comprehensive evaluation of genetic population structure ......population structure for anadromous...

Contents lists available at ScienceDirect

Fisheries Research

journal homepage: www.elsevier.com/locate/fishres

Comprehensive evaluation of genetic population structure for anadromousriver herring with single nucleotide polymorphism data

Kerry Reida,b, Eric P. Palkovacsa, Daniel J. Hasselmana,1, Diana Baetscherb,c, Jared Kibeled,Ben Gahagane, Paul Bentzenf, Meghan C. McBridef, John Carlos Garzab,c,⁎

a Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, CA 95060, USAb Fisheries Ecology Division, Southwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, Santa Cruz, CA95060, USAc Department of Ocean Sciences, University of California, Santa Cruz, CA 95064, USAdNational Center for Ecological Analysis and Synthesis, University of California, Santa Barbara, CA 93101, USAeMassachusetts Division of Marine Fisheries, Gloucester, MA 01930, USAfMarine Gene Probe Laboratory, Biology Department, Dalhousie University, Halifax, NS B3H 4R2, Canada

A R T I C L E I N F O

Handled by J Viñas

Keywords:AlosaAlewifeBlueback herringMixed stock analysisPopulation genetic structureSingle nucleotide polymorphisms

A B S T R A C T

Anthropogenic activities are placing increasing pressure on many species, particularly those that rely on morethan one ecosystem. River herring (alewife, Alosa pseudoharengus and blueback herring, A. aestivalis collectively)are anadromous fishes that reproduce in rivers and streams of eastern North America and migrate to the westernAtlantic Ocean. Here, we use data from single nucleotide polymorphisms (SNPs) to provide a comprehensiveanalysis of population structure for both species of river herring throughout their native ranges. We sampledriver herring spawning runs in rivers from Newfoundland to Florida, examining a total of 108 locations, andgenotyping over 8000 fish. We identified geographic population groupings (regional genetic groups) in eachspecies, as well as significant genetic differentiation between most populations and rivers. Strong correlationsbetween geographic and genetic distances (i.e., isolation by distance) were found range-wide for both species,although the patterns were less consistent at smaller spatial scales. River herring are caught as bycatch infisheries and estimating stock proportions in mixed fishery samples is important for management. We assessedthe utility of the SNP datasets as reference baselines for genetic stock identification. Results indicated highaccuracy of individual assignment (76–95%) to designated regional genetic groups, and some individual po-pulations, as well as highly accurate estimates of mixing proportions for both species. This study is the first toevaluate genetic structure across the entire geographic range of these species and provides an important foun-dation for conservation and management planning. The SNP reference datasets will facilitate continued multi-lateral monitoring of bycatch, as well as ecological investigation to provide information about ocean dispersalpatterns of these species.

1. Introduction

Natural populations that are affected by anthropogenic activitiesrequire monitoring and management to avoid demographic and otherrisks. Genetic data allow accurate evaluation of population structureand patterns of migration, which is critical for the identification ofdemographic independence and appropriate management units(Palsbøll et al., 2007). When extensive population structure exists, itcan be used with genetic stock identification (GSI) techniques so thatindividuals sampled away from their natal areas, or in mixed

aggregations, can be identified to their demographic and genetic unit oforigin (Milner et al., 1981; Rannala and Mountain, 1997; Andersonet al., 2008). This is particularly relevant for anadromous fishes, as theyspawn in freshwater, migrate long distances from their natal rivers andstreams to the ocean and then return, and are often encountered inmixed stock aggregations while at sea. Genetic data from reference“baseline” databases of established population units can allow for thedetermination of which stocks are present in a mixed sample and inwhat proportions (Milner et al., 1981; Seeb et al., 2007; Clemento et al.,2014).

https://doi.org/10.1016/j.fishres.2018.04.014Received 8 December 2017; Received in revised form 12 April 2018; Accepted 19 April 2018

⁎ Corresponding author at: Fisheries Ecology Division, Southwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, 110McAllister Way, Santa Cruz, CA 95060, USA.

1 Present address: Columbia River Inter-Tribal Fish Commission, Hagerman, ID 83332, USA.E-mail address: [email protected] (J.C. Garza).

Fisheries Research 206 (2018) 247–258

0165-7836/ Published by Elsevier B.V.

T

http://www.sciencedirect.com/science/journal/01657836https://www.elsevier.com/locate/fishreshttps://doi.org/10.1016/j.fishres.2018.04.014https://doi.org/10.1016/j.fishres.2018.04.014mailto:[email protected]://doi.org/10.1016/j.fishres.2018.04.014http://crossmark.crossref.org/dialog/?doi=10.1016/j.fishres.2018.04.014&domain=pdf

Population structure for anadromous fishes is typically understoodby sampling populations in freshwater spawning habitat. Populations ofanadromous fishes often show signals of hierarchical structure andpatterns of isolation by distance, due to high rates of homing to natalrivers, with migration usually to proximate river basins (Garza et al.,2014, Ozerov et al., 2017). This restricted gene flow among river ba-sins, and among tributaries within larger river systems, leads to popu-lation structure even when individuals move thousands of kilometersover their lifetimes. The resulting population structure allows in-dividuals sampled in the ocean to be assigned back to rivers and re-gional stocks of origin using GSI techniques (Anderson et al., 2008;Seeb et al., 2007; Clemento et al., 2014). Such information can provideinsight about differential exploitation of populations or regional de-mographic units and patterns of marine migration and distribution inspace and time (e.g., Larson et al., 2012; Bradbury et al., 2016;Anderson et al., 2017).

River herring is the collective name given to alewife (Alosa pseu-doharengus) and blueback herring (A. aestivalis). These anadromoussister species, native to eastern North America and the northwesternAtlantic Ocean, have similar life-history characteristics, includingspawning in freshwater during spring and spending two to five years inthe marine environment, where they undertake migrations along thecontinental shelf, following food resources and schooling with otherspecies such as Atlantic herring and Atlantic mackerel (Turner et al.,2016, 2017), and then return to their natal rivers to spawn (Scott andCrossman, 1973). Hasselman et al. (2014) and McBride et al. (2014)documented hybridization between alewife and blueback herring

populations spawning in the same rivers. River herring are of sig-nificant ecological and conservation concern due to declining popula-tions and the effects of habitat loss, pollution and harvest (Limburg andWaldman, 2009; Atlantic States Marine Fisheries Commission [ASMFC]2012; Palkovacs et al., 2014; McBride et al., 2015; Hasselman et al.,2016).

Previous population genetic studies of river herring have providedimportant insights into the species biology, conservation, and man-agement (McBride et al., 2014; Palkovacs et al., 2014; Turner et al.,2015, Hasselman et al., 2014; Hasselman et al., 2016; Ogburn et al.,2017). Palkovacs et al. (2014), examining populations from rivers southof the US-Canada border, identified three regional genetic units ofalewife and four genetic units of blueback herring. McBride et al.(2014), examining populations from Canadian rivers, detected weakdifferentiation among populations of alewife. These studies providedimportant information to facilitate the conservation and managementof river herring, but derived their genetic data from microsatellitemarkers. Despite their high variability and extensive use in the study offish and wildlife over the last several decades, microsatellites haveimportant limitations when applied to fisheries management. Primaryamong them are a lack of portability across laboratories and instru-ments, which prevents the integration of datasets without extensivestandardization efforts (Seeb et al., 2007, Clemento et al., 2011, Seebet al., 2011). To overcome this limitation, single nucleotide poly-morphism (SNP) genetic markers have been developed to assess po-pulation structure and employ GSI techniques for the study of ana-dromous fishes and other migratory species. SNP markers can be

Fig. 1. Maps of sampling locations from coastal rivers for river herring (A) sampling sites for alewife (B) blueback herring. Sampling location codes correspond toTable 1.

K. Reid et al. Fisheries Research 206 (2018) 247–258

248

consistently genotyped across laboratories and instruments, makingthem robust tools for GSI analyses.

Here, we use recently developed panels of SNP markers for alewifeand blueback herring (Baetscher et al., 2017) to provide a compre-hensive evaluation of population structure for these species. We buildon previous work and increase the geographic range and number ofrivers sampled for each species to include nearly all of the freshwaterspawning ranges for both species. We then used simulations to assesspower for stock identification at multiple hierarchal levels and to de-monstrate the utility of these SNP markers to identify demographicallyconnected groups of populations for the purposes of management andconservation.

2. Materials and methods

2.1. Sampling

Tissue samples from anadromous alewife and blueback herring wereobtained from across the species’ geographic ranges (i.e.,Newfoundland to Florida; Fig. 1). Specimens were identified to speciesbased on a combination of external morphological characteristics andperitoneal color (Jordan and Evermann, 1896; Scott and Crossman,1973). Muscle or fin tissue samples were obtained from adult and ju-venile fish captured in spawning rivers above the influx of salt water.Temporal replicates for a subset of locations were collected to evaluatetemporal genetic variation. The alewife samples included a total of 137collections (n=6783) and the blueback herring samples were from 54collections (n=2502). Tissue samples were either preserved in 95%ethanol or stored dry on Whatman® blotting paper in coin envelopesuntil DNA extraction.

2.2. Laboratory protocols

Genomic DNA was extracted with DNeasy 96 Tissue kits using aBioRobot 3000 (Qiagen, Inc.). Samples were genotyped using species-specific SNP assays described by Baetscher et al. (2017), and included93 SNPs for alewife (Suppl. Table 1A) and 96 SNPs for blueback herring(Suppl. Table 1B). Genotyping was done with 96.96 Dynamic SNPGenotyping Arrays on an EP1 system using SNP Type assays (FluidigmCorporation) according to manufacturer’s specifications. SNP genotypeswere called using the Fluidigm Genotyping Analysis Software v. 2.1.1.

Previous studies of river herring revealed low rates of species mis-identification in the field, and low frequencies of hybridization betweenanadromous alewife and blueback herring (McBride et al., 2015;Hasselman et al., 2016). We screened for misidentified individuals andhybrid specimens by genotyping known “pure” and “hybrid” specimenswith both of the species-specific SNP panels. Misidentified specimensand potential hybrid individuals were identified by both stock assign-ment and by extremely low heterozygosity at the SNP loci, as very fewloci are reciprocally polymorphic in both species, with polymorphismgenerally restricted to a single species (Clemento et al., 2014). In-dividuals missing data from more than 10% of the genotyped loci,misidentified individuals, and potential hybrids were excluded fromfurther analyses.

2.3. Data analyses

2.3.1. Data conformance to model assumptions and temporal replicatesTests for linkage disequilibrium (LD) and departures from Hardy-

Weinberg equilibrium (HWE) expectations were performed withGENEPOP v. 4.2 (Rousset, 2008) using default parameters for all tests.Sequential Bonferroni adjustments were used to determine significancelevels (Holm, 1979; Rice, 1989). Genic differentiation was assessedamong temporal collections to assess the stability of populationsthrough time.

2.3.2. Genetic diversity and differentiationObserved and expected heterozygosity and percent polymorphic

loci were calculated for each collection using Microsatellite Toolkit v.3.1 (Park, 2001). Allelic heterogeneity among rivers was assessed withgenic tests in GENEPOP using default parameters. Tests were combinedacross loci using Fisher’s method. Due to small sample sizes, the Con-necticut EMR 2011 collection was combined with Connecticut Rivercollections from the same year and James River collections (Herring,Walker and Chickahominy) were combined by year (2011 and 2015)for summary statistic estimates in blueback herring. Overall and pair-wise FST values (θ; Weir and Cockerham, 1984) were estimated usingFstat v. 2.3.9.2 (Goudet, 2001). To account for variation in geneticdiversity among populations, standardized estimates of differentiation(F'ST) were calculated as F'ST= FST/FST(max) (Hedrick, 2005) usingRECODEDATA v. 0.1 (Meirmans, 2006). FSTAT was used to estimateFST(max) for each pairwise comparison.

2.3.3. Population genetic structureGenetic relationships among populations were examined with un-

rooted neighbor-joining (NJ) dendrograms using the DA distance (Neiet al., 1983). Bootstrapping over loci (10,000 replicates) was performedusing POPTREEW (Takezaki et al., 2014) to examine stability of re-lationships. The Bayesian model-based clustering method implementedin STRUCTURE v. 2.3.4 (Pritchard et al., 2000; Falush et al., 2003) wasused to infer the number of genetic clusters represented by the collec-tions. A burn-in of 50,000 replicates, followed by 150,000 replicates ofthe MCMC simulation, was performed employing the admixture modeland correlated allele frequencies among populations, with no geo-graphic prior. Ten iterations were performed for each value of K(number of clusters)= 2–10, allowing an estimation of the most likelynumber of clusters by consistency of patterns indicating “geographicbarriers” (Rosenberg et al., 2005). The results were visualized usingCLUMPP v. 1.1.2 (Jakobsson and Rosenberg, 2007) and DISTRUCT v.1.1 (Rosenberg, 2004).

To further characterize population structure, Discriminate Analysisof Principal Components (DAPC; Jombart et al., 2010) was performedwith the package ‘adegenet’ v. 2.0.1 (Jombart, 2008) in R 3.3.1 (R CoreTeam, 2016). The variance of each dataset was broken into principalcomponents and the most likely number of biologically meaningfulclusters described by the data was determined using the change inBayesian Information Criterion (BIC), as recommended by Jombartet al. (2010). These clusters were used to inform the DAPC analysis.

Finally, pairwise genetic differentiation (FST) between identifiedregional genetic groups for both species was calculated using Weir andCockerham’s (1984) estimator implemented in GENETIX (Belkhir et al.,1996–2004), with 10,000 permutations to quantify the amount of di-vergence between groups.

2.3.4. Isolation by distanceTests for potential correlations between geographic distance and

genetic differentiation, so-called isolation by distance (IBD), wereconducted across the entire range of each species using a Mantel testwith 10,000 permutations (Pearson’s correlation) in the R-package‘vegan’ (Oksanen et al., 2007, 2013). Differentiation was evaluated withboth pairwise F'ST values and their linearized counterparts (F'ST/1-F'ST)following Rousset (1997). Geographic distances between sampling lo-cations were estimated using a swimmable distance method (i.e., theshortest distance through rivers and ocean without crossing land).These distances were estimated for each pair of sampling locationsusing a novel network theory approach and implemented using theNetworkX open source Python library (Hagberg et al., 2008) which isavailable from an open source license code repository (https://github.com/jkibele/pyriv).

First, a network of potential ocean paths was created from a sim-plified coastline polygon of the study area. All coastline polygon ver-tices were represented as network nodes and every possible line


249

https://github.com/jkibele/pyrivhttps://github.com/jkibele/pyriv

between two nodes was considered as a potential edge. Only those linesthat did not cross the land polygon were incorporated as network edges.Next, hydrography network shapefiles were obtained (United StatesForest Service; Nagel et al., 2017) and amended where necessary tointersect the coastline network. In cases where river features were notfound within four km of site locations, a straight line to the coast wasadded to the network if that location was within four km of the coast.Where site locations were more than four km from a river feature andmore than four km from the coast, rivers were drawn manually withreference to OpenStreetMap raster basemaps (http://www.openstreetmap.org/copyright). The completed river network was thencombined with the ocean path network. Geographic paths were calcu-lated between each pair of sites using the NetworkX implementation ofDijkstra’s shortest path algorithm (Dijkstra, 1959) weighted by distanceand saved to shapefile format. Distances were then derived from theseline features.

2.3.5. GSI baseline simulationsThe utility of the SNP panels to assign individuals back to collection

and regional genetic unit of origin was evaluated using a self-assign-ment analysis (Rannala and Mountain, 1997) implemented in the mixedstock analysis program GSI_SIM (Anderson et al., 2008). A simulationapproach implemented in GSI_SIM was then used to evaluate mixingproportion estimates for regional genetic units following the approachpreviously described by Hasselman et al. (2016), with regional geneticunit designations based on the observed patterns of genetic populationstructure for each species. This involved simulations of 50 mixtureswith different proportions of the constituent stocks, using Dirichletdistributions= 1.5. From each of these 50 mixtures, four samples of1000 fish each were taken and mixing proportions then estimated usingmaximum likelihood. A correlation was then assessed between esti-mated mixing proportions and “True” mixing proportions from the si-mulations, which allowed for the detection of potential biases. Becauseof a clear signal of non-native ancestry (see below), due to documentedtranslocation events (McBride et al., 2015), we removed the DresdenMills collections from the reference datasets for both species and alsoremoved the Sewall Pond collection from the alewife dataset for allsimulations, mixing proportion estimation and individual assignments.

3. Results

3.1. Data conformance to model assumptions and temporal replicates

One locus from each species was removed (alewife: Aps_4413,blueback herring: Aae_1454) as they showed deviations from HWE andmissing data for several collections. The final dataset for alewife con-sisted of genotypes for 92 SNPs from 137 collections at 99 locations(n=5678), and included one or multiple temporal replicates for 28locations (Table 1A). The final blueback herring dataset consisted of 95SNPs genotyped for 54 collections from 42 locations (n=2247), andincluding one or multiple temporal replicates for 10 locations(Table 1B). Most collections of alewife and blueback herring sampledover multiple years (ranging from two to five years) showed temporalstability. These results were broadly supported by the finding of allelichomogeneity between temporal replicates with genic tests. However,there was allelic heterogeneity between alewife collections from theSaco River in 2010 and 2015, and between blueback herring collectionsfrom the James River and Cape Fear River in 2010 and 2011 (Suppl.Tables 2A, 2B), although these temporal collections still grouped to-gether in cluster analyses (Fig. 2).

3.2. Genetic diversity and differentiation

The genotype data were used to evaluate relative levels of geneticdiversity in river herring populations. Unbiased heterozygosity fell in arelatively narrow range for alewife collections (Table 1A) with values

from 0.214 (Otter Pond, Newfoundland) to 0.281 (Saint John River,New Brunswick). The range was wider for blueback herring collections(Table 1B), and varied from 0.246 (St. John’s River, Florida) to 0.362(James River, Virginia). The proportion of polymorphic loci for alewiferanged from 0.64 (Otter Pond, Newfoundland) to 1.0 (Eastern River,Maine) (Table 1A), and for blueback herring from 0.86 (St. John’sRiver, Florida) to 1.0 (Connecticut River) (Table 1B). The alewife col-lection from Otter Pond also had the lowest number of alleles per locus(1.64) of any anadromous alewife or blueback herring collection(Table 1A).

For alewife, 87.8% (8299/9453) of pairwise tests of allelic hetero-geneity were significant (Suppl. Table 2A) and standardized pairwiseestimates of genetic differentiation (F'ST) for alewife ranged from−0.026 to 0.311 (FST=−0.005–0.188) (Suppl. Table 3A). Non-sig-nificant differentiation was primarily among temporal replicates andbetween neighboring drainages, but there were also a few instancesbetween more distant locations. For example, lack of differentiationwas observed between several collections within the Gulf of St.Lawrence, among numerous collections in northern New England, andbetween collections from Chesapeake Bay to the Albemarle Sound(Suppl. Tables 2A, 3A).

For blueback herring, 90.1% (1241/1378) of pairwise tests weresignificant (Suppl. Table 2B) and F'ST for blueback herring ranged from−0.005 to 0.397 (FST=−0.004–0.212) (Suppl. Table 3B). Non-sig-nificant differentiation was found primarily among temporal replicatesand between neighboring drainages, including most comparisons fromwithin Chesapeake Bay, and among collections from Albemarle Sound(Suppl. Tables 2B, 3B).

3.3. Population genetic structure

The neighbor joining trees (Suppl. Fig. 1A, B) indicated that mostriver herring collections cluster by geography, but there was generallylow bootstrap support for branching relationships, due to the relativelylow genetic divergence among populations. The STRUCTURE analysissupported four distinct groups for alewife with most clustering patternscoinciding with geography. Populations at the boundaries of the ge-netically distinct groups showed some degree of admixture, which re-flects gene flow between adjacent regional groups. Several collectionsnot at genetic group boundaries also appeared admixed− Argyle Brookin Nova Scotia as a mixture of groups from Canada and Northern NewEngland, while Dresden Mills and Sewell Pond within the KennebecRiver basin of Maine had a mixture of Northern New England andSouthern New England ancestry. Argyle Brook is at the southern tip ofNova Scotia, making natural gene flow with populations in Mainelikely, whereas admixture at Dresden Mills and Sewell Pond is likely theresult of anthropogenic stocking (McBride et al., 2015).

For blueback herring, the clustering analyses indicated five regionalgenetic units (Fig. 2B). As with alewife, blueback herring also hadpatterns of gene flow at the boundaries of genetic groups and this wasparticularly prevalent in the Mid-Atlantic region. Additionally, atK=2, there was a clear gradient of ancestry in the Mid-Atlantic regionbetween the Northern and Southern genetic units. There were alsoseveral collections that appeared admixed, including Dresden Mills, andthe James, Connecticut and Metedeconk Rivers (Fig. 2B).

Significant isolation by distance (IBD) was evident across the rangesof both alewife (R2= 0.408, p < 0.001) and blueback herring(R2= 0.465, p < 0.001) (Fig. 3), demonstrating that gene flow be-tween basins is greater the closer they are to each other. However,much of the range-wide pattern in both species was due to comparisonsbetween regional genetic groups, as significant correlations betweengeographic and genetic distance were not present within all of the re-gional groups and were generally not as strong as at the range widelevel (Suppl. Fig. 2). The exception was in the southernmost geneticgroup, where distances between populations are greater and IBD wasstrong and significant in both species. DAPC found similar clustering


250

http://www.openstreetmap.org/copyrighthttp://www.openstreetmap.org/copyright

Table 1ASampling locations for alewife and summary statistics from 92 SNP loci.

Sampling location Code Cluster Latitude Longitude Year Sample Size HE HO % poly. loci

Garnish River GAR CAN 47.2267 −55.3452 2015 35 0.28 0.27 96.7Otter Pond OTT CAN 51.0872 −56.8668 2014 29 0.21 0.26 64.1Miramichi River MIR CAN 46.9665 −65.5783 2011 41 0.26 0.25 95.7Richibucto River RIC CAN 46.6579 −64.8628 2011 46 0.26 0.24 91.3Tidnish River TID CAN 45.9768 −64.0447 2011 30 0.26 0.26 87.0River Phillip RPH CAN 45.8389 −63.7576 2011 40 0.25 0.27 93.4a

Wallace River WAL CAN 45.8116 −63.5158 2011 31 0.24 0.24 93.5Waughs River WAU CAN 45.6969 −63.2685 2011 28 0.26 0.26 92.3a

Hillsborough River HIL CAN 46.3515 −62.8713 2011 37 0.25 0.25 95.7Tracadie Bay TRA CAN 46.3896 −62.9909 2011 42 0.25 0.24 93.5Margaree River MAR CAN 46.4167 −61.0815 2011 42 0.24 0.23 93.5Bras d'Or Lakes BRA CAN 45.8464 −60.8211 2011 39 0.27 0.27 92.4West River WES CAN 44.9265 −62.5445 2011 38 0.27 0.27 94.6Sullivan Pond Outlet SUL CAN 44.6715 −63.5633 2011 47 0.26 0.26 97.8Sackville River SAK CAN 44.7308 −63.6620 2011 45 0.27 0.27 93.5LaHave River LHV CAN 44.3967 −64.5369 2011 46 0.27 0.27 94.6Argyle Brook ARG CAN 43.7930 −65.8669 2011 45 0.26 0.26 91.3Kiack Brook KIA CAN 43.8209 −65.9468 2011 46 0.26 0.26 93.5Tusket River TUS CAN 43.8633 −65.9816 2011 48 0.26 0.26 94.6Gaspereau River GAS CAN 45.0773 −64.3160 2011 43 0.28 0.27 92.4Shubenacadie River SHU CAN 44.9318 −63.5343 2011 45 0.26 0.26 94.6Peticodiac River PET CAN 46.0537 −64.8417 2011 40 0.25 0.25 92.4Saint John River SJR CAN 45.9535 −66.8651 2011 41 0.28 0.28 96.7St. Croix River (Dennis Stream) SCDEN NNE 45.1938 −67.2593 2004 42 0.26 0.26 92.4St. Croix River (Milltown Dam) SCMIL NNE 45.1776 −67.2938 2004 40 0.25 0.24 90.2Little River LIT NNE 44.9722 −67.0894 2010 27 0.25 0.24 84.6a

Dennys River DEN NNE 45.0385 −67.3577 2010 29 0.25 0.26 92.42015 46 0.25 0.24 88.0

East Machias River EMA NNE 44.7566 −67.3624 2010 40 0.25 0.25 90.22015 48 0.24 0.24 94.4a

Narraguagus River NAR NNE 44.4906 −68.0050 2010 22 0.27 0.26 87.02015 45 0.25 0.25 92.4

West Bay Pond WBP NNE 44.4882 −68.0460 2010 42 0.25 0.24 92.3a

Mt. Desert Island MDI NNE 44.3591 −68.3466 2010 40 0.25 0.25 91.3Union River UNI NNE 44.5437 −68.4288 2009 46 0.25 0.24 90.2

2011 48 0.26 0.25 93.5Bagaduce River (Wight Pond) WIG NNE 44.4509 −68.6731 2009 41 0.26 0.25 93.5

2011 45 0.25 0.24 90.22015 43 0.26 0.25 94.6

Bagaduce River (Walker Pond) BWAL NNE 44.3395 −68.6872 2015 45 0.25 0.24 89.1Bagaduce River (Pierce Pond) PIE NNE 44.4808 −68.7259 2015 45 0.25 0.24 90.2Orland River ORL NNE 44.5697 −68.7430 2010 72 0.27 0.27 97.8

2011 43 0.25 0.25 92.4Penobscot River (Milford) PMIL NNE 44.9404 −68.6443 2015 45 0.25 0.24 94.6Penobscot River (Souadabscook Stream) PSOU NNE 44.7491 −68.8326 2009 46 0.26 0.26 93.5

2010 68 0.26 0.26 95.72011 45 0.27 0.26 95.7

Penobscot River (Veazie Dam) PVEA NNE 44.8327 −68.7006 2009 37 0.26 0.25 92.42010 44 0.26 0.26 93.52011 44 0.27 0.25 93.5

Penobscot River (Blackman Stream) PBLA NNE 44.8860 −68.6476 2015 45 0.27 0.27 96.7St. George River STG NNE 44.2383 −69.2766 2008 46 0.25 0.25 93.5

2010 40 0.24 0.25 93.52011 47 0.24 0.25 90.2

Medomak River Falls MDK NNE 44.0959 −69.3780 2010 23 0.24 0.25 89.0a

2015 30 0.26 0.27 88.0Damariscotta River DAM NNE 44.0605 −69.5260 2009 33 0.24 0.26 84.8

2010 70 0.25 0.25 93.52011 39 0.25 0.25 85.9

Sheepscot River SHE NNE 44.0763 −69.6014 2010 40 0.25 0.26 90.2Kennebec River (Sewell Pond) SEW NNE 43.8691 −69.7823 2009 48 0.25 0.25 95.7

2011 39 0.25 0.24 91.3Eastern River (Dresden Mills Dam) DRE NNE 44.1065 −69.7292 2008 20 0.26 0.26 91.3

2009 20 0.26 0.26 91.32010 76 0.27 0.28 100.02011 40 0.27 0.26 92.4

Seven Mile Stream (Webber Pond Dam) WEB NNE 44.4028 −69.6717 2009 31 0.25 0.26 89.12010 84 0.25 0.25 93.52011 42 0.24 0.24 93.5

Sebasticook River SEB NNE 44.5802 −69.5542 2009 45 0.25 0.25 93.52010 36 0.25 0.26 92.42011 44 0.25 0.24 91.3

Kennebec River (Lockwood Dam) LOC NNE 44.5458 −69.6272 2009 45 0.25 0.25 93.52011 28 0.26 0.25 89.1

(continued on next page)


251

patterns for both species (Suppl. Fig. 3) and indicated that, in bothspecies, the geographically southernmost group is the most distinctfrom the other populations.

3.4. Identification of regional genetic units

The genetic clustering analyses identified regional genetic units for

alewife (Fig. 2A) and blueback herring (Fig. 2B). These regional geneticunits for alewife are as follows: Canada (CAN) from Garnish River andOtter Pond in Newfoundland to the Saint John River in New Brunswick;Northern New England (NNE) from the St. Croix River to the MerrimackRiver; Southern New England (SNE) from the Parker River to the Carll’sRiver; and Mid-Atlantic (MAT) from the Hudson River to the AlligatorRiver. The designations for these genetic units of blueback herring are

Table 1A (continued)


Androscoggin River AND NNE 43.9204 −69.9670 2009 48 0.25 0.24 92.42010 80 0.25 0.25 94.62011 47 0.24 0.23 91.3

Presumpscot River PRE NNE 43.7202 −70.2723 2010 27 0.26 0.25 88.0Saco River SAC NNE 43.4953 −70.4461 2010 31 0.27 0.26 91.3

2015 23 0.26 0.25 92.4Cocheco River COC NNE 43.1964 −70.8738 2011 23 0.25 0.24 89.1

2012 44 0.25 0.25 91.3Oyster River OYS NNE 43.1309 −70.9187 2011 26 0.26 0.25 89.1Lamprey River LAM NNE 43.0809 −70.9339 2011 32 0.26 0.27 91.3

2012 19 0.25 0.25 89.1Exeter River EXE NNE 42.9811 −70.9444 2015 42 0.26 0.25 93.5Merrimack River MER NNE 42.7016 −71.1645 46 0.26 0.25 95.7Parker River PAR SNE 42.7501 −70.9292 2015 46 0.27 0.27 94.6Mystic River MYS SNE 42.4153 −71.1381 38 0.25 0.25 91.3Back River BAC SNE 42.2154 −70.9231 2015 46 0.27 0.27 93.5Town Brook TOW SNE 41.9543 −70.6642 2011 45 0.25 0.25 93.5Stony Brook STN SNE 41.7445 −70.1125 2015 52 0.25 0.24 91.3Herring River HER SNE 41.6819 −70.1224 2015 42 0.26 0.25 92.4Mashpee River MAS SNE 41.6158 −70.4792 2015 42 0.26 0.27 94.6Coonamessett River COON SNE 41.5853 −70.5726 2015 41 0.25 0.24 89.1Monument River MON SNE 41.7368 −70.6245 2011 45 0.26 0.26 95.7Nemasket River NEM SNE 41.8907 −71.0846 2010 38 0.26 0.27 92.4

2012 38 0.26 0.27 94.6Nonquit Pond NON SNE 41.5607 −71.1968 2012 39 0.26 0.27 94.6Gilbert-Stuart Brook GIL SNE 41.5184 −71.4463 2011 43 0.25 0.24 94.6Saugatucket River SAU SNE 41.4521 −71.4954 2012 44 0.27 0.25 93.5Thames River THA SNE 41.4662 −72.0702 2012 33 0.27 0.26 97.8Bride Lake BRI SNE 41.3278 −72.2373 2012 43 0.26 0.26 95.7

2014 47 0.25 0.25 91.32015 17 0.26 0.26 85.9

Dodge Pond DOD SNE 41.3275 −72.1986 2014 51 0.27 0.26 93.52015 47 0.26 0.26 94.6

Latimer Brook LAT SNE 41.3689 −72.2020 2013 36 0.26 0.26 90.2Connecticut River CON SNE 41.4791 −72.4986 2011 31 0.25 0.26 92.4Connecticut River (Wethersfield Cove) WTH SNE 41.7252 −72.6558 2015 23 0.27 0.27 92.4Mattabessett River MAT SNE 41.5907 −72.6668 2015 48 0.26 0.26 92.4Mill Creek − Mary Steube MCR SNE 41.3506 −72.3023 2015 124 0.26 0.25 98.9Mill Brook MBR SNE 41.3426 −72.3113 2014 45 0.26 0.25 91.3Quinnipiac River QUI SNE 41.3635 −72.8770 2009 20 0.26 0.26 93.5

2012 20 0.26 0.26 89.1Housatonic River HOU SNE 41.2412 −73.0979 2012 11 0.25 0.23 83.5a

Pequonnock River PEQ SNE 41.1969 −73.1864 2012 33 0.27 0.26 90.2Mianus River MIA SNE 41.0490 −73.5853 2012 36 0.25 0.25 90.2Peconic River PEC SNE 40.9172 −72.6534 2009 45 0.26 0.25 92.4Carlls River CAR SNE 40.7038 −73.3276 2009 38 0.25 0.24 89.1Hudson River HUD MAT 41.2121 −73.9427 2012 47 0.26 0.25 93.5Hudson River (Black Creek) BLA MAT 41.8245 −73.9590 2015 47 0.24 0.24 96.7Delaware River DEL MAT 39.7274 −75.4912 2011 38 0.24 0.23 91.3Nanticoke River NAN MAT 38.4516 −75.8357 2011 36 0.23 0.23 84.8Choptank River CHOP MAT 38.7124 −75.9975 2015 41 0.23 0.23 85.9Northeast River NRT MAT 39.5985 −75.9456 2012 48 0.22 0.22 91.3Susquehanna River SUS MAT 39.6345 −76.1555 2012 29 0.23 0.22 82.6

2015 45 0.22 0.22 88.0Patuxent River PAT MAT 38.5875 −76.6779 2012 46 0.22 0.21 80.4

2015 45 0.23 0.23 92.4Potomac River POT MAT 38.9314 −77.1168 2015 37 0.24 0.23 87.0Rappahannock River RAP MAT 37.9176 −76.8189 2011 42 0.23 0.24 88.0James River JAM MAT 37.3727 −76.8985 2015 44 0.23 0.22 85.9Chowan River CHO MAT 36.1860 −76.7309 2011 44 0.22 0.22 85.6a

2015 44 0.22 0.22 88.0Roanoke River ROA MAT 35.8265 −76.8586 2011 47 0.22 0.22 89.1Alligator River ALL MAT 35.6977 −76.1368 2011 50 0.22 0.22 84.8

HE=Unbiased heterozygosity; HO= observed heterozygosity.a Includes data from 90 (EMA2015 & CHO) or 91 SNP loci.


252

as follows: Canada-Northern New England (CAN-NNE) from the Mar-garee River to the Kennebec River; Mid-New England (MNE) from theOyster River to the Parker River; Southern New England (SNE) from theMystic River to Gilbert-Stuart Pond; Mid-Atlantic (MAT) from theConnecticut River to the Neuse River; and South Atlantic (SAT) fromthe Cape Fear River to the St. John’s River. All pairwise comparisons ofgenetic differentiation between regional genetic groups in both specieswere highly significant (Suppl. Tables 4A, 4B).

3.5. GSI simulations

There was broad concordance between the simulated and estimatedmixing proportions for each regional genetic unit for both alewife andblueback herring (Figs. Figure 4A, Figure 5A ). These simulations

indicated that the genetic baselines provide accurate mixing proportionestimates for alewife and blueback herring encountered in mixturesamples back to their larger genetic units. This pattern was also sup-ported by the self-assignment analysis of populations back to largergeographical units (Figs. Figure 4B, Figure 5B). For alewife, the accu-racy of self-assignment of individuals to their regional genetic group oforigin was 90% to the CAN regional group, 91% to the NNE, 86% to theSNE and 92% to the MAT regional group (Suppl. Table 5A). For blue-back herring, individual assignments to regional genetic group of originwere similarly high, with 89% accuracy to the CAN-NNE regionalgroup, 76% to the MNE, 90% to the SNE, 89% to the MAT and 95% tothe SAT regional group (Suppl. Table 5B). Simulation results and self-assignment tests revealed that assignment to population/river of originwas not as accurate as assignment to regional genetic units, with some

Table 1BSampling locations for blueback herring and summary statistics from 95 SNP loci.


Margaree River MAR CAN-NNE 46.4167 −61.0815 2011 41 0.30 0.28 96.8Peticodiac River PET CAN-NNE 46.0537 −64.8417 2011 32 0.27 0.30 88.2a

Saint John River SJR CAN-NNE 45.9535 −66.8651 2011 43 0.28 0.28 96.8East Machias River EMA CAN-NNE 44.7509 −67.3815 2010 46 0.29 0.29 95.8Orland River ORL CAN-NNE 44.5697 −68.7430 2009 34 0.27 0.28 91.6

CAN-NNE 2015 25 0.27 0.26 93.7St. George River STG CAN-NNE 44.2019 −69.2764 2010 48 0.29 0.30 95.7a

Eastern River – Dresden Mills Dam DRE CAN-NNE 44.1065 −69.7292 2015 46 0.30 0.29 97.9Kennebec River – Lockwood Dam LOC CAN-NNE 44.5458 −69.6272 2015 46 0.30 0.30 98.9Sebasticook River SEB CAN-NNE 44.5802 −69.5542 2015 45 0.29 0.28 96.8Oyster River OYS MNE 43.1309 −70.9187 2015 34 0.29 0.27 95.8Exeter River EXE MNE 42.9811 −70.9444 2011 47 0.29 0.29 94.7

2015 20 0.30 0.28 92.6Parker River PAR MNE 42.7501 −70.9292 2015 46 0.29 0.28 95.8Mystic River MYS SNE 42.4153 −71.1381 39 0.28 0.29 95.8Herring River HER SNE 41.6819 −70.1224 2015 47 0.27 0.28 88.4Coonamessett River COON SNE 41.5853 −70.5726 2015 23 0.28 0.28 86.3Monument River MON SNE 41.7368 −70.6245 2011 47 0.28 0.29 90.5Gilbert-Stuart Brook GIL SNE 41.5184 −71.4463 2011 46 0.28 0.28 92.5a

Connecticut River CON MAT 41.4791 −72.4986 2011 43 0.29 0.30 100.0Connecticut River (Wethersfield Cove) WCV MAT 41.7252 −72.6558 2015 29 0.29 0.29 96.8Connecticut River (Farmington River) FAR MAT 41.8735 −72.6501 2015 21 0.29 0.31 93.7Quinnipiac River QUI MAT 41.3635 −72.8770 2012 28 0.29 0.29 96.8Mianus River MIA MAT 41.0490 −73.5853 2012 28 0.29 0.29 96.8Hudson River HUD MAT 41.2121 −73.9427 39 0.29 0.29 97.9Hudson River ESO MAT 42.0741 −73.9472 2015 46 0.29 0.30 96.8Metedeconk River MET MAT 40.0613 −74.1250 2015 34 0.29 0.28 94.7Delaware River DEL MAT 39.7274 −75.4912 2011 47 0.29 0.29 96.8a

Nanticoke River NAN MAT 38.4516 −75.8357 2011 21 0.29 0.29 94.72012 27 0.30 0.30 94.7

Choptank River CHOP MAT 38.7124 −75.9975 2015 46 0.29 0.29 98.9Susquehanna River SUS MAT 39.6345 −76.1555 2012 43 0.29 0.29 94.7

2015 47 0.30 0.30 95.8Patuxent River PAT MAT 38.5875 −76.6779 2012 44 0.30 0.30 96.8

2015 46 0.30 0.30 95.8Potomac River POT MAT 38.9314 −77.1168 2011 51 0.29 0.28 98.9Rappahannock River RAP MAT 37.9176 −76.8189 2011 43 0.30 0.31 97.9York River YOR MAT 37.5161 −76.7903 2011 48 0.29 0.29 97.9James River JAM MAT 37.3727 −76.8985 2011 47 0.29 0.29 94.7

2015 24 0.33 0.37 90.5Chowan River CHO MAT 36.1860 −76.7309 2011 45 0.29 0.29 96.8a

Roanoke River ROA MAT 35.8265 −76.8586 2011 48 0.30 0.29 98.9Neuse River NEU MAT 35.3240 −77.3428 2011 47 0.30 0.29 96.8a

2015 45 0.30 0.29 96.8Cape Fear River – Rice/Town CF SAT 34.1285 −77.9515 2011 91 0.29 0.29 97.9

2012 46 0.30 0.29 94.72015 45 0.29 0.28 96.8

Santee River SAN SAT 33.2401 −79.5001 2011 44 0.28 0.27 92.4a

2015 42 0.27 0.27 91.5a

Savannah River SAV SAT 32.2314 −81.1460 2011 48 0.27 0.26 94.72015 47 0.27 0.26 92.6

Altamaha River ALT SAT 31.4728 −81.6145 2011 47 0.27 0.28 94.7St. John's River STR SAT 29.6033 −81.6022 2011 47 0.25 0.25 86.3

2015 48 0.26 0.25 87.4

HE=Unbiased heterozygosity; HO= observed heterozygosity.a Includes data from 92 (SAN2011), 93 or 94 (DEL, CHO & SAN2015) SNP loci.


253

populations exhibiting considerable bias in assignments (Suppl. Fig. 4A,B).

4. Discussion

We describe here comprehensive evaluations of genetic populationstructure for alewife and blueback herring using single nucleotidepolymorphism (SNP) data from almost 8000 fish that encompass nearlytheir entire geographic ranges on the Atlantic coast of North America.We also describe how these SNP genotype datasets function as referencebaseline databases for genetic stock identification that allow accurateassignment of individuals back to their regional genetic group of originand the estimation of stock proportions in mixed fisheries with highaccuracy. We extended our understanding of genetic structure for thesesister species by including population samples from essentially all of themajor areas with river herring populations in their native range, andimproved the ability to use genetic stock identification techniques tofacilitate ecological investigation and management monitoring of riverherring.

Defining patterns of population structure and migration for riverherring is imperative for continued management and conservation ef-forts. Identification of genetic population structure provides an under-standing of ecological and demographic interactions between popula-tions and regional groups, and is critical for predicting responses tomanagement actions. Genetic tools that allow accurate estimation ofstock contributions to fisheries provide important capabilities for

monitoring exploited species, as well as methods to identify age-specificand stock-specific migration patterns within the marine and freshwaterenvironments. The utility of genetic methods to facilitate such effortshas been highlighted by recent studies indicating that certain fisheries(e.g., Atlantic herring, Clupea harengus, and Atlantic mackerel, Scomberscombrus) have major bycatch of alewife, blueback herring andAmerican shad (A. sapidissima) (Bethoney et al., 2017). A dispropor-tionate amount of the river herring bycatch is from the SNE alewifegroup and the MAT blueback herring group, which includes popula-tions around Long Island Sound (Hasselman et al., 2016). These re-gional genetic groups have also experienced among the most severerecent population declines (Palkovacs et al., 2014; Hasselman et al.,2016).

4.1. Inferring regional genetic units of river herring from populationstructure

Our primary goal here was to evaluate genetic differentiation ofriver herring populations throughout their native ranges, which extendfrom Newfoundland to North Carolina for alewife and from Nova Scotiato Florida for blueback herring, using recently described sets of SNPmarkers (Baetscher et al., 2017). Our results largely coincide with thelarge-scale genetic patterns observed in previous studies of alewife andblueback herring; however, prior studies only covered portions of theranges of the two species (McBride et al., 2014; Palkovacs et al., 2014).

Extensive sampling of natal rivers, including most basins with

Fig. 2. Bayesian clustering analysis for (A) alewife and (B) blueback herring. Codes correspond to Table 1 and each K produced consistent patterns across teniterations.

Fig. 3. Regression between pairwise genetic distance (F’ST) and geographic distance (km) for river herring. (A) alewife (R2=0.408; P < 0.001) and (B) bluebackherring (R2=0.465; P < 0.001).


254

Fig. 4. Assessment of reporting group assignment using simulations and self-assignment tests for alewife. (A) Correlation of simulated and estimated mixing pro-portions by designated reporting unit. (B) Proportion of self-assignment of individuals within rivers to designated reporting groups.

Fig. 5. Assessment of reporting group assignment using simulations and self-assignment tests for river herring. (A) Correlation of simulated and estimated mixingproportions by designated reporting unit. (B) Proportion of self-assignment of individuals within rivers to designated reporting groups.


255

demographically important river herring populations, provided in-creased resolution that enabled us to define transitions between re-gional genetic groups more clearly than in previous studies. In alewife,multiple analyses supported the identification of four primary geneti-cally distinct units range-wide (Fig. 2A; Suppl. Fig. 2A). At the northernextent of the sampling, populations from Otter Pond and Garnish Riverin Newfoundland to the Saint John River in New Brunswick (Canada)were identified as a regional genetic group, consistent with the resultsof McBride et al. (2014). There was not strong support for additionalregional structuring among rivers in the Canadian portion of the range,as found by McBride et al. (2014) with a distinct set of molecularmarkers, although there was significant differentiation between manypopulations. The Northern New England (NNE) regional genetic groupextended from the St. Croix River at the Canada/US internationalboundary to the Merrimack River, which was again consistent withprevious work (Palkovacs et al., 2014), but substantially refined thetransition due to much denser population sampling. The Southern NewEngland (SNE) regional genetic group extended from the Parker Riverto the Carll’s River on Long Island. The Mid-Atlantic (MAT) regionalgenetic group encompassed all populations from the Hudson River tothe Alligator River. Although the Hudson River populations occur in thetransition zone between the SNE and MAT genetic groups, we desig-nated them as part of the MAT group, as the majority of individuals andtheir ancestry assigned to the MAT group (Figs. Figure 2A, Figure 4B),whereas Palkovacs et al. (2014) and Hasselman et al. (2016) previouslyassigned the Hudson to SNE. Pairwise FST values between the alewiferegional genetic groups ranged from 0.007 to 0.022 and were all highlysignificant (p < 0.001, Suppl. Table 4A).

Blueback herring populations were divided into five regional ge-netic groups (Fig. 2B). The clustering with Structure at K=2 found ageographic cline in ancestry between two major genetic groups, withone corresponding to the southern-most populations, the (SouthAtlantic) SAT regional group, and the other including all other popu-lations. At values of K=3, 4, 5 the three additional boundaries becameapparent, separating the Mid-Atlantic (MAT), Southern New England(SNE), Mid-New England (MNE) and Canada-Northern New England(CAN-NNE) genetic groups from one another, respectively. These ge-netic patterns were also generally supported by the DAPC analysis(Suppl. Fig. 2B), in which clusters mostly corresponded to geographicgroups. Only one of the boundaries between regional genetic groups,the MNE and SNE at the Parker River, was coincident between alewifeand blueback herring. The composition of blueback herring geneticgroups was generally consistent with those described by Palkovacs et al.(2014). However, with the addition of multiple, additional populationsin the northern part of the range, the newly designated CAN-NNE ge-netic group included some (East Machias and St. George River) but notall of the populations that previously were included in the NNE geneticgroup from Palkovacs et al. (2014). Pairwise FST values between theseregional genetic groups were all highly significant (p < 0.001) andranged from 0.026 to 0.114 (Suppl. Table 4B).

4.2. Connectivity across genetic group boundaries

The extent of gene flow between regional genetic groups inferred inthis study is substantially greater than previously estimated (Palkovacset al., 2014), with clearly transitional populations present in manyareas. This is likely a result of increased sampling from populations inbetween those that constituted boundaries of genetic groups in previousstudies. However, there are some discrepancies that could also be duepartially to differences in the genetic datasets. For example, the Con-necticut River blueback herring populations appear transitional/ad-mixed between the SNE and MAT groups with the SNP data, but weremore clearly part of the MAT stock in previous work with microsatellitedata (Palkovacs et al., 2014), which could be the influence of rare al-leles that are characteristic of such data. Additionally, we found severalinstances of populations with clearly admixed ancestry, which was

likely a consequence of stocking events. A previous study focusing onriver herring stocking practices in the northern part of the range foundthat the Sewell Pond and Dresden Mills populations in the KennebecRiver were distinct from the rest of the populations in that region(McBride et al., 2015), a finding confirmed in this study for both alewifeand blueback herring (Figs. 2, 4 and 5). These populations have anearlier spawn time than other populations in surrounding basins(McBride et al., 2015), suggesting that they maintain characteristicsfrom their likely genetic group of origin (SNE). River herring have agradient of spawning time across the range, with spawning startingearlier (March) in the south and later (June) in Canada (Scott andCrossman, 1973). Spawning time has been shown to be highly heritablein other anadromous fish (Hendry and Day, 2005; Abadía-Cardosoet al., 2013). While there are some additional alewife stocking eventsthat have been documented, none involved donor and recipient popu-lations that are in different regional genetic groups (Personal Commu-nication, J. Sheppard, MDMF) and are therefore unlikely to have af-fected boundaries of regional genetic groups.

Another process that likely contributes to the admixture observed intransition zones is metapopulation dynamics. Local extirpations or ex-treme reductions in size of smaller populations in such systems can leadto recolonization from adjacent populations, some of which will be inother genetic groups, on the boundaries of regional population sub-divisions. The patterns of correlation between geographic and geneticdistances observed in this study are also consistent with this. McBrideet al. (2015) found significant isolation by distance in regions withlimited stocking, but not in regions with extensive stocking. Strongpatterns of range-wide isolation by distance were found here for bothspecies (Fig. 3), consistent with the findings of Palkovacs et al. (2014)and McBride et al. (2014), and highlighting the importance of strayingand gene flow in shaping observed population structure, by maintainingconnectivity and avoiding or mitigating local extirpations. However,within regional genetic groups, the patterns of correlations were lessconsistent, indicating that gene flow is less restricted to adjacent basinsat smaller spatial scales. Such patterns of isolation by distance havebeen observed in other networks of populations of anadromous fish,particularly salmonids (e.g. Garza et al., 2014).

4.3. Mixed stock and assignment analyses

The analysis of population structure that we provide was used toidentify regional genetic groups for river herring throughout the nativegeographic ranges of both species and establish the utility of these SNPdata as comprehensive reference datasets (baselines) for genetic stockidentification. These alewife and blueback herring baselines includedata from much more densely sampled populations (alewife n= 99 andblueback herring n=42) than in previous studies and unify geneticdata collection from the northern-most parts of the species ranges withthose from the central and southern-most parts of the ranges. This ap-proach allowed us to more clearly identify transitions between geneticgroups than was previously possible and substantially increase con-fidence in range-wide diversity estimates, which will likely prove usefulfor future research. Using SNPs for genetic stock identification baselinesfacilitates collaborative research for species that are migratory or foundin large geographic areas, as they are more economical to genotype andamenable to high-throughput genotyping for large-scale fisheries orgenome-scale projects (Seeb et al., 2011; Clemento et al., 2014).

We found high individual assignment and mixing proportion accu-racy to regional genetic units (Figs. 4 and 5) with these SNP baselinesthat avoid the slight biases observed in previous work with micro-satellites (Hasselman et al., 2016). However, accuracy of assignment ofindividuals to population and river of origin was low in most cases(Suppl. Fig. 4A, B), indicating that proximate rivers are usually notdemographically independent or genetically differentiated, which islikely a consequence of straying. Gene flow and straying among prox-imate rivers promotes admixture and is a mechanism to recolonize


256

available habitat after local extirpation. In addition, gene flow supportsthe maintenance of effective population size and population viability.At larger spatial scales, signals of genetic subdivision indicate regions ofrestricted gene flow, which are potentially driven by environmental andhabitat differences, including temperature gradients, throughout thespecies’ geographic ranges.

Demographic independence of populations based on genetic diver-gence has been proposed as an appropriate method to define manage-ment units (Palsbøll et al., 2007). In species with life-history traitsleading to high gene flow (such as anadromous and marine species),identifying management units is more complex, as large effective po-pulation sizes, variance in reproductive success and migration can leadto low estimates of genetic differentiation (Hauser and Carvalho, 2008;Waples et al., 2008). As such, a robust, comprehensive understanding ofpopulation structure, such as provided in this study, is essential foreffective, large-scale conservation and management efforts of riverherring. The two SNP reference datasets that we describe provide theability to accurately assign individuals back to regional stock of originand estimate stock proportions of mixtures, such as fishery bycatch,which will allow monitoring and evaluation of such conservation andmanagement efforts, and extend our knowledge of the ecology, evolu-tion and behavior of both alewife and blueback herring.

Data accessibility

Baseline databases including genotypes and metadata for bothspecies is available on DRYAD.

Acknowledgements

We thank the many organizations and individuals who conductedsample collection. Cassondra Columbus, Elena Correa, and EllenCampbell provided assistance in the laboratory. Travis Apgar providedassistance with figures. Eric Anderson and the Linking LifestagesWorkgroup, particularly Alison Bowden and Mike Armstrong, providedinsightful discussion and advice. This work was supported by theNational Fish and Wildlife Federation (NFWF 0104.14.041425),Atlantic States Marine Fisheries Commission (ASMFC 15-0105, 15-0102), The Nature Conservancy, Wildlife Research Institute NortheastRegional Conservation Needs Grants Program, National ScienceFoundation (NSF-DEB 1556378), Pew Charitable Trust, MassachusettsDepartment of Marine Fisheries and NOAA Cooperative Institute forMarine Ecosystems and Climate. The distance calculations resultedfrom the SASAP Working Group, funded by the Gordon and BettyMoore Foundation, and conducted at the National Center for EcologicalAnalysis and Synthesis, a Center funded by the University of California,Santa Barbara, and the State of California.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in theonline version, at https://doi.org/10.1016/j.fishres.2018.04.014.

References

Abadía-Cardoso, A., Anderson, E.C., Pearse, D.E., Garza, J.C., 2013. Large-scale parentageanalysis reveals reproductive patterns and heritability of spawn timing in a hatcherypopulation of steelhead (Oncorhynchus mykiss). Mol. Ecol. 22, 4733–4746.

Anderson, E.C., Waples, R.S., Kalinowski, S.T., 2008. An improved method for predictingthe accuracy of genetic stock identification. Can. J. Fish. Aquat. Sci. 65, 1475–1486.

Anderson, E.C., Ng, T.C., Crandall, E.D., Garza, J.C., 2017. Genetic and individual as-signment of tetraploid green sturgeon with SNP assay data. Con. Genet. 18,1119–1130.

ASMFC (Atlantic States Marine Fisheries Commission), 2012. River herring benchmarkstock assessment. Stock Assessment Report No. 12-02 of the Atlantic States MarineFisheries Commission Vol. 1 Washington, DC.

Baetscher, D.S., Hasselman, D.J., Reid, K., Palkovacs, E.P., Garza, J.C., 2017. Discoveryand characterization of single nucleotide polymorphisms in two anadromous alosinefishes of conservation concern. Ecol. Evol. 7, 6638–6648.

Bethoney, N.D., Schondelmeier, B.P., Kneebone, J., Hoffman, W.S., 2017. Bridges to bestmanagement: effects of a voluntary bycatch avoidance program in a mid-water trawlfishery. Mar. Policy 83, 172–178.

Belkhir, K., Borsa, P., Chikhi, L., Raufaste, N., Bonhomme, F., (1996-2004). GENETIX 4.05, logiciel sous WindowsTM pour la genetique des populations. LaboratoireGenome, Populations, Interactions, CNRS UMR 5000, Universite de Montpellier II,Montpellier, France.

Bradbury, I.R., Hamilton, L.C., Chaput, G., Robertson, M.J., Goraguer, H., Walsh, A.,Morris, V., Reddin, D., Dempson, J.B., Sheehan, T.F., King, T., Bernatchez, L., 2016.Genetic mixed stock analysis of an interceptor Atlantic salmon fishery in theNorthwest Atlantic. Fish. Res. 174, 234–244.

Clemento, A.J., Abadía-Cardoso, A., Starks, H.A., Garza, J.C., 2011. Discovery andcharacterization of single nucleotide polymorphisms in Chinook salmon,Oncorhynchus tshawytscha. Mol. Ecol. Resour. 11, 50–66.

Clemento, A.J., Crandall, E.D., Garza, J.C., Anderson, E.C., 2014. Evaluation of a singlenucleotide polymorphism baseline for genetic stock identification of Chinook Salmon(Oncorhynchus tshawytscha) in the California Current large marine ecosystem. Fish.Bull. 112, 112–130.

Dijkstra, E.W., 1959. A note on two problems in connexion with graphs. NumerischeMathematik 1, 269–271.

Falush, D., Stephens, M., Pritchard, J.K., 2003. Inference of population structure usingmultilocus genotype data: linked loci and correlated allele frequencies. Genetics 164,1567–1587.

Garza, J.C., Gilbert-Horvath, E.A., Spence, B.C., Williams, T.H., Fish, H., Gough, S.A.,Anderson, J.H., Hamm, D., Anderson, E.C., 2014. Population structure of steelhead incoastal California. Trans. Am. Fish. Soc. 143, 134–152.

Goudet, J., 2001. FSTAT, A Program to Estimate and Test Gene Diversities and FixationIndices (version 2.9.3).

Hagberg, A.A., Chult, D.A., Swart, P.J., 2008. Exploring Network Structure, Dynamics,and Function Using NetworkX (No. LA-UR-08-05495; LA-UR-08-5495). Los AlamosNational Laboratory (LANL).

Hasselman, D.J., Argo, E.E., McBride, M.C., Bentzen, P., Schultz, T.F., Perez-Umphrey,A.A., Palkovacs, E.P., 2014. Human disturbance causes the formation of hybridswarm between two naturally sympatric fish species. Mol. Ecol. 23, 1137–1152.

Hasselman, D.J., Anderson, E.C., Argo, E.E., Bethoney, N.D., Gephard, S.R., Post, D.M.,Schondelmeier, B.P., Schultz, T.F., Willis, T.V., Palkovacs, E.P., 2016. Genetic stockcomposition of marine bycatch reveals disproportional impacts on depleted riverherring genetic stocks. Can. J. Fish. Aquat. Sci. 73, 951–963.

Hauser, L., Carvalho, G.R., 2008. Paradigm shifts in marine fisheries genetics: ugly hy-potheses slain by beautiful facts. Fish Fish. 9, 333–362.

Hedrick, P.W., 2005. A standardized genetic differentiation measure. Evolution 59,1633–1638.

Hendry, A.P., Day, T., 2005. Population structure attributable to reproductive time: iso-lation by time and adaptation by time. Mol. Ecol. 14, 901–916.

Holm, S., 1979. A simple sequentially rejective multiple test procedure. Scand. J. Stat. 6,65–70.

Jakobsson, M., Rosenberg, N.A., 2007. CLUMPP: a cluster matching and permutationprogram for dealing with label switching and multimodality in analysis of populationstructure. Bioinformatics 23, 1801–1806.

Jombart, T., Devillard, S., Balloux, F., 2010. Discriminant analysis of principal compo-nents: a new method for the analysis of genetically structured populations. BMCGenet. 11, 94.

Jombart, T., 2008. Adegenet: a R package for the multivariate analysis of genetic markers.Bioinformatics 24, 1403–1405.

Jordan, D.S., Evermann, B.W., 1896. The Fishes of North and Middle America. Bull. U.S.Natl. Mus. 47, Part I.

Larson, W.A., Utter, F.M., Myers, K.W., Templin, W.D., Seeb, J.E., Guthrie, I.I.I., Bugaev,C.H., Seeb, A.C., 2012. Single-nucleotide polymorphisms reveal distribution andmigration of Chinook salmon (Oncorhynchus tshawytscha) in the Bering Sea and NorthPacific Ocean. Can. J. Fish. Aquat. Sci. 70, 128–141.

Limburg, K.E., Waldman, J.R., 2009. Dramatic declines in North Atlantic diadromousfishes. Bioscience 59, 955–965.

McBride, M.C., Willis, T.V., Bradford, R.G., Bentzen, P., 2014. Genetic diversity andstructure of two hybridizing anadromous fishes (Alosa pseudoharengus, Alosa aesti-valis) across the northern portion of their ranges. Conserv. Genet. 15, 1281–1298.

McBride, M.C., Hasselman, D.J., Willis, T.V., Palkovacs, E.P., Bentzen, P., 2015. Influenceof stocking history on the genetic structure of anadromous alewife (Alosa pseudo-harengus) in Maine rivers. Conserv. Genet. 16, 1209–1223.

Meirmans, P.G., 2006. Using the AMOVA framework to estimate a standardized geneticdifferentiation measure. Evolution 60, 2399–2402.

Milner, G.B., Teel, D.J., Utter, F.M., Burley, C.L., 1981. National Marine Fisheries Service,Columbia River Stock Identification Study: Validation of Genetic Method. FinalReport to Bonneville Power Administration (Contract 1980BP18488), BPA ReportDOE/BP-18488-1.

Nagel, D., Wollrab, S., Parkes-Payne, S., Peterson, E., Isaak, D., Ver Hoef, J., 2017.National Stream Internet Hydrography Datasets for Spatial-Stream-Network (SSN)Analysis. Rocky Mountain Research Station, U.S. Forest Service Data Archive, FortCollins, CO.

Nei, M., Tajima, F., Tateno, Y., 1983. Accuracy of estimated phylogenetic trees frommolecular data. J. Mol. Evol. 19, 153–170.

Ogburn, M.B., Spires, J., Aguilar, R., Goodison, M.R., Heggie, K., Kinnebrew, E.,McBurney, W., Richie, K.D., Roberts, P.M., Hines, A.H., 2017. Assessment of riverherring spawning runs in a Chesapeake Bay coastal plain stream using imaging sonar.Trans. Am. Fish. Soc. 146, 22–35.

Oksanen, J., Kindt, R., Legendre, P., O’Hara, B., Stevens, M.H.H., Oksanen, M.J., Suggests,M.A.S.S., 2007. The Vegan Package. Community Ecology Package 10. pp. 631–637.


257

https://doi.org/10.1016/j.fishres.2018.04.014http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0005http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0005http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0005http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0010http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0010http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0015http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0015http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0015http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0020http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0020http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0020http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0025http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0025http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0025http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0030http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0030http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0030http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0040http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0040http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0040http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0040http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0045http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0045http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0045http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0050http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0050http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0050http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0050http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0055http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0055http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0060http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0060http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0060http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0065http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0065http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0065http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0070http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0070http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0075http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0075http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0075http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0080http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0080http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0080http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0085http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0085http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0085http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0085http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0090http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0090http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0095http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0095http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0100http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0100http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0105http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0105http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0110http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0110http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0110http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0115http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0115http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0115http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0120http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0120http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0125http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0125http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0130http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0130http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0130http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0130http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0135http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0135http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0140http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0140http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0140http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0145http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0145http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0145http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0150http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0150http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0155http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0155http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0155http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0155http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0160http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0160http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0160http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0160http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0165http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0165http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0170http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0170http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0170http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0170http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0175http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0175

Oksanen, J., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O’Hara, R.B., Simpson,G.L., Solymos, P., Stevens, M.H.H., Wagner, H., Oksanen, M.J., 2013. Package ‘vegan’.Community Ecology Package, Version, 2. pp. 9.

Ozerov, M., Vähä, J.P., Wennevik, V., Niemelä, E., Svenning, M.A., Prusov, S., Fernandez,Diaz, Unneland, R., Vasemägi, L., Falkegård, A., Kalske, M., 2017. Comprehensivemicrosatellite baseline for genetic stock identification of Atlantic salmon (Salmo salarL.) in northernmost Europe. ICES J. Mar. Sci. 74, 2159–2169.

Palkovacs, E.P., Hasselman, D.J., Argo, E.E., Gephard, S.R., Limburg, K.E., Post, D.M.,Schultz, T.F., Willis, T.V., 2014. Combining genetic and demographic information toprioritize conservation efforts for anadromous alewife and blueback herring. Evol.Appl. 7, 212–226.

Palsbøll, P.J., Berube, M., Allendorf, F.W., 2007. Identification of management units usingpopulation genetic data. Trends Ecol. Evol. 22, 11–16.

Park, S.D.E., 2001. The Excel Microsatellite Toolkit, Version 3.1. Animal GenomicsLaboratory, University College Dublin, Ireland.

Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure usingmultilocus genotype data. Genetics 155, 945–959.

R Core Team, 2016. R: A Language and Environment for Statistical Computing. RFoundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/.

Rannala, B., Mountain, J.L., 1997. Detecting immigration by using genotypes. Proc. Natl.Acad. Sci. U. S. A. 94, 9197–9201.

Rice, W.R., 1989. Analyzing tables of statistical tests. Evolution 43, 223–225.Rosenberg, N.A., Mahajan, S., Ramachandran, S., Zhao, C., Pritchard, J.K., Feldman,

M.W., 2005. Clines, clusters, and the effect of study design on the inference of humanpopulation structure. PLoS Genet. 1 (6), e70.

Rosenberg, N.A., 2004. DISTRUCT: a program for the graphical display of populationstructure. Mol. Ecol. Notes 4, 137–138.

Rousset, F., 1997. Genetic differentiation and estimation of gene flow from F-statisticsunder isolation by distance. Genetics 145, 1219–1228.

Rousset, F., 2008. Genepop'007: a complete reimplementation of the Genepop softwarefor Windows and Linux. Mol. Ecol. Resour. 8, 103–106.

Scott, W.B., Crossman, E.D., 1973. Freshwater fishes of Canada. J. Fish. Res. Board. Can.184, 1–966.

Seeb, L.W., Antonovich, A., Banks, M.A., Beacham, T.D., Bellinger, M.R., Blankenship,S.M., Campbell, M.R., Decovich, N.A., Garza, J.C., Guthrie, I.I.I., Lundrigan, C.M.,Moran, T.A., Narum, P., Stephenson, S.R., Supernault, J.J., Teel, K.J., Templin, D.J.,Wenburg, W.D., Young, J.K., Smith, S.F., 2007. Development of a standardized DNAdatabase for Chinook salmon. Fisheries 32, 540–552.

Seeb, J.E., Carvalho, G., Hauser, L., Naish, K., Roberts, S., Seeb, L.W., 2011. Single-nu-cleotide polymorphism (SNP) discovery and applications of SNP genotyping innonmodel organisms. Mol. Ecol. Resour. 11, 1–8.

Takezaki, N., Nei, M., Tamura, K., 2014. POPTREEW: web version of POPTREE forconstructing population trees from allele frequency data and computing some otherquantities. Mol. Biol. Evol. 31, 1622–1624.

Turner, S.M., Limburg, K.E., Palkovacs, E.P., 2015. Can different combinations of naturaltags identify river herring natal origin at different levels of stock structure? Can. J.Fish. Aquat. Sci. 72, 845–854.

Turner, S.M., Manderson, J.P., Richardson, D.E., Hoey, J.J., Hare, J.A., 2016. Using ha-bitat association models to predict Alewife and Blueback Herring marine distribu-tions and overlap with Atlantic Herring and Atlantic Mackerel: can incidental catchesbe reduced? ICES J. Mar. Sci. 73, 1912–1924.

Turner, S.M., Hare, J.A., Richardson, D.E., Manderson, J.P., 2017. Trends and potentialdrivers of distribution overlap of river herring and commercially exploited pelagicmarine fishes on the Northeast US Continental Shelf. Mar. Coast. Fish. 9, 13–22.

Waples, R.S., Punt, A.E., Cope, J.M., 2008. Integrating genetic data into management ofmarine resources: how can we do it better? Fish Fish. 9, 423–449.

Weir, B.S., Cockerham, C.C., 1984. Estimating F-statistics for the analysis of populationstructure. Evolution 38, 1358–1370.


258

http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0180http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0180http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0180http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0185http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0185http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0185http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0185http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0190http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0190http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0190http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0190http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0195http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0195http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0200http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0200http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0205http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0205https://www.R-project.org/http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0215http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0215http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0220http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0225http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0225http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0225http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0230http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0230http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0235http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0235http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0240http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0240http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0245http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0245http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0250http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0250http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0250http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0250http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0250http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0255http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0255http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0255http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0260http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0260http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0260http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0265http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0265http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0265http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0270http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0270http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0270http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0270http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0275http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0275http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0275http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0280http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0280http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0285http://refhub.elsevier.com/S0165-7836(18)30113-9/sbref0285

Comprehensive evaluation of genetic population structure for anadromous river herring with single nucleotide polymorphism dataIntroductionMaterials and methodsSamplingLaboratory protocolsData analysesData conformance to model assumptions and temporal replicatesGenetic diversity and differentiationPopulation genetic structureIsolation by distanceGSI baseline simulations

ResultsData conformance to model assumptions and temporal replicatesGenetic diversity and differentiationPopulation genetic structureIdentification of regional genetic unitsGSI simulations

DiscussionInferring regional genetic units of river herring from population structureConnectivity across genetic group boundariesMixed stock and assignment analyses

Data accessibilityAcknowledgementsSupplementary dataReferences

comprehensive evaluation of genetic population structure ......population structure for anadromous...

Documents