Evaluating salmon hatchery supplementation programs requires assessing not only program objectives but identifying potential risks to wild populations as well. Such evaluations can be hampered by difficulty in distinguishing between hatchery- and wild-born returning adults. Here, we conducted three years (2011–2013) of experimental hatchery supplementation of sockeye salmon in Auke Lake, Juneau, Alaska where a permanent weir allows sampling and genotyping of every returning adult (2008–2019). We identified both hatchery- and wild-born returning adults with parentage assignment, quantified the productivity (adult offspring/spawner) of hatchery spawners relative to that of wild spawners, and compared run timing, age, and size at age between hatchery- and wild-born adults. Hatchery-spawning females produced approximately six to 50 times more returning adults than did naturally spawning females. Supplementation had no discernable effect on run timing and limited consequences for size at age, but we observed a distinct shift to younger age at maturity in the hatchery-born individuals in all three brood years. The shift appeared to be driven by hatchery-born fish being more likely to emigrate after one, rather than two, years in the lake but the cause is unknown. In cases when spawning or incubation habitat is limiting sockeye salmon production, hatchery supplementation can be effective for enhancing the number of returning adult fish but not without the risk of phenotypic change in the recipient population, which can be an undesired outcome of hatchery supplementation. This study adds to a growing body of evidence suggesting that phenotypic change within a single generation of captive spawning might be widespread in salmon hatchery programs.

From 2008 to 2019, each returning adult sockeye salmon that ascended to Auke Lake was captured at the Auke Creek weir, on the outlet of Auke Lake, and identified based on the appearance of the snout and vent as a female, regular-sized male, or “jack” (precocious male ≤ 400 mm mid-eye to fork length) male.  Each salmon had an axillary process removed and stored in 95% ethanol for subsequent DNA extraction and genotyping. Fish were sampled every day during the run, and all were sampled for genotypes. A sub-sample (~12% of the total run) was sampled each year for age (from scales) and length (mid-eye to fork) data. Sampling was intended to be representative across the entire run.

Over three successive years (2011–2013), sockeye salmon were artificially spawned and their progeny were reared to the early fry stage prior to release into Auke Lake. A small number of adults (<50 per year, averaging 2.5% of the total run annually) were captured at the weir, retained for broodstock, held at the hatchery in tanks (≈ 6 weeks) until mature, and subsequently spawned. The target for broodstock was 30 females and 15 males with a 2:1 breeding design (each male crossed with two females). Fertilized eggs were placed in vertical flow incubators fed by water from a 2.3 m deep intake in Auke Lake. Emergent fry were ponded and given starter feed (Pacific Bio-Products, Inc.) at the manufacturer’s size- and temperature-dependent recommendations for optimal growth; this captive-rearing period lasted for ≈ 4 weeks. Each year, ≈ 50,000 fry (≈ 0.3 g) were released into several locations in the limnetic zone of Auke Lake, where they were allowed to feed and grow alongside wild-born individuals and emigrate volitionally to the ocean after one or two years in freshwater. Sockeye salmon return to Auke Creek 3–6 years after birth (Kovach et al. 2014), so returning adults that were born in captivity would return over the years 2014–2019.

Genotyping was conducted at the Gene Conservation Lab of the Alaska Department of Fish and Game. Genomic DNA was extracted using DNeasy® 96 Tissue Kit (Qiagen). Samples were genotyped at 9 short tandem repeat (STR) and 45 single nucleotide polymorphism (SNP) markers. Taqman assays and the Fluidigm Biomark 96.96 and 48.48 Dynamic Arrays were used to genotype the SNPs, following the methods outlined in Seeb et al. (2009). Amplification of STR loci was carried out in 10 µL reaction volumes and PCR fragments were identified on an Applied Biosystems 3730 capillary DNA sequencer. PCR bands were visualized and separated into bin sets using AB GeneMapper software v4.0 and automated binning was subsequently confirmed or corrected manually. A subsample (8 of every 95 individuals) was re-extracted and genotyped to quantify genotyping error rate.

Seeb, J. E., Pascal, C. E., Ramakrishnan, R. Seeb, L. W. (2009). SNP genotyping by the 5′-nuclease reaction: Advances in high-throughput genotyping with nonmodel organisms. In: A. Komar (Ed.), Methods in Molecular Biology: Single-nucleotide Polymorphisms, 2nd edition (pp. 277-292). New York: Humana Press.