Using parentage-based tagging to estimate survival of Chinook salmon fry in a large storage reservoir

• Published in: Environmental biology of fishes 2024-07, Vol.107 (7), p.735-754
• Main Authors: Pope, Adam C., Kock, Tobias J., Perry, Russell W., Cogliati, Karen M., O’Malley, Kathleen G., Murphy, Christina A., Hance, Dalton J., Fielding, Scott D.
• Format: Article
• Language: English
• Subjects: Animal Systematics/Taxonomy/Biogeography; Biomedical and Life Sciences; Creeks & streams; Demography; Design; Design parameters; Environment; Estimates; Field study; Fish; Fish hatcheries; Fisheries; Freshwater & Marine Ecology; Freshwater fishes; Fry; Juveniles; Life Sciences; Marking; Migration; Nature Conservation; Oncorhynchus tshawytscha; Parameter estimation; Parameter identification; Parameter robustness; Population studies; Populations; Probability; Reservoir storage; Reservoirs; Robust design; Salmon; Sampling; Sampling designs; Storage reservoirs; Survival; Tracking; Wildlife conservation; Zoology
• ISSN: 0378-1909; 1573-5133
• DOI: 10.1007/s10641-024-01564-9

Research efforts focusing on salmonid populations have highlighted the need to better understand demographic parameters for the fry and parr life stages. Monitoring these small fish presents a challenge because negative effects from handling and tagging can bias subsequent parameter estimates. Removal models and associated sampling designs represent one class of mark-recapture models with potential to be applied to very small juvenile salmon, yet existing methods associated with removal studies are not well-suited for all study environments. For example, populations residing in large storage reservoirs may yield low capture probabilities when subjected to removal sampling, making unbiased estimation of survival using traditional removal models difficult. To address this limitation, we developed a sampling design and associated model using parentage-based tagging in hatchery-raised juvenile Chinook salmon ( Oncorhynchus tshawytscha ) to estimate survival over a 2-year study period in a large storage reservoir in western Oregon, USA. Individual fish were identified to family groups, serving as replicate batch marks in a robust design removal model framework. Results from a simulation suggested that parameter estimates were unbiased even at very low capture probabilities, although the use of model constraints (i.e., covariates or constant parameter values) was necessary to achieve this. Model fitting to field data supported a trend in survival over time, with survival increasing with time since release in the first study year but decreasing in the second.