Genetic parentage reveals the (un)natural history of Central Valley Hatchery steelhead

Populations composed of individuals descended from multiple distinct genetic lineages often feature significant differences in phenotypic frequencies. We considered hatchery production of steelhead, the migratory anadromous form of the salmonid species Oncorhynchus mykiss, and investigated how differences among genetic lineages and environmental variation impacted life history traits. We genotyped 23,670 steelhead returning to the four California Central Valley hatcheries over nine years from 2011–2019, confidently assigning parentage to 13,576 individuals to determine the age and date of spawning, and rates of iteroparity and repeat spawning within each year. We found steelhead from different genetic lineages showed significant differences in adult life-history traits despite inhabiting similar environments. Differences between coastal and Central Valley steelhead lineages contributed to significant differences in age at return, timing of spawning, and rates of iteroparity amongst programs. In addition, adaptive genomic variation associated with life history development in this species varied among hatchery programs and was associated with the age of steelhead spawners only in the coastal lineage population. Environmental variation likely contributed to variation in phenotypic patterns observed over time, as our study period spanned both a marine heatwave and a serious drought in California. Our results highlight evidence of a strong genetic component underlying known phenotypic differences in life history traits between two steelhead lineages.

Methods
SNP loci, genotyping, and basic population genetics analysis

Samples were genotyped with a panel of 96 biallelic SNP markers (Abadía-Cardoso et al. 2013), including a Y chromosome-linked marker to determine genetic sex (Brunelli et al. 2008). However, the marker composition of the panel varied slightly over time, with 92 loci genotyped across all years of the study; markers not typed across all years were removed from downstream analyses (Table S1). All individuals were genotyped using TaqMan assays (Applied Biosystems) on 96.96 Dynamic Genotyping Arrays with the EP1 Genotyping System (Fluidigm Corporation) following the manufacturer’s protocols. Two negative controls were included in each array, and genotypes were called using SNP GENOTYPING ANALYSIS SOFTWARE V 3.1.1 (Fluidigm).

To evaluate genotyping error rates for each SNP marker we inferred parent-offspring trios using parentage analysis (see below) and estimated the minimum genotyping error rate expected to produce the Mendelian incompatibilities observed at each marker across the trios. Of the 23,670 genotyped samples, 83 yielded low-quality genotypes after the initial round of genotyping (indicated primarily by large fractions of missing genotypes). These samples were re-genotyped. Any individuals missing more than 10% of loci (fewer than 82 successful genotype calls) were identified and removed.