from winter spawning to summer foraging areas (range 60–395 km) was detected for the 4 tagged fish that were at liberty for more than 90 days. Light-based longitude was the primary geolocation variable for detecting migrations with precision (root mean square error) estimates of 0.56 degrees during winter and 1.3 degrees during the summer. Simulation studies confirmed the effectiveness of model framework modifications and generated guidelines for use in specific applications. Conclusions This study demonstrates that post-spawning migrations of Pacific cod in the Aleutian Islands can be detected and characterized using PSAT data. Initial insights into migrations, summer foraging areas, and associated development of appropriate analysis tools will support future Pacific cod movement studies in the Aleutian Islands as well as other regions of Alaska. The adaptations to the HMM presented here will benefit current and future research on demersal fish in other regions as well as fish species that occupy areas with convoluted shorelines or island chain topography.

Keywords

• Geolocation
• Hidden Markov model
• Fish migration
• Satellite tags
• Pop-up Satellite Archival Tags
• Viterbi algorithm