4 Discussion
This study provides further insight to the spatial distribution of genetic variation underlying migration timing in a broad range of steelhead populations. Genetic relationships were characterized for neutral and candidate markers for 113 populations, supporting previous findings of population structure and demonstrated strong differences between major lineages. We determined linkage blocks for 13 candidate markers associated with migration timing and found different heterozygote haplotypes were found to be predominant in coastal versus inland lineages. Environmental drivers of candidate variation revealed the importance of temperature and precipitation to selection on variation for migration in this system. Overall, this study provides extensive geographic variation for candidate markers associated with migration timing that is expected to be important for conservation applications in this species (Waples & Lindley, 2018).
Patterns of genetic variation among steelhead populations were highly distinct between neutral and candidate markers. Neutral structure was consistent with previous studies with various marker types that largely correspond to geographic population structure and significant heterogeneity in environmental conditions at collections sites (Blankenship et al., 2011; Matala et al., 2014; Micheletti et al., 2018b). For example, the Clearwater River steelhead have consistently showed a distinct genetic signal from others in the Snake River basin regardless of marker type (Narum et al., 2008; Campbell et al., 2012; Matala et al., 2014; Micheletti et al., 2018b). Additionally, the neutral markers provided further resolution than previous studies for the inland lineage especially for populations in the Yakima River drainage that were distinct from the rest of the populations in the middle Columbia River. The distinct neutral patterns in the Clearwater and Yakima River drainages were likely due to different levels of genetic influence from hatchery programs (Blankenship et al., 2011). Populations in the Yakima River have likely been minimally influenced by hatchery programs as there has been no direct hatchery program for steelhead in this subbasin. Similarly, large stretches of the Clearwater River basin, including the Selway and Lochsa Rivers, are managed exclusively for wild fish (Nielsen et al., 2009; Campbell et al., 2012). Presence of dams and historical rapid population declines may also differentiate sub-basins of inland steelhead from one another (Blankenship et al., 2011; Matala et al., 2014). Finally, the Klickitat River is positioned in a geographically intermediate location at the eastern base of the Cascades and considered to be on the boundary of the coastal and inland lineages. The intermediate status of the Klickitat River collections was evident in the neutral PCA population structuring which is consistent with previous studies (e.g., Micheletti et al., 2018b). This intermediate signal was also observed in two other populations, Fifteenmile Creek and Mill Creek, which may indicate gene flow with steelhead in the Klickitat R. or admixture. In contrast to geographical patterns observed at neutral loci, the candidate PCA divided collections by their predominant adult migration timing. The Skamania stock was a useful reference for the extreme extent of fixed genetic variation for premature alleles due to artificial selection for early migration timing over several decades in this hatchery program. At the other end of the spectrum, the mature genotype was predominant in most collections, while the heterozygote collections were dispersed across the basin, but with divergent ratios of haplotypes between coastal and inland lineages. The presence of genetic variation for premature alleles in the inland lineage suggests that some populations of steelhead (i.e., those in the Salmon R. drainage) may exhibit phenotypic variation for early and late arrival timing to spawning grounds as shown by Micheletti et al. (2018b).
Haplotype blocks of markers with the greatest association with one another and with the migration timing phenotype improve ability to evaluate genetic variation associated with migration timing across the landscape. In addition to LD assessments, we evaluated differences between average genotype frequencies with fewer candidate markers. Marker 9 had the most similar average genotype frequencies to markers 8-12 for all genotypes and markers 8-12 had the greatest LD in all collections. This finding suggests that marker 9 could be useful under circumstances of limited genotyping abilities. This same marker was also helpful at distinguishing patterns in steelhead arrival timing to spawning grounds (Micheletti et al., 2018b). However, it is still beneficial to assess collections with entire haplotype blocks when possible, to generate numerous haplotype combinations instead of only three genotypes gained from a single marker.
We observed significant association between multiple environmental variables and candidate markers when examined across lineages, which was expected given that environmental conditions vary significantly across the Columbia River basin landscape. We found migration distances, temperature variables, and precipitation variables had the strongest association to adaptation for all collections which was consistent with previous landscape genomics analyses (Micheletti et al., 2018b). Migration distance traveled between the Pacific Ocean and spawning sites ranged from 60 to 1,400 km, presenting a vast difference between coastal and inland lineages in energetic allocation before spawning (Olsen et al., 2011; Hecht et al., 2015). Migration distance was not significantly associated with migration timing within each lineage suggesting that variation at candidate markers is not highly distinct among populations at small geographic scales. Significant association of temperature with candidate markers was not surprising since fish rely on environmental temperatures to regulate body temperatures and trigger migratory behavior (Jonsson 1991; Sykes et al., 2009), and extreme temperatures can inhibit cardiac and metabolic proficiencies (Chen et al., 2018). Further, genetic disparities in thermal tolerance when encountering temperature barriers has been found to contribute to local adaptation in salmonids (Eliason et al., 2011; Narum et al., 2013; Muñoz et al., 2015). Finally, the significance of precipitation with variation at candidate markers is expected to be important since precipitation conditions can impact survival and selection on genes associated with thermal tolerance when flow is low (Heath et al., 2002) and water temperatures are elevated (Narum et al., 2013). In contrast, when precipitation is high and stream flow is powerful, conditions may become energetically costly for migrating steelhead, but also provide cues for migration to spawning grounds (Keefer et al. 2014; 2018). Significantly associated environmental variables within each lineage were more limited than across lineages of steelhead, and largely reflected regional differences in precipitation within the coastal lineage and temperature within the inland lineage.
In this study, we assessed the spatial distribution of candidate haplotype frequencies because selective pressures on steelhead migration are disparate across the heterogeneous landscape. The coastal lineage contained steelhead maturing both in the ocean and streams, whereas inland lineage steelhead only matured in streams. Initial studies (Hess et al., 2016; Prince et al., 2017; Thompson et al., 2019) identified and associated greb1L genotypes with freshwater entry, while Micheletti et al. (2018a) revealed a greater greb1L association with arrival timing to spawning grounds. We also detected more than one genotype present in inland collections, further supporting an association with arrival timing to spawning grounds introduced by Micheletti et al. (2018a). Our study incorporated more collections and more candidate markers associated with migration timing than previous studies, which allowed us to determine haplotypes to describe the spatial pattern of mature and premature genotypes across the Columbia River basin in greater detail. Coastal collections exhibited greater genetic diversity at candidate markers, but greater influence of premature alleles from Skamania and other hatchery stocks. In the inland lineage, the mature genotype was detected at high frequency despite all inland steelhead maturing in freshwater, supporting findings by Micheletti et al. (2018a). Variation in the second haplotype block, which includes markers in the intergenic region, indicates that inland populations retain genetic variation that may allow for variable timing in arrival to spawning grounds. However, further studies are needed that dissect arrival phenotypes and the association at candidate markers atgreb1L and rock1 .
From a management perspective, detailing the distribution of migration run timing has direct conservation implications. Early migrating fish spend less time feeding in the nutrient rich ocean, resulting in less opportunities for growth and potential for decreased reproductive success. Further, more time in freshwater systems exposes early migrators to thermal stress, disease, and greater risk for impacts of climate change and selective fisheries (Quinn et al., 2015). Thus, steelhead with this early migration pattern have increased odds of extirpation and may require greater conservation efforts (Prince et al., 2017). Previous findings (Micheletti et al., 2018a) were bolstered by this study that indicate greater genetic diversity at candidate genes for inland collections than previously understood. Effective conservation efforts to maintain or increase genetic variation underlying migration timing is expected to provide broader life history diversity for populations to endure stochastic environments. Thus, maintenance of genetic diversity associated with migration timing across the Columbia River basin may be a key to promote resilient steelhead populations that are able to recover from anthropogenic impacts.
Data Accessibility:Genotype data are available in Dryad at doi:10.5061/dryad.jh9w0vt80.
Competing Interests: None declared
Acknowledgements: Thanks to all tribes and agencies that provided samples, laboratory staff involved in sample processing (CRITFC, IDFG), Funding from Bonneville Power Administration grant number 2008-907-00.
Author Contributions: SRN designed and directed the study. EEC analyzed the data. All authors interpreted the results and wrote the manuscript.