Reliability of trans‐generational genetic mark–recapture (tGMR) for enumerating Pacific salmon

• Published in: Evolutionary applications 2024-02, Vol.17 (2), p.e13647-n/a
• Main Authors: Rosenbaum, Samuel W., May, Samuel A., Shedd, Kyle R., Cunningham, Curry J., Peterson, Randy L., Elliot, Brian W., McPhee, Megan V.
• Format: Article
• Language: English
• Subjects: abundance estimation; Animal reproduction; Bias; Breeding success; close‐kin mark–recapture; dropout; Enumeration; Estimates; Fisheries management; Habitats; individual‐based model; Migration; Pacific salmon; Population; Population studies; Rivers; Salmon; Sampling; Simulation; Spawning; Special Issue; Success; tGMR; trans‐generational genetic mark–recapture; Violations
• ISSN: 1752-4571; 1752-4571
• DOI: 10.1111/eva.13647

As Pacific salmon (Oncorhynchus spp.) decline across much of their range, it is imperative to further develop minimally invasive tools to quantify population abundance. One such advancement, trans‐generational genetic mark–recapture (tGMR), uses parentage analysis to estimate the size of wild populations. Our study examined the precision and accuracy of tGMR through a comparison to a traditional mark–recapture estimate for Chilkat River Chinook salmon (O. tshawytscha) in Southeast Alaska. We examined how adult sampling location and timing impact tGMR by comparing estimates derived using samples collected in the lower river mainstem to those using samples obtained in upriver spawning tributaries. Results indicated that tGMR estimates using a representative sample of mainstem adults were most concordant with, and 3% more precise than, the traditional mark–recapture estimate for this stock. Importantly, the timing and location of adult sampling were found to impact abundance estimates, depending on what proportion of the population dies or moves to unsampled areas between downriver and upriver sampling events. Additionally, we identified potential sources of bias in tGMR arising from violations of key assumptions using a novel individual‐based modeling framework, parameterized with empirical values from the Chilkat River. Simulations demonstrated that increased reproductive success and sampling selectivity of older, larger individuals, introduced negative bias into tGMR estimates. Our individual‐based model offers a customizable and accessible method to identify and quantify these biases in tGMR applications (https://github.com/swrosenbaum/tGMR_simulations). We underscore the critical role of system‐specific sampling design considerations in ensuring the precision and accuracy of tGMR projects. This study validates tGMR as a potentially useful tool for improved population enumeration in semelparous species.