draft...draft 20 abstract 21 pink salmon, the most abundant pacific salmon, have an obligate...

Draft

SNP data describe contemporary population structure and

diversity in allochronic lineages of pink salmon (Oncorhynchus gorbuscha)

Journal: Canadian Journal of Fisheries and Aquatic Sciences

Manuscript ID cjfas-2017-0023.R1

Manuscript Type: Article

Date Submitted by the Author: 30-Jun-2017

Complete List of Authors: Tarpey, Carolyn; University of Washington, School of Aquatic and Fishery

Sciences Seeb, James; University of Washington, School of Aquatic and Fishery Sciences McKinney, Garrett; University of Washington, School of Aquatic and Fishery Sciences Templin, William; Alaska Department of Fish and Game, Bugaev, Alexander; Kamchatka Fishery and Oceanography Research Institute, Sato, Shunpei; Hokkaidoku Suisan Kenkyujo Seeb, Lisa; University of Washington, School of Aquatic and Fishery Sciences

Is the invited manuscript for

consideration in a Special Issue? :

N/A

Keyword: GENETICS < General, MOLECULAR ECOLOGY < General, GEOGRAPHICAL DISTRIBUTION < General, Last Glacial Maximum, Glacial Refugia

https://mc06.manuscriptcentral.com/cjfas-pubs

Canadian Journal of Fisheries and Aquatic Sciences

Draft

SNP data describe contemporary population structure and diversity in allochronic 1

lineages of pink salmon (Oncorhynchus gorbuscha) 2

3

Authors: Carolyn M. Tarpey¹a, James E. Seeb¹

b, Garrett J. McKinney

1c, William D. 4

Templin2d

, Alexander Bugaev3e

, Shunpei Sato4f

, and Lisa W. Seeb¹g

5

Affiliation: ¹University of Washington, School of Aquatic and Fishery Sciences, Box 6

355020 Seattle, Washington, 98195, USA 7

2Alaska Department of Fish and Game, 333 Raspberry Road, Anchorage, Alaska, 8

99518, USA 9

3 Kamchatka Fishery and Oceanography Research Institute (KamchatNIRO), 10

18, Naberezhnaya Str. Petropavlovsk-Kamchatsky 683602, Russia

11

4 Hokkaido National Fisheries Research Institute, Japan Fisheries Research and 12

Education Agency, 2-2 Nakanoshima, Toyohira-ku, Sapporo 062-0922, Hokkaido, 13

Japan 14

Email: [email protected];

[email protected];

[email protected]; 15

[email protected];

[email protected];

[email protected]; 16

[email protected] 17

Corresponding author: Lisa Seeb (206) 685-3723 18

19

Page 1 of 47



Draft

Abstract 20

Pink salmon, the most abundant Pacific salmon, have an obligate two-year life cycle 21

that leads to reproductively isolated even- and odd-year lineages. Using new and 22

existing data, we examined the genetic structure of both lineages across their 23

distributional range by genotyping 16 681 SNPs for 383 individuals originating from 24

seven pairs of even- and odd-year populations. Distinct differences in standing pools 25

of genetic variation were identified between the lineages; we observed higher levels 26

of heterozygosity, allelic richness, and significantly more private alleles in the odd-27

year lineage. However, the patterns of population structure were concordant 28

between lineages: the Asian and northern Alaska populations displayed little 29

differentiation but differed significantly from populations in southcentral Alaska and 30

the Pacific Northwest. Our population structure results, in context of known 31

paleoecological information, suggest that both lineages occupied a northern 32

Beringial refugium as well as a Cascadian refugium in North America during the 33

Last Glacial Maximum. These results highlight the influence of historical patterns of 34

habitat availability on contemporary population structure and support the hypothesis 35

of a pre-glacial origin of the lineages.36

Page 2 of 47



Draft

3

Introduction 37

An enduring goal in evolutionary biology is to describe the dynamics that result in population 38

structure, species distribution, and genetic diversity. Groups of organisms that use the same 39

habitat sequentially, separated in time (allochronic species; Alexander and Bigelow 1960), are an 40

intriguing resource to investigate the drivers of population genetic structure and speciation 41

(Taylor and Friesen 2017). These completely isolated groups use the same habitat in tandem, 42

allowing for the study of replicated wild systems and the role of temporal divergence in 43

speciation. Most known allochronic species are periodical insects such as the pine processionary 44

moth (Santos et al. 2007) and aphids (Abbot and Withgott 2004). Vertebrate allochrony as well 45

as yearly allochrony are rare (Taylor and Friesen 2017). 46

Pink salmon Oncorhynchus gorbuscha are the only Pacific salmon and one of the few vertebrates 47

that display allochrony. Pink salmon are anadromous and semelparous, homing to tributaries of 48

the North Pacific Ocean and Bering Sea ranging from Japan in Asia to Washington State in 49

North America (Ruggerone et al. 2010). Their obligate two-year life cycle leads to 50

reproductively isolated even- and odd-year lineages (Anas 1959, Turner and Bilton 1968). These 51

reproductively isolated lineages often spawn in the same locations offset by one year. However, 52

one lineage can be much more abundant than the other in a given location, and relative 53

abundance can shift from one lineage to the other through time (Krkošek et al. 2011, NPAFC 54

2016). Although dominance fluctuates, in recent years the odd-year lineage has been dominant 55

in Asia (Krkošek et al. 2011, Gordeeva 2014). In North America, the pattern is more static: the 56

even-year lineage is dominant in northern drainages while the odd-year lineage dominates in the 57

Page 3 of 47



Draft

4

southern drainages; some rivers at northern and southern extremes are inhabited by only one 58

lineage (Krkošek et al. 2011, Beacham et al. 2012). 59

There are significant phenotypic and life history differences between the lineages even though 60

the two lineages often occupy similar or identical spawning habitats: different gill raker counts 61

(Beacham et al. 1988), different temperature optima (Beacham and Murray 1988b, Beacham and 62

Murray 1988a), and differences in juvenile energy allocation and growth rates (Wechter et al. 63

2017). Beamish (2012) hypothesized that, in North America, the two lineages have adopted 64

different feeding strategies in the early marine period where odd-year pink salmon store fewer 65

lipids and have an extended growing period compared to even-year pink salmon. 66

The existence of distinct lineages, combined with differing abundance of each, has precipitated a 67

number of intriguing ecological observations (Ruggerone and Nielsen 2005). Pink salmon may 68

exert top-down control and are thought to drive major ecological shifts between years of high 69

and low abundances (Springer and van Vliet 2014). Seabird diet, body mass, and reproductive 70

success are reduced in odd-numbered years, when pink salmon abundance has been 71

exceptionally high, due to competition for identical prey (Toge et al. 2011, Springer and van 72

Vliet 2014). Similarly, the abundance of pink salmon affects other Pacific salmon species and 73

has been linked to a decline in productivity of sockeye salmon (O. nerka; Ruggerone et al. 2015). 74

The existence of distinct lineages has also been the focus of comparative studies of population 75

genetic structure. Early studies using allozymes, mtDNA, and microsatellites showed that the 76

strongest signal of divergence occurs between the two lineages: even- and odd-year lineages of a 77

single river system are more genetically diverged than populations of the same lineage from 78

adjacent rivers (Aspinwall 1974, Olsen and Seeb 1998, Churikov and Gharrett 2002, Beacham et 79

Page 4 of 47



Draft

5

al. 2012). More recently, genome scans combined with genetic mapping were used to identify 80

genomic regions of divergence and investigate parallel and divergent selection between the 81

lineages. Seeb et al. (2014) identified parallel signatures of selection in populations of two 82

lineages inhabiting the same environments using over 8 000 SNPs derived from restriction-site-83

associated DNA (RAD) sequencing. Limborg et al. (2014), using the same set of loci aligned to 84

a dense linkage map, identified not only genomic regions of parallel selection, but also regions of 85

divergent selection between lineages, suggesting that adaptation in the two lineages may have 86

arisen from different pools of standing genetic variation. 87

Previous studies of the genetic structure of pink salmon have been largely limited to a single 88

continent or single lineage (e.g., Shaklee and Varnavskaya 1994). We can better address two 89

lingering broad-scale questions by examining populations from both lineages across the entire 90

range of the species. How do the lineages compare in their distribution of diversity and genetic 91

structure across the range? Is there evidence of the same pattern of population structure repeated 92

within the lineages that could help illuminate the evolutionary history of the species? To answer 93

these questions, we built upon the genomic studies in North American populations by Seeb et al. 94

(2014) and Limborg et al. (2014) and examined four new paired replicates along the western 95

Pacific Ocean. This design provided consistent data for a comparison of paired populations from 96

throughout Asia and North America to explore the effects of demographic and evolutionary 97

factors influencing genetic population structure in each lineage. 98

Materials and Methods 99

Samples 100

Page 5 of 47



Draft

6

Paired populations (even- and odd-year lineage populations from the same river) were sampled 101

from four regions of Asia: Hokkaido Island and Amur, Magadan, and Kamchatka Provinces 102

(Figure 1). Populations were chosen based on the presence of both even- and odd-year lineages 103

at the site, the number of available samples, and the geographical spread of locations (Table 1). 104

Samples were collected as a part of North Pacific Anadromous Fish Commission cooperative 105

studies (http://www.npafc.org/new/index.html); fin clips were collected from spawning adults 106

and stored in 95% ethanol at room temperature. Thirty-two individuals from each population 107

were used for DNA sequencing. 108

Previously generated sequence data were obtained for paired populations from three regions of 109

North America: (Norton Sound, Prince William Sound, and Puget Sound (Norton Sound, Prince 110

William Sound, and Puget Sound; Seeb et al. 2014). Twenty-four samples had been sequenced 111

from each of the North American populations with the exception of the Norton Sound even-year 112

population which was limited to 20 samples (Table 1). 113

Detailed information on the 396 samples is included in Supplementary File S1. 114

RAD Sequence Data 115

We used the RAD protocol originally described by Miller et al. (2007), modified for genotyping-116

by-sequencing by Baird et al. (2008). DNA extraction and library preparation for sequencing of 117

the Asian samples followed that of the North American samples and were based on the protocol 118

in Everett et al. (2012). The complete protocol is included in Supplementary Information 119

(Protocol 1). We imported the raw DNA sequences for North American populations from Seeb 120

et al. (2014). 121

Page 6 of 47



Draft

7

Genotyping 122

The 100bp filtered raw sequences from the North American populations and the Asian 123

populations were jointly genotyped to provide continuity between studies. 124

Sequence data was analyzed using STACKS v. 1.31 (Catchen et al. 2013). Default settings were 125

used for all STACKS modules with the following exceptions: process_radtags (-c -r -q -t 94), 126

ustacks (-r --model_type bounded --bound_low 0 --bound_high 0.05), cstacks (-n 2). These 127

settings were used for consistency with analyses from Limborg et al. (2014). The STACKS 128

catalog of variation was created using four individuals from each population. The individuals 129

with the highest number of sequence reads were chosen to represent each population, as these 130

would likely also have the highest quality sequence. The catalog also included four female 131

parents of the crosses used to make a linkage map in collaborative studies; two females from 132

Washington (Limborg et al. 2014) and two from southeastern Alaska were included (unpublished 133

data). Locus names in the catalog were standardized for consistency with the earlier study by 134

Limborg et al. (2014). The genotypes of loci that were present in at least 80% of the individuals 135

were exported with the STACKS program populations. 136

The program GENEPOP v4.3 (Rousset 2008) was used to calculate basic diversity indices for 137

each locus by population. SNPs were retained in the final data set based on the following 138

criteria: 1) if there were multiple putative SNPs per locus, then we retained the single SNP with 139

the highest minor allele frequency (MAF); 2) SNPs that had a MAF > 0.05 in at least one 140

population, and 3) SNPs that conformed to Hardy-Weinberg equilibrium expectations (P> 0.05) 141

in at least seven populations (50% of tests). To filter paralogous loci, we calculated the 142

inbreeding coefficient (FIS) per locus (Robertson and Hill 1984) using GENEPOP v4.3 (Rousset 143

Page 7 of 47



Draft

8

2008), and retained those SNPs that had a global FIS > -0.5 (Hohenlohe et al. 2013). Finally, we 144

excluded individuals that genotyped at less than 80% of the final set of loci. 145

We used all retained SNPs for the analyses. We assumed that the non-neutral SNPs within the 146

dataset were a very small percentage of the total number of SNPs and that their signal would be 147

overwhelmed by the neutral SNPs. Analysis of these North American populations by Seeb et al. 148

(2014) found that 2% of the 8,036 loci they used were non-neutral; using different methods but 149

the same loci, Limborg et al. (2014) found a comparable 2.7% were non-neutral. Further, 150

Limborg et al. (2014) found that the neighbor-joining trees based on either neutral-only or all 151

SNPs were nearly identical. 152

Summary Statistics and Genetic Diversity Measures 153

Several diversity indices were calculated for each population using R, version 3.2.1 (R Core 154

Team 2015). We calculated the mean observed heterozygosity (Ho) within each population and 155

compared the values for the populations within the even-year lineage to those within the odd-156

year lineage using the nonparametric Mann-Whitney test (Mann and Whitney 1947) in R. We 157

also calculated the standardized observed heterozygosity per individual (HS_obs) within each 158

population with the R function genhet (Coulon 2010); the mean values of HS_obs for each 159

population were compared between lineages as above with the nonparametric Mann-Whitney test 160

in R. HS_obs is a conservative estimate of the observed heterozygosity of an individual averaged 161

over all the loci used, standardized by the mean observed heterozygosity of the genotyped loci. 162

HS_obs is standardized to account for individuals that are not genotyped at the same loci to ensure 163

that the heterozygosity is measured on the same scale (Coltman et al. 1999, Amos 2005). We 164

used the package hierfstat (Goudet and Jombart 2015) in R, to determine the rarefied allelic 165

Page 8 of 47



Draft

9

richness per locus by population as another measure of the genetic diversity in the populations 166

and calculated average values for each population and lineage; mean values for the populations 167

were compared between lineages as above using the nonparametric Mann-Whitney test in R. 168

Comparison of the distributions of private alleles between lineages was conducted in R using a 169

two sample Kolmogorov-Smirnov test. 170

Population Structure 171

Several methods were used to evaluate population structure within and between lineages. To get 172

a broad view of the relationship of the 14 populations, we constructed a Neighbor-Joining (N-J) 173

tree (Saitou and Nei 1987) using the DA distance (Nei 1983) with 10 000 bootstraps of the loci, 174

using the program POPTREE2 (Takezaki et al. 2010). We also estimated per locus FST, pairwise 175

FST between populations, and FST for groupings of populations based on Weir and Cockerham 176

(1984). Pairwise and per locus global FST estimates were obtained using GENEPOP. Overall 177

FST and FST of six different groups of populations along with their respective confidence 178

intervals were calculated with GENETIX v.4.05.2. The six groupings were as follows: 1) overall 179

FST of all 14 populations, 2) pairwise FST comparing populations of different lineages at the same 180

location, 3) all even-year lineage populations, and 4) all odd-year populations. We estimated FST 181

values separately for each lineage to characterize the magnitude of differentiation between the 182

populations within each lineage. Finally, we estimated the FST for groups of populations within 183

each lineage to compare the magnitude of the differentiation. The N-J tree informed the 184

grouping of the populations tested within each lineage, as a clear parallel population structure 185

was evident in the tree. The populations were grouped within each lineage as follows: 5) even-186

year lineage populations Prince William Sound and Puget Sound, called the Gulf of Alaska 187

Page 9 of 47



Draft

10

(GoA) populations, and Hokkaido, Amur, Magadan, Kamchatka and Norton Sound populations, 188

which we called the Asia/Bering Sea (Asia/Ber) populations and 6) odd-year lineage populations 189

grouped into the same Asia/Ber and GoA populations. Confidence intervals (95%) were 190

estimated using 1 000 bootstraps. 191

We used an individual-based analysis of principal components (PCA) that included all 14 192

populations to explore the relationships among populations. The eigenvalues were calculated in 193

PLINK v1.90 (Chang et al. 2015) which uses a variance-standardized relationship matrix; the 194

input was pairwise relatedness between individuals. We extracted the top 20 principal 195

components using the default settings with the following exceptions (flags: --allow-extra-chr --196

allow-no-sex). The top three principal components were visualized in R, using the package 197

ggplot2 (Wickham 2009). 198

Three analyses of molecular variance (AMOVA) were conducted in Arlequin v.3.5.1.2 199

(Excoffier and Lischer 2010), using 10 000 permutations, to test for hierarchical population 200

structure. The first AMOVA (A) evaluated the distribution of genetic diversity within and 201

between the two lineages. The second hierarchy was tested with two AMOVAs (B and C), each 202

evaluated the distribution of diversity within each lineage; populations within lineage were 203

grouped as follows: i) GoA: Puget Sound and Prince William Sound, and ii) Asia/Ber: Hokkaido, 204

Amur, Magadan, Kamchatka and Norton Sound. 205

Results 206

Bioinformatics and Genotyping 207

Page 10 of 47



Draft

11

Combining the new sequences for the Asian populations with the existing sequences for the 208

North American populations resulted in a total of 903.5 million retained reads for 396 209

individuals (Supplementary File S2). On average, each individual from Asia had two million 210

reads, and each individual from North America had 2.8 million reads. From these retained reads, 211

the Stacks pipeline identified 666 351 loci, of which 16 681 putative SNPs remained after 212

filtering (Supplementary File S2). Thirteen individuals, possibly of low sample quality, were 213

genotyped at fewer than 80% of the putative 16 681 SNPs and were dropped from the analysis. 214

These included five individuals from Magadan even-year population, four from Magadan odd-215

year population, two from Norton Sound even-year population, and one each from the 216

Kamchatka even- and odd-year populations. Final sample sizes for each population are listed in 217

Table 2. 218

Diversity within populations 219

We estimated diversity within each of the 14 populations and compared the results between the 220

lineages (Table 2). We found significant difference in the distributions of the average observed 221

heterozygosity (Ho) per population with 0.158 estimated in the even-year lineage and 0.167 in 222

the odd-year lineage (P-value = 0.0055, one-tailed). Additionally, for each comparison of the 223

paired populations, the population within the even-year lineage had lower average observed and 224

expected (He) heterozygosities (Table 2). The individual standardized observed heterozygosity 225

(Hs_obs) followed the same pattern seen in Ho and He (Table 2, Figure 2). In nearly every 226

location, the populations of the even-year lineage had lower mean values of Hs_obs than those in 227

the odd-year lineage. The overall mean of Hs_obs was 0.966 in the even-year and 1.024 in the 228

odd-year lineage, and the distributions in the two groups differed significantly (P-value = 0.0055, 229

Page 11 of 47



Draft

12

one-tailed). The mean allelic richness was also significantly lower for the even-year lineage 230

(1.60 for the even-year lineage and 1.65 for the odd-year lineage; P-value = 0.0189, one-tailed; 231

Table 2). We also tallied lineage-specific private alleles: some loci were variable in one lineage 232

but were fixed for a single allele in the other. The number of private alleles in the odd-year 233

lineage was greater than the number of private alleles in the even-year lineage (2 300 vs. 1 184). 234

In addition, odd-year private alleles had overall greater allele frequencies than even-year private 235

alleles (Figure 3). 236

Lower values for average Ho and He were seen in both lineages in the most southern populations 237

at the end of the ranges on both continents (Puget Sound and Hokkaido populations). Similarly, 238

the lowest allelic richness values were observed at the southern extent of the ranges. 239

Diversity among populations 240

The N-J tree shows a very distinct division between the lineages, with all branches having strong 241

bootstrap support. All but one node was supported by 100% of the bootstrap samplings (Figure 242

4). Within both lineages, the most divergent branch of the tree separated the GoA populations of 243

Prince William Sound and Puget Sound from the Asia/Ber group including all Asian populations 244

plus the Norton Sound populations from northwestern Alaska. The branch lengths for the GoA 245

group were longer in the odd-year lineage than the even-year lineage, reflecting the greater 246

differentiation seen in the pairwise FST for these populations. 247

The principal component analyses provide additional perspectives, but like the N-J tree, the 248

greatest amount of variation was explained by the division between the lineages (percent of 249

variation explained by principal component 1: 30.24%) (Figure 5A, B). The second greatest 250

Page 12 of 47



Draft

13

amount of variation in the data was explained by differentiation between the populations in the 251

odd-year lineage (principal component 2: 13.83%, Figure 5A, C). The differentiation appears to 252

be dominated by the differences between the GoA populations within each lineage, as the 253

Asia/Ber populations cluster more closely within lineages. Comparing the first principal 254

component to the third shows that the differentiation among populations in the even-year lineage 255

explains the third greatest amount of variation in the data (principal component 3: 8.73%, Figure 256

5B, C), and comparing the second and third principal components more clearly illustrates the 257

relationship of the populations within each lineage by removing the visual pattern of the 258

differences between the even and odd-year lineages that was shown in the first principal 259

component (Figure 5C). To compare the relationship of all 14 populations in three dimensions 260

simultaneously, the centroids of each population were plotted in a three dimensional PCA using 261

the R package scatterplot3d (Ligges 2017) (Figure S1). 262

Non-hierarchical diversity analyses were conducted to quantify the distribution of variation. We 263

estimated an FST value of 0.060 over all populations (Table 3A). Populations within the odd-264

year lineage had a significantly higher FST (0.033) than populations within the even-year lineage 265

(0.026) when estimates were conducted with all populations within a lineage (Table 3B). The 266

FST for the even-year Asia/Ber populations (0.011) and the GoA populations (0.030) differed 267

significantly, but the difference was smaller than that observed in the odd-year lineage: Asia/Ber 268

(0.008) GoA (0.052) (Table 3B). The estimate of FST for the odd-year Asia/Ber populations 269

(0.008) was smaller than that of the even-year lineage (0.011). The opposite pattern was seen in 270

the GoA populations, where the even-year lineage (0.030) had a smaller estimate of FST than the 271

odd-year lineage (0.052). Pairwise comparisons of FST for all 14 populations ranged from 0.001 272

Page 13 of 47



Draft

14

to 0.148 with the largest values generally resulting from comparisons between lineages (Figure 273

S2). 274

Hierarchical analyses were used to partition the genetic variation and estimate the distribution of 275

variation. The first AMOVA (Table 4A), estimated that a total of 5.72% of the variation was 276

explained by differences between the lineages with 2.87% attributed to differences among 277

populations within the lineages. Both estimates were significantly different from zero (P< 278

0.0001). 279

Separate AMOVAs were conducted for each lineage to evaluate the partitioning of genetic 280

variation between the GoA and Asia/Ber groups within lineage (Table 4B and C). In both 281

AMOVAs, differences among the groupings explained the largest amount of the variation, 282

followed by variation attributed to differences among populations within the groups. This was a 283

recurrent pattern observed in all the analyses, seen in both the diversity and the differentiation 284

statistics. 285

Discussion 286

The geographic overlap of allochronic lineages of pink salmon provides a rare opportunity to 287

examine the repeatability of evolutionary processes. We used a paired sampling design across 288

the entire range to provide a detailed comparison of the distribution of diversity in each lineage. 289

In general, population structure was similar between lineages, likely a reflection of similar 290

demographic histories. However, the extent and distribution of individual and population 291

diversity varied significantly, suggesting differences in the origin and age of the lineages. 292

Population structure within lineages 293

Page 14 of 47



Draft

15

The paired study design allowed us to compare patterns of population structure across the 294

species' range. Two predominant relationships were consistently supported by all analyses. The 295

greatest signal of divergence among the 14 populations was between the lineages, a result 296

consistent with earlier studies. The second predominant relationship was a strong parallel signal 297

of little differentiation between Asian and northern Alaska populations (Asia/Ber), but 298

significant differentiation between populations in northern Alaska and those from southcentral 299

Alaska and the Pacific Northwest of North America (GoA). This pattern was evident in both 300

lineages and is likely the effect of the isolation of the Asia/Ber populations (Hokkaido, Amur, 301

Magadan, Kamchatka and Norton Sound) from the GoA populations (Prince William Sound and 302

Puget Sound) in different glacial refugia. The historical reproductive isolation of the Asia/Ber 303

and GoA populations, adaptations to distinct refugial habitats, and differences incurred through 304

subsequent post-glacial colonization, would all have undoubtedly led to separate demographic 305

histories within the isolated groups. These drivers would explain the variation in pattern of 306

genetic differentiation seen in the two groups of populations. For example, in contrast to the 307

homogeneity within the lineages in Asia/Ber, the GoA populations show a greater genetic 308

differentiation within lineages. In North America, the greatest differentiation among populations 309

is in the odd-year lineage, a pattern that was also observed by Beacham et al. (2012) with 310

microsatellite loci. 311

For the Asian populations, earlier allozyme studies found significant population structure in the 312

even-year lineage but only weak population structure in the odd-year lineage (Noll et al. 2001, 313

Hawkins et al. 2002). Moreover, these studies found evidence for an isolation-by-distance 314

relationship among populations in both lineages with the strongest relationship observed in the 315

even-year lineage. We also found higher FST values among the even-year lineage than among 316

Page 15 of 47



Draft

16

the odd-year lineage, but the limited number of populations precluded an isolation-by-distance 317

analysis. 318

We observed an additional trend that the southern-most populations on both sides of the Pacific 319

Ocean exhibited the largest pairwise FST values between lineages at a single location. The 320

cumulative effect of random genetic drift on the populations at the edge of the range is a likely 321

explanation for the pattern of greater values of pairwise FST at the extremes of the range. 322

Diversity within lineages 323

There were consistent significant differences in population diversity between the lineages; these 324

differences may offer clues to differences in the pools of standing genetic variation and genetic 325

backgrounds. Limborg et al. (2014) found significantly higher observed heterozygosity in the 326

odd-year lineage for loci considered candidates for selection and hypothesized that adaptation 327

arose from different pools of standing genetic variation in the two lineages. In contrast, we 328

found significant differences in heterozygosity across the full range of loci, not just limited to a 329

subset of highly diverged loci. The ability to detect the higher heterozygosity across the full set 330

of loci is likely due to the inclusion of the Asian populations which had a strong signal of higher 331

observed heterozygosity in the odd-year lineage. Like Limborg et al. (2014), when we limited the 332

comparison of observed heterozygosity between the two lineages to three North American 333

populations (using all 16 681 SNPs), we did not find a significant difference between the 334

lineages (results not shown). 335

Our results can be compared to previous allozyme studies of Asian populations of pink salmon: 336

those studies surveyed a larger number of populations but with far fewer loci (<40). Hawkins et 337

Page 16 of 47



Draft

17

al. (2002) focused on odd-year pink salmon and contrasted results to the allozyme study of even-338

year pink salmon of Noll et al. (2001). Hawkins et al. (2002) found that the even-year lineage 339

had a significantly higher average heterozygosity (0.059) than the odd-year lineage (0.044). The 340

contradictory findings between the early allozyme studies and the significant findings of our 341

genomic (RAD) study are not easily reconciled. There are large differences in number and type 342

of loci (genome wide coverage vs. 40 enzyme coding loci most often important in oxidative 343

metabolism), but the allozyme studies had a larger number of populations with denser 344

geographic coverage. However, we believe our dataset provides very strong evidence that 345

heterozygosities are greater in the odd-year lineage. Surveys of additional Asian populations 346

with genomic methods could provide additional evidence. 347

A striking finding of our study was consistent and significant differences in observed individual 348

heterozygosities; individuals of the odd-year lineage have significantly greater observed 349

heterozygosity than individuals of the even-year lineage. Moreover, there are more private 350

alleles in the odd-year population than private alleles in the even-year populations. The allele 351

frequencies of odd-year private alleles are also significantly greater than the frequencies of the 352

even-year private alleles. These results likely reflect a combination of factors including a 353

potentially older age of the odd-year lineage or a past reduction in population size in the even-354

year lineage. 355

These trends are interesting in light of previous studies linking population diversity with adaptive 356

potential in pink salmon lineages. In a study of transplantation of both lineages from the 357

Kamchatka Peninsula to northern Europe, habitat outside their native range, the odd-year lineage 358

was able to readily establish self-sustaining populations, while the even-year lineage was not 359

Page 17 of 47



Draft

18

(Gordeeva and Salmenkova 2011). The pattern of routine failure of even-year pink salmon 360

transplants led Gordeeva and Salmenkova (2011) to propose that the lineages may differ in 361

adaptive plasticity and that the even-year lineage may have arisen from the odd-year lineage. 362

Further, the authors suggest that the even-year lineage may have lost a part of its genetic 363

variation with which adaptive potential is associated. Our results are consistent with these 364

hypotheses. 365

Phylogeographic relationship of the lineages 366

Northern Alaskan populations of many species share affinities with Asian populations due to the 367

transient geographical connection between the Asian and North American continents (Pielou 368

1991). The habitat along the coastal North Pacific Ocean has changed drastically with the 369

oscillating climate and resulting cycles of glaciation during the last million years (Brigham-370

Grette et al. 2003). The LGM was a period during the most recent glacial cycle with the highest 371

global ice volume (Ehlers and Gibbard 2007) and is thought to have reached its peak 372

approximately 26.5 thousand years ago (Clark et al. 2009). Most of eastern Asia and western 373

Alaska were ice-free, whereas extensive ice sheets covered the northern latitudes of North 374

America down to 40°N in some parts of the continent (Hewitt 2000, Brigham-Grette et al. 2003, 375

Gualtieri et al. 2005). These glaciers reduced the volume of water in the oceans, causing the sea 376

level to fall between 120 to 130m lower than present (Hopkins 1973, Rohling et al. 1998). The 377

lower coastline exposed the continental shelf connecting the Asian and North American 378

continents (Bering Land Bridge) and provided vast, ice-free habitat in Beringia (Hultén 1937). 379

The dynamic habitat of the North Pacific Rim has influenced the contemporary population 380

structure of many species of varying taxa (McPhail and Lindsey 1986a, Hoffman et al. 2006) 381

Page 18 of 47



Draft

19

including salmonids (Gharrett et al. 1988, Smith et al. 2001, Moran et al. 2013, Petrou et al. 382

2013). 383

The origin of the two pink salmon lineages has consistently been dated prior to the LGM. 384

Estimates of lineage divergence vary widely from approximately 26 thousand years ago and up 385

to one million years ago (Brykov et al. 1996, Churikov and Gharrett 2002), and the origin(s) of 386

the lineages and yearly allochrony has been debated in the literature. Various authors have 387

hypothesized that the even-year lineage was ancestral (e.g., Churikov and Gharrett 2002), the 388

odd-year lineage was ancestral (Gordeeva and Salmenkova 2011), or that the ancestral lineage 389

could have varied across the range with multiple independent origins (Churikov and Gharrett 390

2002, Hawkins et al. 2002). 391

The parallel population structure that we observe suggests that both lineages were established 392

around the Pacific Rim before vicariance due to Pleistocene glaciation, and the reduced diversity 393

at the extremes of the range on both continents support a northern origin for pink salmon. 394

Subsequently, evidence suggests that the contemporary Asian and Norton Sound populations of 395

both lineages are descended from pink salmon populations that persisted in Asia and the Bering 396

land bridge during the LGM. These refugial populations were likely not directly impacted by 397

glacial ice; instead, their habitat was primarily altered by lower coastlines and a drastic increase 398

in the amount of available habitat. 399

In addition to Beringia, there is evidence that Cascadia was a second major glacial refugium 400

(McPhail and Lindsey 1986b, Pielou 1991, Shafer et al. 2010) (Figure 1). With the advance of 401

the Cordilleran ice sheets in North America during the LGM, terrestrial species were forced 402

either northwest of the ice to the available habitat in Beringia or south of the ice, where much of 403

Page 19 of 47



Draft

20

the continent was ice-free (McPhail and Lindsey 1986b, Pielou 1991). Ocean-dependent species 404

in the northern latitudes relied on the Pacific Northwest for refuge south of the North American 405

ice sheets (Westlake and O'Corry-Crowe 2002, Hoffman et al. 2006) and as well as ice-free 406

regions of the Queen Charlotte Islands in British Columbia (Warner et al. 1982, Smith et al. 407

2001, Beacham et al. 2009). 408

Many hypothesize that allochrony and the divergence of even- and odd-year lineages was 409

accelerated through adaptation to the environment of distinct historical refugia (Aspinwall 1974, 410

Churikov and Gharrett 2002, Beacham et al. 2012). Their hypothesis is based on the 411

asymmetrical abundance distribution in North America where the even-year lineage is most 412

abundant in northern populations and the odd-year lineage is most abundant in southern 413

populations. We observed parallel signals across lineages in both continents when comparing 414

population structure in even- and odd-year populations across Asia and North America although 415

the odd-year populations were more divergent than the even-year populations in North America. 416

This suggests that the contemporary population structures of both lineages were shaped by 417

similar phylogeographic processes in the recent past and that both lineages shared two main 418

refugia during the LGM, Beringia in the north and Cascadia in the south. Other potential refugia 419

important for pink salmon include the Queen Charlotte Islands in British Columbia as well as the 420

southern Asian mainland and the islands of Japan (Beacham et al. 2009). 421

Within both lineages, the least amount of differentiation occurs between the Asia/Ber 422

populations, which may be evidence for a history of gene flow between those populations. There 423

is evidence that pink salmon have the highest straying rate of the Pacific salmon: early migration 424

to saltwater immediately after hatching suggest that pink salmon may have less opportunity for 425

Page 20 of 47



Draft

21

imprinting on their freshwater rearing environment, resulting in higher straying rates as returning 426

adults (Quinn 2005). Pink salmon are also rapid colonizers of newly emerging habitats (Milner 427

and Bailey 1989, Pess et al. 2012). There was an increase in the available habitat during the Last 428

Glacial Maximum (LGM) (Figure 1) as well as alternative river paths that joined the mouths of 429

Western Alaska rivers into a common drainage that flowed into the Bering Sea through the 430

Bering Land Bridge (Hopkins 1973, Knebel and Creager 1973, Lindsey and McPhail 1986). 431

This increase in habitat placed northwestern Alaska drainages into closer proximity to the 432

mouths of the rivers in Asia and may have resulted in connectivity and subsequent gene flow 433

among the Beringian populations. 434

The greatest amount of genetic differentiation within both lineages occurs between the GoA 435

populations. Populations within lineages may have been isolated within the Cascadia refugia by 436

glacial ice and experienced differing demographic processes of random genetic drift and post-437

glacial recolonization; or inherent differences in standing genetic variation could account for the 438

differing population structure within the lineages in North America. Additional populations 439

from North America would help to clarify the relationship among populations within each 440

lineage and evaluate whether the odd-year survived in more southern ice-free areas compared to 441

the even-year lineage. 442

Allochrony and pink salmon 443

The significant differences in observed heterozygosity, allelic richness among populations, and 444

the distribution of private alleles provide strong evidence for an overall reduction of genetic 445

diversity in the even-year compared to the odd-year lineage. These differences support the 446

hypothesis that the two allochronic lineages exhibit distinct genetic backgrounds and pools of 447

Page 21 of 47



Draft

22

standing variation. The differences appear consistent across the range with implications for 448

adaptive divergence and geographic distribution. 449

Looking into a future with a warming climate (Karl et al. 2009 , Nielsen et al. 2013), we 450

anticipate the response of the lineages to such pressures will often be correlated. Similar 451

responses to climate change have already been observed in both lineages with changes to 452

migration timing thought to better match new environmental conditions. Kovach et al. (2013) 453

observed earlier migration timing in both lineages, and evidence for the genetic basis of the 454

change in the odd-year lineage is accumulating (Kovach et al. 2012, Manhard et al. 2017). 455

However, the differing genomic backgrounds may also result in divergent adaptive responses and 456

differential range shifts of the lineages. Pink salmon are currently the most abundant Pacific 457

salmonid with dramatic increases in catches in the odd-year lineage particularly in the southern 458

portion of the range in North America (NPAFC 2012). Recent studies by Wechter et al. (2017) 459

on growth of juvenile pink salmon in the Bering Sea support the hypothesis of Beamish (2012) 460

that the odd-year lineage allocates more energy to growth rather than to energy storage; this life 461

history strategy may be particularly successful in warming conditions. Limborg et al. (2014) 462

identified genomic regions of differentiation between the lineages. Additional studies may help 463

identify the genomic regions important to lineage differentiation and response to climatic 464

variability. These genomic data should increase our understanding of whether current warming 465

trends in the southern portions of the range favor odd-year genotypes as well as better predict 466

how abundance and distribution of even- and odd-year populations might change in arctic and 467

subarctic regions of North America and Asia in response to warming conditions. 468

469

Page 22 of 47



Draft

23

Acknowledgements 470

Asian samples were provided through exchanges promoted by North Pacific Anadromous Fish 471

Commission; North American samples were provided by the Alaska Department of Fish and 472

Game and the Washington Department of Fish and Wildlife. We would like to thank members 473

of the Seeb Laboratory, especially Ryan Waples and Morten T. Limborg, for their advice and 474

contribution to the bioinformatics, and Carita Pascal for assistance in the laboratory. CT was 475

partially funded by Richard T. Whiteleather, Fisheries B.S. 1935 Endowed Scholarship from the 476

School of Aquatic and Fishery Sciences, University of Washington. Additional funding was 477

provided through Cooperative Agreement Number COOP-13-085 from the Alaska Department 478

of Fish and Game.479

Page 23 of 47



Draft

24

References 480

Abbot, P., and J.H. Withgott. 2004. Phylogenetic and molecular evidence for allochronic 481

speciation in gall-forming aphids (Pemphigus). Evolution 58(3): 539-553. 482

Alexander, R.D., and R.S. Bigelow. 1960. Allochronic speciation in field crickets, and a new 483

species, Acheta veletis. Evolution 14: 334-346. 484

Amos, W. 2005. IRmacroN3 documentation. URL: www.zoo.cam.ac.uk/directory/william-485

amos/. Accessed June, 2016. 486

Anas, R.E. 1959. Three-year-old pink salmon. Journal of Fisheries Research Board of Canada 487

16(1): 91-94. 488

Aspinwall, N. 1974. Genetic analysis of North American populations of the pink salmon, 489

Oncorhynchus gorbuscha, possible evidence for the neutral mitation-random drift hypothesis. 490

Evolution 28(2): 295-305. 491

Baird, N.A., P.D. Etter, T.S. Atwood, M.C. Currey, A.L. Shiver, Z.A. Lewis, . . . E.A. Johnson. 492

2008. Rapid SNP discovery and genetic mapping using sequenced RAD markers. PloS one 493

3(10): e3376. 10.1371/journal.pone.0003376 494

Beacham, T.D., J.R. Candy, K.D. Le, and M. Wetklo. 2009. Population structure of chum 495

salmon (Oncorhynchus keta) across the Pacific Rim, determined from microsatellite analysis. 496

Fish. Bull. 207: 244-260. 497

Beacham, T.D., B. McIntosh, C. MacConnachie, B. Spilsted, and B.A. White. 2012. Population 498

structure of pink salmon (Oncorhynchus gorbuscha) in British Columbia and Washington, 499

determined with microsatellites. Fisheries Bulletin 110: 242-256. 500

Beacham, T.D., and C.B. Murray. 1988a. Genetic analysis of growth and maturity in pink 501

salmon (Oncorhynchus gorbuscha). Genome 30: 529-535. 502

Beacham, T.D., and C.B. Murray. 1988b. Variation in developmental biology of pink salmon 503

(Oncorhynchus gorbuscha) in British Columbia. Can. J. Zool. 66: 2634-2648. 504

Beacham, T.D., R.E. Withler, C.B. Murray, and L.W. Barner. 1988. Variation in Body Size, 505

Morphology, Egg Size, and Biochemical Genetics of Pink Salmon in British Columbia. 506

Transactions of the American Fisheries Society 117(2): 109-126. 10.1577/1548-507

8659(1988)117<0109:vibsme>2.3.co;2 508

Page 24 of 47



Draft

25

Beamish, R.J. 2012. Observations and speculations on the reasons for recent increased in pink 509

salmon production. International Workshop on Explanations for the High Abundance of Pink 510

and Chum Salmon and Future Trends. North Pacific Anadromous Fish Commission Technical 511

Report No. 8: 1-8. http://www.npafc.org/new/pub_technical8.html 512

Brigham-Grette, J., L.M. Gualtieri, O.Y. Glushkova, T.D. Hamilton, D. Mostoller, and A. Kotov. 513

2003. Chlorine-36 and C-14 chronology support a limited Last Glacial Maximum across central 514

Chukotka, northeastern Siberia, and no Beringian ice sheet. Quaternary Research 59(3): 386-398. 515

10.1016/s0033-5894(02)00000-0 516

Brykov, V.A., N. Polyakova, L.A. Skurikhina, and A.D. Kukhlevsky. 1996. Geographical and 517

temporal mitochondrial DNA variability in populations of pink salmon. Journal of Fish Biology 518

48: 899-909. 519

Catchen, J., P.A. Hohenlohe, S. Bassham, A. Amores, and W.A. Cresko. 2013. Stacks: an 520

analysis tool set for population genomics. Molecular Ecology 22(11): 3124-3140. 521

10.1111/mec.12354 522

Catchen, J.M., A. Amores, P. Hohenlohe, W. Cresko, and J.H. Postlethwait. 2011. Stacks: 523

building and genotyping loci de novo from short-read sequences. Genetics 1: 171-182. 524

10.1534/g3.111.000240/-/DC1 525

Chang, C.C., C.C. Chow, L.C. Tellier, S. Vattikuti, S.M. Purcell, and J.J. Lee. 2015. Second-526

generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 4: 7. 527

10.1186/s13742-015-0047-8 528

Churikov, D., and A.J. Gharrett. 2002. Comparative phylogeography of the two pink salmon 529

broodlines: an analysis based on a mitochondrial DNA genealogy. Molecular Ecology 11: 1077-530

1101. 531

Clark, P.U., A.S. Dyke, J.D. Shakun, A.E. Carlson, J. Clark, B. Wohlfarth, . . . A.M. McCabe. 532

2009. The Last Glacial Maximum. Science 325(5941): 710-714. 10.1126/science.1172873 533

Coltman, D.W., J.G. Pilkington, J.A. Smith, J. M., and J.M. Pemberton. 1999. Parasite-Mediated 534

selection against inbred Soay sheep in a free-living, island population. Evolution 53(4): 1259-535

1267. 536

Coulon, A. 2010. genhet: an easy-to-use R function to estimate individual heterozygosity. URL: 537

http://www.aureliecoulon.net/research/ac-computer-programs.html. Mol Ecol Resour 10(1): 167-538

169. 10.1111/j.1755-0998.2009.02731.x 539

Page 25 of 47



Draft

26

Ehlers, J., and P.L. Gibbard. 2007. The extent and chronology of Cenozoic Global Glaciation. 540

Quaternary International 164-165: 6-20. 10.1016/j.quaint.2006.10.008 541

Everett, M.V., M.R. Miller, and J.E. Seeb. 2012. Meiotic maps of sockeye salmon derived from 542

massively parallel DNA sequencing. BMC Genomics 13: 521. 543

Excoffier, L., and H.E. Lischer. 2010. Arlequin suite ver 3.5: a new series of programs to 544

perform population genetics analyses under Linux and Windows. Mol Ecol Resour 10(3): 564-545

567. 10.1111/j.1755-0998.2010.02847.x 546

Gharrett, A.J., C. Smoot, A.J. McGregor, and P.B. Holmes. 1988. Genetic relationships of even-547

year northwestern Alaskan pink salmon. Trans. Am. Fish. Soc. 117: 536-545. 548

Gordeeva, N.V. 2014. Estimates of population differentiation in pink salmon Oncorhynchus 549

gorbuscha with using of the microsatellite markers may be underestimated due to high 550

population sizes. Journal of Ichthyology 54(5): 347-358. 10.1134/S0032945214030059 551

Gordeeva, N.V., and E.A. Salmenkova. 2011. Experimental microevolution: transplantation of 552

pink salmon into the European North. Evolutionary Ecology 25(3): 657-679. 10.1007/s10682-553

011-9466-x 554

Goudet, J., and T. Jombart. 2015. hierfstat: estimation and tests of hierarchical F-statistics. URL: 555

https://cran.r-project.org/web/packages/hierfstat/hierfstat.pdf. 556

Gualtieri, L., S. Vartanyan, J. Brigham-Grette, and P. Anderson. 2005. Evidence for an ice-free 557

Wrangel Island, northeast Siberia during the Last Glacial Maximum. Boreas 34(3): 264-273. 558

10.1080/03009480510013097 559

Hawkins, S., N.V. Varnavskaya, M.E. A., V.V. Efremov, C.M. Guthrie III, R.L. Wilmot, . . . A.J. 560

Gharrett. 2002. Population structure of odd-broodline Asian pink salmon and its contrast to the 561

even-broodline structure. Journal of Fish Biology 60(2): 370-388. 10.1006/jfbi.2001.1850 562

Hewitt, G. 2000. The genetic legacy of the Quaternary Ice Ages. Nature 405: 907-913. 563

Hoffman, J.I., C.W. Matson, W. Amos, T.R. Loughlin, and J.W. Bickham. 2006. Deep genetic 564

subdivision within a continuously distributed and highly vagile marine mammal, the Steller's sea 565

lion (Eumetopias jubatus). Mol Ecol 15(10): 2821-2832. 10.1111/j.1365-294X.2006.02991.x 566

Hohenlohe, P.A., M.D. Day, S.J. Amish, M.R. Miller, N. Kamps-Hughes, M.C. Boyer, . . . G. 567

Luikart. 2013. Genomic patterns of introgression in rainbow and westslope cutthroat trout 568

illuminated by overlapping paired-end RAD sequencing. Molecular Ecology 22(11): 3002-3013. 569

10.1111/mec.12239 570

Page 26 of 47



Draft

27

Hopkins, D.M. 1973. Sea level history in Beringia during the past 250,000 years. Quanternary 571

Research 3: 520-540. 572

Hultén, E. 1937. Outline of the history of Arctic and Boreal biota during the Quarternary period, 573

Stockholm. 574

Karl, T.R., J.M. Melilo, and T.C. Peterson. 2009. Global climate change impacts in the United 575

States. National Science and Technology Council Cambridge University Press. 576

Knebel, H.J., and J.S. Creager. 1973. Yukon River: Evidence for Extensive Migration during the 577

Holocene Transgression. Science 179(4079): 1230-1232. 10.1126/science.179.4079.1230 578

Kovach, R.P., A.J. Gharrett, and D.A. Tallmon. 2012. Genetic change for earlier migration 579

timing in a pink salmon population. Proceedings of the Royal Society B: Biological Sciences 580

279(1743): 3870-3878. 10.1098/rspb.2012.1158 581

Kovach, R.P., J.E. Joyce, J.D. Echave, M.S. Lindberg, and D.A. Tallmon. 2013. Earlier 582

Migration Timing, Decreasing Phenotypic Variation, and Biocomplexity in Multiple Salmonid 583

Species. PLOS ONE 8(1): e53807. 10.1371/journal.pone.0053807 584

Krkošek, M., R. Hilborn, R.M. Peterman, and T.P. Quinn. 2011. Cycles, stochasticity and 585

density dependence in pink salmon population dynamics. Proceedings of the Royal Society B: 586

Biological Sciences 278(1714): 2060-2068. 10.1098/rspb.2010.2335 587

Ligges, U. 2017. scatterplot3d. p. Plots a three dimensional (3D) point cloud. 588

Limborg, M.T., R.K. Waples, J.E. Seeb, and L.W. Seeb. 2014. Temporally isolated lineages of 589

pink salmon reveal unique signatures of selection on distinct pools of standing genetic variation. 590

The Journal of heredity 105(6): 741-751. 10.1093/jhered/esu063 591

Lindsey, C.C., and J.D. McPhail. 1986. Zoogeography of fishes of the Yukon and Mackenzie 592

basins. In The zoogeography of North American freshwater fishes. Edited by C.H. Hocutt and 593

E.O. Wiley. Wiley, New York. pp. 639-674. 594

Manhard, C.V., J.E. Joyce, and A.J. Gharrett. 2017. Evolution of phenology in a salmonid 595

population: a potential adaptive response to climate change. Canadian Journal of Fisheries and 596

Aquatic Sciences: 1-9. 10.1139/cjfas-2017-0028 597

Mann, H.B., and D.R. Whitney. 1947. On a test of whether one of two random variables in 598

stochastically larger than the other. The Annals of Mathematical Statistics 18(1): 50-60. 599

Page 27 of 47



Draft

28

McPhail, J.D., and C.C. Lindsey. 1986a. Zoogeography of fishes of the Yukon and Mackenzie 600

Basins. In The zoogeography of North American freshwater fishes. Edited by C.H. Hocutt and 601

E.O. Wiley. John Wiley and Sons. pp. 639-674. 602

McPhail, J.D., and C.C. Lindsey. 1986b. Zoogeography of the freshwater fishes of Cascadia (the 603

Columbia system and rivers north to the Stikine). In The zoogeography of North American 604

freshwater fishes. Edited by C.H. Hocutt and E.O. Wiley. John Wiley and Sons. pp. 615-637. 605

Miller, M.R., J.P. Dunham, A. Amores, W.A. Cresko, and E.A. Johnson. 2007. Rapid and cost-606

effective polymorphism identification and genotyping using restriction site associated DNA 607

(RAD) markers. Genome Res 17(2): 240-248. 10.1101/gr.5681207 608

Milner, A.M., and R.G. Bailey. 1989. Salmonid colonization of new streams in Glacier Bay 609

National Park, Alaska. Aquaculture Research 20(2): 179-192. 10.1111/j.1365-610

2109.1989.tb00343.x 611

Moran, P., D.J. Teel, M.A. Banks, T.D. Beacham, M.R. Bellinger, S.M. Blankenship, . . . E. 612

Taylor. 2013. Divergent life-history races do not represent Chinook salmon coast-wide: the 613

importance of scale in Quaternary biogeography. Canadian Journal of Fisheries and Aquatic 614

Sciences 70(3): 415-435. 10.1139/cjfas-2012-0135 615

Nei, M. 1983. Accuracy of estimated phylogenetic trees from molecular data. II. Gene frequency 616

data. Journal of Molecular Evolution 19(2): 153-170. 617

Nielsen, J.L., G.T. Ruggerone, and C.E. Zimmerman. 2013. Adaptive strategies and life history 618

characteristics in a warming climate: Salmon in the Arctic? Environmental Biology of Fishes 619

96(10): 1187-1226. 10.1007/s10641-012-0082-6 620

Noll, C., N.V. Varnavskaya, E.A. Matzak, S.L. Hawkins, V.V. Midanaya, O.N. Katugin, . . . A.J. 621

Gharrett. 2001. Analysis of contemporary genetic structure of even-broodyear populations of 622

Asian and western Alaskan pink salmon, Oncorhynchus gorbuscha. Fishery Bulletin 99(1): 123-623

138. 624

NPAFC. 2016. NPAFC Pacific salmonid catch statistics (updated 29 March 2016). Edited by V. 625

North Pacific Anadromous Fish Commission. 626

Olsen, J.B., and L.W. Seeb. 1998. Genetic interpretation of broad-scale microsatellite 627

polymorphism in odd-year pink salmon. Transactions of the American Fisheries Society 127: 628

535-550. 629

Page 28 of 47



Draft

29

Pess, G.R., R. Hilborn, K. Kloehn, and T.P. Quinn. 2012. The influence of population dynamics 630

and environmental conditions on pink salmon (Oncorhynchus gorbuscha) recolonization after 631

barrier removal in the Fraser River, British Columbia, Canada. Canadian Journal of Fisheries and 632

Aquatic Sciences 69(5): 970-982. 10.1139/f2012-030 633

Petrou, E.L., L. Hauser, R.S. Waples, J.E. Seeb, W.D. Templin, D. Gomez-Uchida, and L.W. 634

Seeb. 2013. Secondary contact and changes in coastal habitat availability influence the 635

nonequilibrium population structure of a salmonid (Oncorhynchus keta). Molecular Ecology 636

22(23): 5848-5860. 10.1111/mec.12543 637

Pielou, E.C. 1991. After the Ice Age: The return of life to glaciated North America. The 638

University of Chicago Press. 639

Quinn, T.P. 2005. The behavior and ecology of Pacific salmon and trout University of 640

Washington Press. 641

R Core Team. 2015. R: a language and environment for statistical computing. R Foundation for 642

Statistical Computing, Vienna, Austria. 643

Robertson, A., and W.G. Hill. 1984. Deviations from Hardy-Weinberg proportions: Sampling 644

variances and use in estimation of inbreeding coefficients. Genetics 107: 703-718. 645

Rohling, E.J., M. Fenton, F.J. Jorissen, P. Bertrand, G. Ganssen, and J.P. Caulet. 1998. 646

Magnitudes of sea-level lowstands of the past 500,000 years. Nature 394: 162-165. 647

Rousset, F. 2008. genepop'007: a complete re-implementation of the genepop software for 648

Windows and Linux. Mol Ecol Resour 8(1): 103-106. 10.1111/j.1471-8286.2007.01931.x 649

Ruggerone, G.T., B.M. Connors, and T.Q. Ii. 2015. Productivity and life history of sockeye 650

salmon in relation to competition with pink and sockeye salmon in the North Pacific Ocean. 651

Canadian Journal of Fisheries and Aquatic Sciences 72(6): 818-833. 10.1139/cjfas-2014-0134 652

Ruggerone, G.T., and J.L. Nielsen. 2005. Evidence for competitive dominance of pink salmon 653

(Oncorhynchus gorbuscha) over other salmonids in the North Pacific Ocean. Reviews in Fish 654

Biology and Fisheries 14(3): 371-390. 10.1007/s11160-004-6927-0 655

Ruggerone, G.T., R.M. Peterman, B. Dorner, and K.W. Myers. 2010. Magnitude and trends in 656

abundance of hatchery and wild pink salmon, chum salmon, and sockeye salmon in the North 657

Pacific Ocean. Marine and Coastal Fisheries 2(1): 306-328. 10.1577/c09-054.1 658

Saitou, N., and M. Nei. 1987. The Neighbor-Joining method: A new method for reconstructing 659

phylogenetic trees. Mol Biol Evol 4(4): 406-425. 660

Page 29 of 47



Draft

30

Santos, H., J. Rousselet, E. Magnoux, M.R. Paiva, M. Branco, and C. Kerdelhue. 2007. Genetic 661

isolation through time: allochronic differentiation of a phenologically atypical population of the 662

pine processionary moth. Proceedings of the Royal Society B: Biological Sciences 274(1612): 663

935-941. 10.1098/rspb.2006.3767 664

Seeb, L.W., R.K. Waples, M.T. Limborg, K.I. Warheit, C.E. Pascal, and J.E. Seeb. 2014. Parallel 665

signatures of selection in temporally isolated lineages of pink salmon. Molecular Ecology 666

23(10): 2473-2485. 10.1111/mec.12769 667

Shafer, A.B., C.I. Cullingham, S.D. Cote, and D.W. Coltman. 2010. Of glaciers and refugia: a 668

decade of study sheds new light on the phylogeography of northwestern North America. Mol 669

Ecol 19(21): 4589-4621. 10.1111/j.1365-294X.2010.04828.x 670

Shaklee, J.B., and N.V. Varnavskaya. 1994. Electrophoretic characterization of odd-year pink 671

salmon (Oncorhynchus gorbuscha) populations from the Pacific Coast of Russian, and 672

comparison with selected North American populations. Canadian Journal of Fisheries and 673

Aquatic Sciences 51(Suppl 1): 158-171. 674

Smith, C.T., R.J. Nelson, C.C. Wood, and B.F. Koop. 2001. Glacial biogeography of North 675

American coho salmon (Oncorhynchus kisutch). Molecular Ecology 10(12): 2775-2785. 676

Springer, A.M., and G.B. van Vliet. 2014. Climate change, pink salmon, and the nexus between 677

bottom-up and top-down forcing in the subarctic Pacific Ocean and Bering Sea. Proceedings of 678

the National Academy of Sciences of the United States of America. 10.1073/pnas.1319089111 679

Takezaki, N., M. Nei, and K. Tamura. 2010. POPTREE2: Software for constructing population 680

trees from allele frequency data and computing other population statistics with Windows 681

interface. Mol Biol Evol 27(4): 747-752. 10.1093/molbev/msp312 682

Taylor, R.S., and V.L. Friesen. 2017. The role of allochrony in speciation. Molecular Ecology 683

26(13): 3330-3342. 10.1111/mec.14126 684

Toge, K., R. Yamashita, K. Kazama, M. Fukuwaka, O. Yamamura, and Y. Watanuki. 2011. The 685

relationship between pink salmon biomass and the body condition of short-tailed shearwaters in 686

the Bering Sea: can fish compete with seabirds? Proceedings of the Royal Society B: Biological 687

Sciences 278(1718): 2584-2590. 10.1098/rspb.201 688

Turner, C.E., and H.T. Bilton. 1968. Another pink salmon (Oncorhynchus gorbuscha) in its third 689

year. Journal of Fisheries Research Board of Canada 25(9): 1993-1996. 690

Page 30 of 47



Draft

31

Warner, B.G., R.W. Mathewes, and J.J. Clague. 1982. Ice-Free Conditions on the Queen 691

Charlotte Islands, British Columbia, at the Height of Late Wisconsin Glaciation. Science 692

218(4573): 675-677. 10.1126/science.218.4573.675 693

Wechter, M.E., B.R. Beckman, A.G. Andrews, A.H. Beaudreau, and M.V. McPhee. 2017. 694

Growth and condition of juvenile chum and pink salmon in the northeastern Bering Sea. Deep-695

Sea Research Part Ii-Topical Studies in Oceanography 135: 145-155. 696

10.1016/j.dsr2.2016.06.001 697

Weir, B.S., and C.C. Cockerham. 1984. Estimating F-statistics for the analysis of population 698

structure. Evolution 38(6): 1358-1370. 699

Westlake, R.L., and G.M. O'Corry-Crowe. 2002. Macrogeographic structure and patterns of 700

genetic diversity in harbor deals (Phocavitulina) from Alaska to Japan. Journal of Mammalogy 701

83(4): 1111-1126. 702

Wickham, H. 2009. ggplot2: elegant graphics for data analysis. URL : http://ggplot2.org/. 703

Springer-Verlag, New York. 704

705

Page 31 of 47



Draft

32

Table 1. Location and collection year for even- and odd-year populations of pink 706

salmon. 707

708

709

710

711

712

713

714

1From Seeb et al. (2014) 715

Location River Collection Year

Lat. Long. even odd

Hokkaido Island Kushiro River 2006 2007 42.980 144.378

Amur Province Amur River 2010 2011 52.938 139.727

Magadan Province Tauy River 2012 2009 59.714 148.929

Kamchatka Province Haylylulya River 2010 2009 57.769 162.485

Norton Sound1 Nome River 1994 1991 64.483 -165.301

Prince William Sound1 Koppen Creek 1996 1991 60.706 -145.898

Puget Sound1 Snohomish River 1996 2003 48.014 -122.181

Page 32 of 47



Draft

33

Table 2. Population genetic statistics for even- and odd-year populations of pink salmon, including final sample number, 716

allelic richness, mean observed heterozygosity (Ho), mean expected heterozygosity (He), and mean individual standardized 717

heterozygosity (HS_obs). 718

719

720

Location N

Allelic

Richness Mean Ho Mean He Mean HS_obs

even odd even odd even odd even odd even odd

Hokkaido 32 32 1.595 1.645 0.159 0.170 0.160 0.170 0.968 1.037

Amur 32 32 1.625 1.677 0.161 0.171 0.163 0.172 0.988 1.051

Magadan 27 28 1.623 1.677 0.161 0.171 0.161 0.171 0.986 1.048

Kamchatka 31 31 1.619 1.679 0.160 0.172 0.161 0.171 0.983 1.056

Norton Sound 18 24 1.617 1.678 0.153 0.170 0.160 0.172 0.937 1.045

Prince William

Sound 24 24 1.593 1.623 0.161 0.162 0.160 0.165 0.988 0.996

Puget Sound 24 24 1.558 1.555 0.149 0.154 0.154 0.156 0.910 0.939

Overall Mean 27 28 1.604 1.648 0.158 0.167 0.160 0.168 0.966 1.024

Page 33 of 47



Draft

34

Table 3. Estimates of FST for: A. all populations, B. by lineage: the even- and odd-721

year lineages and groups of populations (Asia/Bering Sea (Asia/Ber) and Gulf of 722

Alaska (GoA)) within each lineage, and C. pairwise by location. Confidence 723

intervals are based on 1 000 bootstraps. 724

725

Comparison No.

Populations FST 95% CI

Over all populations 14 0.060 (0.059 - 0.062)

By lineage

Odd-year lineage 7 0.033 (0.032 - 0.035)

Odd Asia/Ber 5 0.008 (0.007 - 0.009)

Odd GoA 2 0.052 (0.049 - 0.053)

Even-year lineage 7 0.026 (0.025 - 0.027)

Even Asia/Ber 5 0.011 (0.010 - 0.012)

Even GoA 2 0.030 (0.029 - 0.033)

Pairwise by location

Hokkaido 2 0.088 (0.084 - 0.089)

Amur 2 0.063 (0.062 - 0.066)

Magadan 2 0.060 (0.058 - 0.062)

Kamchatka 2 0.060 (0.059 - 0.063)

Norton Sound 2 0.062 (0.059 - 0.064)

Prince William Sound 2 0.094 (0.089 - 0.095)

Puget Sound 2 0.148 (0.143 - 0.145)

Page 34 of 47



Draft

35

Table 4. Results of AMOVA comparing the amount variation between lineages and 726

between groups of populations. 727

728

729

730

731

732

733

734

*significant at P < 0.05 735

**significant at P < 0.001 736

737

AMOVA Source of Variation Amount of

Variation (%) Fixation Indices

A. Between

lineages

Between lineages 5.72 FST

0.086**

Among populations within lineages 2.87 FSC

0.030**

Within populations 91.41 FCT

0.057**

B. Between

even lineage

groups

Between groups 2.82 FST 0.042**

Among populations within groups 1.40 FSC 0.014*

Within populations 95.78 FCT 0.028**

C. Between

odd lineage

groups

Between groups 4.13 FST

0.056**

Among populations within groups 1.46 FSC

0.015*

Within populations 94.40 FCT

0.041**

Page 35 of 47



Draft

36

Figure 1. Map of the present-day coastline (grey) overlaid with the exposed land 738

during lowered sea levels (120 to 130m lower than present, orange) and glacial cover 739

(blue) during the LGM, approximately 18-26 thousand years ago. Historical 740

coastline adapted from Grant et al. (2012), Brigham-Grette et al. (2003b) and Pielou 741

(1991). Glacial cover adapted from Shafer et al. (2010), Ehlers and Gibbard (2007), 742

and Pielou (1991). 743

Figure 2. Boxplot of individual standardized observed heterozygosity (Hs_obs) across 744

loci within each population. Hs_obs was measured as the proportion of heterozygotes 745

standardized by the mean observed heterozygosity of the genotyped loci. For each 746

population, the thick black lines show the median values, and the box depicts the 747

quartiles of the individual values. Whiskers show 1.5x the interquartile range or the 748

maximum value, with circles representing individuals with Hs_obs that fall outside of 749

that range. 750

Figure 3. Elevated allelic diversity in odd-year pink salmon. The histogram shows 751

allele frequencies for loci that are variable in a single lineage (private alleles). There 752

are more private alleles in the odd-year lineage (dark bars) than in the even-year 753

lineage (light bars). These private alleles also have greater frequency in the odd-year 754

lineage. 755

Figure 4. Unrooted Neighbor-Joining tree based on DA distance (Nei 1983) using 16 756

681 SNPs. All branches had 100% bootstrap support with 10 000 permutations, 757

except where noted. 758

Page 36 of 47



Draft

37

Figure 5. Individual based PCA exploring the relationship of all 14 populations: A) 759

principal components 1 vs 2, B) principal components 1 vs 3, C) principal 760

components 2 vs 3. The even-year lineage individuals are represented by circles and 761

the odd-year lineage individuals are represented by triangles, with each sample 762

location assigned a different color. The percent of variation explained by each 763

principal component is in parenthesis. 764

Page 37 of 47



Draft

38

765

766

Page 38 of 47



Draft

39

767

768

769

770

771

Page 39 of 47



Draft

40

772

773

Page 40 of 47



Draft

41

774

775

Page 41 of 47



Draft

42

776

777

778

779

780

781

Page 42 of 47



Draft

43

Supplemental Files 782

File S1. Detailed information for 396 population samples. 783

File S2. Sequence information for 16 681 loci. 784

Page 43 of 47



Draft

44

785

786

Figure S1. Three-dimensional visualization of population based PCA exploring the relationship 787

between all 14 populations. The centroid of the even-year lineage populations are represented by 788

circles and the centroid of the odd-year lineage populations are represented by triangles, with 789

each sample location assigned a different color. The percent of variation explained by each 790

principal component is in parenthesis. 791

Page 44 of 47



Draft

45

792

Figure S2. Comparisons of pairwise FST between all fourteen populations. Populations are 793

ordered geographically, from the southern Asian populations to the southern North American 794

populations. Orange highlights the pairwise comparisons with the greatest FST. 795

796

797

798

Page 45 of 47



Draft

46

Protocol 1. Detailed RAD sequencing protocol for Asian populations. 799

RAD sequencing 800

Asian samples were sequenced using a bulk pooling and re-sequencing method, whereby the 801

RAD libraries included many individually barcoded samples with the expectation that those that 802

had poor sequence coverage would be re-sequenced. This process should result in more even 803

coverage of retained sequence reads across individuals. Genomic DNA was digested using the 804

restriction site enzyme SbfI, and P1 adaptors were ligated. Each individual was barcoded using 805

6bp long adaptors that differed by at least 2 nucleotides. The DNA was pooled into three 806

libraries of no more than 86 individuals each and purified. The individuals were assigned to the 807

three libraries randomly to avoid any lane effect. 808

The pooled libraries were separated evenly into two subsets and sheared at the same time. The 809

ends of the DNA were repaired, and P2 adaptors were ligated to the sheared DNA. A final PCR 810

with conditions from Everett et al. (2012) amplified the P1 and P2 adaptor-ligated DNA 811

fragments, and the product was verified on a 1% agarose gel. Each library was assessed for 812

quality and concentration of genomic DNA using a Bioanalyzer DNA 1000 kit (Agilent 813

Technologies, Santa Clara, CA), and the final concentration of each library was determined using 814

the accompanying Bioanalyzer software. The prepared libraries were assessed for DNA quality 815

and sequenced on an Illumina HiSeq2000 sequencer (Illumina, San Diego, CA) at the University 816

of Oregon High-Throughput Sequencing facility. 817

Initial Filtering and Re-sequencing 818

Page 46 of 47



Draft

47

The raw sequence data for all fourteen populations were filtered, and polymorphic loci were 819

identified using the software package Stacks v.1.31 (Catchen et al. 2011). For each lane of raw 820

sequence, the program process_radtags was used to trim the terminal nucleotide from the 101 bp 821

raw reads, de-multiplex the individuals, and filter the low quality reads. The barcodes used as 822

sample identifiers were replaced with a unique sample name, and all reads from multiple lanes of 823

sequencing were combined for ease and coherence of downstream analysis. 824

After this initial filtering, samples with fewer than 1.5 million retained reads were re-sequenced 825

using the above protocol and their original barcodes. The resulting raw data from the re-826

sequenced samples was again analyzed using the Stacks program process_radtags, and all 827

sequences were combined for downstream analysis. 828

Page 47 of 47



draft...draft 20 abstract 21 pink salmon, the most abundant pacific salmon, have an obligate...

Documents