Y‐chromosome haplotypes are associated with variation in size and age at maturity in male Chinook salmon

Abstract Variation in size and age at maturity is an important component of life history that is influenced by both environmental and genetic factors. In salmonids, large size confers a direct reproductive advantage through increased fecundity and egg quality in females, while larger males gain a reproductive advantage by monopolizing access to females. In addition, variation in size and age at maturity in males can be associated with different reproductive strategies; younger smaller males may gain reproductive success by sneaking among mating pairs. In both sexes, there is a trade‐off between older age and increased reproductive success and increased risk of mortality by delaying reproduction. We identified four Y‐chromosome haplogroups that showed regional‐ and population‐specific variation in frequency using RADseq data for 21 populations of Alaska Chinook salmon. We then characterized the range‐wide distribution of these haplogroups using GT‐seq assays. These haplogroups exhibited associations with size at maturity in multiple populations, suggesting that lack of recombination between X and Y‐chromosomes has allowed Y‐chromosome haplogroups to capture different alleles that influence size at maturity. Ultimately, conservation of life history diversity in Chinook salmon may require conservation of Y‐chromosome haplotype diversity.


| INTRODUC TI ON
Variation in life history within populations is common across taxa and is often associated with alternative strategies for increasing fitness. This includes partial migration, where some individuals of a population migrate while others remain resident (Chapman, Brönmark, Nilsson, & Hansson, 2011), reproductive morphs that exhibit different mating strategies or sexually selected traits (Johnston et al., 2013;Küpper et al., 2015;Shuster, 1989), age and size at maturity (Gibbons, Semlitsch, Greene, & Schubauer, 1981), and even length of life span such as annual versus perennial plants (Hall, Basten, & Willis, 2006). While life history variation is often assumed to be under the influence of many genes of small effect, examples of a large-effect genes and supergenes (sets of linked genes) influencing life history variation are increasingly being found. These include single genes that influence age at maturity (Barson et al., 2015) and sexually selected traits (Johnston et al., 2013); these also include chromosome inversions contributing to annual versus perennial life history (Twyford & Friedman, 2015) and variation in migration versus residency (Pearse, Miller, Abadia-Cardoso, & Garza, 2014 Variation in life history strategies is exhibited by many salmon species with size and age at maturity being an important component of this variation. In females, large body size confers a direct reproductive advantage through increased fecundity and egg quality (Healey & Heard, 1984). In contrast, variation in size and age at maturity in males can be associated with different reproductive strategies that exhibit frequency-dependent fitness. For example, older larger males gain reproductive success by monopolizing access to females, while younger smaller males gain reproductive success by sneaking in among mating pairs (Berejikian et al., 2010;Healey & Heard, 1984). For both sexes, there is a trade-off to delayed maturation where increased reproductive success is countered by an increased risk of mortality before reproduction. The optimal age at maturity for a population should represent the balance between reproductive benefits and mortality costs of delayed maturation (Healey, 1986), and forces that change these costs or benefits could result in shifts in age composition. Nonetheless, considerable diversity in age at maturation is maintained within many populations, presumably as a bet-hedging strategy to spread risks over the life cycle of these fish.
Age at maturity in salmon is generally thought to be a threshold trait that is dependent either upon reaching a minimum size at age or upon growth rate at key periods (Healey, 1991;Thorpe, 2007).
Environmental factors that influence growth rate have been shown to influence age at maturity in many species. In the wild, studies show correlations between ocean conditions such as temperature (or productivity) and patterns of age at maturity (Otero et al., 2012;Siegel, McPhee, & Adkison, 2017). In experimental settings, age at maturity was manipulated through changing temperature (Harstad et al., 2018;Heath, Devlin, Heath, & Iwama, 1994) or food ration (Larsen et al., 2006;Rowe & Thorpe, 1990).
In addition to environmental effects, multiple lines of evidence show a genetic component to age at maturity. High heritability values suggest considerable genetic variation for age at maturity in several salmon species (Gall, Baltodano, & Huang, 1988;Gjerde, 1984;Hankin, Nicholas, & Downey, 1993;Heath et al., 1994), and quantitative trait locus (QTL) and genome-wide association (GWAS) studies identified genomic regions associated with variation in age at maturity (Barson et al., 2015;Kodama, Hard, & Naish, 2018;Waters et al., 2018). Studies also demonstrated that offspring of alternative male phenotypes exhibit different growth rates (Berejikian, Van Doornik, & Atkins, 2011;Garant, Fontaine, Good, Dodson, & Bernatchez, 2002) with offspring of early maturation phenotypes (grilse and jacks) exhibiting high growth rates. In Atlantic salmon (Salmo salar), individuals with different life histories exhibit differing maturation thresholds that are genetically based (Aubin Horth & Dodson, 2004). Despite these findings, the genetic basis of maturation age in most salmonids remains poorly understood.
One complicating factor is that the genetic basis underlying variation in age at maturity appears to vary among salmonid species and even among populations within a species. In Atlantic salmon, age at maturity is strongly influenced by a single gene (VGLL3) (Ayllon et al., 2015;Barson et al., 2015). This gene exhibits sex-dependent dominance, which facilitates sexually antagonistic selection (Barson et al., 2015). While this gene explained 39% of the phenotypic variability in European Atlantic salmon, studies in North American Atlantic salmon have shown that this association varies by population (Boulding, Ang, Elliott, Powell, & Schaeffer, 2019;Kusche et al., 2017) and this gene has not been shown to have an effect in Pacific salmon (genus Oncorhynchus) .
Chinook salmon (O. tshawytscha) are the largest of the Pacific salmon and follow various life history strategies, spending 0 to 2 years in fresh water and 1 to 4 or more years in the ocean (Riddell et al., 2018). Male Chinook salmon exhibit significant variation in size and age at maturity (Healey, 1991) that is linked to differential reproductive tactics and is likely controlled by both environmental and genetic components (Berejikian et al., 2010;Young, Conti, & Dean, 2013). Many populations throughout North America recently experienced marked declines in size and age at maturity, which may erode life history variation (Lewis, Grant, Brenner, & Hamazaki, 2015;Ohlberger, Ward, Schindler, & Lewis, 2018). Explanations for decreased age at maturity have focused on the impacts of fisheries-induced evolution (Hard et al., 2008;Kendall, Dieckmann, Heino, Punt, & Quinn, 2014) or changing environmental conditions (Siegel et al., 2017); however, these factors alone are not consistent nor sufficient to explain current declines, and it is likely that these declines are driven by multiple complex factors (Ohlberger et al., 2018).
While the genetic control of age at maturity in Chinook salmon is still poorly understood, past studies offer clues to genomic regions that may be associated with maturation age. In particular, Heath, Rankin, Bryden, Heath, and Shrimpton (2002) identified a strong sex-linked component to age at maturity in Chinook salmon, suggesting the influence of genes on the Y chromosome.
The X and Y chromosomes in most salmonid species are morphologically undifferentiated (Davidson, Huang, Fujiki, von Schalburg, & Koop, 2009). Available sequence data suggest that the primary difference between sex chromosomes is an insertion containing the sex-determining gene (SDY, Yano et al., 2013), and in Chinook salmon, a 2.4 Mb male-specific repetitive sequence has been identified (Devlin, Stone, & Smailus, 1998). While the sex-determining gene has been assigned to chromosome 17 (Ots17) in Chinook salmon (Phillips, Park, & Naish, 2013), the region that contains this gene is not in either of the current genome assemblies (Christensen et al., 2018;Narum, Genova, Micheletti, & Maass, 2018), so the exact location of sdY is unknown. The genome assembly by Christensen et al. (2018) was of a female, while the assembly by Narum et al. (2018) used a male Chinook salmon; however, sdY was not assembled as part of a chromosome, likely because of extended repetitive sequence in this region (cf., Devlin et al., 1998).
Despite a lack of large-scale differentiation, the X-and Y-chromosomes could show sequence divergence due to sex-specific patterns of recombination (heterochiasmy, see Sardell & Kirkpatrick, 2020). Recombination in females takes place along the full length of the chromosome, while recombination in males is strongly localized to telomeric regions (Lien et al., 2011;Sakamoto et al., 2000), restricting recombination between the X-and Y-chromosomes. Reduced recombination between sex chromosomes is common across taxa, and in salmonids is supported by a 33 Mb signal of sex association observed in Atlantic salmon (Kijas et al., 2018) and a lack of recombination between the sex-determining region and an allozyme locus on the sex chromosome in Chinook salmon (Marshall, Knudsen, & Allendorf, 2004). In addition to facilitating divergence between sex chromosomes, sex-limited recombination could lead to the formation of different Y-chromosome haplotypes and the capture of adaptive genetic variants (Bergero, Gardner, Bader, Yong, & Charlesworth, 2019). If Y-chromosome haplotypes have sufficiently diverged from the X-chromosome, they may be identified through patterns of extended linkage disequilibrium.
We examined patterns of linkage disequilibrium on the sex chromosome of Chinook salmon to determine whether male-specific haplotype blocks (Y-chromosome haplotypes) existed, and if so, are these haplotypes associated with variation in size and age at maturity which commonly differ between sexes in Chinook salmon. We identified four Y-chromosome haplogroups (groups of similar haplotypes) in Chinook salmon from Alaska that showed regional-and population-specific variation in frequency. These haplogroups showed associations with size at maturity in multiple populations, suggesting that the lack of recombination between X-and Y-chromosomes has allowed genetic variants influencing size and age at maturity to segregate on different Y-chromosome haplogroups.

| RAD Y-chromosome haplotypes
We used existing RADseq data to examine patterns of genetic variation in the sex chromosome. RADseq data for 21 populations of Chinook salmon from Alaska were obtained from: NCBI SRA accessions SRP034950 (Larson, Seeb, Pascal, Templin, & Seeb, 2014) and SRP129894 (McKinney, Waples, Pascal, Seeb, & Seeb, 2018), bioproject PRJNA560365 (McKinney, Pascal, et al., 2020, and raw data used in Dann et al. (2018). Raw data from Dann et al. (2018) are available from those authors upon request. Populations ranged from Cook Inlet to the Upper Yukon River and include a total of 1,082 samples ( Figure 1, Table 1). RADseq data were processed with Stacks V1.7 (Catchen, Hohenlohe, Bassham, Amores, & Cresko, 2013;Catchen, Amores, Hohenlohe, Cresko, & Postlethwait, 2011) using default settings with the following exceptions: process_radtags F I G U R E 1 Frequency of Y-chromosome haplogroups throughout Alaska based on RADseq and GT-seq data. Locations of populations are approximate to prevent overlap of pie charts. Population names are given in Table 1. Note that the location for the Lower Yukon Test Fishery (population 3) indicates where fish were caught on their return to the Yukon River; these samples may represent fish from many populations that spawn throughout the Yukon River TA B L E 1 Populations used in this study (Lower Yukon Test Fishery is a mixture of populations). RADseq data were used for haplogroup discovery, while GT-seq data were used to confirm that haplogroups were male-specific and to expand the geographic distribution of haplogroups   RADseq results were filtered to remove SNPs that were likely to genotype poorly or be uninformative. These categories included collapsed paralogs, SNPs with more that 10% missing data, and SNPs with a minor allele frequency < 0.01. Paralogs are common in salmonid genomes due to an ancestral whole-genome duplication (Allendorf & Thorgaard, 1984). With short-read sequence, the two copies of paralog are often collapsed into a single locus that appears polyploid; these cannot be reliably  (Chang et al., 2015;Purcell et al., 2007), and SNP pairs with r 2 ≥ 0.3 were retained for network analysis. Network analysis was conducted using the igraph package in R (https://igraph.org/r) to identify sets of loci with high LD. Sets of loci that contained at least five SNPs and spanned at least 1 Mb were retained for further analysis. Genotypes for retained SNPs were then phased into haplotypes using fastPhase (Scheet & Stephens, 2006). Putative Y-chromosome haplogroups were identified by clustering haplotypes using heat-map2 in R (Warnes et al., 2020) with the Ward.D clustering algorithm to minimize within-group variance. Haplogroups that appear to be male-specific are hereafter referred to as Y-chromosome haplogroups and assigned names based on the following convention: chromosome number, MH for male haplogroup, followed by a sequential number, for example, Y-chromosome haplogroups on chromosome 17 would be Ots17-MH1, Ots17-MH2, and so on. Allele frequencies within each haplogroup were visualized using logo plots with the ggseqlogo package in R (Wagih, 2017).
Assays for SNPs that were diagnostic for Y-chromosome hap- The Yukon River is part of Western Alaska but was analyzed separately for this study. b Only one male from Pullen Creek was assigned to a Y-chromosome haplogroup.
c Only 11 out of 20 males from Little Port Walter-Unuk River Stock were assigned to a Y-chromosome haplogroup.

TA B L E 1 (Continued)
primer3 (You et al., 2008). Primer design used the consensus RAD sequence for each RADtag unless SNPs occurred within 20 bp of the end of the RADtag. In these cases, the consensus sequences were aligned to the genome using bowtie2 (Langmead & Salzberg, 2012), and genomic sequence flanking the RADtag was added to the consensus sequence for primer design. Default settings for SNP flanking primers were used with the following exceptions: a minimum amplicon size of 75 bp and a maximum amplicon size of 150 bp. Panel optimization followed the methods of ; one round of sequencing using 80 individuals was conducted to identify loci that over-amplified or produced unreliable genotypes.

| Expanded genotyping
Additional samples ranging from Alaska to California were genotyped using the GT-seq panel to establish the geographic distribution of the haplogroups ( Two methods were used to assign GT-seq samples to Y-chromosome haplogroups. First, samples were assigned Y-chromosome haplogroups using the same methods as the RADseq samples: phasing genotypes into haplotypes using fastPhase followed by haplotype clustering using heatmap2. Genotypes from RADseq samples were included in the GT-seq haplotype assignment to assess concordance between the discriminatory ability of the full RADseq marker set and the subset that was successfully developed into GT-seq markers. Second, the expected genotype patterns for males with each haplogroup were constructed assuming fixation of alleles on the X-chromosome and the Y-chromosome. The observed genotypes were then compared with the expected genotypes.
Samples were assigned to a haplogroup if the observed genotypes had less than two mismatches to the expected genotypes.

| Y-chromosome haplotype analyses
Assignments were combined for GT-seq and RADseq samples to characterize the distribution and frequency of Y-chromosome haplogroups. Data for length and age at maturity were obtained from Alaska Department of Fish and Game (ADFG) for a subset of populations in Alaska and compared with Y-chromosome haplogroup data to determine whether there were relationships between Y-chromosome haplogroup and length and age at maturity. Analysis of variance (ANOVA) was used to test the significance of associations between Y-chromosome haplogroups and length; population of origin was included as a covariate. A post hoc Tukey test was then performed to determine whether differences in size distribution were significant between individual haplogroups. The significance of associations between Y-chromosome haplogroups and age at maturity of fish captured in the Lower Yukon River Test Fishery was assessed using an ANOVA with population of origin added as a covariate. There are multiple methods of reporting age in salmon (Koo, 1962); we report freshwater and ocean age for each individual using European notation, so an individual with an age of 1.3 would have spent 1 year growing in freshwater after emergence, followed by 3 years in the ocean, for a total age of 5 years.

| RE SULTS
A total of 448 SNPs from 323 RADtags remained after filtering SNPs with more than 10% missing data or a MAF ≤ 0.01, removing paralogs identified by HDplot, and retaining only SNPs that aligned to chromosome 17. Retained SNPs had an average coverage of 24x.
When multiple SNPs occurred in the same RADtag, these were analyzed as independent SNPs.

| RADseq Y-haplotype discovery
Network analysis of linkage disequilibrium identified three sets of linked SNPs exhibiting long-distance LD (4.5-9.5 Mb each). There were 35 SNPs in these high LD sets, representing 7.8% of the SNPs on Ots17 retained after filters. Individual genotypes for these SNPs were then phased, resulting in two haplotypes per individual.
Clustering of haplotypes resulted in five major groupings that exhibited extended LD up to 17.4 Mb ( Figure 2). Although phenotypic sex was available from only a subset of individuals, phased haplotypes from sexed individuals were present in each of these groupings ( Table 2). One of the groupings contained phased haplotypes from both males and females, suggesting that this represents the X-chromosome (gray cluster, Figure 2). Four of the haplogroups contained phased haplotypes predominantly from male samples (85%-95%), suggesting that these are from the Ychromosome (Table 2, Figure 2). In addition, all individuals with a haplotype in one of the Y-chromosome haplogroups had their second haplotype in one of the X-chromosome groupings.
All Y-chromosome haplogroups were present, at varying frequency, in each geographic region, suggesting that these haplogroups are conserved throughout this broad geographic region SNPs exhibited no allelic variation within the X-chromosome, suggesting that alternative alleles observed on the Y-chromosome haplotypes may be novel genetic variants (e.g., SNP 71572_67; Figure 3).

F I G U R E 2
Plot of haplotype clusters identified by phasing high LD loci from the RADseq dataset. Each individual is represented by two haplotypes corresponding to each chromosome of Ots17. For each SNP, the most common allele is in yellow and the least common allele is in red. Haplotypes were clustered into haplogroups, and six major haplotype clusters were identified; these are denoted by different colors along the sample dendrogram (y-axis). The gray haplogroup represents X-chromosomes, while four haplogroups (pink = Ots17-MH1, green = Ots17-MH2, blue = Ots17-MH3, and purple = Ots17-MH4) represent Ychromosomes; numeric designations only are labeled on plot. SNP position on Ots17 is given on the x-axis. The position for SNPs that were successfully developed into GT-seq assays are colorcoded in red on the x-axis. The relative position of each SNP on chromosome 17 is shown on the top of the x-axis   64086  470610  661484  661535  661560  1743821  1760840  1767113  1848646  1848734  2121609  2250467  2250485  2466819  2802115  2839600  3376211  3551175  3564336  4466954  4590393  4959224  6262360  6310357  6406809  6425153  6491379  6608956  8513051  9202095  9698184  13171055  14896983  15055429  18065098 1 3 2 4 X

| GT-seq Y-chromosome expanded sampling
A GT-seq panel was developed to genotype Y-chromosome haplotype markers for a set of samples representing the North American range of Chinook salmon. A total of 23 of the 35 RAD SNPs passed filtering criteria prior to primer design; of these, 16 RAD SNPs were successfully converted to GT-seq assays ( Figure 3, Table S1).
Samples genotyped with GT-seq were assigned haplogroups using two methods, clustering of haplotypes using the Ward.D algorithm and assignment based on genotypes. Haplogroups Ots17-MH1 and Samples that were genotyped with both RADseq and GT-seq showed high concordance between RADseq and both GT-seq haplogroup assignment methods except for Ots17-MH3 (Table S2).
When RADseq haplogroup assignment was compared with GT-seq haplogroup assignment using Ward.D clustering, 10 of the 12 samples assigned to RADseq Ots17-MH3 were grouped with females.
This is likely because only three of the RADseq SNPs that differentiated this haplotype from the X-chromosome were successfully developed into GT-seq ( Figure 3). In contrast, when GT-seq samples were assigned to haplogroups based on comparing observed to expected genotypes for each haplogroup, 10 of the 12 samples correctly assigned to Ots17-MH3, while two were not assigned to any haplogroup. For samples genotyped only with GT-seq, there were no discrepancies in haplogroup assignment between the clustering and genotype-based methods except for Ots17-MH3. Ots17-MH3 had no individuals assigned using haplotype clustering but had 23 samples assigned based on genotype matching. Genotype-based matching assigned approximately 12% fewer individuals to haplotypes overall than the phased haplotype clustering (Table S2C) but was better able to assign individuals to the Ots17-MH3 haplogroup.
All GT-seq samples assigned to a Y-chromosome haplogroup were genetic males based on the Ots-SEXY-3-1 sex identification assay (Table S3).

The majority of male (phenotypic or genetic) Chinook salmon in
Alaska were assigned to Y-chromosome haplogroups with regional
Age-at-maturity data were available for 177 males from six of the populations in this study (

| D ISCUSS I ON
Variation in life history within populations is common across taxa and is often assumed to be under the influence of many genes of small effect; however, examples of a genes and regions of large effect on life history variation, including single genes, small genomic regions, or chromosome inversions, are increasingly being found (Barson et al., 2015;Johnston et al., 2013;Pearse et al., 2014;Twyford & Friedman, 2015). The genetic mechanism underlying variation in life history has important implications for how life history diversity is maintained under different selective regimes (Hess, Zendt, Matala, & Narum, 2016;Prince et al., 2017), particularly in the case of sexually antagonistic selection where males and females have different phenotypic optima (Barson et al., 2015;Pearse et al., 2019).
Size and age at maturity are ecologically and evolutionarily important traits in Chinook salmon. Numerous studies have examined ongoing declines in age at maturity; however, it has been difficult to disentangle the interactions between environmental and genetic causes of this decline. Size-and age-associated markers and genes have previously been identified in genetic studies of Chinook salmon Waters et al., 2018); however, results were not consistent across populations, and no markers were located on the sex chromosome. We show that a conserved set of Y-chromosome haplotypes is associated with variation in size and age at maturity in Chinook salmon across the Yukon River, Western Alaska, and Cook Inlet. These observations open a new line of research into the genetic basis of age at maturity in salmonids.  Fairweather (Templin, Seeb, Jasper, Barclay, & Seeb, 2011) Ots17-MH4 haplogroups, which were associated with larger fish.

| Range-wide distribution of haplotypes
While we did not have adequate samples with size data to characterize size distributions within regions, our finding is consistent with a long-term analysis of Chinook salmon returns by Lewis et al. (2015).
These authors reported smaller fish on average in Kuskokwim and Nushagak River populations from Western Alaska relative to Yukon River and Cook Inlet populations. In addition, the Cook Inlet populations sampled in this study are near and share a common migration pathway with the Kenai River, which has historically produced large Chinook salmon (Lewis et al., 2015;Schoen et al., 2017). The relationship between size and haplotype also varied by region, with fish from the Yukon River having the smallest difference in sizes between haplotypes and fish from Cook Inlet having the largest difference in sizes ( Figure 4). While the magnitude of difference appears to be largest in Cook Inlet, it is difficult to accurately assess statistical significance due to the low sample size when splitting samples among regions.
Taken together, these results suggest that differing frequencies of Y-chromosome haplotypes may contribute to regional variation in size of Chinook salmon and that the effect of haplotype on size can vary between regions, potentially due to other genetic influences or different environmental conditions. The Ots17-MH3 haplogroup was unusual in that it showed no consistent pattern, with large fish in some regions and small fish in other regions (Figure 4). This haplogroup had the smallest sample size, which may affect the trends. This haplogroup also showed the least differentiation from the X-chromosome based on RADseq data ( Figure 3) and may not contain adaptive variants influencing size or age at maturity.
Recombination is generally restricted to the telomeres in male salmon; however, there is evidence that centromeric recombination does occasionally occur (Sutherland, Rico, Audet, & Bernatchez, 2017). The occurrence of rare recombination events could break up haplotype blocks, leading to the degradation of existing haplotypes and the formation of new haplotypes. Recombination events could explain the differences in extended LD observed among the Ots17-MH-1, Ots17-MH-2, and Ots17-MH-3 haplotypes or the variation in extended LD within the Ots17-MH1 haplotype ( Figure 2). However, the variation in extended LD could also represent sequential fixation of alleles due to drift. The boundaries of the haplogroups may also provide clues to the position of sdY on chromosome 17. All Y-chromosome haplogroups overlapped between 0 and ~5 Mb. If the boundaries of the haplogroups are due to rare recombination with the X-chromosome, then this suggests that the sdY gene may be located near this region.

| Resolving sexual conflict
Different phenotypic optima for males and females are common across species and can create sexual conflict that is difficult to resolve when adaptive loci are on autosomes. One mechanism is for the same alleles to exhibit sex-specific dominance, such as the VGLL3 gene that influences age at maturity in Atlantic salmon (Barson et al., 2015). Another mechanism is to partition adaptive variants between nonrecombining regions of sex chromosomes, such as genes governing coloration and fin morphology in Poecilids; these genes are attractive in males but would increase predation risk in females (Lindholm & Breden, 2002). The existence of Y-chromosome haplotypes demonstrates not only that genetic variation is partitioned between the X-and Y-chromosomes, but that Y-chromosomes have partitioned different genetic variants ( Figure 3). While it is unlikely that the specific SNPs we ob- Hypotheses of population structure and delineation of management units using genetic data are typically based on genome-wide analyses consistent with the assumption that major life history traits are controlled by many genes with small effects. (2018)

| CON CLUS ION
Variation in size and age at maturity is common across taxa and is often associated with alternative strategies for increasing fitness; however, the genetic basis of this variation is largely unknown.
We identified Y-chromosome haplogroups that are associated with size, and likely age, at maturity in Chinook salmon throughout Alaska. These haplogroups were primarily restricted to western and southcentral Alaska Chinook salmon where the most diversity in age at maturity exists, and likely represent a subset of the total diversity across the species range. It is possible that each Chinook salmon lineage has a specific set of haplogroups and relationships between haplotypes and size/age at maturity may differ by lineage.
Y-chromosome haplotypes and their potential effect on life history variation in Chinook salmon may provide a basis to help explain the causes and consequences of the recent declines in size and age of adult Chinook salmon, trends that are most pronounced in the region with the highest haplotype diversity. Ongoing efforts to understand the causes of these declines point to size-specific mortality of maturing fish but also require an unknown evolutionary basis . Our findings reveal a mechanism for the genetic control of changes in size at age and age at maturity in Chinook salmon. Monitoring haplotype diversity may be particularly important as future changes in environmental conditions and selective fishing may lead to further demographic responses in this economically and ecologically important species.

ACK N OWLED G EM ENTS
We would like to thank Chris Habicht, Bill Templin, Wes Larson,

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

DATA AVA I L A B I L I T Y S TAT E M E N T
Raw GT-seq data are available in NCBI SRA Bioprojects PRJNA646992 and PRJNA646245.