PROJECT COORDINATION AND PARTNERSHIPS
The genetics work has been, and continues to be, carried out by Michael Blouin at Oregon State University.  This project was originally coordinated with the Hood River steelhead hatchery and research program, funded by Bonneville Power Administration and administered and implemented by the Oregon Department of Fish and Wildlife and the Warm Springs Tribes (project numbers 198805307, 198805308, 198805304 and 198805303).  These projects included operation and maintenance of the Oak Springs and Parkdale hatchery facilities, and operation and maintenance of the fish collection and handling facility at Powerdale Dam, as well as database management and data analysis on the part of ODFW.   The work continues in cooperation with ODFW biologists, both in continued analysis of the samples that were originally collected at Powerdale Dam and in ongoing experiments conducted mainly at the Oregon Hatchery Research Center (OHRC), an ODFW facility run in partnership with Oregon State University

LOCATION OF PROJECT
Steelhead samples were collected at the former Powerdale Dam, Hood River, under supervision of Rod French, ODFW, who also coordinated aging of scale samples.  All laboratory work and genetics data analysis continues to be conducted in the laboratory of Michael Blouin at Oregon State University.  We continue to use those pedigreed samples from the Hood River in ongoing studies on the genetics of domestication.  We also conduct experimental work at the OHRC using coastal stocks of steelhead to test hypotheses about adaptation to captivity that were generated by the Hood River data.  

BACKGROUND AND RESULTS TO DATE FROM THE HOOD RIVER
The Hood River supports two populations of steelhead, a summer run and a winter run.  They spawned only above the Powerdale Dam, which is a complete barrier to all salmonids.  From 1991 to 2010 every adult passed above the dam was measured, cataloged and sampled for scales or fin snip.  Therefore, we have a DNA sample from every adult steelhead that went over the dam to potentially spawn in the Hood River from 1991 to 2010, when the dam was removed.  Similar numbers of hatchery and wild fish were passed above the dam during the last decade.  During the 1990's "old" domesticated hatchery stocks of each run (multiple generations in the hatchery, out-of-basin origin) were phased out, and conservation hatchery programs were started for the purpose of supplementing the two wild populations (using wild broodstock; hereafter F1 hatchery fish).  The winter-run F1s were started in 1991, and the summer-run F1s were started in 1997.  In a supplementation program such as this, wild-born broodstock are used as parents in the hatchery in an attempt to circumvent the low fitness induced by multiple generations of selection in the hatchery.  This 19 years of samples gave us the ability to estimate, via microsatellite-based pedigree analysis, the relative total reproductive success (adult to adult production) of hatchery (H) and wild (W) fish for two populations (summer and winter), over multiple brood years.  We now have an almost 4-generation pedigree that is complete for all anadromous fish (note, however, that we are missing samples from resident fish that apparently are the parents of many steelhead).  We compared the relative success of two "old" hatchery stocks vs. wild fish (the winter run “Big Creek” stock and the summer run “Skamania” stock), and showed they have much lower total fitness than wild fish when both breed in the wild (Araki et al., 2007a).  In that paper we also concluded that the winter-run F1 were not significantly different from wild fish, based on 3 run years of data.  But in a subsequent analysis based on six run years of data the difference was significant, with the F1 winter run fish averaging about 85% the fitness of wild fish (Araki et al., 2007d).  

One problem with interpreting an observed difference in fitness between fish raised in a hatchery and fish raised in the wild is that the difference can have a genetic and/or environmental basis (because the H fish experienced a very different environment during the juvenile phase).  However, in Araki et al. (2007d) we were able to compare the first generation fish with second-generation hatchery fish raised in the same hatchery.  These data suggest that the second generation fish have ~55% the reproductive fitness of the first generation fish (Araki et al., 2007d).  Because both types of fish experienced identical environments, the difference between them must be genetically based.  This result also suggests that the decline in fitness that results from additional generations of selection in the hatchery can occur very quickly.  

Thus, we have demonstrated a genetically-based effect of hatchery culture that reduces fitness in the wild and that accumulates with each generation of hatchery culture.  Nevertheless, even if captive-bred individuals are genetically different and produce fewer offspring than wild individuals, adding them to a wild population can still give a demographic boost without substantial harm to a wild population that is below carrying capacity if (1) the genetic effects do not persist into the next generation (i.e., natural selection purges the offspring generation of their deleterious alleles before they reproduce), and (2) enough captive-bred individuals are added each generation to make up for their lower productivity.   If the first condition is not true, however, genetic effects will accumulate over time, potentially leading to a downward spiral in the absolute fitness of the supplemented wild population.  Thus, one key question is whether the wild-born descendents of captive-bred fish are less reproductively successful than the descendents of wild fish. In Araki et al. (2009) we analyzed the fitness of wild-born fish as a function of their parentage.  We found that wild-born offspring of two first-generation hatchery fish averaged 37% the fitness of the offspring of wild fish, while offspring of hatchery-by-wild crosses averaged 87% (Araki et al., 2009).  These results suggest that the hatchery genetic load is not purged from the wild-born population after a full generation of natural selection in the wild.  

We subsequently showed that F1 winter-run  hatchery fish make better broodstock than do wild fish (in terms of number of returning adult hatchery offspring produced)(Christie et al., 2012a), while at the same time performing worse in the wild.  We also showed an interesting tradeoff in which wild broodstock that successfully produced many returning adult hatchery offspring, produced offspring that performed poorly in the wild and vice versa (Christie et al., 2012a).  These two pieces of information strongly suggest that strong domestication selection was acting in the hatchery to make fish rapidly adapt to hatchery conditions. Interestingly, a similar pattern has now been observed in Chinook, so the result does not appear to be limited to steelhead (Ford et al., 2012).  

To summarize the pedigree-based work to date on the Hood River, we have shown: (1) the older, multi-generation, summer and winter hatchery stocks from the Hood River had very low fitness relative to wild fish (10-30%).  This result is consistent with results of many other studies on old stocks (Berejikian and Ford, 2004; Araki et al., 2008). (2) first generation winter run fish have significantly lower fitness than wild fish (about 85%), second generation fish do even worse, and the effect is genetically based. (3) The genetic effects of hatchery culture identified for the winter-run stock persist into the first wild-born generation, with the fitness of wild-born fish depending on whether their parents were both wild, both hatchery or one of each.  Again, the common environment experienced by these three types of wild fish suggests a genetic effect.  Finally, the data in Christie et al. (2012a) strongly suggest that the above effects (loss of fitness in the wild) result from rapid adaptation to the hatchery, rather than some generalized genomic deterioration.

In addition to completing our main mission of analyzing the fitness of hatchery and wild fish and their descendents, we have also addressed several other applied and basic questions. Some examples include the effects of hatchery stock and resident fish on inbreeding and the effective size of the Hood River population (Araki et al., 2007b; Christie et al. 2012b, Christie et al. 2013a) and methodological work on methods for fitness estimation (Araki and Blouin, 2005; Christie et al., 2011), parentage assignment (Christie et al. 2013b) and for estimation of effective size (Araki et al., 2007c; Christie et al., 2012c, 2013a ).  We also tested what fraction of missing winter-run parents were residualized hatchery fish, and found that only a very few could have been.  Therefore, residualized hatchery fish are not a significant route of gene flow from the hatchery into the wild steelhead population (Christie et al., 2011).   Furthermore, that work illustrated how important are resident, wild fish to the genetic integrity of the anadromous, steelhead populations.  We also recently reviewed the data from other studies on the relative reproductive success of early-generation hatchery fish and show that reduced fitness of even F1 stocks is a general phenomenon, and not just an unusual feature of steelhead, or the Hood River population (Christie et al., 2014).  A full bibliography of our work resulting from BPA funding can be found in Project Attachments (file: “Bibliography to 2016”).

CURRENT AND FUTURE WORK
As described above, there now exists substantial evidence that even early-generation hatchery salmonids have lower  fitness than wild fish in the wild, and strong evidence from steelhead that the effect is genetic and owing to adaptation to captivity.  So we believe that now the most important research question is to figure out how to modify hatchery culture conditions to reduce the rate of adaptation to captivity.  

Because size at release is positively correlated with survival at sea, one plausible hypothesis to explain rapid domestication in hatcheries is that hatcheries select for physiological or behavioral traits that promote fast growth in captivity (this might be especially true for steelhead, which are raised to smolting in one year versus the normal two years that they take in the wild).   If those favored traits are maladaptive in the wild, then that could explain why hatchery fish quickly evolve to have lower reproductive success than natural-origin fish in the wild environment.  Support for this hypothesis comes from scale-aging and pedigree data that show the average size at smolting of steelhead that survived to return as adults to the Hood River was much larger than the average size of all smolts from the same cohort at release (unpublished data; manuscript under review).  Therefore, in our experimental work we have been using size at smolting as our measure of performance in the hatchery.

The two main questions we aim to answer are: (1) what traits are under selection in the hatchery? And (2) what hatchery conditions cause strong selection on those traits?  Answering these questions would then tell us how one might modify hatchery culture conditions to reduce the selection pressures that cause such rapid domestication.  This would allow managers to produce hatchery fish that are more like wild fish and thus pose less genetic risk to wild populations.  We are taking four main approaches to answer these questions.  The first three involve raising full sib families of steelhead together in the same tanks, measuring each fish’s size at smolting, and then sorting them back into families via DNA to obtain family average growth rates.  The 4th is a purely genomics approach.

(1) The first approach has been to vary environmental conditions in the hatchery that we suspect might exacerbate selection.  Then we test whether the modified conditions produce fish that show less variation among families in performance in the hatchery than do the standard conditions.  Under standard conditions we see large variance among families in size at release, which means large opportunity for selection.  If we can find environmental conditions that reduce that variance, then there would be less selection among families, and a slower rate of adaptation to the hatchery.  This experimental work has been going on at the Oregon Hatchery Research Center (originally funded by BPA, but now with matching funding from the ODFW).  Environmental conditions we have been studying include crowding levels (i.e. density), feed type and feeding methods, and environmental complexity (including flowing water in circular tanks).  To date we have rejected the hypothesis that increased crowding in the hatchery increases the opportunity for domestication selection (Thompson and Blouin, 2015).  Tests of the other environmental conditions are ongoing.  

(2) The second approach has been to test candidate behavioral and physiological traits that might be under selection.  To date we have shown that fast growing families tend to be more dominant than slow growing families (Thompson and Blouin, 2016), and are currently testing whether other measures of boldness and activity correlate with family growth rate.

(3) The third approach involves looking at genome-wide patterns of gene expression in siblings of each family that is being raised in the growth rate experiments (here gene expression means how actively each gene is being transcribed to make proteins).  Then we test whether expression levels of any sets of genes predict which families grow the fastest.  The idea is that identifying suites of genes that predict growth rate under hatchery conditions might point us to the traits that are under selection.  

(4) The fourth approach is to do genome-wide scans of wild fish and first-generation hatchery fish in order to identify genes that may have responded to selection.  The idea is that, as in (3), finding such genes may point us to the traits that were under selection.  There are two ways to do genome scans, gene expression analysis and genome-wide association studies.  
(a) Gene expression: We recently compared genome-wide expression levels in the offspring of hatchery and wild fish from the Hood River, and then repeated that experiment using H and W fish from the Siletz River in order to see if the patterns were repeatable.  We found hundreds of genes are differentially expressed between offspring of H and W fish from the Hood River, and that the difference could not be a maternal effect (Christie et al., 2015).  We found a similar, but much less dramatic effect in Siletz River fish (unpub. data).  We are currently re-analyzing both data sets together to see whether similar metabolic pathways appear to be under selection in both populations, even if the same particular genes were not.
(b) GWAS: The second genome-scan approach is to do a genome wide association study using hatchery fish and wild fish as the two phenotypes.  The approach here is to look at hundreds of thousands of polymorphic markers (SNPs = single nucleotide polymorphisms) throughout the genome of each fish, and then ask if SNPs in any particular chromosomal regions show unusually strong allele frequency differences between the H and W fish samples.  Such a result would suggest that some gene in that region was under strong selection in the group that went through the hatchery.   Then the next step would be to identify and characterize all the genes that are in the region of statistical association with phenotype, and study their function to figure out which gene was under selection.  We recently collected the raw data for such a GWAS on wild and hatchery Hood River fish using funding from ODFW.  An initial analysis showed at least two very significant genomic regions of association.  We now need to do confirmatory analysis of those results to verify they are not an artifact, and then do bioinformatics analysis of the genes that reside in those significant genomic regions.

REFERENCES CITED
Araki, H. and M.S. Blouin. 2005. Unbiased estimation of relative reproductive success of different groups: evaluation and correction of bias caused by parentage assignment errors. Molecular Ecology, 13:4907-4110.

Araki, H., W.R. Ardren, E. Olsen, B. Cooper and M.S. Blouin. 2007a. Reproductive success of captive-bred steelhead trout in the wild: evaluation of three hatchery programs in the Hood River. Conservation Biology 21:181-190.

Araki, H., R.S. Waples, W.R. Ardren, B. Cooper and M.S. Blouin. 2007b. Effective population size of steelhead trout: influence of variance in reproductive success, hatchery programs, and genetic compensation between life-history forms. Molecular Ecology 16:953-966

Araki, H., R.S. Waples and M.S. Blouin. 2007c. A potential bias in the temporal method for estimating Ne in admixed populations under natural selection. Molecular Ecology 16: 2261–2271

Araki, H., B. Cooper and M.S. Blouin. 2007d. Genetic effects of captive breeding cause a rapid, cumulative fitness decline in the wild. Science 318: 100-103.

Araki, H., B. Berejikian, M. Ford, and M.S. Blouin. 2008 Fitness of hatchery-reared salmonids in the wild. Evolutionary Applications 1:342-355.

Araki, H., B. Cooper and M.S. Blouin. 2009. Carry-over effect of captive breeding reduces reproductive fitness of wild-born descendents n the wild. Biology Letters doi: 10.1098/rsbl.2009.0315

Berejikian, B. A., and M. J. Ford. 2004. Review of the Relative Fitness of Hatchery and Natural Salmon. U.S. Dept. Commer., NOAA Tech. Memo. NMFS-NWFSC-61. 28 p. Northwest Fisheries Science Center, Seattle, WA.

Blouin, M.S. V. Thuillier, B. Cooper, V. Amarasinghe, L. Cluzel, H. Araki and C. Grunau. 2010. No evidence for large differences in genomic methylation between wild and hatchery steelhead trout (Oncorhynchus mykiss). Canadian Journal of Fisheries and Aquatic Sciences. 67: 217-224.

Christie, M., M. Ford and M.S. Blouin. 2014.  On the reproductive success of early-generation hatchery fish in the wild.  Evolutionary Applications, 7:883-896.

Christie, M. R., M. L. Marine, R. A. French, and M. S. Blouin. 2012a. Genetic adaptation to captivity can occur in a single generation. Proc Natl Acad Sci U S A 109:238-242.

Christie MR, RA French, ML Marine and MS. Blouin. 2012b Effective size of a wild salmonid population is greatly reduced by hatchery supplementation.  Heredity, 109, 254–260

Christie MR, RA French, ML Marine and MS. Blouin. 2012c Effective size of a wild salmonid population is greatly reduced by hatchery supplementation.  Heredity, 109, 254–260

Christie MR, RA French, ML Marine and MS. Blouin. 2013a. Does inbreeding cause the reduced fitness of captive-born individuals in the wild? J. Heredity, doi: 10.1093/jhered/est076

Christie MR, Tennessen JA, Blouin MS 2013b. Bayesian parentage analysis with systematic accountability of genotyping error, missing data, and false matching. Bioinformatics 10.1093/bioinformatics/btt039

Christie, M.R., M.L. Marine and M.S. Blouin. 2011. Who are the missing parents? Grandparentage analysis identifies multiple sources of gene flow into a wild population. Molecular Ecology, 20, 1263–1276

Christie, MR, ML Marine, SE Fox, RA French and MS Blouin. 2015. A single generation of domestication heritably alters expression at hundreds of genes.  Nature Communications doi:10.1038/ncomms10676

Ford,  M., A. Murdoch, and S. Howard. 2012. Early male maturity explains a negative correlation in reproductive success between hatchery spawned
salmon and their naturally spawning progeny. Conservation Letters. DOI: 10.1111/j.1755-263X.2012.00261.x

Thompson, NF and MS Blouin 2015. The effects of high rearing density on the potential for domestication selection in hatchery culture of steelhead (Oncorhynchus mykiss).  Canadian Journal of Fisheries and Aquatic Sciences.  72:1-6.

Thompson NF, Blouin MS. 2016. Family dominance level measured during the fry stage weakly influences family length at smolting in hatchery reared steelhead (Oncorhynchus mykiss). Transactions of the American Fisheries Society 145: 1282-1289.