Significant habitat degradation, reduced connectivity, and total isolation of suitable habitat are possible outcomes of anthropogenic barriers in freshwater environments for numerous species (Bunn and Arthington, 2002). Anthropogenic barriers can cause significant population declines, extirpations, and elimination of unique life-histories, which can have cascading ecological effects for the system across taxa (Brenkman et al., 2008; Bellmore et al., 2019; Zarri et al., 2022). Barrier removal has occurred for decades to improve accessibility and reestablish the necessary habitats for migratory animals such as salmon (Pess et al., 2012; Pess et al., 2014; Bellmore et al., 2019), which has in turn has successfully restored connectivity and contributed to population gains for key taxa in formerly blocked habitats (Brenkman et al., 2012; McHenry et al., 2017; Duda et al., 2020; Hess et al., 2020; Fraik et al., 2021; Pess et al., In Review ). Understanding the extent to which the removal of these large barriers may ultimately lead to restored biological communities remains an important area of study (Grummer et al., 2019; Tamario et al., 2019).

Necessary ecological conditions for movement must exist for aquatic species to reestablish in their historical habitat after barrier removal. Aquatic species respond, move and disperse across riverscapes in different ways, as a function of their migration capabilities (Crozier et al., 2008; Dodson et al., 2013; Pess et al., 2014). Some species remain in one freshwater habitat for their entire lives, whereas others move between different freshwater environments (e.g., rivers and lakes) or into marine habitats (Quinn, 2018). This is exemplified in salmonids which undergo physiological transformations to migrate (Nichols et al., 2008; Hecht et al., 2012; Dodson et al., 2013; Kendall et al., 2015) and have significant intraspecific variation in their migration life-history forms. Challenges to upriver movement in freshwater environments for salmonid species include the distance to new habitat, environmental conditions, physical capacity to reach the habitat, life-history traits required to exploit the habitat, and physical barriers (Pess, 2009; Duda et al., 2020). These challenges may impact both the local population connectivity of those species still present, and the species that may reestablish the river network after local extirpation. Evaluating the impacts of barrier removal therefore requires consideration of the ecological characteristics of migratory corridors, the breadth of life-history forms expressed and the standing genetic variation underlying these life-history traits in extant populations (Gaggiotti et al., 2004; Anderson et al., 2010; Pess et al., 2012; Anderson et al., 2013).

Juvenile and adult ocean migration life-history forms of salmonids have a genetic (Nichols et al., 2008; Hecht et al., 2012; Pearse et al., 2019) and environmental basis (Docker and Heath, 2003; Heath et al., 2010; Kendall et al., 2015). In particular, there is a strong association of alleles in and near the GREB1L and ROCK1 genes with ocean migration return-timing in several salmonid species (Hess et al., 2016; Prince et al., 2017; Micheletti et al., 2018; Thompson et al., 2019; Collins et al., 2020; Willis et al., 2020; Willis et al., 2021; Tigano and Russello, 2022). Previous work suggests early freshwater adult return-timing for anadromous salmonid species is an adaptation for fish migrating longer distances as they return to freshwater during higher flows, when water temperatures are cooler, producing more energetically optimal conditions compared to those experienced by later returning adult fish (Quinn et al., 2016; Waples et al., 2022). This knowledge of the genetic basis, in association with the ecological requirements, of these life-history forms is crucial for understanding the response to migratory corridor restoration. Monitoring and assessing these diverse life-history forms of salmonids, however, can be quite challenging due to variable river sampling conditions during these distinct freshwater entry-times.

Landscape genetics is a tool that allows us to examine the role specific environmental conditions have on genetic connectivity among these life history forms across a defined area (Storfer et al., 2007; Manel and Holderegger, 2013; Storfer et al., 2018). While developed specifically for terrestrial studies, landscape genetics has been applied to riverscapes to identify the impacts of shared environmental features on the movement, gene flow and structure of aquatic populations (Whiteley et al., 2004; Davis et al., 2018; Kelson et al., 2020; Rougemont et al., 2023). These methods have been employed to identify unique, population specific responses (Narum et al., 2008; Sidharthan et al., 2022) as well as inter-specific responses to shared, environmental stressors (Whiteley et al., 2004; Goldberg and Watis, 2010; Olsen et al., 2010; Emel et al., 2019). Using riverscape genetic methods, we can test the impacts of barrier removal on spatial patterns of species with life-history specific genetic variation.

Research is underway to document the implications of barrier removal on aquatic life-history diversity in the Elwha River watershed (Brenkman et al., 2019; Duda et al., 2020; Duda et al., 2021). Located in the Olympic peninsula of western Washington state, the Elwha River basin occupies 883 rkm² and was once one of the more productive watersheds in the Pacific Northwest. This watershed is home to the five Pacific Salmon species (from the genus Oncorhynchus), as well as bull trout (Salvelinus confluentus), coastal cutthroat trout (Oncorhynchus clarkii), Steelhead/rainbow trout (Oncorhynchus mykiss), Pacific lamprey (Entosphenus tridentatus) and numerous other freshwater fish (Brenkman et al., 2008; Pess, 2009). The construction of the Elwha River (1912) and Glines Canyon (1927) dams in the lower and middle portions of the main stem without fish passage blocked upstream and downstream connectivity for fishes, leading to declines of several aquatic species (DOI, 1996; Winans et al., 2008). In particular, life-history variants of salmonid species that previously occupied the river appeared to be extirpated after dam construction, including many species’ early return-timing life history forms (Brenkman et al., 2008; Pess, 2009). As a result of the Elwha River Restoration Act, dam removal of the Elwha River dams began in 2011. Following the Glines Canyon Dam removal in August 2014, a rockfall occurred in Glines Canyon and created at least a partial barrier to fish. This blockage was addressed in late 2015 with selective blasting which reopened the channel for fish passage (Ritchie et al., 2018). Over time, Chinook salmon (Oncorhynchus tshawytscha), Steelhead and numerous other salmonid species have ascended the river upstream of the former Glines Canyon dam, expanding their spatial distributions into the upper watershed (Duda et al., 2020; Duda et al., 2021; McHenry et al., 2022; Peters et al., 2022). However, the rates and extents of fish returning to the upper river varied, particularly across species with distinct life-history forms (Brenkman et al., 2019; Pess et al., In Review ). One remaining question is how reestablishing connectivity in the Elwha River may affect salmonid species’ distinct life-history forms and their underlying genetic variation in different ways.

In this study, we used population and riverscape genetics to investigate the distribution of neutral and return-timing genetic variation within O. tshawytscha and O. mykiss sampled from the Elwha River. First, we investigated neutral genetic structure for each species and explored associations with time and space. Second, we tested for shifts in frequencies of alleles associated with early ocean return-timing for each species. Finally, we employed riverscape genetics to test for differences in gene flow of neutral and return-timing genetic variants in each species that could be explained by freshwater environmental variation. Specifically, we compared the relationships of genetic variation with models of river distance to models of at-site and between-site environmental conditions, with the distance models representing the null hypothesis of isolation-by-distance. Overall, this study represented a unique opportunity to understand potential differences between neutral and return-timing genetic variation through space and time after large-scale dam removal, with important implications for broad restoration efforts related to connectivity.

In this study, we leveraged the removal of two large dams on the Elwha River to understand how habitat restoration impacted the return and reestablishment of distinct ocean migrating life-history forms of O. tshawytscha and O. mykiss to their historical distribution. While we observed no difference in the proportion of early return-timing alleles in O. tshawytscha across the watershed, we observed significant differences for O. mykiss across the watershed for samples collected both pre- and post-dam removal. Specifically, the proportion of early return-timing alleles was greater in O. mykiss sampled above dam locations than below dam locations. Using riverscape genetics analyses, we found that return-timing genetic variation was more strongly influenced by environmental differences compared to neutral variation. By exploring the relationships of at- and between-site environmental characteristics, we identified significant patterns of spatial genetic variation in O. mykiss. For both species, the patterns and processes are still actively shifting in response to these restoration efforts; however, we show here early evidence of the importance of considering adaptive potential of ocean return-timing and neutral genetic variation in understanding species responses to restoration efforts.

Lack of population genetic structure and decreases in genetic diversity observed in O. mykiss may be, in part, an artifact of the study design. Previous work documenting Elwha River O. mykiss using microsatellite and tens of thousands of SNP loci found that both dams and anadromous barriers generated significant genetic structure pre-dam removal (Winans et al., 2016). Post-dam removal, structure appeared to decrease as gene flow resumed between formerly allopatric populations (Fraik et al., 2021). While we know the natural-origin Steelhead populations in the upper river have significantly increased post-dam removal (Denton et al., 2022b; Peters et al., 2022; Pess et al., In Review ). We observed diminished N_b above the former dams, but an increase below the former dams. Biologically, this might reflect down river movements of resident rainbow trout after dam removal and greater overall admixture. Alternatively, this could be an artifact of sampling location and design since Steelhead may not have been sampled at spawning sites. Ascertainment bias in the genotyping panels utilized for this study could also explain deviations from previous work. One half of the neutral loci in the O. mykiss GTseq panel (92 SNPs) were selected to delineate genetic stocks at a river basin or watershed level, across reporting groups in the Columbia River and across the US West Coast (Narum et al., 2010; Abadía-Cardoso et al., 2011; Limborg et al., 2012). Given the geographic focus of our study is considerably smaller than the target spatial scale of the panels, we may be observing higher genetic covariance amongst the SNP genotypes of fish from the Elwha River watershed, in line with the intended purpose of the panel.

Similar to O. mykiss, we did not observe significant evidence of genetic structure in O. tshawytscha. The reestablishing O. tshawytscha in the Elwha are heavily dominated by hatchery spawners that were generated from the small, natural-origin population that remained below the dams decades after dam removal (McHenry et al., 2022). We have evidence that eleven of the O. tshawytscha sampled in the Elwha River might represent strays (McKinney et al., In Prep ) and may encompass greater genetic diversity. As the neutral loci in the O. tshawytscha panel were designed to delineate genetic stocks, we may be capturing the introduction of straying fish admixing with Elwha River endemic O. tshawytscha. While there is limited data to determine how genetic diversity and structure changed over time ( Appendix Table 2 ), these genotypes provide an important baseline for O. tshawytscha as limited genetic studies have been published on Elwha and surrounding Puget Sound populations to date (McKinney et al., In Prep ; Ruckelshaus et al., 2006; Winans et al., 2008).

While we did not find evidence of significant neutral genetic structure between locations, there were significant deviations in the return-timing allele frequencies across the watershed in O. mykiss, but not O. tshawytscha ( Figure 6 ). In O. mykiss, there were higher frequencies of the early return-timing haplotypes and alleles in fish sampled above the former dams compared to below the dams post-dam removal ( Figures 5 , 6 ). In the Elwha River, both early return-timing life-history forms of each species were extirpated following dam construction (DOI, 1996). The resident form of O. mykiss, rainbow trout, was able to persist in the middle and upper watershed, while O. tshawytscha and the anadromous form of O. mykiss, Steelhead, were limited to seven river kilometers of spawning habitat below the dams (McHenry et al., 2017). While Steelhead were not able to return to spawn above the dams, these land-locked O. mykiss harbored high proportions of the early return-timing alleles and maintained high levels of standing genetic variation river-wide (Hiss and Wunderlich, 1994; Winans et al., 2016; Fraik et al., 2021; Fraik et al., 2022). Steelhead reestablished in the upper river before O. tshawytscha (Duda et al., 2020; Duda et al., 2021), with the first early return-timing Steelhead detected in 2013 (Personal Comm. J. McMillan) and Chinook detected in 2022 (Personal Comm. R. Peters). A high proportion of these summer Steelhead ascending the former dammed river have been shown to be descendants of the land-locked, above-dam ancestries (Fraik et al., 2022). The preference for summer Steelhead to spawn in colder, snow-melt fed water (Busby et al., 1996) may also suggest there is pleiotropy at the return-timing genetic locus for local adaptation (Micheletti et al., 2018).

Riverscape genetic models indicated that patterns of return-timing variation in O. mykiss were affected by sampling site environmental differences. Variation in frequencies of early return-timing alleles in adult Steelhead and juvenile O. mykiss at sampling sites were best explained by habitat and full models. Pool frequency was a significant covariate in the habitat model for adult Steelhead, suggesting that spawning Steelhead adults with different early return-timing genotypes may be differentially impacted by spawning ground habitat characteristics. Similarly, juvenile O. mykiss return-timing, as well as neutral genetic variation, was best explained by habitat variables (pool frequency and canopy) across rearing habitat. When considering all O. mykiss cumulatively, the significance of these models diminished. Consistent influence of habitat characteristics across age classes and genetic response variables suggests that these covariates may be biologically significant, reflecting age class specific responses.

Broadly, we found that geographic distance best explained patterns of neutral gene flow both at-site (IBE) and between-site (IBR) for most species’ models. The full and habitat IBE models were the optimal neutral models for Steelhead and juvenile O. mykiss, but not for O. tshawytscha. The lack of environmental influence observed in the O. tshawytscha models most likely reflects the limited, or lack of, environmental heterogeneity among O. tshawytscha sampling sites. While water temperature is usually a significant environmental factor influencing growth, migration and movement in salmonids (Brewitt and Danner, 2014; Kendall et al., 2015; Armstrong et al., 2021; Pitman et al., 2021), temperature variation throughout the Elwha is not physiologically limiting (Myrick and Cech, 2005; Siegel et al., 2022; Siegel et al., 2023) and probably not impacting gene flow. Alternatively, these differences could be the by-product of how O. mykiss were sampled at specific sampling sites at known age classes compared to O. tshawytscha, which were sampled across river reaches post-spawning. For example, MWMT stream temperature standard deviation for O. mykiss was 0.44 Celsius (n = 19, mean = 16.1 Celsius, range = 15.1–16.5 Celsius) and O. tshawytscha was 0.37 Celsius (n = 9, mean = 16.6 Celsius, range = 16.3–17.4 Celsius), reflecting a narrow range of environmental variation across the sampling locations. However, these differences in stream temperature could also reflect the statistical uncertainty around the derivation of environmental point estimates from stream reaches rather than geographic points. Regardless, we may simply lack the statistical power to detect the influence of the environment on neutral variation due to a more homogenous resistance surface.

In this study, we explicitly did not conduct any direct comparisons between O. tshawytscha and O. mykiss because of several important differences in the founding populations, sampling methods, and stage of recovery of the two species in the Elwha River. First, the founding Elwha O. tshawytscha is not only smaller than the O. mykiss population but represents lower return-timing genetic diversity compared to O. mykiss ( Figures 4 , 5 ). Both species populations are bolstered by hatchery programs that were founded from below dam, natural-origin Elwha broodstock that do not necessarily select for diversity at the GREB1L/ROCK1 locus. While these hatchery fish make up a large proportion of the reestablishing O. tshawytscha population, they make up less of the Steelhead population (Pess et al., In Review ; Denton et al., 2022a; Denton et al., 2022b; McHenry et al., 2022; Peters et al., 2022). Reestablishing Steelhead possess ancestry from below dam, similar to those of hatchery fish, as well as formerly above and in-between dam ancestries that contain greater return-timing genetic variation (Fraik et al., 2021). Second, the temporal and spatial extent of sampling was more limited for O. tshawytscha than for O. mykiss. The relatively short time frame of sampling for O. tshawytscha and the majority of samples coming from the lower reaches of the Elwha is a direct result of this species being in an early stage of reestablishment in the Elwha River. While these sampling limitations do impede the power of temporal and spatial analyses for O. tshawytscha, our study was able to document contemporary patterns, or the lack thereof, of genetic variation in the study area. For example, while IBD models were the most optimal for O. tshawytscha, the coefficients were not significantly different from zero reflecting no clear association or return-timing variation with river kilometer. Thus, we may be observing an artifact of a founder effect generated by genetic stochasticity rather than IBD (Gaggiotti et al., 2004; Weigel et al., 2013). In contrast, we observed a positive association between geographic distance and return-timing genetic distance and resistance in O. mykiss, regardless of age class. These results may occur because distance can represent a proxy for the cumulative differences in environmental conditions experienced by each return-timing life-history as individuals move further upriver from the ocean (Rougemont et al., 2023), and our sampling sites for O. mykiss extended over twice the distance upstream compared to O. tshawytscha sampling sites.

Our study provided a framework for testing the impacts of a significant environmental change, large barrier removal, on the recovery of two important species and their key life-history forms. The reservoir of neutral and return-timing genetic variation maintained in resident rainbow trout above the dams appeared to allow for rapid, robust reestablishment of diverse O. mykiss life-history forms river wide post-dam removal. While we did not have phenotype data to characterize recovery of each life-history form in the watershed, we were able to infer how adaptive potential underlying these life-history forms is distributed (Hess et al., 2016; Prince et al., 2017; Micheletti et al., 2018; Narum et al., 2018; Thompson et al., 2019; Collins et al., 2020; Ford et al., 2020; Koch and Narum, 2020; McKinney, 2020; Willis et al., 2020; Willis et al., 2021; Tigano and Russello, 2022; Waples et al., 2022). We observed species and age-class specific patterns of genetic distances structured among sites that shows how at- and between-site environmental heterogeneity may influence O. mykiss. Given that O. tshawytscha were extirpated from the river above the former dams, we are still in the early stages of observing their reestablishment in the upper watershed. We were able to test biologically informative hypotheses about environmental heterogeneity that influence species, population and life-history specific gene flow. However, differences in the geographic scale, time frame and sampling methodologies for each species must be considered in the interpretation of results in this study. Future studies should continue to monitor O. tshawytscha to study the long-term impacts of different demographic and environmental factors on reestablishment.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI BioProject, PRJNA1021324.

TS, AKF and GM conceptualized the study, organized the course planning and solicited help from the 2022 Landscape Genetics Distributed Graduate Seminar course. JHA, KMN, TRS, GM and GP coordinated the field planning, study design, and metadata collection for both species. KD, JHM, JHA, HC, MM, and TB coordinated and collected, aggregated and disseminated the field samples and metadata for both species. DK, KMN, AKF, GM, and TRS conducted and oversaw the lab work and genotyping. AKF and GM genotyped the genetic data. AKF, TS, and GM oversaw and directed the graduate work. AHF and GP contributed to the environmental data collection, aggregation and analyses. KJL, YS, and JYJ performed all data analyses, data visualization and results presentation. KJL, YS, JYJ, and AKF contributed to the writing. All authors contributed to the articleand approved the submitted version.