Human disturbance causes the formation of a hybrid swarm ... · parapatry, hybridization between forest elephant (Loxodonta cyclotis) and savannah elephant (L. africana) threatens

Human disturbance causes the formation of a hybridswarm between two naturally sympatric fish species

DANIEL J . HASSELMAN,* EMILY E. ARGO,* MEGHAN C. MCBRIDE,† PAUL BENTZEN,†

THOMAS F. SCHULTZ,‡ ANNA A. PEREZ-UMPHREY‡ and ERIC P. PALKOVACS*

*Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, CA 95060, USA, †Marine Gene Probe

Laboratory, Biology Department, Dalhousie University, Halifax, NS B3H 4R2, Canada, ‡Marine Conservation Molecular

Facility, Duke University Marine Laboratory, Beaufort, NC 28516, USA

Abstract

Most evidence for hybrid swarm formation stemming from anthropogenic habitat dis-

turbance comes from the breakdown of reproductive isolation between incipient spe-

cies, or introgression between allopatric species following secondary contact. Human

impacts on hybridization between divergent species that naturally occur in sympatry

have received considerably less attention. Theory predicts that reinforcement should

act to preserve reproductive isolation under such circumstances, potentially making

reproductive barriers resistant to human habitat alteration. Using 15 microsatellites, we

examined hybridization between sympatric populations of alewife (Alosa pseudoharen-gus) and blueback herring (A. aestivalis) to test whether the frequency of hybridiza-

tion and pattern of introgression have been impacted by the construction of a dam that

isolated formerly anadromous populations of both species in a landlocked freshwater

reservoir. The frequency of hybridization and pattern of introgression differed mark-

edly between anadromous and landlocked populations. The rangewide frequency of

hybridization among anadromous populations was generally 0–8%, whereas all land-

locked individuals were hybrids. Although neutral introgression was observed among

anadromous hybrids, directional introgression leading to increased prevalence of ale-

wife genotypes was detected among landlocked hybrids. We demonstrate that habitat

alteration can lead to hybrid swarm formation between divergent species that naturally

occur sympatrically, and provide empirical evidence that reinforcement does not

always sustain reproductive isolation under such circumstances.

Keywords: Alosa, anadromous, hybrid swarm, hybridization, introgression, landlocked

Received 30 October 2013; revision received 13 January 2014; accepted 17 January 2014

Introduction

Much of the world’s biodiversity is of recent evolution-

ary origin and is sustained by divergent adaptation in

heterogeneous environments (Rundle & Nosil 2005;

Seehausen et al. 2008). However, rugged adaptive land-

scapes that promote and maintain diversification in nat-

ure can be smoothed by human activities, hindering

adaptive radiation (Hendry et al. 2006; De Le�on et al.

2011) and contributing to the loss of evolutionary

lineages through ‘reverse speciation’ (Seehausen 2006;

Taylor et al. 2006; Seehausen et al. 2008).

Most evidence for reverse speciation comes from

studies of introgressive hybridization between incipient

fish species that have recently (i.e. postglacially)

diverged in sympatry through disruptive selection in

heterogeneous environments (e.g. threespine stickleback

benthic–limnetic pairs (Taylor et al. 2006); Laurentian

Great Lakes ciscoes (Seehausen et al. 2008); European

whitefish (Vonlanthen et al. 2012); African cichlids

(Seehausen et al. 1997)). Although reproductive isolation

between incipient species may be incomplete, diver-

gence can be reinforced through prezygotic (e.g. assor-

tative mating) and postzygotic (e.g. reduced hybridCorrespondence: Daniel J. Hasselman, Fax: (831) 459 3833;

E-mail: [email protected]

© 2014 John Wiley & Sons Ltd

Molecular Ecology (2014) 23, 1137–1152 doi: 10.1111/mec.12674

fitness) isolating mechanisms when gene flow is low

(Gow et al. 2006). However, the loss of environmental

heterogeneity can relax disruptive selection and elimi-

nate reproductive isolating mechanisms, facilitating

introgressive hybridization and the collapse of incipient

species into a hybrid swarm (Seehausen et al. 2008).

Indeed, several studies have shown how habitat

homogenization through human activities (e.g. species

introductions, eutrophication) can weaken assortative

mating and facilitate reverse speciation for multiple

incipient species (Todd & Stedman 1989; Seehausen

et al. 1997; Behm et al. 2010).

Introgressive hybridization stemming from anthropo-

genic habitat disturbance also threatens the integrity of

species with deeper evolutionary histories (Wiegand

1935; Anderson 1948; Rhymer & Simberloff 1996). For

example, introgression between grey wolf (Canis lupus)

and coyote (C. latrans), species that diverged 1.4–

2.1 million years ago, occurs across a broad range

where forests have been converted to agricultural land

(Lehman et al. 1991). Similarly, despite a 2.6-million-

year evolutionary history of ecologically maintained

parapatry, hybridization between forest elephant

(Loxodonta cyclotis) and savannah elephant (L. africana)

threatens species integrity in regions of ongoing defor-

estation in Africa (Roca et al. 2005). However, the for-

mation of a hybrid swarm resulting from the

breakdown of reproductive isolation between divergent

species that naturally occur in sympatry is not well

established (Seehausen 2006). Indeed, under these

conditions, theory predicts that reinforcement should

sustain reproductive isolation, potentially making repro-

ductive barriers resistant to habitat disturbance.

The classic definition of reinforcement implies selec-

tion against hybrids through a variety of intrinsic (e.g.

hybrid inviability, sterility, behavioural dysfunctions) or

extrinsic (e.g. lowered fitness associated with a specific

set of ecological conditions) mechanisms (see Rundle &

Schluter 1998). However, this restrictive definition does

not take into account processes that reflect classic crite-

ria for reinforcement (see Servedio & Noor 2003). Here,

we refer to reinforcement in the ‘broad sense’ (sensu

Servedio & Noor 2003), where reinforcement constitutes

‘an increase in prezygotic isolation between hybridizing

[species] in response to any type of selection against

interspecific matings, regardless of whether hybrids

themselves [exhibit reduced fitness]’. Under this defini-

tion, reinforcement could manifest as reproductive iso-

lation via spatiotemporal differences in breeding for

divergent species in sympatry.

River herrings (alewife, Alosa pseudoharengus, and

blueback herring, A. aestivalis) provide an opportunity

to empirically examine hybridization between divergent

lineages that naturally occur in sympatry, and in

response to anthropogenic habitat disturbance. Alewife

and blueback herring are anadromous (migratory sea-

run) species native to the Atlantic coast of North Amer-

ica. Although they co-occur in rivers over much of their

historical range (alewife: Labrador to South Carolina;

blueback herring: Gulf of St. Lawrence to Florida),

reproductive isolation is maintained by differences in

spawning time (temperature dependent) and spawning

habitat (reviewed by Loesch 1987). In terms of timing,

alewife spawn at cooler temperatures (5–10 °C) than

blueback herring (10–15 °C) and typically begin spawn-

ing 3–4 weeks earlier. However, peak spawn timing

only differs by 2–3 weeks, resulting in considerable

temporal overlap in freshwater (Loesch 1987). In terms

of spawning habitat, alewife preferentially select lentic

(i.e. still-water) habitats for spawning, whereas blueback

herring prefer lotic (i.e. flowing-water) habitats (Loesch

1987). Prior molecular research revealed that these spe-

cies diverged up to 1 million years ago (Faria et al.

2006), but also detected a shared mitochondrial DNA

haplotype, suggesting either incomplete lineage sorting

(Chapman et al. 1994) or introgressive hybridization

(Faria et al. 2006).

A dam constructed on the Roanoke River in 1953

created novel conditions for river herring in Kerr Reser-

voir (North Carolina/Virginia, USA). Alewife have

repeatedly established landlocked populations in lakes

(Palkovacs et al. 2008), and landlocked blueback herring

populations persist in several reservoirs following intro-

duction (Prince & Barwick 1981; Guest & Drenner

1991). To our knowledge, Kerr Reservoir is the only

instance where these species have become landlocked

in sympatry. Further, the absence of fish passage has

prevented the immigration of anadromous alewife and

blueback herring for approximately 60 years. Thus, Kerr

Reservoir provides a unique opportunity to examine

the effects of anthropogenic habitat disturbance on the

frequency of hybridization and pattern of introgression

between divergent species that naturally occur in symp-

atry. Alewife and blueback herring populations have

declined dramatically in recent decades (Limburg &

Waldman 2009; Hall et al. 2012), making these species a

conservation concern (Atlantic States Marine Fisheries

Commission 2012; Palkovacs et al. 2013). Evidence for

human impacts on hybridization could hold important

implications in setting future conservation guidelines

(Allendorf et al. 2001) and for prioritizing river restora-

tion projects through dam removal.

Using 15 microsatellite loci, we (i) examine the extent

and spatial distribution of hybridization between anadro-

mous alewife and blueback herring across both species’

ranges, including Kerr Reservoir, (ii) determine whether

the frequency of hybridization and pattern of introgres-

sion differ between anadromous and landlocked


1138 D. J . HASSELMAN ET AL.

scenarios and (iii) resolve whether anthropogenic habitat

disturbance can result in the formation of a hybrid

swarm between divergent species that naturally occur in

sympatry.

Materials and methods

Sample collections

Sampling was conducted across the ranges of both spe-

cies, and targeted a minimum of 50 individuals per col-

lection (i.e. river�year�1). Individuals were initially

assigned to species based on diagnostic peritoneal col-

oration (Jordan & Evermann 1896; Scott & Crossman

1973; Messieh 1977), but were ultimately identified as

alewife, blueback herring or hybrid using Bayesian clus-

tering analyses (see below). Collections were replicated

for a subset of rivers over successive years and pooled

following an analysis of molecular variance (AMOVA) that

indicated nonsignificant (P > 0.05) genetic variation

among years within rivers (Palkovacs et al. 2013). Sam-

ples were obtained in freshwater from adult specimens

on their spawning run, with the exception of those

collected from Veazie Dam and Souadabscook Falls

(Penobscot River) that comprised young-of-the-year

specimens. Sampling effort from 2005 to 2012 provided

tissue for 5358 alewife and 1529 blueback herring

from 56 to 30 anadromous populations, respectively, as

well as 76 and 43 presumed alewife and blueback

herring, respectively, from Kerr Reservoir (Table 1;

Fig. 1). Tissue was preserved in 95% ethanol until DNA

extraction.

Laboratory protocols

DNA extraction and genotyping. Laboratory procedures

were conducted at separate institutions (i.e. Dalhousie

University and Duke University Marine Laboratory) as

parts of independent studies of river herring genetic

variation in the northeastern USA and Canada (McBride

2013) and the USA (Labbe 2012; Palkovacs et al. 2013).

Details regarding DNA isolation and genotyping proto-

cols were reported previously (Palkovacs et al. 2013;

McBride 2013). We examined the variation at 21 poly-

morphic microsatellites: 15 loci developed for alewife

(Ap010, Ap033, Ap037, Ap038, Ap047, Ap058, Ap070,

Ap071) and blueback herring (Aa046, Aa070, Aa074,

Aa081, Aa082, Aa091, Aa093) (A’Hara et al. 2012) that

were common across studies and six additional loci

developed for alewife (Aps1, Aps2A; Bentzen & Paterson

2005), blueback herring (Aa039; A’Hara et al. 2012) and

American shad (A. sapidissima) [AsaD042, AsaC249

(Julian & Bartron 2007), Asa8 (Waters et al. 2000)] that

were examined by McBride (2013) alone. The data sets

of McBride (2013) and Palkovacs et al. (2013) were

combined after a correction factor for allele scores had

been applied for each locus common to both studies.

Briefly, small (30 ll) aliquots of pure genomic DNA

for alewife (n = 46) and blueback herring (n = 44) of

known genotype originally scored at Dalhousie Uni-

versity were genotyped at the Duke University Marine

Laboratory following the protocol of Palkovacs et al.

(2013). These data were then used to establish a cor-

rection factor to adjust the allele scores of the McBride

(2013) data set.

Data analyses

Overview. Following conformance of the data to model

assumptions, we simulated data for purebred alewife

and blueback herring and various hybrid classes (F1, F2

and backcrosses) using confirmed purebred individuals

from the empirical data set (identified using Bayesian

methods; see below). We then analysed the simulated

data using two Bayesian methods to determine the

appropriate threshold (Tq) for designating individuals

as either purebred or admixed. Next, we applied Tq to

the empirical data set to identify purebred and admixed

individuals, examined the extent of hybridization across

the ranges of alewife and blueback herring and deter-

mined whether the frequency of hybridization differed

between anadromous and landlocked scenarios. We

then used the genomic clines method (Gompert &

Buerkle 2009) to identify whether the pattern of intro-

gression differed among anadromous and landlocked

hybrids.

Data conformance to model assumptions. Genotyping arte-

facts were assessed using MICROCHECKER version 2.2.3

(Van Oosterhout et al. 2004). Departures from Hardy–

Weinberg equilibrium (HWE) and linkage disequilib-

rium (LD) were assessed within rivers (where n > 20)

for each species using GENEPOP version 4.0.6 (Rousset

2007) with default parameters for all tests. Sequential

Bonferroni adjustments were used to judge significance

levels for all simultaneous tests (Holm 1979; Rice 1989).

Selective neutrality of the microsatellite markers was

confirmed previously (Palkovacs et al. 2013).

Simulations and hybrid identification. Hybrid identifica-

tion depends on the genetic markers used and the

degree of differentiation between parental species

(Anderson & Thompson 2002; V€ah€a & Primmer 2006).

To assess the utility of each microsatellite locus to dis-

criminate alewife and blueback herring, we identified

purebred individuals using Bayesian clustering methods

(see below) and estimated the Kullback–Leibler diver-

gence (Kullback & Leibler 1951) following Anderson &


HYBRIDIZATION AND INTROGRESSION IN ALOSINES 1139

Table 1 Sampling locations and sample sizes (N) for alewife and blueback herring collected from across the species’ ranges from

2005 to 2012. Specimens were initially classified to species at the time of collection using peritoneal coloration and were later con-

firmed (or adjusted) based on results of Bayesian clustering analysis of microsatellite data

Watershed Code Latitude 2005 2008 2009 2010 2011 2012 Unknown Total N

Alewife

Miramichi River MIR 47.1419 54 54

Richibucto River RIC 46.6914 63 63

Tidnish River TID 45.9900 63 63

River Phillip RPH 45.7928 57 57

River John RJO 45.7539 117 117

Hillsborough River HIL 46.2403 60 60

Tracadie Bay TRA 46.3828 60 60

Margaree River MAR 46.4300 51 51

Bras d’Or Lakes BRA 45.8536 156 156

West River WES 44.9269 51 51

Sullivan Pond Outlet SUL 44.9269 52 52

Sackville River SAK 44.7292 71 71

LaHave River LHV 44.2928 56 56

Medway River MED 44.1339 51 51

Mersey River MER 44.0419 57 57

Tusket River TUS 43.8756 155 155

Gaspereau River GAS 45.0894 58 58

Shubenacadie River SHU 45.2603 51 51

Petitcodiac River PET 45.9053 49 49

Saint John River SJR 45.2594 50 50

St. Croix River STC 45.1911 111 25 136

Dennys River DEN 44.9089 30 30

East Machias River EMA 44.5419 59 59

Narraguagus River NAR 44.4906 30 30

West Bay Pond WBP 44.3583 30 30

Mt. Desert Island MDI 44.5011 29 29

Union River UNI 44.4408 59 28 49 136

Penobscot River PEN 44.4408 213 232 230 675

St. George River STG 43.9592 50 132 51 233

Damariscotta River DAM 44.0294 60 85 53 198

Sheepscot River SHE 44.0219 28 28

Kennebec River KEN 43.7519 44 323 321 311 999

Androscoggin River AND 43.9147 59 86 51 196

Presumpscot River PRE 43.7153 28 28

Saco River SAC 43.4669 29 29

Piscataqua River PIS 43.0664 47 74 121

Mystic River MYS 42.3439 20 49 69

Monument River MON 41.7244 46 46

Town Brook TOW 41.9625 49 49

Taunton River TAU 41.6094 89 89

Gilbert-Stuart GIL 41.4472 44 44

Thames River THA 41.3256 36 30 66

Bride Brook BRI 41.3006 34 27 61

Connecticut River CON 41.2828 7 26 33

Quinnipiac River QUI 41.2722 1 25 35 61

Housatonic River HOU 41.1708 13 25 38

Pequonnock River PEQ 41.1736 25 25

Mianus River MIA 41.0156 25 32 57

Hudson River HUD 40.6958 13 48 61

Delaware River DEL 39.1122 45 45

Nanticoke River NAN 38.1689 58 58

Rappahannock River RAP 37.4933 62 62

James River JAM 36.9686 1 1



Thompson (2002) and as implemented in the ‘flexmix’

package in R (R Development Core Team 2010). We also

estimated a multilocus global measure of FST (h; Weir &

Cockerham 1984) between the species using FSTAT ver-

sion 2.9.3.2 (Goudet 2001).

To maximize the accuracy of delineating purebred

and hybrid individuals, we conducted admixture analy-

ses using two Bayesian clustering methods imple-

mented in STRUCTURE version 2.3.3 (Pritchard et al. 2000;

Falush et al. 2003) and NEWHYBRIDS version 1.0 (Anderson

& Thompson 2002). Each method probabilistically

assigns multilocus genotypes to clusters using a Markov

chain Monte Carlo (MCMC) simulation procedure that

provides estimates from the posterior distribution that

reflects the membership of each individual. For STRUC-

TURE, the posterior probability (q) describes the propor-

tion of an individual genotype originating from each of

K clusters. For NEWHYBRIDS, q describes the probability

that an individual belongs to each of six different geno-

type frequency classes (i.e. parental purebreds, F1

hybrid, F2 hybrid and backcrosses). For both programs,

we ran 10 iterations of a parameter set that included a

burn-in of 50 000 steps followed by 250 000 replicates

of the MCMC simulation. For STRUCTURE, we set K = 2

(i.e. assuming two species contributed to the gene pool

of the sample) and employed the admixture model and

correlated allele frequencies. For NEWHYBRIDS, we used

uninformative priors for both the allele frequency and

admixture distributions.

An important consideration when using Bayesian

methods to identify admixed individuals is the choice of

optimal threshold value (Tq) for the q associated with

their assignment as purebred or hybrid (V€ah€a & Primmer

2006). For both methods, we initially used q ≥ 0.9

Table 1 Continued

Watershed Code Latitude 2005 2008 2009 2010 2011 2012 Unknown Total N

Chowan River CHO 36.0811 55 55

Roanoke River ROA 35.9228 50 50

Kerr Lake KER 36.5702 76 76

Alligator River ALL 35.9125 49 49

Watershed Code Latitude 2008 2009 2010 2011 2012 Unknown Total N

Blueback Herring

Miramichi River MIR 47.1419 1 1

River John RJO 45.7539 8 8

Margaree River MAR 46.4300 47 47

Petitcodiac River PET 45.9053 48 48

Saint John River SJR 45.2594 50 50

East Machias River EMA 44.5419 58 58

Penobscot River PEN 44.4408 36 36

St. George River STG 43.9592 46 46

Kennebec River KEN 43.7519 6 2 8 16

Exeter River EXE 43.0664 41 41

Mystic River MYS 42.3439 10 58 68

Monument River MON 41.7244 51 51

Gilbert-Stuart River GIL 41.4472 38 38

Connecticut River CON 41.2828 34 62 46 142

Quinnipiac River QUI 41.2722 30 25 55

Hudson River HUD 40.6958 48 31 79

Delaware River DEL 39.1122 51 51

Nanticoke River NAN 38.1689 24 33 57

Rappahannock River RAP 37.4933 58 58

James River JAM 36.9686 97 97

Chowan River CHO 36.0811 12 59 71

Roanoke River ROA 35.9228 50 50

Kerr Lake KER 36.5702 44 44

Neuse River NEU 35.0764 65 65

Cape Fear River CFE 33.9286 57 57

Santee-Cooper River SAN 33.1833 62 62

Savannah River SAV 32.0481 52 52

Altamaha River ALT 31.3147 53 53

St .John’s River STJ 30.4081 37 34 71



(Pritchard et al. 2000) to designate purebreds (i.e. crite-

rion #3 for NEWHYBRIDS; Burgarella et al. 2009). We then

randomly chose 100 individuals identified as purebred

alewife and blueback herring from across their range,

and simulated five data sets of parental and various

hybrid classes using HYBRIDLAB version 1.0 (Nielsen et al.

2006) to determine Tq. For each of the five simulated data

sets, we generated profiles for 200 alewife and blueback

herring. From these 400 simulated parental genotypes,

we generated 100 F1 hybrids that were used to generate

100 F2 hybrids, and 100 F1 backcrossed alewife and blue-

back herring.

Simulated data sets were analysed using STRUCTURE

and NEWHYBRIDS (as described above), and results were

summarized across iterations using CLUMPP version 1.1.2

(Jakobsson & Rosenberg 2007) and visualized with DI-

STRUCT version 1.1 (Rosenberg 2004). These results were

used to determine Tq and to develop classification rules

for assigning individuals from the empirical data set

as purebred or hybrid. For STRUCTURE, we assessed the

q value of each K cluster for every simulated parental

and hybrid class. In NEWHYBRIDS, assignment to a specific

hybrid class may be uncertain, with q split among geno-

type frequency classes (Burgarella et al. 2009). There-

fore, we summed q across hybrid genotype frequency

classes, because we were only concerned with the iden-

tification of hybrids, not distinguishing among hybrid

classes per se. We then assessed the q value for every

simulated parental and hybrid class.

To determine Tq, we used the simulated data set to

examine how the proportion of misassigned parental

and hybrid individuals using STRUCTURE and NEWHYBRIDS

changed in response to a shifting q (0.005 increments)

from 0.85 to 1.00. The resulting intersection between the

proportion of misassigned parental and hybrid individ-

uals minimizes overall misassignments and was taken

as Tq. We then used this optimal threshold to determine

the error rate for parental and hybrid classification and

to establish confidence bounds on our assignments for

the empirical data set.

The empirical data set was then analysed using STRUC-

TURE and NEWHYBRIDS (as described above). Results were

summarized across iterations using CLUMPP (Jakobsson

& Rosenberg 2007) and visualized with DISTRUCT (Rosen-

berg 2004). Hybrids were identified based on the classi-

fication rules established through the determination of

Tq (see Results; Table S1, Supporting Information), and

the frequency and extent of hybridization across the

ranges of both species were examined. Factorial corre-

spondence analysis (FCA) was used to visualize the

relative similarity of multilocus allele frequencies

among purebred alewife and blueback herring, and

anadromous and landlocked hybrids using GENETIX 4.05

(Belkhir et al. 2004).

Analysis of genomic clines. To examine whether patterns

of introgression differed among landlocked and anadro-

mous hybrids, we used the genomic clines method

(Gompert & Buerkle 2009) implemented in the R pack-

age ‘introgress’ (Gompert & Buerkle 2010; R Develop-

ment Core Team 2010). For this analysis, we only

selected specimens that were identified by both STRUC-

TURE and NEWHYBRIDS as hybrids. The proportion of blue-

back herring ancestry among hybrids was assessed

using a maximum-likelihood estimator of hybrid index

that calculates genome-wide admixture based on the

Fig. 1 Map of the Atlantic coast of North

America displaying sampling locations

for alewife and blueback herring. River

names associated with each sampling

location code are provided in Table 1.



proportion of alleles inherited from each parental spe-

cies (Gompert & Buerkle 2009). Parental species were

represented by the 100 purebred alewife and blueback

herring used in prior HYBRIDLAB simulations. Multiallelic

microsatellite data were reduced to biallelic classifica-

tion, without the loss of information or distortion of the

relationship between the parental species, using ‘intro-

gress’. Hybrids were then recognized as interallelic

class (interspecific) heterozygotes (Aa/Ab) or homozyg-

otes (Aa/Aa, Ab/Ab) (i.e. both alleles from the same alle-

lic class). The probabilities of observing each of these

possible allelic classes at each locus (i.e. genomic clines;

Gompert & Buerkle 2009) were then predicted using

multinomial regression of the observed genotypes on

hybrid index.

We identified loci that deviated from expectations of

neutral introgression by comparing the likelihoods of

the regression model to that of a neutral model, given

the observed data. Expected genomic clines under neu-

tral introgression were generated using 10 000 runs of a

parametric procedure implemented in ‘introgress’.

Genomic clines for the landlocked and anadromous

hybrid zones were then fitted with a logistic regression

using the observed data to estimate probabilities of

observing homozygote and heterozygote interallelic clas-

ses as a function of hybrid index (Gompert & Buerkle

2009). Significant deviations from neutral expectations

for genomic clines were adjusted for multiple compari-

sons using the false discovery rate (FDR; Benjamini &

Hochberg 1995). We then summarized deviations from

neutrality on the basis of whether interallelic class ho-

mozygotes and heterozygotes were more or less com-

mon than expected under neutrality, and determined

whether locus-specific patterns of introgression differed

among landlocked and anadromous hybrids.

Results

Data conformance to model assumptions

Evidence for null alleles resulted in the exclusion of loci

for both alewife (Aa082, Ap037, Ap047, Ap070) and blue-

back herring (Aa081, Ap058) prior to further analyses (Mi-

crochecker). Remaining loci were retained as evidence

for null alleles was sporadic among loci and rivers. Exact

tests revealed that genotypic frequencies were largely in

accordance with HWE for both alewife (P > 0.05; sequen-

tial Bonferroni correction for 55 comparisons) and blue-

back herring (P > 0.05; sequential Bonferroni correction

for 25 comparisons). HWE departures for alewife and

blueback herring remained for 25 and 11 locus–river

comparisons, respectively, and were due to heterozygote

deficiencies from sporadic null alleles. Exact tests of

LD revealed that loci were physically unlinked and

statistically independent (P > 0.05; sequential Bonferroni

correction for 2762 and 1270 comparisons for alewife and

blueback herring, respectively).

Simulations and hybrid identification

Global FST between alewife and blueback herring was

0.352. Although no locus was diagnostic, the Kullback–

Leibler divergence revealed several loci that would be

informative in species discrimination when used

together (e.g. KL ≥ 6.0; Fig. S1, Supporting Information).

In Bayesian analyses, q-values for randomly chosen ale-

wife and blueback herring used in simulations ranged

from 0.94–1.00 (STRUCTURE) and 0.99–1.00 (NEWHYBRIDS),

exceeding Tq (see below) and confirming that we had

correctly selected pure strain individuals for simulating

parental and hybrid classes. Further, allele frequency

distributions for the randomly chosen individuals were

representative of each species [P > 0.05; two-sample

Kolmogorov–Smirnov test; SYSTAT version 11 (SPSS, Inc.

2004)].

Across the five simulated data sets, the assignment of

purebred individuals to their correct parental class

using either Bayesian method was highly accurate

(Fig. 2). STRUCTURE unambiguously discriminated F1 and

F2 hybrids from purebreds (Fig. S2a, Supporting Infor-

mation), but the increased variance (by an order of

magnitude) of F2 hybrids reduced assignment accuracy

using NEWHYBRIDS (Fig. S2b, Supporting Information).

Both Bayesian methods had difficulty with advanced

introgression and could not unambiguously discrimi-

nate all backcrossed alewife and blueback herring from

purebred individuals and F2 hybrids (Fig. S2, Support-

ing Information). Similar results have been observed

even using diagnostic markers (Gow et al. 2007).

Using STRUCTURE, the lowest proportion of misas-

signed purebreds (0.007) and hybrids (0.006) across the

five simulated data sets occurred when Tq = 0.935. For

NEWHYBRIDS, the lowest proportion of misassigned pure-

breds (0.004) and hybrids (0.004) occurred when

Tq = 0.870. Using these optimal thresholds, we estab-

lished a set of classification rules for delineating pure-

breds and hybrids from the empirical data set (Table

S1, Supporting Information). Using these classification

rules, we detected hybrids from across the range of

river herring. Across the species’ ranges, 162 anadro-

mous specimens (2.4%) were identified as hybrids, with

Bayesian methods producing consistent results for 90

specimens. Every specimen from Kerr Reservoir

(n = 119) was identified as a hybrid (Figs 3 and 4). Dis-

crepancies between Bayesian methods occurred pre-

dominantly where individuals identified as purebreds

by NEWHYBRIDS were identified as hybrids by STRUCTURE

(n = 67), and more infrequently vice versa (n = 5). The



proportion of hybrids in anadromous populations gen-

erally ranged from 0.00 to 0.08, but there were notable

exceptions (Fig. 4). The greatest proportions of anadro-

mous hybrids were observed for the Petitcodiac River

(0.227, 0.216) and Margaree River (0.133, 0.112) using

STRUCTURE and NEWHYBRIDS, respectively (Table S2, Sup-

porting Information). Both Bayesian methods identified

identical proportions of hybrids from the St. John’s

River (0.056), Altamaha River (0.019) and Savannah

River (0.019), despite these rivers being beyond the

southern range limit for alewife. No purebred alewives

were detected south of the range limit.

Two factors in FCA explained 85.6% of the genetic

variation among purebred and hybrid river herring and

demonstrated the clear separation of landlocked

hybrids (Fig. 5). While Axis 1 (44.35%) delineated pure-

bred anadromous alewife and blueback herring and

revealed an intermediate genetic composition of anadro-

mous hybrids, Axis 2 (41.25%) isolated landlocked

hybrids from anadromous hybrids and purebred river

herrings. Interestingly, landlocked hybrids did not exhi-

bit Axis 1 scores similar to anadromous hybrids, but

values that exceeded those of purebred alewife (Fig. 5).

This prompted an examination of the distribution of

probability values (q) among genotype frequency classes

in NEWHYBRIDS for hybrids and revealed a signifi-

cantly (P < 0.001; two-sample KS test) higher frequency

of F1 and F2 individuals among anadromous than

among landlocked hybrids. Landlocked hybrids were

more deeply introgressed with alewife, consistent with

FCA results (data not shown) and genomic clines analy-

sis (see below).

Analysis of genomic clines

The genomic clines method revealed striking differences

in patterns of blueback herring ancestry and patterns of

introgression between anadromous and landlocked

hybrids (Fig. S3, Supporting Information). While the

hybrid index for anadromous hybrids ranged from 0.00

to 1.00, the fraction of the landlocked hybrid genome

that was derived from blueback herring never exceeded

0.78. Further, an approximately equal proportion of ale-

wife and blueback herring ancestry was observed

among anadromous hybrids, but there was little blue-

back herring ancestry among landlocked hybrids; only

two individuals exhibited blueback herring ancestry

≥0.5 (Fig. S3b, Supporting Information). Cumulatively,

these results support neutral introgression of blueback

herring genotypes among anadromous hybrids, but

directional introgression leading to increased prevalence

of alewife genotypes in Kerr Reservoir.

The genomic clines method also revealed marked het-

erogeneity in locus-specific patterns of introgression for

anadromous and landlocked hybrids. For anadromous

hybrids, variation at four loci (Aa046, Aa070, Aa093,

Ap033) deviated from neutral introgression based on

genome-wide admixture and remained significant after

FDR correction (P < 0.015) (Fig. 6a). For three of these

markers (Aa046, Aa070, Aa093), the total probability for

the heterozygous genotype was increased relative to

neutral expectations (Fig. 6a). For landlocked hybrids,

eight of the nine loci (except Aa046) deviated from neu-

tral introgression and remained significant after FDR

correction (P < 0.039) (Fig. 6b), with highly variable pat-

terns of deviation from neutral introgression among

markers. Nonetheless, for several of the deviant loci (i.e.

Ap071, Ap010, Aa070), the total probability for heterozy-

gous genotype was decreased relative to neutral expec-

tations (Fig. 6b). We also observed significant

differences in the patterns of introgression between

anadromous and landlocked hybrids for two of five

markers that could be directly compared (i.e. Aa074,

Ap033).

Fig. 2 Results of Bayesian clustering

analyses for five simulated data sets (ale-

wife, blueback herring, F1, F2, F1 back-

crosses) using STRUCTURE (K = 2; number

of species) and NEWHYBRIDS (K = 6; num-

ber of genotype frequency classes). Indi-

viduals are represented by a thin vertical

line which is partitioned into K-coloured

segments representing an individual’s

estimated membership fractions from

each of the identified clusters. Black lines

separate individuals from different geno-

type frequency classes (labelled below).



Discussion

Most evidence for the formation of hybrid swarms

resulting from anthropogenic habitat disturbance comes

from studies of either the breakdown of reproductive

isolation between incipient species that have recently

diverged in sympatry (e.g. Todd & Stedman 1989;

Seehausen et al. 1997; Behm et al. 2010; Vonlanthen et al.

2012), or introgression between allopatric species fol-

lowing secondary contact (e.g. Echelle & Connor 1989;

Walters et al. 2008; McDevitt et al. 2009). Our rangewide

study of hybridization between alewife and blueback

herring demonstrates that human activities can also

impact the formation of hybrid swarms between diver-

gent species that naturally occur in sympatry, providing

empirical evidence that reinforcement does not always

sustain reproductive isolation under such circum-

stances.

Spatial extent of hybridization

Despite varying frequencies of hybridization across

their range (Figs 3 and 4), our study reveals that species

integrity is maintained in drainages where anadromous

alewife and blueback herring occur sympatrically. This

is consistent with hybridization studies of European

alosines (A. fallax and A. alosa; Alexandrino et al. 2006;

Coscia et al. 2010; Jolly et al. 2011) with similar diver-

gence times (Bentzen et al. 1993; Faria et al. 2006), and

with studies of other hybridizing species (e.g. Helianthus

sp.; Strasburg & Rieseberg 2008). The frequency of

hybridization among anadromous populations generally

ranged from 0 to 8% (Table S2, Supporting Informa-

tion). Whether this approximates the expected fre-

quency of hybridization under pristine environmental

conditions is uncertain. Many of the rivers examined

have dams that may impact the spatiotemporal distri-

bution of spawning migrations through restricted

Fig. 3 Results of Bayesian clustering analyses for rangewide empirical data for alewife and blueback herring using STRUCTURE (K = 2;

number of species) and NEWHYBRIDS (K = 6; number of genotype frequency classes). Individuals are represented by a thin vertical line

which is partitioned into K-coloured segments representing an individual’s estimated membership fractions from each of the identi-

fied clusters. Kerr Reservoir (KER) is denoted because it showed a substantially greater number of hybrids than other sampling loca-

tions.

Fig. 4 Spatial distribution of the proportion of hybrids identi-

fied using two Bayesian clustering analyses. MAR, PET and

KER (see Table 1) are denoted because they exhibited a sub-

stantially greater proportion of hybrids than other sampling

locations.

Fig. 5 Factorial correspondence analysis revealed two factors

that explained 85.6% of the genetic variation among purebred

and hybrid river herring and demonstrated the clear separation

of landlocked hybrids in Kerr Reservoir from alewife, blueback

herring and anadromous hybrids.



(a)

(b)

Fig. 6 Genomic clines plots generated in

‘introgress’ for (a) anadromous (n = 90)

and (b) landlocked (Kerr Reservoir)

(n = 119) river herring hybrids for nine

microsatellite loci common to both data

sets. The name of each locus and P-value

for the test of departure from neutral

introgression is provided [P < 0.015 and

P < 0.039 indicate significance for anad-

romous and landlocked hybrids, respec-

tively, after FDR correction]. Solid

coloured clines represent the 95% confi-

dence intervals for hybrids with homozy-

gous (Aa/Aa or Ab/Ab; dark green) and

heterozygous (Aa/Ab; light green) geno-

mic clines given neutral introgression.

The solid and dashed lines give the esti-

mated genomic cline based on the

observed homozygous and heterozygous

allelic classes, respectively. Circles indi-

cate the raw allelic class data (Aa/Aa

along the top, Aa/Ab in the centre and

Ab/Ab along the bottom), with counts of

each allelic class along the right vertical

axis. The hybrid index quantifies the

fraction of alleles derived from blueback

herring across the nine microsatellite loci.



habitat access (Hall et al. 2011) and altered spawning

cues (i.e. water temperature; Loesch 1987), and may

facilitate hybridization (Boisneau et al. 1992; Maitland &

Lyle 2005). Spawning alewife and blueback herring are

known to co-occur where upstream migration is

blocked by dams (Loesch 1987), and this may provide

opportunities for interspecific matings. However, the

Delaware River has no mainstem dams and exhibited

5–6% hybridization, suggesting that some level of

hybridization may be natural.

Although we do not know the extent to which pre-

and postzygotic reproductive isolating mechanisms

maintain species integrity, the general absence of deeply

introgressed (i.e. backcrossed) individuals in anadro-

mous populations may indicate an important role for

selection against hybrids under natural conditions.

In contrast to the maintenance of species integrity

among anadromous populations, our study reveals that a

hybrid swarm has become established in Kerr Reservoir

following dam construction that has prevented immigra-

tion of purebred alewife or blueback herring since 1953.

All individuals in Kerr Reservoir were identified as

hybrids (Table S2, Supporting Information; Fig. 4) and

were deeply introgressed – likely via multiple genera-

tions of backcrossing. This finding suggests that both

pre- and postzygotic reproductive isolating mechanisms

may have broken down and that there may not be selec-

tion against hybrids in this altered environment. In unal-

tered drainages, differential spawning habitat selection

exhibited by these species may limit opportunities for

interspecific matings (Loesch 1987). However, the lotic

spawning habitats preferred by blueback herring are not

available in Kerr Reservoir, potentially leading to

increased interspecific matings. Decreased spatial segre-

gation may be accompanied by altered (temperature-

dependent) migratory and spawning cues, possibly

increasing temporal spawning overlap within Kerr Reser-

voir as well. The exact mechanisms at play require fur-

ther investigation. Nonetheless, our study demonstrates

that anthropogenic habitat alterations can breakdown

reproductive isolation between distinct evolutionary lin-

eages that naturally occur in sympatry.

Although hybridization in temperate fishes as a con-

sequence of habitat disturbance is not a new concept

(Hubbs 1955), prior examinations were largely limited

to young evolutionary lineages of freshwater species, or

intraspecific comparisons of anadromous and resident

salmonids (Utter 2001). While hybridization between

anadromous species placed in novel freshwater envi-

ronments has been previously reported (Rosenfield et al.

2000), their collapse as a hybrid swarm has not been

documented. Our study reinforces the contention that

reproductive isolating mechanisms may be incomplete

during the first 2–5 million years after speciation

(Coyne & Orr 2004) and that a substantial portion of

global biodiversity may be susceptible to anthropogeni-

cally induced hybridization (Rhymer & Simberloff 1996;

Seehausen et al. 2008).

Elevated frequencies of hybridization in the Petitcodi-

ac River (~22%) and Margaree River (~12%) (Fig. 4)

were unexpected, but may be explained by a combina-

tion of migratory barriers, decreased abundances and

range-edge effects. A causeway with ineffective fish

passage was constructed near the head of tide in the

Petitcodiac River in 1968. This barrier completely

blocked the access of migratory fishes to upstream

spawning habitat and dramatically reduced the abun-

dance of alewife and blueback herring in the drainage

(Locke et al. 2003). Increased spatiotemporal overlap of

spawning adults below the causeway coupled with

population declines may have increased the chances of

interspecific matings (Reyer 2008). Although dams are

not present on the mainstem Margaree River, river her-

ring have been heavily exploited in this drainage (R.G.

Bradford, personal communication), and low popula-

tion abundances may have increased opportunities for

hybridization. Further, this drainage approximates the

northern range edge of blueback herring where popula-

tion density is low and heterospecific matings may be

more likely (Reyer 2008). However, we did not observe

elevated levels of hybridization for other northern

populations (i.e. Miramichi River, River John). Elevated

levels of hybridization reported for European alosines

(Jolly et al. 2011, 2012) have been attributed to similar

factors (Boisneau et al. 1992; Maitland & Lyle 2005),

but the role of shifting phenologies (Ellis & Vokoun

2009) on the frequency of hybridization requires

attention.

Hybrids detected in rivers from South Carolina to

Florida are beyond the southern range limit for alewife

and could be strays; our sampling of adult specimens

does not preclude their origin in more northerly drain-

ages. However, there is a paucity of evidence for

increased straying rates of hybrid fishes, especially

anadromous species (Scribner et al. 2001; but see Gilk

et al. 2004). Alternatively, these hybrids could be the

progeny of stray purebred alewife that reproduced with

blueback herring in southern rivers. These remain

untested hypotheses that require attention.

Detecting hybrids

Accurate identification of hybrids using molecular

methods depends on the markers used and the degree

of differentiation between parental species (Anderson &

Thompson 2002; V€ah€a & Primmer 2006). Nonetheless,

studies continue to employ a default Tq (0.90) that was

originally derived from simulations (V€ah€a & Primmer



2006); rarely has Tq been assessed on an individual

study basis. Employing this arbitrary threshold in our

study would have inflated our estimates of hybridiza-

tion. Our simulations revealed that misclassifications

were minimized using a Tq of 0.87 (NEWHYBRIDS) and

0.935 (STRUCTURE). This value for STRUCTURE approximates

that used in hybridization studies of European alosines

(0.94; Jolly et al. 2012). We advocate a simulation

approach (as outlined herein) for setting the most

appropriate Tq for discriminating purebred and hybrid

specimens on an individual study basis.

Hybrid divergence in isolation

Although hybridization can threaten biodiversity (Rhymer

& Simberloff 1996), genetic admixture in novel and per-

turbed environments can generate diversity by increasing

the adaptive potential of admixed individuals with novel

genetic variation (e.g. Buerkle et al. 2000; Barton 2001;

Lexer et al. 2003). Our study demonstrates that landlocked

hybrids are genetically distinct from anadromous

hybrids, alewife and blueback herring (Fig. 5). Indeed,

landlocked hybrids are nearly as differentiated from ale-

wife (FST = 0.30) as alewife are from blueback herring

(FST = 0.38). This may have resulted from an initial popu-

lation bottleneck followed by hybridization, and the

effects of drift in isolation from anadromous congeners. A

number of landlocked European alosine populations have

become adapted to lacustrine habitats (Faria et al. 2006)

and are genetically distinct from their anadromous cong-

eners (e.g. A. f. killarensis; Jolly et al. 2012). Our data

suggest that Kerr Reservoir constitutes an admixed popu-

lation in a novel environment that is in the process of

divergence.

Contrasting patterns of introgression

The distribution of blueback herring ancestry observed

among anadromous hybrids (Fig. S3a, Supporting Infor-

mation) is consistent with neutral introgression. This

contrasts with landlocked hybrids, where the decreased

prevalence of blueback herring ancestry (Fig. S3b, Sup-

porting Information) suggests directional introgression

of alewife genotypes in Kerr Reservoir. Although the

proportion of the two species that became landlocked is

unknown, available evidence suggests that blueback

herring were more abundant in the Roanoke River than

alewife when the dam was constructed (Carnes 1965).

Thus, it is unlikely that our result simply reflects a

chance colonization event where alewife were numeri-

cally dominant and where the introgressed gene pool is

largely of alewife origin.

For hybrids, the extent of introgression at individ-

ual loci is a consequence of the fitness effects of

genotype combinations. Contrasts among markers per-

mit the identification of loci that lower hybrid fitness

and contribute to reproductive isolation, or that

increase fitness and promote adaptive introgression

(Gompert & Buerkle 2010). Although our analyses

revealed some concordance in locus-specific patterns

of introgression between anadromous and landlocked

hybrids, we also observed loci that exhibited mark-

edly different patterns of introgression and that may

be subject to varying selection in different settings

(i.e. environments and genetic backgrounds; Nolte

et al. 2009). Differential patterns of introgression have

been previously reported for sculpin (Cottus spp.)

hybrid zones, where discordance in locus-specific pat-

terns has been attributed to differing extrinsic factors

(i.e. local ecological conditions may impose different

selection pressures on admixed genotypes), differing

underlying genetic architecture of reproductive isola-

tion and adaptation and/or the influence of stochas-

ticity and drift in (small) hybrid populations (Nolte

et al. 2009). One or more of these factors may contrib-

ute to the discordance in patterns of introgression

that we observe.

Patterns of non-neutral introgression can be catego-

rized based on their correspondence with different

models of genotypic effects in hybrids (Nolte et al.

2009). For anadromous hybrids, loci that exhibited

increased probability for heterozygous (Aa/Ab) allelic

classes are consistent with overdominance, while

decreased probability for homozygous (Aa/Aa, Ab/Ab)

allelic classes is consistent with negative selection

(Fig. 6a). Conversely, for landlocked hybrids, loci that

exhibited decreased probability for heterozygous allelic

classes are consistent with positive selection, while

increased probability of homozygous allelic classes is

consistent with adaptive introgression (Fig. 6b).

Although different patterns of introgression are

apparent between anadromous and landlocked hybrids,

we hesitate to invoke any particular mode of selection

pending analyses of a larger suite of molecular markers.

The genomic clines method requires a sufficient number

of loci distributed broadly across the genome to ensure

that estimates of genome-wide admixture are represen-

tative of neutral introgression (Gompert & Buerkle

2009). Although the loci used in this study are not

linked, without a linkage map it will be unclear to what

extent loci demonstrating similar patterns of introgres-

sion are independent, and it will not be possible to

determine the proportion of the genome experiencing

different forms of selection (Gompert & Buerkle 2009).

The investigation of contrasting patterns of introgres-

sion between anadromous and landlocked hybrids

warrants further study using advanced genomic

approaches.



Conclusions

Our study reveals that anthropogenic habitat changes

can breakdown reproductive isolation between diver-

gent evolutionary lineages that naturally occur in symp-

atry. While sympatric populations of anadromous

alewife and blueback herring maintain species integrity

across their range, reproductive isolation has broken

down in Kerr Reservoir, leading to the formation of a

hybrid swarm. The construction of dams constitutes a

dramatic alteration to habitat and may disrupt pro-

cesses that sustain reproductive isolation under natural

conditions. We posit that decreased opportunities for

spatiotemporal spawning segregation in Kerr Reservoir

have increased the likelihood of interspecific matings

and have lead to the breakdown of reproductive isola-

tion. The detection of deeply introgressed hybrids in

Kerr Reservoir indicates that hybrids are not disfa-

voured in this altered environment. This contrasts with

a general absence of deeply introgressed hybrids in

anadromous populations, suggesting an important role

for selection against hybrids under natural conditions.

Our results show that reinforcement may be insufficient

to sustain reproductive isolation in the face of some

types of human disturbance. The benefits of dam

removal for the restoration of anadromous fishes have

been largely discussed in the context of replenishing

historic spawning runs (e.g. Hasselman & Limburg

2012). Our study provides evidence that dam removal

could also help maintain species integrity for alewife

and blueback herring.

Acknowledgements

We are grateful to all those who contributed sampling effort to

our ongoing coast-wide genetic studies of river herring ecol-

ogy, evolution and conservation. B. Wynne (North Carolina

Wildlife Resources Commission), D. Michaelson (Virginia

Department of Game and Inland Fisheries) and Casey Seelig

(Dominion Environmental Biology) assisted with the sampling

of Kerr Reservoir. E. M. Labbe, S. Blienbry and L. Thornton

assisted with microsatellite data collection, and T. M. Apgar

assisted with Fig. 1. We also thank three anonymous reviewers

and the associate editor for constructive comments that

strengthened the quality of this paper. This research was

supported by an NSERC Discovery Grant to P. Bentzen, a

National Fish and Wildlife Foundation Grant to E.P. Palkovacs

and North Carolina SeaGrant funding to E.P. Palkovacs and

T.F. Schultz.

References

A’Hara SW, Amouroux P, Argo EE et al. (2012) Permanent

genetic resources added to Molecular Ecology Resources

Database 1(August), pp. 2011–30, September 2011. Molecular

Ecology Resources, 12, 185–189.

Alexandrino P, Faria R, Linhares D et al. (2006) Interspecific

differentiation and intraspecific substructure in to closely

related clupeids with extensive hybridization. Journal of Fish

Biology, 69(Supplement B), 242–259.Allendorf FW, Leary RF, Spruell P, Wenburg JK (2001) The

problems with hybrids: setting conservation guidelines.

Trends in Ecology & Evolution, 16, 613–622.

Anderson E (1948) Hybridization of the habitat. Evolution, 2,

1–9.

Anderson EC, Thompson EA (2002) A model-based method for

identifying species hybrids using multilocus genetic data.

Genetics, 160, 1217–1229.Atlantic States Marine Fisheries Commission (2012) River her-

ring benchmark stock assessment. Volume 1. Stock Assess-

ment Report No. 12-02 of the Atlantic States Marine

Fisheries Commission. Washington, DC.

Barton NH (2001) The role of hybridization in evolution. Molec-

ular Ecology, 10, 551–568.Behm JE, Ives AR, Boughman JW (2010) Breakdown in post-

mating isolation and the collapse of a species pair through

hybridization. The American Naturalist, 175, 11–26.

Belkhir K, Borsa P, Chikhi L, Raufaste N, Bonhomme F (2004)

GENETIX 4.05, logiciel sous Windows TM pour la genetique

des populations.

Benjamini Y, Hochberg Y (1995) Controlling the false disco-

very rate: a practical and powerful approach to multiple

testing. Journal of the Royal Statistical Society Series B, 57,

289–300.

Bentzen P, Paterson IG (2005) Genetic analyses of freshwater

and anadromous alewife (Alosa pseudoharengus) populations

from the St. Croix River, Maine/New Brunswick. Final

Report to Maine Rivers, 3 Wade Street, Augusta, ME 04330.

Bentzen P, Leggett WC, Brown GG (1993) Genetic relationships

among the shads (Alosa) revealed by mitochondrial DNA

analysis. Journal of Fish Biology, 43, 909–917.Boisneau P, Mennesson-Boisneau C, Guyomard R (1992) Electro-

phoretic identity between allis shad, Alosa alosa (L.), and twaite

shad, A. fallax (Lacepede). Journal of Fish Biology, 40, 731–738.

Buerkle CA, Morris RJ, Asmussen MA, Rieseberg LH (2000)

The likelihood of homoploid hybrid speciation. Heredity, 84

(Pt 4), 441–451.Burgarella C, Lorenzo Z, Jabbour-Zahab R et al. (2009) Detec-

tion of hybrids in nature: application to oaks (Quercus suber

and Q. ilex). Heredity, 102, 442–452.

Carnes WC (1965) Survey and Classification of the Roanoke River

Watershed North Carolina. North Carolina Wildlife Resources

Commission, Raleigh, North Carolina, 17p.

Chapman R, Patton J, Eleby B (1994) Comparison of mitochon-

drial DNA variation in four alosid species as revealed by the

total genome, the NADH dehydrogenase I and cytochrome b

regions. In: Genetics and Evolution of Aquatic Organisms (ed.

Beaumont A), pp. 29–263. Chapman and Hall, London, UK.

Coscia I, Rountree V, King JJ et al. (2010) A highly permeable

species boundary between two anadromous fishes. Journal of

Fish Biology, 77, 1137–1149.Coyne JA, Orr HA (2004) Speciation. Sinauer Associates, Sun-

derland, Massachusetts.

De Le�on LF, Raeymaekers JAM, Bermingham E et al. (2011)

Exploring possible human influences on the evolution of

Darwin’s finches. Evolution, 65, 2258–2272.



Echelle AA, Connor PJ (1989) Geographically extensive genetic

introgression after secondary contact between two pupfish

species (Cyprinodon, cyprinodontidae). Evolution, 43, 717–727.

Ellis D, Vokoun JC (2009) Earlier spring warming of coastal

streams and implications for alewife migration timing. North

American Journal of Fisheries Management, 29, 1584–1589.Falush D, Stephens M, Pritchard JK (2003) Inference of popu-

lation structure using multilocus genotype data: Linked

loci and correlated allele frequencies. Genetics, 164, 1567–

1587.

Faria R, Weiss S, Alexandrino P (2006) A molecular phyloge-

netic perspective on the evolutionary history of Alosa spp.

(Clupeidae). Molecular Phylogenetics and Evolution, 40, 298–304.

Gilk SE, Wang IA, Hoover CL et al. (2004) Outbreeding depres-

sion in hybrids between spatially separated pink salmon,

Oncorhynchus gorbuscha, populations: marine survival, hom-

ing ability, and variability in family size. Environmental Biol-

ogy of Fishes, 69, 287–297.Gompert Z, Buerkle CA (2009) A powerful regression-based

method for admixture mapping of isolation across the gen-

ome of hybrids. Molecular Ecology, 18, 1207–1224.

Gompert Z, Buerkle CA (2010) Introgress: a software package

for mapping components of isolation in hybrids. Molecular

Ecology Resources, 10, 378–384.Goudet J (2001) FSTAT, a program to estimate and test gene

diversities and fixation indices (version 2.9.3).

Gow JL, Peichel CL, Taylor EB (2006) Contrasting hybridiza-

tion rates between sympatric three-spined sticklebacks high-

light the fragility of reproductive barriers between

evolutionarily young species. Molecular Ecology, 15, 739–752.

Gow JL, Piechel CL, Taylor EB (2007) Ecological selection

against hybrids in natural populations of sympatric three-

spine sticklebacks. Journal of Evolutionary Biology, 20, 2173–2180.

Guest WC, Drenner R (1991) Relationship between feeding of

blueback herring and the zooplankton community of a Texas

reservoir. Hydrobiologia, 209, 1–6.Hall CJ, Jordaan A, Frisk MG (2011) The historic influence of

dams on diadromous fish habitat with a focus on river her-

ring and hydrologic longitudinal connectivity. Landscape Ecol-

ogy, 26, 95–107.Hall CJ, Jordaan A, Frisk MG (2012) Centuries of anadromous

forage fish loss: consequences for ecosystem connectivity

and productivity. BioScience, 62, 723–731.

Hasselman DJ, Limburg KE (2012) Alosine restoration in the

21st century: challenging the status quo. Marine and Coast

Fisheries: Dynamics, Management, and Ecosystem Science, 4,

174–187.

Hendry AP, Grant PR, Rosemary Grant B et al. (2006) Possible

human impacts on adaptive radiation: beak size bimodality

in Darwin’s finches. Proceedings of the Royal Society B-Biologi-

cal Sciences, 273, 1887–1894.

Holm S (1979) A simple sequentially rejective multiple test pro-

cedure. Scandinavian Journal of Statistics, 6, 65–70.

Hubbs C (1955) Hybridization between fish species in nature.

Systematic Zoology, 4, 1–20.

Jakobsson M, Rosenberg NA (2007) CLUMPP: a cluster match-

ing and permutation program for dealing with label switch-

ing and multimodality in analysis of population structure.

Bioinformatics, 23, 1801–1806.

Jolly MT, Maitland PS, Genner MJ (2011) Genetic monitoring

of two decades of hybridization between allis shad (Alosa

alosa) and twaite shad (Alosa fallax). Conservation Genetics, 12,

1087–1100.Jolly MT, Aprahamian MW, Hawkins SJ et al. (2012) Popula-

tion genetic structure of protected allis shad (Alosa alosa) and

twaite shad (Alosa fallax). Marine Biology, 159, 675–687.

Jordan DS, Evermann BW (1896) The fishes of North and Mid-

dle America. Bulletin of the US National Museum, 47, 1–1240.

Julian SE, Bartron ML (2007) Microsatellite DNA markers for

American shad (Alosa sapidissima) and cross-species amplifi-

cation within the family Clupeidae. Molecular Ecology Notes,

7, 805–807.

Kullback S, Leibler RA (1951) On information and sufficiency.

The Annals of Mathematical Statistics, 22, 79–86.

Labbe EM (2012) Influence of stocking history and geography

on the population genetics of alewife (Alosa pseudoharengus)

in Maine rivers. M.Sc. Thesis. University of Southern Maine.

Portland, ME.

Lehman N, Eisenhawer A, Hansen K et al. (1991) Introgression

of coyote mitochondrial DNA into sympatric North Ameri-

can gray wolf populations. Evolution, 45, 104–119.Lexer C, Welch ME, Durphy JL, Rieseberg LH (2003) Natural

selection for salt tolerance quantitative trait loci (QTLs) in

wild sunflower hybrids: implications for the origin of

Helianthus paradoxus, a diploid hybrid species. Molecular

Ecology, 12, 1225–1235.

Limburg KE, Waldman JR (2009) Dramatic declines in North

Atlantic diadromous fishes. BioScience, 59, 955–965.Locke A, Hanson JM, Klassen GJ et al. (2003) The damming of

the Petitcodiac River: species, populations, and habitats lost.

Northeastern Naturalist, 10, 39–54.

Loesch JG (1987) Overview of life history aspects of anadro-

mous alewife and blueback herring in freshwater habitats.

American Fisheries Society Symposium, 1, 89–103.Maitland PS, Lyle AA (2005) Ecology of Allis Shad Alosa alosa

and Twaite Shad Alosa fallax in the Solway Firth, Scotland.

Hydrobiologia, 534, 205–221.

McBride M (2013) Population structure of river herring (alewife,

Alosa pseudoharengus and blueback herring, A. aestivalis)

examined using neutral genetic markers. M.Sc. Thesis, Dalhousie

University, Halifax, Nova Scotia.

McDevitt AD, Mariani S, Hebblewhite M et al. (2009) Survival

in the Rockies of an endangered hybrid swarm from

diverged caribou (Rangifer tarandus) lineages. Molecular Ecol-

ogy, 18, 665–679.

Messieh S (1977) Population structure and biology of alewife

(Alosa pseudoharengus) and blueback herring (A. aestivalis) in the

St. John River, NB. Environmental Biology of Fishes, 2, 195–210.Nielsen EE, Bach LA, Kotlicki P (2006) Hybridlab (version 1.0):

a program for generating simulated hybrids from population

samples. Molecular Ecology Notes, 6, 971–973.

Nolte AW, Gompert Z, Buerkle CA (2009) Variable patterns of

introgression in two sculpin hybrid zones suggest that geno-

mic isolation differs among populations. Molecular Ecology,

18, 2615–2627.

Palkovacs EP, Dion KB, Post DM, Caccone A (2008) Indepen-

dent evolutionary origins of landlocked alewife populations

and rapid parallel evolution of phenotypic traits. Molecular

Ecology, 17, 582–597.



Palkovacs EP, Hasselman DJ, Argo EE et al. (2013) Combining

genetic and demographic information to prioritize recovery

efforts for anadromous alewife and blueback herring. Evolu-

tionary Applications, 7, 212–226.Prince ED, Barwick DH (1981) Landlocked blueback herring in

two South Carolina reservoirs: reproduction and suitability

as stocked prey. North American Journal of Fisheries Manage-

ment, 1, 41–45.Pritchard JK, Stephens M, Donnelly P (2000) Inference of popu-

lation structure using multilocus genotype data. Genetics,

155, 945–959.

R Development Core Team (2010) R: A Language and Environ-

ment for Statistical Computing. R Foundation for Statistical

Computing, Vienna, Austria. ISBN 3-900051-07-0, URL

http://www.R-project.org.

Reyer HU (2008) Mating with the wrong species can be right.

Trends in Ecology & Evolution, 23, 289–292.

Rhymer M, Simberloff D (1996) Extinction by hybridization

and introgression. Annual Reviews of Ecology and Systematics,

27, 83–109.Rice WR (1989) Analyzing tables of statistical tests. Evolution,

43, 223–225.Roca AL, Georgiadis N, O’Brien SJ (2005) Cytonuclear genomic

dissociation in African elephant species. Nature Genetics, 37,

96–100.

Rosenberg NA (2004) DISTRUCT: a program for the graphical

display of population structure. Molecular Ecology Notes, 4,

137–138.

Rosenfield JA, Todd T, Greil R (2000) Asymmetric hybridiza-

tion and introgression between pink salmon and Chinook

salmon in the Laurentian Great Lakes. Transactions of the

American Fisheries Society, 129, 670–679.

Rousset F (2007) Genepop’007: a complete re-implementation

of the genepop software for Windows and Linux. Molecular

Ecology Notes, 8, 103–106.Rundle HD, Nosil P (2005) Ecological speciation. Ecology Let-

ters, 8, 336–352.Rundle HD, Schluter D (1998) Reinforcement of stickleback

mate preferences: sympatry breeds contempt. Evolution, 52,

200–208.

Scott WB, Crossman ED (1973) Freshwater fishes of Canada. J.

Fish. Res. Bd. Can. No. 184.

Scribner KT, Page KS, Bartron ML (2001) Hybridization in

freshwater fishes: a review and case studies and cytonuclear

methods of biological inference. Reviews in Fish Biology and

Fisheries, 10, 293–323.

Seehausen O (2006) Conservation: losing biodiversity by

reverse speciation. Current Biology, 16, 334–337.

Seehausen O, van Alphen J, Witte F (1997) Cichlid fish diver-

sity threatened by eutrophication that curbs sexual selection.

Science, 277, 1808–1811.Seehausen O, Takimoto G, Roy D, Jokela J (2008) Speciation

reversal and biodiversity dynamics with hybridization in

changing environments. Molecular Ecology, 17, 30–44.

Servedio MR, Noor MAF (2003) The role of reinforcement in

speciation: theory and data. Annual Review of Ecology, Evolu-

tion, and Systematics, 34, 339–364.SPSS, Inc. (2004) SYSTAT version 11.0. SPSS, Inc., Chicago, Illi-

nois.

Strasburg JL, Rieseberg LH (2008) Molecular demographic

history of the annual sunflowers Helianthus annuus and H.

petiolaris-large effective population sizes and rates of long-

term gene flow. Evolution, 62, 1936–1950.Taylor EB, Boughman JW, Groenenboom M et al. (2006) Specia-

tion in reverse: morphological and genetic evidence of the

collapse of a three-spined stickleback (Gasterosteus aculeatus)

species pair. Molecular Ecology, 15, 343–355.Todd N, Stedman RM (1989) Hybridization of ciscoes (Coreg-

onus spp.) in Lake Huron. Canadian Journal of Zoology, 67,

1679–1685.

Utter F (2001) Patterns of subspecific anthropogenic introgres-

sion in two salmonid genera. Reviews in Fish Biology and Fish-

eries, 10, 265–279.V€ah€a JP, Primmer CR (2006) Efficiency of model-based Bayes-

ian methods for detecting hybrid individuals under different

hybridization scenarios and with different numbers of loci.

Molecular Ecology, 15, 63–72.Van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P

(2004) MICRO-CHECKER: software for identifying and cor-

recting genotyping errors in microsatellite data. Molecular

Ecology Notes, 4, 535–538.Vonlanthen P, Bittner D, Hudson AG et al. (2012) Eutrophica-

tion causes speciation reversal in whitefish adaptive radia-

tions. Nature, 482, 357–362.

Walters DM, Blum MJ, Rashleigh B et al. (2008) Red shiner

invasion and hybridization with blacktail shiner in the upper

Coosa River, USA. Biological Invasions, 10, 1229–1242.Waters JM, Epifanio JM, Gunter T, Brown BL (2000) Homing

behaviour facilitates subtle genetic differentiation among

river populations of Alosa sapidissima: microsatellites and

mtDNA. Journal of Fish Biology, 56, 622–636.

Weir BS, Cockerham CC (1984) Estimating F-statistics for

the analysis of population structure. Evolution, 38, 1358–

1370.

Wiegand K (1935) A taxonomist’s experience with hybrids in

the wild. Science, 81, 161–166.

D.J.H. designed the study with E.P.P., conducted the

analyses and wrote the manuscript. E.E.A. M.C.M. and

A.A.P.-U. collected data and contributed to manuscript

revisions. E.P.P., T.F.S. and P.B. funded the project and

contributed to manuscript revisions.

Data accessibility

Microsatellite data used in this manuscript: DRYAD

Digital Repository. doi:10.5061/dryad.8v0c3.

All necessary input files required to replicate our

genomic clines analysis using the R package ‘intro-

gress’: DRYAD Digital Repository doi:10.5061/dryad.

ft48k.

Supporting information

Additional supporting information may be found in the online ver-

sion of this article.



Fig. S1 Allele frequency distributions for alewife ( ) and blue-

back herring ( ) for 15 loci examined in this study. Estimates

of the Kullback–Leibler divergence between the species for

each locus are located in the top-right of each panel.

Fig. S2 Mean (�SD) of q-values for five simulated data sets of

six hybrid categories (ALE: alewife, BBH: blueback herring, F1,

F2 and F1 backcrosses) analysed using (a) STRUCTURE (K = 2;

number of species) and (b) NEWHYBRIDS (K = 6; number of geno-

type frequency classes). The horizontal dashed line indicates

q = 0.90.

Fig. S3 Ancestry plots generated in ‘introgress’ for (a) anadro-

mous (n = 90) and (b) landlocked (Kerr Reservoir) (n = 119)

river herring hybrids for nine microsatellite loci common to

both data sets. Dark green blocks indicate hybrids that are

homozygous for alewife allelic classes (Aa/Aa), light green

blocks indicate hybrids that are homozygous for blueback her-

ring allelic classes (Ab/Ab), and intermediate green blocks cor-

respond to hybrids that are interclass heterozygotes (Aa/Ab).

White blocks indicate missing data. The plot to the right in

each panel indicates the proportion of each individual’s gen-

ome that has blueback herring ancestry; equivalent to the

hybrid index. Individuals are sorted, with those that have

genomic compositions resembling alewife at the bottom and

increasing similarity to blueback herring toward the top.

Table S1 Classification rules for delineating purebred and

hybrid specimens based on the optimal threshold (Tq) deter-

mined using Bayesian analyses of five simulated data sets.

Table S2 Proportion of hybrids detected by river using STRUC-

TURE and NEWHYBRIDS.



Documents

Human disturbance causes the formation of a hybrid swarm ... · parapatry, hybridization between forest elephant (Loxodonta cyclotis) and savannah elephant (L. africana) threatens