Empirical validation of grandparent inference to quantify reproduction of hatchery-origin fish in the wild

Abstract Objective We sought to validate grandparentage analysis as a method to quantify the reproduction of hatchery-origin steelhead Oncorhynchus mykiss and fall Chinook Salmon O. tshawytscha in the wild by identifying natural-origin individuals with hatchery broodstock grandparents for each species in the Snake River basin. Methods Known steelhead and fall Chinook Salmon grandparent–grandchild trios were genotyped with amplicon sequencing panels, and grandparent relationships were inferred from the genotypes (381 and 1,185 loci, respectively, for each species). False-positive and false-negative error rates were calculated by comparing inferred relationships to the known pedigree. We then demonstrated application of the technique by performing grandparentage analysis on steelhead and fall Chinook Salmon samples that were systematically collected at Lower Granite Dam on the Snake River, Washington, during spawning migrations. The grandparent assignments were integrated into an escapement estimation process to determine the number of natural-origin fish with zero, one, or two hatchery-origin parents and to estimate the effective proportion of hatchery-origin spawners. Results Low empirical error rates were achieved in both species, with false-negative rates from 0.01 to 0.1 and per-comparison false-positive rates for trios with no correct grandparents from 3 × 10−6 to 1 × 10−8. The incorporation of hatchery metadata (sex, spawn day, and crossing records) allowed larger analyses to be completed while maintaining a low overall number of errors. Applying the method to brood year 2016 steelhead adults in the Snake River basin revealed that of 13,502 natural-origin fish passing upstream of Lower Granite Dam, 10,690, 2,099, and 713 had zero, one, and two hatchery-origin parents. For fall Chinook Salmon returning to spawn in 2021, an estimated 6,135 natural-origin fish returned, with 4,110, 1,469, and 556 having zero, one, and two hatchery-origin parents from the 2012–2014 year-classes. Conclusions The low empirical error rates demonstrated that grandparent inference can be applied using currently available genotyping techniques to quantify reproduction of hatchery-origin fish in the wild. This enables analyses to evaluate trends in the effective proportion of hatchery-origin spawners over time and space to complement existing efforts to manage impacts of hatchery programs on natural populations of anadromous salmonids.