Genomics and conservation units: The genetic basis of adult migration timing in Pacific salmonids

Abstract It is now routinely possible to generate genomics‐scale datasets for nonmodel species; however, many questions remain about how best to use these data for conservation and management. Some recent genomics studies of anadromous Pacific salmonids have reported a strong association between alleles at one or a very few genes and a key life history trait (adult migration timing) that has played an important role in defining conservation units. Publication of these results has already spurred a legal challenge to the existing framework for managing these species, which was developed under the paradigm that most phenotypic traits are controlled by many genes of small effect, and that parallel evolution of life history traits is common. But what if a key life history trait can only be expressed if a specific allele is present? Does the current framework need to be modified to account for the new genomics results, as some now propose? Although this real‐world example focuses on Pacific salmonids, the issues regarding how genomics can inform us about the genetic basis of phenotypic traits, and what that means for applied conservation, are much more general. In this perspective, we consider these issues and outline a general process that can be used to help generate the types of additional information that would be needed to make informed decisions about the adequacy of existing conservation and management frameworks.


| INTRODUC TI ON
Advances in genomics technologies have revolutionized many areas of biology, and are beginning to be taken up in conservation (Benestan et al., 2015;Bernatchez, 2016;Garner et al., 2016;Hendricks et al., 2018;Moore, Bourret et al., 2014;Shafer et al., 2015;Waters et al., 2018). Identification of intraspecific conservation units (CUs) is one of the most common applications of genetic data. Two levels of CUs are typically identified (Moritz, 1994): Evolutionarily Significant Units (ESUs), which represent important evolutionary/ecological components of the species as a whole, and Management Units (MUs), which typically represent demographically independent units that merit separate management. Several frameworks for defining ESUs have been proposed (reviewed by Fraser & Bernatchez, 2001), but most can be characterized in terms of the relative importance they place on two major axes of diversity: isolation and adaptation (Waples, 2006; Figure 1). Prior to the last decade, genetic data for markers thought to be selectively neutral were used primarily to characterize the isolation axis, whereas insights regarding adaptations had to rely primarily on proxies, such as ecological features of the species' habitats, or phenotypic traits that reflect combined effects of genetics and the environment.
Genomics data potentially can help to quantify both types of diversity (Funk, McKay, Hohenlohe, & Allendorf, 2012). The ability to assay tens or hundreds of thousands of DNA markers greatly enhances power to infer historical demography and patterns of connectivity, which in turn increases resolution along the isolation axis (e.g., Benestan et al., 2015). Genomics methods also can identify genes associated with phenotypes thought to be adaptive. As demonstrated by Moore, Bourret et al. (2014), there is no guarantee that this new information will suggest any major changes to conservation units defined on the basis of neutral markers. On the other hand, it is inevitable that sometimes different patterns of genetic affinity will be implied by neutral and putatively adaptive markers.
Given that the goal of identifying and protecting CUs is the conservation of intraspecific biodiversity, in at least some of these cases it might be necessary to consider whether to revise existing conservation units based on the genomics data, and if so how best to do so.
A recent example of this latter issue involves conservation units of Pacific salmon (Oncorhynchus spp.) and steelhead (the anadromous form of rainbow trout, O. mykiss). Roughly one-third of the Pacific salmon and steelhead populations that existed in the coterminous United States ca 1800 have been extirpated (Gustafson et al., 2007), and about half of those that remain are federally listed as threatened or endangered under the U.S. Endangered Species Act (ESA) (http:// www.westcoast.fisheries.noaa.gov/protected_species/salmon_ steelhead/salmon_and_steelhead_listings/salmon_and_steelhead_ listings.html). These populations are protected under a provision of the ESA that allows listing of Distinct Population Segments (DPSs) of vertebrate species. Pacific salmon populations or groups of populations are considered DPSs if they meet the criteria to be an ESU (NMFS 1991;Waples, 1991). In Canada, formal assessments under the Species at Risk Act (SARA) are not complete, but a num- In the United States, the framework used to identify listable units for Pacific salmon places roughly equal weight on reproductive isolation and adaptation (Waples, 1991(Waples, , 2006 Figure 1). Atlantic salmon and steelhead are dealt with under a broader DPS framework that applies to all vertebrate species (USFWS and NMFS, 1996), but its two DPS criteria (discreteness and significance) are largely parallel to those in the salmon policy. After initially experimenting with a different approach for defining DUs under SARA (Green, 2005), in 2009 Canada adopted a framework that is essentially identical to the discreteness/significance approach used to define DPSs in the United States (COSEWIC 2018).
Proxies most commonly used for adaptive differences/significance have been phenotypic and life history traits and distinctive ecological features of the habitat (Waples, 2006). Of the former, adult migration timing (season of entry into fresh water to begin the spawning migration) has received particularly careful scrutiny, not only because of clear evidence for a genetic basis (Carlson & Seamons, 2008), but also because this trait has been widely used to define harvest management and artificial propagation programs in both the United States and Canada. The strongest adult migration contrast is between early returning populations (summer steelhead or spring Chinook salmon, O. tshawytscha), which enter fresh water when they are immature and hold for many months before spawning, and late-returning populations (winter steelhead or fall Chinook salmon), which mature in the ocean and spawn soon after entering fresh water. Maintaining diversity among populations in migration timing and other life history traits has been shown to be important for long-term persistence and sustainability (Moore, Yeakel, Peard, Lough, & Beere, 2014;Schindler et al., 2010).
A considerable body of evidence has suggested that differences in adult migration timing often have resulted from parallel adaptations and hence do not define separate evolutionary lineages. Studies using presumably neutral genetic markers have shown that, in most coastal drainages from California to Washington, Chinook salmon and steelhead populations from the same river that have different run timing are genetically more similar to each other than are populations from different rivers that have the same run timing (Figure 2).
This pattern was first documented with allozymes (Chilcote, Crawford, & Leider, 1980;Waples, Teel, Myers, & Marshall, 2004) and has subsequently been confirmed in microsatellites and singlenucleotide polymorphisms (SNPs) (Arciniega et al., 2016). These results have led to the conclusion that run-timing diversity has arisen many times within each species by a process of parallel evolution. It F I G U R E 1 A general framework for evaluating strength of evidence in support of ESUs or other types of conservation units. Widely used ESU concepts focus on two axes of intraspecific diversity (isolation and adaptation) but differ in the relative importance assigned to each. Moritz's (1994) reciprocal monophyly of mtDNA concept focused almost entirely on isolation; the exchangeability concept proposed by Crandall, Bininda-Emonds, Mace, and Wayne (2000) placed more emphasis on adaptation; and the frameworks developed by Waples (1991) and Dizon, Lockyer, Perrin, Demaster, and Sisson (1992) placed roughly equal weight on each factor. Until recently, information regarding isolation generally relied on molecular genetic data, whereas inferences about adaptations typically had to be based on proxies such as ecology, behavior, life history, and other phenotypic traits. Recent advances in genomics technology for non-model species now make it possible to identify genes associated with traits thought to be adaptive-but is this sufficient to adequately characterize this axis?  is generally assumed that the late-maturing life history is the more general form, and the early returning form evolves from standing genetic variation only when suitable habitat/environmental conditions are present (e.g., seasonal access to high-elevation habitats and suitable locations for summer holding). This view is consistent with the reigning paradigm in quantitative genetics, which has been that most phenotypic traits are controlled by many genes of small effect. A classic example of this is the genetic architecture of height in humans. Studies of relatives consistently indicate that about 80% of variation in human height is due to additive genetic factors, yet the top ~10,000 SNPs most strongly associated with human height still explain only about 30% of the phenotypic variance (Wood et al., 2014). and because of the latter's relatively high overall abundance, the ESU is not federally protected.
The new genomics results raise some important questions about conservation and management priorities-in particular, the relative importance in conservation of focusing on (a) preserving certain phenotypes, or (b) conserving key ecological/evolutionary processes that are ultimately responsible for creating diversity (Moritz, 2002). If evolution of different life histories is a common parallel process, it is particularly important to conserve the ability of natural ecological and evolutionary processes to produce  inversion that is strongly associated with expression of anadromy in steelhead/rainbow trout. Barson et al. (2015) found a single gene strongly associated with age at maturity in Atlantic salmon, but this was thought to be an exception because it appears to be important in resolving sexual conflict. Veale and Russello's (2017) range-wide study of sockeye salmon found that reproductive ecotype (spawning either along lakeshore beaches or in streams) is controlled by a single locus. Pritchard et al. (2018) found evidence for diversifying selection in multiple regions of the genome in Atlantic salmon. A comprehensive synthesis of published and unpublished association studies for phenotypic traits in salmonids would be a valuable asset in evaluating this issue.

| D ISCUSS I ON
If strong associations with one or a few genes prove to be relatively common for phenotypic/life history traits in salmon, it would raise the following question with respect to the example consid-  (Quinn, Unwin, & Kinnison, 2000), but also because this trait is widely used in management and conservation planning. But this would represent a novel, perhaps unprecedented, approach to applied conservation. Funk et al. (2012) cautioned against focusing on individual traits in defining ESUs; instead, they recommended that both neutral and adaptive genes be used, and that adaptive significance be assessed using a larger suite of markers ("outlier loci") that show evidence for effects of natural selection across the genome. In this context, Micheletti, Matala, Matala, and Narum (2018)  Under the ESA, the federal government takes a broader perspective that focuses on ensuring that major components of ecological/evolutionary diversity (salmon and steelhead ESUs) are conserved.
How might this picture change in light of new genomics data?
If one takes the large-effect-gene hypothesis to its extreme, such that expression of a key phenotypic trait requires a specific gene that is only found in populations that express that phenotype, then evolution of a new population with that phenotype would require immigration of the key genes from another population that already expresses the trait. This could be a real conservation concern, es-  Thompson et al. (2018) showed that a higher resolution analysis of the GREB1L region can produce stronger phenotype-genotype associations in Chinook salmon. This suggests that our understanding of the relative frequency and importance of large-effect genes in salmon is likely to continue to evolve in the near future as sampling across the genome becomes more comprehensive.
Compiling more information regarding Question 1 is essential to determine whether the paradigm for thinking about evolution of migration timing diversity needs a major adjustment. If alleles associated with early migration timing are widely distributed in populations that do not have the early migration phenotype, then the parallel evolution paradigm might still be largely applicable in practice, even though the explanatory mechanism would require updating. That is, parallel evolution of early migrating populations from nearby latemigrating populations based on standing genetic variation might still be the norm, but each realization of this scenario would involve a mix of the same few alleles of large effect and potentially different combinations of many more alleles of smaller effect, such as those found by Brieuc, Ono, Drinan, and Naish (2015).
Conversely, if alleles associated with early migration timing are only found in early migrating populations, then it would be important to ensure that populations with that phenotype persist. What as well as others that can be imagined. In the end, the effectiveness of conservation and management efforts depends not only on how CUs are defined, but rather on the interaction between the CU structure and the framework for assessing extinction risk. That is, any of the CU frameworks in Figure 2  Most or all of the key questions identified above would be useful in evaluating the relevance of large-effect genes for these other applications. Furthermore, although we have focused on a specific example involving Pacific salmonids, the issue of how best to use genomics data in conservation is of much broader relevance, as evidenced by lively discussions in the scientific literature (Funk et al., 2012;McMahon, Teeling, & Höglund, 2014;Pearse, 2016;Prado-Martinez et al., 2013;Primmer, 2009). As demonstrated by Bay et al. (2018), phenological traits roughly comparable to adult migration timing are of considerable ecological and evolutionary importance to a wide range of taxa.
Applied conservation is hard because biology is complex and messy, and every practical application involves interactions with human customs, laws, and institutions that can differ greatly from place to place. This means that each new application typically introduces new wrinkles that have to be considered as special cases. But there is something that is more general and more broadly applicable to a wide range of issues: The process involved in figuring out how best to apply genomic data (or any new kind of data) to a real-world Over time, as the number of real-world applications of genomics continues to grow, so too will our understanding of the role genomics information can play in 21st Century conservation and management.
We are still in the early stages of this with respect to application of genomics data. In the meantime, this overall effort can be enhanced by a systematic approach to work through each applied problem to provide managers and decision makers with as much relevant information as possible upon which to base their decisions.

CO N FLI C T O F I NTE R E S T
None declared.