Pyganodon (Bivalvia: Unionoida: Unionidae) phylogenetics: A male-and female-transmitted mitochondrial DNA perspective

Molecular Phylogenetics and Evolution 63 (2012) 430–444

Contents lists available at SciVerse ScienceDirect

Molecular Phylogenetics and Evolution

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

Pyganodon (Bivalvia: Unionoida: Unionidae) phylogenetics:A male- and female-transmitted mitochondrial DNA perspective

Hélène Doucet-Beaupré a,⇑, Pierre U. Blier a, Eric G. Chapman b, Helen Piontkivska c, France Dufresne a,Bernard E. Sietman d, Renee S. Mulcrone c, Walter R. Hoeh c,⇑a Département de Biologie, Université du Québec à Rimouski, 300 Allée des Ursulines, Rimouski, Québec, Canada G5L 3A1b Department of Entomology, University of Kentucky, S225 Agricultural Science Center N Lexington, KY 40546-0091, USAc Department of Biological Sciences, Kent State University, Kent, OH 44242, USAd Minnesota Department of Natural Resources, Saint Paul, MN 55155, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 13 June 2011Revised 17 January 2012Accepted 20 January 2012Available online 2 February 2012

Keywords:UnionoidaPhylogeographyDoubly uniparental inheritancePyganodonFreshwater musselsSequence-based species delimitation

1055-7903/$ - see front matter � 2012 Elsevier Inc. Adoi:10.1016/j.ympev.2012.01.017

⇑ Corresponding authors.E-mail addresses: [email protected]

[email protected] (P.U. Blier), ericgchapman@[email protected] (H. Piontkivska), [email protected]@state.mn.us (B.E. Sietman), [email protected]

Species boundaries, evolutionary relationships and geographic distributions of many unionoid bivalvespecies, like those in the genus Pyganodon, remain unresolved in Eastern North America. Because unio-noid bivalves are one of the most imperiled groups of animals in the world, understanding the geneticvariation within and among populations as well as among species is crucial for effective conservationplanning. Conservation of unionoid species is indispensable from a freshwater habitat perspective butalso because they possess a unique mitochondrial inheritance system where distinct gender-associatedmitochondrial DNA lineages coexist: a female-transmitted (F) mt genome and a male-transmitted (M)mt genome that are involved in the maintenance of separate sexes (=dioecy). In this study, 42 populationsof Pyganodon sp. were sampled across a large geographical range and fragments of two mitochondrialgenes (cox1 and cox2) were sequenced from both the M- and F-transmitted mtDNA genomes. Our resultssupport the recency of the divergence between P. cataracta and P. fragilis. We also found two relativelydivergent F and M lineages within P. grandis. Surprisingly, the relationships among the P. grandis speci-mens in the F and M sequence trees are not congruent. We found that a single haplotype in P. lacustrishas recently swept throughout the M genotype space leading to an unexpectedly low diversity in theM lineage in that species. Our survey put forward some challenging results that force us to rethinkhybridization and species boundaries in the genus Pyganodon. As the M and F genomes do not always dis-play the same phylogeographic story in each species, we also discuss the importance of being careful inthe interpretation of molecular data based solely on maternal transmitted mtDNA genomes. The involve-ment of F and M genomes in unionoid bivalve sex determination likely played a role in the genesis of theunorthodox phylogeographic patterns reported herein.

� 2012 Elsevier Inc. All rights reserved.

1. Introduction

Unionoid bivalves, also called freshwater mussels, are an impor-tant but neglected component of freshwater aquatic ecosystems.They are long-lived partially infaunal filter feeders that can befound in most permanent freshwater bodies. North America boaststhe richest unionoid fauna in the world with nearly 300 currentlyrecognized species and subspecies, representing two families(Unionidae and Margaritiferidae) in the United States and Canada(Bogan, 1993; Turgeon et al., 1988, 1998; Williams et al., 1993).Unionoid bivalves have a unique and complex life cycle. They are

ll rights reserved.

m (H. Doucet-Beaupré), pier.com (E.G. Chapman), opiontqc.ca (F. Dufresne), bernard.

(W.R. Hoeh).

ovoviviparous and depend on host species for reproduction sincetheir life cycle typically includes a short obligate parasitic larvalstage on fish or amphibians hosts (Bauer and Wächtler, 2000). Asadults, they have limited dispersal abilities and the release of sper-matozoa from males occurs in the open water of lakes, rivers andstreams, which is the main dispersal mode along with the parasiticlarval stage (Nagel, 2000). The insular nature and diversity of fresh-water habitats (Strayer, 2006), their limited dispersal opportunities(Bilton et al., 2001), their high degree of variation and endemismand association with major drainage systems with relatively longgeologic histories (Burch, 1975; Burky, 1983; McMahon, 1991) aswell as their unique and complex combination of life history traits(Haag and Staton, 2003) has deeply influenced the observed genet-ic diversity and geographic variation in unionoid bivalves.

Unionoid bivalves are also one of the most imperiled groups ofanimals in the world (Lydeard et al., 2004), with 70% of the

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 431

recognized species in North America considered extinct, endan-gered, threatened or of special concern (Williams et al., 1993).Unionoids are generally sensitive to ecosystem deterioration andhave declined for various reasons such as watershed alterations,changes in water quality resulting from organic enrichment andchemical contamination, and the recent introduction of the exoticzebra mussel, Dreissena polymorpha (Nalepa, 1994; Schloesser andNalepa, 1994), the quagga mussel, Dreissena bugensis, and the Asianclam, Corbicula fluminea (McMahon, 1983; Ricciardi et al., 1998).Effective conservation planning for unionoids includes protectingdistinct species and populations and requires an understandingof genetic variation within and among populations for proper spe-cies and subspecies delineation. Unionoid species identification isusually based on conchological characteristics but phenotypicplasticity in shell morphology renders the differentiation of partic-ular species problematic (e.g., species within the genus Pyganodon;Cyr et al., 2007). Alternatively, molecular markers can be used for amore accurate estimate of species diversity within unionoid lin-eages and to investigate population genetics and phylogeography.

Despite its unquestionable relevance to freshwater mussel con-servation, the study of unionoid genetics is of pivotal importance inregard to this taxon’s unique mitochondrial inheritance system.Unionoid bivalves possess a distinct form of mitochondrial DNA(mtDNA) inheritance, termed ‘‘doubly uniparental inheritance’’(DUI; for review, see Breton et al., 2007; Passamonti and Ghiselli,2009). In these organisms, distinct gender-associated mitochon-drial DNA lineages coexist: a female-transmitted (F) genome anda male-transmitted (M) genome. Under DUI, female bivalves trans-mit their mitochondria (F mtDNA) to both sons and daughters, asin standard maternal inheritance, but males pass on their mito-chondria (via sperm carrying M mtDNA) to only sons (e.g., Bretonet al., 2007; but see Chakrabarti et al., 2007; Obata et al., 2006,2007). At the organismal level, male bivalves with DUI are thusheteroplasmic and contain both M and F genomes. Evidence has re-cently been presented that is consistent with the hypothesis thatthe DUI system in unionid and margaritiferid bivalves functionsin the maintenance of separate sexes (=dioecy; Breton et al.,2011). The bivalve order Unionoida is the only freshwater taxonwhere DUI is present; the two other bivalve lineages possessingDUI are marine (Mytiloida and Veneroida). This nontraditionaland unique mode of mitochondrial inheritance provides the oppor-tunity to study the mitochondrial DNA evolution from a differentand privileged point of view and could be very useful in answeringmany systematic and phylogenetic questions that cannot be effec-tively solved by studying taxa with the classical animal mitochon-drial transmission mode (i.e., strict maternal inheritance [SMI])(Passamonti and Ghiselli, 2009). Specifically, studies of intra- andinterspecific genetic variation for both M and F genomes poten-tially offer complementary views on species limits, evolutionaryrelationships, and levels of population variation (Hoeh et al.,2002; Krebs, 2004). Furthermore, extensive phylogenetic analysisof M and F genomes might shed light on molecular evolutionaryprocesses affecting these unique mtDNA genomes.

The genus Pyganodon (Unionidae: Unioninae) (previouslyincluded in the genus Anodonta; see Hoeh, 1990) contains severalconchologically similar species whose evolutionary relationshipsand distributions in North America remain uncertain. Clarke(1981) recognized four species in this genus distributed across Can-ada, namely P. cataracta (Say, 1817), P. fragilis (Lamarck, 1819), P.grandis (Say, 1829) and P. simpsoniana (Lea, 1861). In a revised clas-sification of the Eastern North American Anodonta, Hoeh (1990)indicated that Pyganodon includes two other species, P. lacustris(Lea, 1857) and P. gibbosa (Say, 1824) without mentioning P. simp-soniana (which could be conspecific with P. grandis, Clarke, 1981).In a recent survey of the genus based on morphological charactersand DNA sequences from nuclear (ITS1&2), F mitochondrial and M

mitochondrial loci (cox1 and 16S) from southern Quebec, Cyret al. (2007) (i) confirmed the distinction between P. fragilis and P.cataracta, (ii) corroborated the presence of P. grandis but not P. simp-soniana and (iii) found two additional Pyganodon mt lineages thatthey were unable to identify.

The aims of the present study are (1) to clarify the systematicand phylogenetic relationships among Pyganodon lineages innortheastern North America using F and M mtDNA sequences fromcytochrome c oxidase subunits I and II (cox1 and cox2); (2) to com-pare the phylogenetic signal obtained from M and F mitochondrialgenomes in this group and (3) to place the pattern of genetic var-iation within and among lineages in a broad geographical and his-torical context.

2. Material and methods

2.1. Populations sampled

Males and females, representing four Pyganodon species, werecollected in northeastern North America from different popula-tions in Newfoundland and Labrador, Quebec, Michigan,Minnesota, Wisconsin, New Jersey, Indiana, Maine, Maryland,Massachusetts and Pennsylvania (Table 1 and Fig. 1). Mussels werequickly transported to the laboratory and dissected. Tissues werefrozen using liquid nitrogen in the field or whole mussels were pre-served in 95–100% non-denatured ethanol in the field. Microscopicexamination of gonadal tissue allowed for rapid determination ofmussel gender (by the presence of eggs or sperm/sperm morulae).Dissected tissues samples were stored either at �80 �C or in95–100% ethanol at �20 �C.

2.2. DNA extraction, amplification and sequencing

Total genomic DNA extractions were performed on the dissectedtissues using a QIAGEN DNEasy Extraction Kit (QIAGEN Inc., Missis-sauga, Canada) following the manufacturer’s animal tissue protocol.Extracted DNA quality and quantity were examined using 0.7% aga-rose gel electrophoresis and the DNA were stored at �20 �C. Fourdistinct gene regions were analyzed in the present study: Fcox1,Mcox1, Fcox2, and Mcox2e. The complete F cytochrome c oxidasesubunit II (Fcox2), Mcox2e and partial F and M cytochrome c oxidasesubunit I (cox1) gene regions were amplified and directly sequencedwithout cloning. The following primer pair was used to amplifycox1: LCO22me2/HCO700dy2 (Walker et al., 2006). These primersare complementary to very conserved regions of animal mtDNA(Folmer et al., 1994) and have been shown to reliably PCR amplifymtDNA from unionoid bivalves (Hoeh et al., 1996, 1998; Boganand Hoeh, 2000; Hoeh et al., 2001, 2002) and they facilitated theproduction of both Fcox1 (from ovarian or mantle tissue-basedDNA) and Mcox1 (from testicular tissue-based DNA) amplicons.The primer pair UNIOND3-155F (50-AGHSCKTTTGARTGYGGKTTTGA-30)/FCOIPygR (50-TGCCARTAACAARTAYAAAGTA-30) was usedto amplify the Fcox2 gene from ovarian or mantle tissue-basedDNA. Additionally, a third primer pair (MCOIIh 35F 50-TTTATRCCTRTKKTGTGTRGARGCTGT-30/PygMcox2eR 50-TAYAATCTTYCAATRTCYTTATGATT-30) was used to amplify the hypervariable extensionregion of Mcox2 (=Mcox2e) (e.g., Chapman et al., 2008; Walkeret al., 2006) from testicular tissue-based DNA. Some primers had a50 M13 tail added for sequencing purposes. All amplifications wereperformed in 50-ll volumes of a solution containing 1X Qiagen buf-fer, 2.5 mM MgCl2, 200 lM of each dNTP, 0.5 lM of each primer,2.5 U Qiagen Taq (QIAGEN Inc., Mississauga, Canada) and �1 lg oftemplate DNA. Thermal cycling conditions for cox1 and cox2 ampli-fication were as follows: 94 �C for 3 min, followed by 35 cycles of94 �C for 1 min, 54 �C for 1 min and 72 �C for 90 s and a finalextension at 72 �C for 5 min. The resulting PCR products were gel

Table 1Description and geographical coordinates of sampling sites for Pyganodon lacustris, P. cataracta, P. fragilis and P. grandis and numbers of individual per morphospecies sampled (seeFig. 1 and 9–12) documented in each site and morphospecies. L.: Lake, R.:River, P.:Pond, Res.: Reservoir, Co.: County, Gr: P. Grandis, Fr: P.fragilis, Ca: P.cataracta, La: P.lacustris.

State/province (Country) Exact site Region Lat. Long. No. on map Species (N)

Female Male

Indiana (USA) Mill Creek Jackson Co. 41.5664 �86.5490 39 Gr (3) –Maine (USA) Indian L. Washington Co. 45.4174 �69.2979 32 Ca (2) Fr (1)Maine (USA) Tilson farm P. Hampshire Co 42.3903 �72.5327 29 Ca (3) Ca (3)Maryland (USA) Loch Raven Res. Baltimore Co. 39.4177 �76.5383 36 Ca (1) –Michigan (USA) Beach pool Emmet Co. 45.7372 �84.8019 34 La (5) –Michigan (USA) Cedar R. Gladwin Co. 43.9932 �84.3704 19 Gr (4) Gr (1)Michigan (USA) Fish L. Calhoun Co. 42.0494 �85.8600 37 Gr (2) Gr (3)Michigan (USA) Lyons L. Kalamazoo Co 42.1884 �84.9668 28 Gr (8) Gr (5)Michigan (USA) Mona L. Muskegon Co. 43.1767 �86.2602 18 Gr (2) Gr (1)Michigan (USA) Pogey L. Mecosta Co. 43.7962 �93.4821 17 Gr (8) –Michigan (USA) Prairie L. Barry Co. 41.8586 �85.4037 27 Gr (5) –Minnesota (USA) Big Fork R. Koochiching Co. 47.8004 �93.5724 38 Gr (1) –Minnesota (USA) Black Duck L. Koochiching Co. 48.2087 �92.8094 9 La (11) La (6)Minnesota (USA) Bowstring R. Itasca Co. 47.4979 �93.7389 33 La (4) –Minnesota (USA) Deer L. Itasca Co. 47.8253 �93.3748 11 La (7) Gr (8) La(12) Gr (2)Minnesota (USA) Elephant L. Koochiching Co. 48.1819 �92.7373 8 La (4) La (6)Minnesota (USA) Mississippi R. Goodhue Co. 44.5269 �92.3297 16 Gr (4) Gr (1)Minnesota (USA) Pelican L. St. Louis Co. 48.0633 �92.8321 10 La (9) Gr (3) La (5) Gr (4)Minnesota (USA) Pfeiffer L. St. Louis Co. 47.7515 �92.4771 12 La (4) Gr (5) La (1) Gr (1)Minnesota (USA) Prairie R. Itasca Co. 47.2391 �93.4821 13 La (5) Gr (13) La (9) Gr (6)Minnesota (USA) Rice R. Aitkin Co. 46.5326 �93.3200 14 La (9) Gr (7) La(26) Gr (6)Minnesota (USA) Twin Lakes Itasca Co. 45.2483 �94.2126 15 La (2) –New Jersey (USA) Mantua Creek Gloucester Co. 39.7088 �75.0923 35 Ca (6) –Nfld (Canada) Burin L. Eastern 47.0504 �55.1632 7 Fr (1) Fr (1)Nfld. (Canada) Birds pond Whittbourne 47.4162 �53.5281 31 Fr (8) Fr (4)Nfld. (Canada) Goose Bay Labrador 53.2945 �60.3618 30 Ca (1) Fr (5) Fr (5)Pennsylvania (USA) Pickering Creek Chester Co. 40.1034 �75.5337 26 Ca (6) Fr (1) Fr (3)Quebec (Canada) Atikamakusch L. Nord du Québec 53.5044 �77.7186 1 Gr (8) Gr (9)Quebec (Canada) Bellevue L. Capitale-Nationale 47.1833 �72.2333 4 – Ca (5)Quebec (Canada) Elgin L. Estrie 45.7483 �71.3361 6 – Fr (1)Quebec (Canada) Macpès L. Bas-Saint-Laurent 48.3119 �68.4869 3 Fr (9) Fr (9)Quebec (Canada) Matapédia L. Bas-Saint-Laurent 48.5588 �67.5855 2 Fr (3) Fr (11)Quebec (Canada) Sainte-Marie L. Outaouais 45.9561 �75.9305 5 Gr (2) Gr (3)Wisconsin (USA) Annabelle L. Vilas Co. 46.2206 �89.6787 22 La (6) Gr (2) La (2)Wisconsin (USA) Chetak L. Sawyer Co 45.7253 �91.4936 20 La (1) Gr (3) La (1) Gr (1)Wisconsin (USA) Kentuck L. Vilas Co. 45.9847 �89.0120 21 La (6) La (4)Wisconsin (USA) Mendota L. Dane Co. 43.1224 �89.4042 24 Gr (3) Gr (1)Wisconsin (USA) North Twin L. Vilas Co. 46.0665 �89.0887 25 La (4) Gr (6) La (4) Gr (4)Wisconsin (USA) St. Germain R. Vilas Co 45.9184 �89.5324 23 La (1) Gr (6) La (5) Gr (6)

-120o -100o -80o -60o30o35o40o45o50o

55o

60o

65o

70o

1

234

5 67

89 10

11

12

131415 1617

181920

2122

23

2425

262728 29

30

31

32

3334

37

38

3536

39

Fig. 1. Sampling sites of Pyganodon spp. (black .) and geographical occurrence ofPyganodon spp. (yellow j) based on Global Biodiversity Information Facility (GBIF;Biodiversity occurrence data provided by: Canadian Museum of Nature, EMANProvider, New Brunswick Museum, National Museum of Natural History andUniversity of Kansas Biodiversity Research Center (Accessed through GBIF DataPortal, data.gbif.org, 2010-03-26). Sites are numbered as in Table 1. The maximumextent of the proglacial lakes (dashed line) is outlined.

432 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

purified using Qiagen QIAquick Gel Extraction Kit (Qiagen, Valencia,California, USA) or Promega Wizard PCR Kits (Promega, Madison,

Wisconsin, USA) and sequenced at the Sequencing Platform(3730XL, Applied Biosystems) of McGill University (Québec,Canada; using the PCR primers) or at Geneway Research (Hayward,California, USA; using M13 primers).

2.3. Phylogenetic reconstruction

All sequences were trimmed and assembled with the programBioEdit (v.7.0.4.1, Hall, 1999) and multiple sequence alignmentswere made using Clustal W (Larkin et al., 2007) in MEGA v. 4.0(Tamura et al., 2007), using default parameters and Dialign v.2.2.1 (Subramanian et al., 2008) followed by manual alignmentadjustments. Phylogenetic trees for the F and M genomes of Pygan-odon were separately estimated using Bayesian inference (BI) andmaximum likelihood (ML) analyses of concatenated nucleotide se-quences from the cox1 and cox2 protein-coding genes. For the F se-quence trees, sequences from 232 Pyganodon individuals and fiveoutgroup sequences representing three Anodonta species (A. ber-ingiana, A. woodiana and A. californiensis) were aligned using Clus-tal W in MEGA 4.0 and manually reviewed. In the same way,sequences from 184 Pyganodon individuals and those representingthree outgroup species (A. beringiana, A. woodiana and A. californi-ensis) were aligned to produce the M sequence trees. The analyzedF-matrix had 1389 nucleotide positions (comprising 678 nucleo-tides from cox2 and 711 nucleotides from cox1) and the M-matrix1245 nucleotides positions (comprising 585 nucleotides from cox2

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 433

and 660 nucleotides from cox1). These matrices are available fromWRH.

Phylogenetic analyses were conducted using Bayesian inference(BI) via Mr. Bayes (v. 3.1.2; Huelsenbeck and Ronquist, 2001;Ronquist and Huelsenbeck, 2003). Two independent simultaneousanalyses were done using the GTR + G + I substitution model(Rodriguez et al., 1990) for both the F and M sequence matrices.Searches were conducted for five million generations with sixsearch chains each. The molecular data were partitioned by generegion and by codon position (=two gene regions, each with threecodon positions for the cox1 and cox2 partitions) yielding a total ofsix partitions, and saving a total of 100,000 trees (one tree savedevery 100 generations in each of the two analyses) from both theF and M sequence analyses. To allow each partition to have itsown set of parameter estimates, revmat, tratio, statefreq, shape,and pinvar were all unlinked during the analyses. The BI analyseswere terminated when the average standard deviation of the splitfrequencies fell below 0.020 and 0.011 for the F and M analyses,respectively. The 60,000 post-burn-in trees (determined by exam-ination of the log probability of observing the data � generationplots) were used to generate the F and M majority-rule consensustrees. To allow each gene to evolve independently at its own rate,the option prset ratepr = variable was employed as per the recom-mendations of Marshall et al. (2006). Best maximum likelihood(ML) F and M trees were generated using GARLI (v. 0.951; Zwickl,2006) with no data partitioning using default settings except forthe following: autoterminate run 1000,000 generations post lastimproved topology, lnL increase for significantly better topol-ogy = 0.0001 and score improvement threshold = 0.0005. Garliwas also used to generate a 200-replicate ML majority-rule boot-strap (Felsenstein, 1985) trees using default settings except forthe following: lnL increase for significantly better topology = 0.001and score improvement threshold = 0.005, based on analyses of theF and M sequence matrices with no data partitioning.

2.4. Sequence-based species delimitation

We implemented two different empirical methods to delimitspecies, including a general mixed Yule-coalescent method(GMYC) and a statistical parsimony network analysis. The generalmixed Yule coalescent model (Fontaneto et al., 2007; Pons et al.,2006) is a statistical procedure for estimating species boundariesdirectly from phylogenetic trees. Equations describing two typesof lineage evolution, speciation (a stochastic birth-only or Yule coa-lescent model) and a model that describes populations (a neutralcoalescent model; microevolution) are combined and applied onan ultrametric tree. The procedure identifies a threshold value forthe transition from coalescent to speciation branching patterns.Location of the transition point on the phylogenetic tree definesthe species boundary. A standard log-likelihood ratio test is thenused to assess whether the stepped model provides a significantlybetter fit than the null model of no such shift in branching process.The analysis was carried out using R package SPLITS (Species Limitsby Threshold Statistics) available at http://r-forge.r-project.org/projects/splits/. To build the clock-constrained tree, we used theBayesian consensus tree as described before. From this tree, iden-tical haplotype and out-group sequences were removed and con-verted to an ultrametric tree using penalized likelihood in r8s (v.1.7; Sanderson, 2003), with the optimal smoothing parameterdetermined by cross validation between 10�5 and 105.

For the second algorithmic method of sequence-based speciesdelineation, patterns of sequence variation were analyzed for thepresence of species level groups by identifying independent net-works using parsimony (Templeton et al., 1992) as implementedin the software TCS v. 1.21 (Clement et al., 2000). This procedurepartitions the data into independent networks of closely related

haplotypes connected by changes with a <95% probability of beingnon-homoplasmic. Alignment gaps were treated as missing data.

We measured divergence (uncorrected p-distance) within andamong lineages using MEGA v. 4. To test whether genetic differen-tiation among putative species was significant, we used analysis ofmolecular variance (AMOVA; Excoffier et al., 1992) implemented inARLEQUIN v. 3.11 (Excoffier et al., 2005). For combined cox1 andcox2 sequence from each Pyganodon lineage, standard measuresof polymorphic sites, nucleotide diversity and sequence diversitywere obtained from the PGEToolbox for MATLAB (Cai, 2008) andARLEQUIN v. 3.11 (Excoffier et al., 2005).

3. Results

In this study, DNA samples for the cox1 gene fragment withinthe genus Pyganodon were analyzed to produce 224 F haplotypesand 178 M haplotypes. We also obtained 107 F haplotypes and164 M haplotypes of a fragment of the cox2 gene. Four morpholog-ical species (hereafter called morphospecies) were identified anddesignated as P. cataracta, P. fragilis, P. lacustris and P. grandis(Fig. 2). The molecular divergence between morphospecies is pre-sented in Table 2 for F haplotypes and in Table 3 for M haplotypes.The further subdivisions within morphospecies are based on phy-logenetics analysis.

3.1. Phylogenetic analysis

The BI and ML phylogenetic analyses produced similar topolo-gies for F and M data sets. Therefore, in the following discussion,we used as our point of reference the majority-rule consensus phy-logenetic trees obtained by Bayesian inference (Fig. 2; for a de-tailed structure of the F and M trees, see Figs. 3–8). In the F tree,two relatively differentiated clades were found (hereafter LFAand LFB; see Fig. 4) in P. lacustris morphospecies (Table 2 andFig. 3). The branching patterns in the M tree were not the samewhere P. lacustris was comprised of a single clade containing verysimilar haplotypes. However, in the M tree, the P. lacustris cladeis sister to the P. cataracta + P. fragilis lineage as in the F tree. Fur-thermore, in the F tree, two, relatively differentiated clades (here-after named GFA and GFB) were found in P. grandis rendering thespecies paraphyletic (Figs. 2 and 5 and Table 2). The presence oftwo, non-sister clades in the F mt genomes obtained from P. grandismorphospecies indicates cryptic mt lineages. Specifically, the P.grandis clade GFA is sister to the F clade containing P. cataracta + P.fragilis + P. lacustris morphospecies. The larger P. grandis F clade(GFB) is sister to the monophyletic group containing all otherPyganodon F sequences. However, in the M tree (Figs. 2 and 8), P.grandis is monophyletic but the lineage is comprised of threeclades (hereafter GMA, GMB and GMC). Divergence among thosethree clades ranges from 10.13% to 12.51% (Table 3). The P. grandisM clade is sister to the monophyletic group containing all otherPyganodon M sequences.

The two other morphospecies, P. cataracta and P. fragilis, arereciprocally monophyletic and the P. cataracta + P. fragilis clade issister to P. lacustris in both the F and M trees (Fig. 2). However,there is relatively little genetic divergence between the P. cataractaand P. fragilis F and M sequences (Tables 2 and 3) such that nodalsupport values are robust for species’ monophyly only in the Mtree.

3.2. Sequence-based species delimitation

Analysis of branch length dynamics were also applied to betterdelimit putative species. In lineage-through-time plots, the shift inbranching rates is clear in both M and F analyses. Furthermore, for

Fig. 2. Bayesian tree of the F and M Pyganodon complex inferred from cox1 and cox2. ML bootstrap values (below) and BI posterior probabilities (x100, above) are indicated atthe branches. Boxes indicate to which species the specimens were identified morphologically prior to the genetic analysis.

Table 2Morphospecies and divergence in percentage (%) between F.

P. cataracta P. fragilis P. lacustris LFA P. lacustris LFB P. grandis GFA

P. fragilis 0.43P. lacustris LFA 4.46 4.10P. lacustris LFB 5.05 4.66 1.64P. grandis GFA 8.38 7.99 10.28 11.01P. grandis GFB 9.58 9.43 9.16 10.69 9.60

Table 3Morphospecies divergence in percentage (%) between M.

P. cataracta P. fragilis P. lacustris P. grandis GMA P. grandis GMB

P. fragilis 1.81P. lacustris 5.20 4.71P. grandis GMA 11.25 10.30 11.58P. grandis GMB 10.93 9.99 11.41 12.51P. grandis GMC 12.54 12.63 12.82 10.13 10.52

434 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

Fig. 3. Bayesian majority-rule consensus tree of the F Pyganodon complex. Terminal labels indicate collection site and abbreviation of US states or Canadian provinces.GenBank accession numbers come from Cyr et al. (2007). The first column to the right of the tree indicates species as identified by morphology. The second column refers tothe name of clades (the first letter of the acronym represent the morphospecies, the second represent the genome (F vs. M)). The third column shows the boundariesidentified by the likelihood analysis (GMYC method). The fourth column indicates membership in statistical parsimony networks. In the last column, two tones of grayrepresent a connected network divided in two distinct subnetworks.

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 435

both analyses, the GMYC model of lineage evolution was statisti-cally a better fit to the datasets than the null model of uniform coa-lescent branching. Likelihood values for the F analysis werelogL = �25.362 for the GMYC model and logL = �63.228 for thenull model; 2DlogL = 75.732, v2 test, d.f. = 3, p � 0.001. For theM analysis, likelihood values were logL = �7.259 for the GMYCmodel and logL = �55.281 for the null model; 2DlogL = 96.043,v2 test, d.f. = 3, p � 0.001. For the F sequences, the analysisrevealed a total of six species (interval for the number of clusters6–8) and for the M sequences, a total of six species with a confi-dence interval of six.

The clusters identified by these shifts closely match with thoseidentified by our second algorithmic sequence-based method forspecies delineation, namely the network analysis (see the corre-spondence between columns 3 and 4 in Figs. 3–8). Statistical parsi-mony network construction within the 95% connection limitconducted separately for F and M sequences resulted in five andfour unconnected networks respectively (Figs. 3–8). Both F and Mhaplotype networks regroup P. cataracta and P. fragilis morphospe-cies into one independent network formed by two connected sub-networks. Similarly, while the GMYC method based on Msequences resolve P. grandis morphospecies into three putative spe-cies, the parsimony network analysis revealed two unconnected

networks but one of them is actually constituted of two subnet-works that correspond to GMA and GMB clades on the tree.

3.3. Biogeography

P. cataracta (Fig. 9) (clades CF and CM) and P. fragilis (Fig. 10)(clades FF and FM) occupy the eastern part of the area of studywhile the two other species are sympatric, mostly around theGreat Lakes (Figs. 11 and 12). There are two localities (GooseBay, Labrador and Pickering Creek, Pennsylvania) where some fe-male individuals clustered in P. fragilis and some others clusteredin P. cataracta putative species based on mtDNA sequences butwere indistinguishable according to their morphologies. P. lacus-tris and P. grandis were often found in the same locations in oursurvey. The range of P. grandis (clades GFB and GMC) extendsnorth to the James Bay region while we found P. cataracta (cladeCF) and P. fragilis (clades FF and FM) in the Labrador region. The Fgenome of the P. lacustris morphospecies displays two distinctmitochondrial lineages (clades LFA and LFB) and the geographicaldistributions of these two F clades show a clear structure wherethe Wisconsin, Michigan and Quebec P. lacustris populations(LFB) are differentiated from those in Wisconsin and Minnesota(LFA) while the homogeneous M genome is widespread over the

Fig. 4. Bayesian majority-rule consensus tree of the F Pyganodon complex. Terminal labels indicate collection site and abbreviation of US states or Canadian provinces.GenBank accession numbers come from Cyr et al. (2007). The first column to the right of the tree indicates species as identified by morphology. The second column refers tothe name of clades (the first letter of the acronym represent the morphospecies, the second represent the genome (F vs. M) and the third represents a subclade (A or B)). Thethird column shows the boundaries identified by the likelihood analysis (GMYC method). The fourth column indicates membership in statistical parsimony networks.

436 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

distribution range of the morphospecies (see Fig. 11). One P. gran-dis F genome clade, GFB, is widely distributed around the GreatLakes region and is also found in the James Bay region (Fig. 12).The second P. grandis F genome clade, GFA, was less commonand was detected in specimens from the Rice and Pelican River

drainages in Minnesota as well as in the southwestern part ofthe St-Lawrence drainage basin (Fig. 12). The three clades fromanalyses of P. grandis M genomes clearly depict a geographicstructure. The GMC clade is broadly distributed to the east of LakeMichigan while GMA and GMB are principally found to the west

Fig. 5. Bayesian majority-rule consensus tree of the F Pyganodon complex. Terminal labels indicate collection site and abbreviation of US states or Canadian provinces.GenBank accession numbers come from Cyr et al. (2007). The first column to the right of the tree indicates species as identified by morphology. The second column refers tothe name of clades (the first letter of the acronym represents the morphospecies, the second represent the genome (F vs. M) and the third represent a subclade (A, B or C)). Thethird column shows the boundaries identified by the likelihood analysis (GMYC method). The fourth column indicates membership in statistical parsimony networks.

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 437

Fig. 6. Bayesian majority-rule consensus tree of the M Pyganodon complex. Terminal labels indicate collection site and abbreviation of US states or Canadian provinces.GenBank accession numbers come from Cyr et al. (2007). The first column to the right of the tree indicates species as identified by morphology. The second column refers tothe name of clades (the first letter of the acronym represents the morphospecies, the second represent the genome (F vs. M)). The third column shows the boundariesidentified by the likelihood analysis (GMYC method). The fourth column indicates membership in statistical parsimony networks. In the last column, two tones of grayrepresent a connected network divided in two distinct subnetworks.

438 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

of Lake Michigan (Fig. 12). The GMB clade is spatially clusteredprincipally in Wisconsin while GMA is found in Minnesota(Fig. 12).

3.4. Variation in F and M genomes within and among clades

For the F genomes, most of the variation was found within pop-ulations (54.33%; Table 4). Variation among populations withinclades accounted for 24.45% leaving 21.22% of the total variationamong clades. For the M genomes, the partitioning of variationwas different. As for the F genomes, most of the variation wasfound within populations (48.22%; Table 4) but there was little var-iation among populations within clades (6.92%). The variationamong clades comprised the remaining 44.85%. All values weresignificant (AMOVA p < 0.001; 10,000 permutations).

P. cataracta (CF) and P. fragilis (FF) diverged from one anotherby only 0.43% (uncorrected p-distance) according to the F data(Table 2). The M genome was, however, nearly four times more

divergent between these two morphospecies/clades than their Fcounterpart (Table 3). Genetic distance between the two F cladesof P. lacustris (LFA � LFB) was 1.64% and both diverged from P. cat-aracta and P. fragilis by �4.50% and from P. grandis clades (GFA andGFB) by �9–11%. Values for these comparisons were similar butslightly higher for the M genomes, i.e., �4.9% and �11–12% respec-tively. The genetic distances between P. grandis clades were thehighest within a single morphospecies averaging 9.60% and12.51% for F and M genomes, respectively. The patristic distancesbetween the P. grandis clades and all other clades were the largestin our study.

4. Discussion

We studied maternally and paternally inherited mtDNAsequence variation in the Pyganodon complex from northeasternNorth America. While we identified four species based onmorphology, namely Pyganodon grandis, P. lacustris, P. fragilis and

Fig. 7. Bayesian majority-rule consensus tree of the M Pyganodon complex. Terminal labels indicate collection site and abbreviation of US states or Canadian provinces.GenBank accession numbers come from Cyr et al. (2007). The first column to the right of the tree indicates species as identified by morphology. The second column refers tothe name of clades (the first letter of the acronym represents the morphospecies, the second represent the genome (F vs. M)). The third column shows the boundariesidentified by the likelihood analysis (GMYC method). The fourth column indicates membership in statistical parsimony networks.

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 439

Fig. 8. Bayesian majority-rule consensus tree of the M Pyganodon complex inferred from cox1 and cox2. ML bootstrap values (below) and BI posterior probabilities (x100,above) are indicated at the branches. Terminal labels indicate collection site and abbreviation of US states or Canadian provinces. The first column to the right of the treeindicates species as identified by morphology. The second column refers to the name of clades (the first letter of the acronym represents the morphospecies, the secondrepresent the genome (F vs. M) and the third (if necessary) represents a subclade (A, B or C)). The third column shows the boundaries identified by the likelihood analysis(GMYC method). The fourth column indicates membership in statistical parsimony networks. In the last column, two tones of gray represent a connected network divided intwo distinct subnetworks.

440 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

P. cataracta, the phylogenetic analysis of F and M mitochondrial se-quences revealed the presence of six divergent clades, which hasforced us to reconsider the actual number of species. However,our data do not lead to a simple classification in six species as thereis extensive disagreement between maternally and paternallyinherited genes with respect to inter- and intraspecific relation-ships in the P. lacustris and P. grandis group.

Sequence-based species delineation is particularly usefulfor taxa in which phenotypic plasticity makes morphological

identification difficult. DNA sequence-based taxonomy includesseveral quantitative methods such as population aggregation anal-ysis (PAA; Davis and Nixon, 1992), cladistic haplotype aggregation(Brower, 1999) and the Wiens and Penkrot methods (WP; Wiensand Penkrot, 2002). These procedures aggregate populations lack-ing discrete differences into a single species. In these methods, spe-cies are recognized based on fixed nucleotide differences unique tosets of populations, which are defined a priori based on morpho-logical, geographical or ecological information (Sites and Marshall,

-95o -90o -85o -80o -75o -70o -65o -60o -55o

36o38o40o42o44o46o48o50o52o54o

56o

419(C)

26

29

32

3536

4119119(

30

Fig. 9. Geographic distribution of Pyganodon cataracta morphospecies cladesinferred from F (left side of the symbol) and M genomes (right side of the symbol).Color codes and clade names follow Figs. 3 and 6. An empty half circle indicates thatno data exist for that location. Sites are numbered as in Table 1. Symbol (C) meansdata from Cyr et al. (2007). The purple zone indicates area of the 36 localities fromCyr et al. (2007). (For interpretation of the references to color in this figure legend,the reader is referred to the web version of this article.)

-95o -90o -85o -80o -75o -70o -65o -60o -55o

36o38o40o42o44o46o48o50o52o54o

56o

23

6

7

26

31

32

30

Fig. 10. Geographic distribution of Pyganodon fragilis morphospecies cladesinferred from F (left side of the symbol) and M genomes (right side of the symbol).An empty half circle indicates that no data exist for that location. Color codes andclade names follow Figs. 3 and 6. Sites are numbered as in Table 1. Symbol (C)means data from Cyr et al. (2007). (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

-95o -90o -85o -80o -75o -70o -65o -60o -55o

36o38o40o42o44o46o48o50o52o54o

56o

8910

1112

131415

20 212223

25

33

34

Fig. 11. Geographic distribution of Pyganodon lacustris morphospecies cladesinferred from F (left side of the symbol) and M genomes (right side of the symbol).An empty half circle indicates that no data exist for that location. Color codes andclade names follow Figs. 4 and 7. Sites are numbered as in Table 1. Symbol (C)means data from Cyr et al. (2007). (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

-95o -90o -85o -80o -75o -70o -65o -60o -55o36o38o40o42o44o46o48o50o52o54o

56o

1

5

10

1112

131415

1617

1819

20 212223

24

25

2728

34

37

38

39

Fig. 12. Geographic distribution of Pyganodon grandis morphospecies cladesinferred from F (left side of the symbol) and M genomes (right side of the symbol).An empty half circle indicates that no data exist for that location. Color codes andclade names follow Figs. 5 and 8. Sites are numbered as in Table 1. Symbol (C)means data from Cyr et al. (2007). (For interpretation of the references to color inthis figure legend, the reader is referred to the web version of this article.)

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 441

2004). Given the possibility of cryptic diversity (Cyr et al., 2007),morphological similarities and possibly sympatric clades, definingsets of populations a priori to fit into those procedures could intro-duce a strong bias in species delineation. We, therefore, rely onquantitative methods that do not require prior assumptions ofpopulation boundaries. GMYC and the Templeton parsimonymethods gave a total of six putative species from analyses of theF sequences and also a total of six putative species from analysesof the M sequences. However, the predicted number of distinctputative species within each of the recognized morphospecies isnot congruent between M and F genomes.

4.1. Recent speciation of P. cataracta and P. fragilis

The degree of reproductive isolation between P. fragilis and P.cataracta has been a subject of controversy (Clarke and Rick,1963; Kat, 1983, 1986; Hoeh, 1990). They are morphologically verysimilar and difficult to distinguish based on the shell characteris-tics (Cyr et al., 2007; Hanlon and Smith, 1999). Because beak sculp-ture is subject to erosion, the number of beak loops overlaps

(Clarke, 1981), periostracum color varies with habitat (Hanlonand Smith, 1999) and stomach anatomy does not reliably distin-guish them (Hanlon and Smith, 1999), identification of these spe-cies based on these conchological and anatomical criteriaremains questionable. However, both of our mitochondrial data-sets distinguish two putative species with relatively robust MLand BI nodal support values present in the F tree (93%/1.00 and57%/1.00 for the F and M tree, respectively). In addition, the GMYCspecies delimitation analysis clearly indicates a specific distinctionbetween those clades while the Templeton parsimony analysistends toward the same conclusion. Therefore, our data support adistinction between P. fragilis and P. cataracta, but these speciesappear to have diverged relatively recently from their commonancestor.

In some drainages (i.e., Goose Bay drainage [Labrador] and Pick-ering Creek [Pennsylvania]), we could not distinguish distinct spe-cies based on morphology. However, mtDNA from differentindividuals of a population clustered in both the P. cataracta andP. fragilis mt clades and multiple Pickering Creek P. cataracta indi-viduals’ M sequences were nearly identical to those from some P.fragilis individuals from Newfoundland. This puzzling distribution

Table 4Hierarchical AMOVA for the M and F data sets from Pyganodon morphospecies/clades.

Male Female

Source of variation df Var % df Var %Among clades 5 119.40 44.85* 5 26.10 21.22Among population within clades 37 18.43 6.92* 47 30.07 24.45Within populations 136 128.38 48.22* 175 66.82 54.33

* p < 0.001.

442 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

of haploytpes could be due to either (1) incomplete lineage sortingor (2) interspecific hybridization. The apparent recency of diver-gence between these two species makes either hypothesis plausi-ble. However, the relatively large geographic distance betweenthese populations and the fact that they are relatively distant froma potential hybrid zone for these two species, best support the for-mer hypothesis.

As reported in Cyr et al. (2007), the M sequence divergence be-tween P. cataracta and P. fragilis was considerably greater than thatobserved in their F sequences. In fact, the M sequences were moredivergent than the F sequences in all interspecific comparisons ofPyganodon. However, the P. cataracta–P. fragilis M divergence/Fdivergence ratio is, by far, the largest in this study. Cyr et al.(2007) suggest that the divergence in the female genomes has beenerased following a relatively ancient hybridization event. An alter-native explanation for this pattern could be ongoing speciationwhere both species were isolated from each other in different refu-gia during the last glaciation period. Possibly, the M mitochondrialgenome evolved relatively faster than the F genome during theearly stages of speciation. Consistent with this hypothesis, the Mgenome could, at this time, be under a more relaxed selective con-straint (Stewart et al., 1996; Zouros, 2000) which would allowmore mutations to accumulate.

4.2. Non-monophyly/cryptic speciation in P. grandis

Cyr et al. (2007) identified predominantly only one lineage for P.grandis. We documented two relatively distinct (�10% uncorrectedsequence divergence) F lineages in the P. grandis morphospeciesrendering this species paraphyletic and basal in the genus Pygan-odon. One P. grandis F genome clade (GFA) was less common andwas detected in specimens from two river drainages in northernMinnesota and in the southwestern part of the St. Lawrence Riverdrainage basin in Quebec. Therefore, the geographic distribution ofthe GFA clade appears discontinuous. This could be accurate or,alternatively, GFA might be more widely distributed to the northof the Great Lakes across the Ontario region (Canada), an areanot covered by our sampling. If this latter hypothesis is confirmed,the distribution of GFA could correspond to that of P. simpsoniana, aspecies that has been categorized as a subspecies of P. grandis(Clarke, 1981). Despite its description in the literature, our datadoes not allow us to confirm the existence of this species. Haplo-types from one of the unidentified Pyganodon F lineages of Cyret al. (2007) (therein called species lineage B) reside in our GFAclade. Our second P. grandis F clade, GFB, was the most widely dis-tributed around the Great Lakes region and was also found in theJames Bay region. This situation, with two relatively divergent mtlineages within one species, is highly unusual as intraspecificdivergence in animal mtDNA is usually 10 times lower than inter-species divergence (Hebert et al., 2004). The level of F mtDNAsequence divergence within the morphospecies P. grandis is ofthe same magnitude as that observed between any F clades of P.grandis and P. lacustris and higher than that of all other interspecificF lineage comparisons in Pyganodon.

In the P. grandis M genomes, there are not two but three highlydivergent clades (GMA, GMB and GMC) or three clades with two

being relatively similar (GMA and GMB) but, together, divergentfrom the third clade (GMC). The M genomes possess slightly great-er levels of intraspecific variation than do the F genomes of thismorphospecies. Unlike the situation in the F tree, the three P. gran-dis M clades form a monophyletic group. A greater level of intra-specific variation in the M compared to the F genomes is inaccordance with what has been previously demonstrated in otherDUI systems (e.g. Mytilus: Stewart et al., 1995; Sutherland et al.,1998, Tapes [Venerupis]: Passamonti and Scali, 2001).

This presence of divergent F and M genomes in P. grandis is con-sistent with the hypothesis that interspecific hybridization has oc-curred and that mt genes from distinct species have introgressedinto P. grandis. When regarding the geographical pattern of distri-bution, it appears that both F genomes could be geographicallypaired with different M genomes and vice versa. For example,GFA occurs with GMC in Quebec, but also occurs with GMA in Min-nesota. These facts point more toward a complex of species thatshare both M and F genomes. It is surprising that such sharing couldoccur with a high level of divergence between GFA and GFB groupsin F genomes and GMA, GMB and GMC groups in M genomes. Yetanother possibility is that P. grandis originated via the fusion of pre-viously divergent gene pools (for both mtDNA and nuDNA). The Cyret al. (2007) nuclear sequence data (from ITS) indicate that their‘‘species lineage B’’ and ‘‘P. grandis’’ F genomes are found associatedwith a singular nuclear background. This finding is consistent withthe hypotheses that (1) P. grandis is a single species and (2) mtDNAfrom other species have introgressed into P. grandis. The fact thatthe branching patterns for the P. grandis F and M genomes are sig-nificantly different support this hypothesis as the five mtDNA lin-eages we see in P. grandis could have been the result of acomingling of gene pools from three species, e.g., the ‘‘original’’ P.grandis which contributed an F and an M genome, a P. lacustris-likespecies to account for the P. grandis F lineage that is sister to P. lacus-tris + P. cataracta + P. fragilis lineages and a P. grandis-like species(possibly P. simpsoniana) to account for one of the three P. grandisM lineages that are sister lineages. Introgressive hybridizationcould have been facilitated by the dramatic ecological disturbancescaused by the filling and draining of the ancient proglacial LakeOjibway and Lake Agassiz, which occurred toward the close of theWisconsinan glacial episode (Barber et al., 1999; Fisher and Smith,1994; Veillette, 1994). The filling of these large basins would havefacilitated contiguity between previously isolated river drainagesthus allowing divergent Pyganodon populations to exchange genesand individuals to disperse widely across the considerable breadthof these ancient lakes. If the above scenario occurred, i.e., augmen-tation of gene pools via interspecific hybridization, it could explainthe protean nature of P. grandis, possibly the most conchologicallyvariable unionid species in North America.

4.3. Selective sweep in P. lacustris M genome

The P. lacustris F genome displays two geographically distinctclades while the homogeneous M genome has a widespread geo-graphic distribution. Even though we cannot rule out the possibil-ity that a divergent M genome was not sampled, it appears thatthe M and F genomes of P. lacustris have undertaken different

H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444 443

evolutionary trajectories where the F genomes could have beenisolated from each other long enough to diverge. A single M lineageappears to have recently invaded the M genome space in this spe-cies inside the territory under study. This could be due to a recentmutation which could have conferred an adaptive advantage tothis mitotype over the other and could have erased past divergenceduring its dispersal. This M mitotype could also have been bottle-necked during the last glacial period leaving the entire territoryfor it to spread without any need of advantageous mutation(Smietanka et al., 2009). In any case, our observations with P. lacus-tris demonstrate that two relatively divergent F mitochondrial gen-omes can function in concert with a single M genome.

5. Conclusion: evolution of M and F genomes in Pyganodon

Portions of mitochondrial genomes have long been used to esti-mate the evolutionary history of populations and species. All stud-ies to date using bivalve M mt genomes to assess populationstructure postulate that greater genetic diversity will lead to easierinference of fine-scale population differentiation (Smietanka et al.,2009). However, our study shows that this is not as straightfor-ward as one might have previously thought and highlights thatM mitochondrial genomes do not necessarily follow the same evo-lutionary paths as the conspecific F mitochondrial genomes. Whilethe M and F genomes revealed quite similar genetic patterns for P.fragilis and P. cataracta, the M genomes of P. lacustris, unlike their Fcounterpart, are almost homogeneous and P. grandis consists of anamalgam of divergent M and F lineages, which are not concor-dantly structured relative to geography.

Previous population genetic studies that used sequences fromboth M and F genomes have also yielded some paradoxical results.For example, in the marine mussel Musculista senhousia, Passamonti(2007) found an inverted variability pattern with the F genomebeing more variable than the M genome. Smietanka et al. (2009)also reported increased geographic differentiation in the F lineages,relative to the conspecific M lineages, of European Mytilus spp. Bystudying European mytilid mussel populations and by reinterpret-ing data from Riginos et al. (2004), Smietanka et al. (2009) came toan appealing explanation for the disparity among M and F genomes.They concluded that the overall pattern of polymorphism was con-sistent with periodic selective sweeps within the M lineage, leadingto fixation of different variants in different populations, which wasalso consistent with more relaxed selection pressures in the M line-age in comparison with the F lineage. Our results in Pyganodon areparticularly interesting in regard to this hypothesis because theyhighlight the possible/alternative scenarios that might occur undersuch a regime. Indeed, a selective sweep involving the M genomemight have occurred recently in P. lacustris which has led to the lossof mitochondrial variation within populations and erased any geo-graphical structure whereas the F genome data for this species re-veal differentiation into two geographically structured clades. Ifwe assume that selective sweeps happen periodically and repeat-edly in M lineages and more rarely in F lineages, the spatial distri-bution of F haplotypes is more likely to represent ancient‘‘history’’ whereas M haplotype distributions might reflect more re-cent population history (but without necessarily allowing infer-ences on fine-scale population structure if a sweep recentlyoccurred). Following a sweep, the relaxation of selective constraintson the M molecule would rapidly re-establish a high level of poly-morphism, which in return makes the sweep barely discernible.

Acknowledgments

We thank Dr. R.A. Krebs and an anonymous reviewer for sug-gestions that improved this article. We thank P. Boucher, A.K. Fritts,

J.A. Bury and P.L. Phelps for their help collecting specimens. Thiswork was supported by research grants from the National Sciencesand Engineering Research Council (NSERC, to PU Blier) and the Na-tional Science Foundation (NSF DEB-0237175, to WR Hoeh), andthe Minnesota Department of Natural Resources and US Fish andWildlife Service through Minnesota’s State Wildlife GrantsProgram (SWG to BE Sietman). H. Doucet-Beaupré was financiallysupported by a NSERC scholarship and a NSERC fellowship, respec-tively. EG Chapman is supported by the Kentucky AgriculturalExperiment Station State Project KY008043.

References

Barber, D.C., Dyke, A., Hillaire-Marcel, C., Jennings, A.E., Andrews, J.T., Kerwin, M.W.,Bilodeau, G., McNeely, R., Southon, J., Morehead, M.D., Gagnon, J.M., 1999.Forcing of the cold event of 8200 years ago by catastrophic drainage ofLaurentide lakes. Nature 400, 344–348.

Bauer, G., Wächtler, K., 2000. Ecology and Evolution of the Freshwater MusselsUnionoida. Springer-Verlag, Berlin.

Bilton, D.T., Freeland, J.R., Okamura, B., 2001. Dispersal in freshwater invertebrates.Annu. Rev. Ecol. Evol. Syst. 32, 159–181.

Bogan, A.E., 1993. Freshwater bivalve extinctions (Mollusca: Unionoida): a searchfor causes. Am. Zool. 33, 599–609.

Bogan, A.E., Hoeh, W.R., 2000. On becoming cemented: evolutionary relationshipsamong the genera in the freshwater bivalve family Etheriidae (Bivalvia:Unionoida). Geological Society, London, Special Publications 177, 159–168.

Breton, S., Beaupre, H.D., Stewart, D.T., Hoeh, W.R., Blier, P.U., 2007. The unusualsystem of doubly uniparental inheritance of mtDNA: isn’t one enough? TrendsGenet. 23, 465–474.

Breton, S., Stewart, D.T., Shepardson, S., Trdan, R.J., Bogan, A.E., Chapman, E.G.,Ruminas, A.J., Piontkivska, H., Hoeh, W.R., 2011. Novel protein genes in animalmtDNA: a new sex determination system in freshwater mussels (Bivalvia:Unionoida)? Mol. Biol. Evol. 28, 1645–1659.

Brower, A.V.Z., 1999. Delimitation of phylogenetic species with DNA sequences: acritique of Davis and Nixon’s population aggregation analysis. Syst. Biol. 48,199–213.

Burch, J.B., 1975. Freshwater Unionacean Clams (Mollusca: Pelecypoda) of NorthAmerica. Malacological Publications, Hamburg, Michigan.

Burky, A.J., 1983. Physiological ecology of freshwater bivalves. In: Russell-Hunter,W.D. (Ed.), The Mollusca. Academic Press, Orlando, Florida.

Cai, J.J., 2008. PGEToolbox: a matlab toolbox for population genetics and evolution.J. Hered. 99, 438–440.

Chakrabarti, R., Walker, J.M., Chapman, E.G., Shepardson, S.P., Trdan, R.J., Curole, J.P.,Watters, G.T., Stewart, D.T., Vijayaraghavan, S., Hoeh, W.R., 2007. Reproductivefunction for a C-terminus extended, male-transmitted cytochrome c oxidasesubunit II protein expressed in both spermatozoa and eggs. Febs Lett. 581,5213–5219.

Chapman, E.G., Piontkivska, H., Walker, J.M., Stewart, D.T., Curole, J.P., Hoeh, W.R.,2008. Extreme primary and secondary protein structure variability in thechimeric male-transmitted cytochrome c oxidase subunit II protein infreshwater mussels: evidence for an elevated amino acid substitution rate inthe face of domain-specific purifying selection. BMC Evol. Biol..

Clarke, A.H., 1981. The Freshwater Molluscs of Canada. National Museums ofCanada, Ottawa (Ontario).

Clarke, A.H. Jr., Rick, A.M., 1963. Supplementary Records of Unionidacea from NovaScotia with a discussion of the identity of Anodonta fragilis Lamarck.Contributions to Zoology National Museum of Canada Bulletin, Department ofNorthern Affairs and National Resources. Ottawa (Ontario), pp. 15-27.

Clement, M., Posada, D., Crandall, K.A., 2000. TCS: a computer program to estimategene genealogies. Mol. Ecol. 9, 1657–1659.

Cyr, F., Paquet, A., Martel, A.L., Angers, B., 2007. Cryptic lineages and hybridization infreshwater mussels of the genus Pyganodon (Unionidae) in northeastern NorthAmerica.. Can. J. Zool. – Rev. Can. Zool. 85, 1216–1227.

Davis, J.I., Nixon, K.C., 1992. Populations, genetic-variation, and the delimitation ofphylogenetic species. Syst. Biol. 41, 421–435.

Excoffier, L., Smouse, P.E., Quattro, J.M., 1992. Analysis of molecular varianceinferred from metric distances among DNA haplotypes – application to humanmitochondrial-DNA restriction data. Genetics 131, 479–491.

Excoffier, L., Laval, G., Schneider, S., 2005. Arlequin ver. 3.0: an integrated softwarepackage for population genetics data analysis. Evol. Bioinform. Online 1, 47–50.

Felsenstein, J., 1985. Confidence limits on phylogenies: An approach using thebootstrap. Evolution 39, 783–791.

Fisher, T.G., Smith, D.G., 1994. Glacial Lake Agassiz – its northwest maximum extentand outlet in Saskatchewan (Emerson Phase). Quat. Sci. Rev. 13, 845–858.

Folmer, O., Black, M., Hoeh, W., Lutz, R., Virjenhoek, R., 1994. DNA primers foramplification of mitochondrial cytochrome c oxidase subunit I from diversemetazoan invertebrates. Mol. Mar. Biol. Biotechnol. 3, 294–299.

Fontaneto, D., Herniou, E.A., Barraclough, T.G., Ricci, C., Melone, G., 2007. On thereality and recognisability of asexual organisms: morphological analysis of themasticatory apparatus of bdelloid rotifers. Zool. Scr. 36, 361–370.

Haag, W.R., Staton, J.L., 2003. Variation in fecundity and other reproductive traits infreshwater mussels. Freshwater Biol. 48, 2118–2130.

444 H. Doucet-Beaupré et al. / Molecular Phylogenetics and Evolution 63 (2012) 430–444

Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor andanalysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 41, 95–98.

Hanlon, S.D., Smith, D.G., 1999. An attempt to detect Pyganodon fragilis (Mollusca:Unionidae) in Maine. Northeast. Nat. 6, 119–132.

Hebert, P.D.N., Stoeckle, M.Y., Zemlak, T.S., Francis, C.M., 2004. Identification of birdsthrough DNA barcodes. PLoS Biol. 2, 1657–1663.

Hoeh, W.R., 1990. Phylogenetic relationships among eastern North AmericanAnodonta (Bivalvia: Unionidae). Malacol. Rev. 23, 63–82.

Hoeh, W.R., Stewart, D.T., Sutherland, B.W., Zouros, E., 1996. Cytochrome c oxidasesequence comparisons suggest an unusually high rate of mitochondrial DNAevolution in Mytilus (Mollusca: Bivalvia). Mol. Biol. Evol. 13, 418–421.

Hoeh, W.R., Frazer, K.S., Naranjo-Garcia, E., Black, M.B., Berg, D.J., Guttman, S.I.,1998. Correlation between mating system and distribution of genetic variationin Utterbackia (Bivalvia: Unionidae). J. Shellfish Res. 17, 1383–1393.

Hoeh, W.R., Bogan, A.E., Heard, W.H., 2001. A phylogenetic perspective on theevolution of morphological and reproductive characteristics in the Unionoida.In: Bauer, G., Wächlter, K. (Eds.), Ecology and Evolution of the FreshwaterMussels Unionoida. Springer-Verlag, Berlin., pp. 257–280.

Hoeh, W.R., Stewart, D.T., Guttman, S.I., 2002. High fidelity of mitochondrial genometransmission under the doubly uniparental mode of inheritance in freshwatermussels (Bivalvia: Unionoidea). Evolution 56, 2252–2261.

Huelsenbeck, J.P., Ronquist, F.R., 2001. MrBayes: Bayesian inference of phylogeny.Bioinformatics 17, 754–755.

Kat, P.W., 1983. Genetic and morphological divergence among nominal species ofNorth American Anodonta (Bivalvia, Unionidae). Malacologia 23, 361–374.

Kat, P.W., 1986. Hybridization in a unionid faunal suture zone. Malacologia 27, 107–125.

Krebs, R.A., 2004. Combining paternally and maternally inherited mitochondrialDNA for analysis of population structure in mussels. Mol. Ecol. 13, 1701–1705.

Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan, P.A., McWilliam,H., Valentin, F., Wallace, I.M., Wilm, A., Lopez, R., Thompson, J.D., Gibson, T.J.,Higgins, D.G., 2007. Clustal W and Clustal X version 2.0. Bioinformatics 23,2947–2948.

Lydeard, C., Cowie, R.H., Ponder, W.F., Bogan, A.E., Bouchet, P., Clark, S.A., Cummings,K.S., Frest, T.J., Gargominy, O., Herbert, D.G., Hershler, R., Perez, K.E., Roth, B.,Seddon, M., Strong, E.E., Thompson, F.G., 2004. The global decline of nonmarinemollusks. Bioscience 54, 321–330.

Marshall, D.C., Simon, C., Buckley, T.R., 2006. Accurate branch length estimation inpartitioned Bayesian analyses requires accommodation of among-partition ratevariation and attention to branch length priors. Syst. Biol. 55, 993–1003.

McMahon, R.F., 1983. Ecology of an invasive pest bivalve, Corbicula. In: Russel-Hunter, W.D. (Ed.), The Mollusca, vol. 6. Academic Press, New York, USA.

McMahon, R., 1991. Mollusca: Bivalvia. In: Covich, J.H.T.a.A.P. (Ed.), Ecology andClassification of North American Freshwater Invertebrates. Academic Press, SanDiego, USA.

Nagel, 2000. Testing hypotheses on the dispersal and evolutionary history offreshwater mussels (Mollusca: Bivalvia: Unionidae). J. Evol. Biol. 13, 854–865.

Nalepa, T.F., 1994. Decline of native unionid bivalves in Lake St Clair afterinfestation by the zebra mussel, Dreissena polymorpha. Can. J. Fish. Aquat. Sci.51, 2227–2233.

Obata, M., Kamiya, C., Kawamura, K., Komaru, A., 2006. Sperm mitochondrial DNAtransmission to both male and female offspring in the blue mussel Mytilusgalloprovincialis. Dev. Growth Diff. 48, 253–261.

Obata, M., Sano, N., Kawamura, K., Komaru, A., 2007. Inheritance of two M typemitochondrial DNA from sperm and unfertilized eggs to offspring in Mytilusgalloprovincialis. Dev. Growth Diff. 49, 335–344.

Passamonti, M., 2007. An unusual case of gender-associated mitochondrial DNAheteroplasmy: the mytilid Musculista senhousia (Mollusca Bivalvia). BMC Evol.Biol..

Passamonti, M., Ghiselli, F., 2009. Doubly uniparental inheritance. Twomitochondrial genomes, one precious model for organelle DNA inheritanceand evolution. DNA Cell Biol. 28, 79–89.

Passamonti, M., Scali, V., 2001. Gender-associated mitochondrial DNA heteroplasmyin the venerid clam Tapes philippinarum (Mollusca Bivalvia). Curr. Genet. 39,117–124.

Pons, J., Barraclough, T.G., Gomez-Zurita, J., Cardoso, A., Duran, D.P., Hazell, S.,Kamoun, S., Sumlin, W.D., Vogler, A.P., 2006. Sequence-based speciesdelimitation for the DNA taxonomy of undescribed insects. Syst. Biol. 55,595–609.

Ricciardi, A., Nevesm, R.J., Rasmussen, J.B., 1998. Impending extinctions of NorthAmerica freshwater mussels (Unionoida) following the zebra mussels(Dreissena polymorpha) invasion. J. Anim. Ecol. 67, 613–619.

Riginos, C., Hickerson, M.J., Henzler, C.M., Cunningham, C.W., 2004. Differentialpatterns of male and female mtDNA exchange across the Atlantic Ocean in theblue mussel, Mytilus edulis. Evolution 58, 2438–2451.

Rodriguez, F., Oliver, J.L., Marin, A., Medina, J.R., 1990. The general stochastic-modelof nucleotide substitution. J. Theor. Biol. 142, 485–501.

Ronquist, F., Huelsenbeck, J.P., 2003. MRBAYES 3: Bayesian phylogenetic inferenceunder mixed models. Bioinformatics 19, 1572–1574.

Sanderson, M.J., 2003. R8s: inferring absolute rates of molecular evolution anddivergence times in the absence of a molecular clock. Bioinformatics 19, 301–302.

Schloesser, D.W., Nalepa, T.F., 1994. Dramatic decline of unionid bivalves in offshorewaters of western Lake Erie after infestation by the zebra mussel, Dreissenapolymorpha. Can. J. Fish. Aquat. Sci. 51, 2234–2242.

Sites, J.W., Marshall, J.C., 2004. Operational criteria for delimiting species. Annu.Rev. Ecol. Evol. Syst. 35, 199–227.

Smietanka, B., Burzynski, A., Wenne, R., 2009. Molecular population genetics of maleand female mitochondrial genomes in European mussels Mytilus. Mar. Biol. 156,913–925.

Stewart, D.T., Saavedra, C., Stanwood, R.R., Ball, A.O., Zouros, E., 1995. Male andfemale mitochondrial-DNA Lineages in the blue mussel (Mytilus edulis) speciesgroup. Mol. Biol. Evol. 12, 735–747.

Stewart, D.T., Kenchington, E.R., Singh, R.K., Zouros, E., 1996. Degree of selectiveconstraint as an explanation of the different rates of evolution of gender-specific mitochondrial DNA lineages in the mussel Mytilus. Genetics 143, 1349–1357.

Strayer, D.L., 2006. Challenges for freshwater invertebrate conservation. J. N. Am.Benthol. Soc. 25, 271–287.

Subramanian, A., Kaufmann, M., Morgenstern, B., 2008. DIALIGN-TX: greedy andprogressive approaches for segment-based multiple sequence alignment.Algorithms Mol. Biol. 3, 6.

Sutherland, B., Stewart, D., Kenchington, E.R., Zouros, E., 1998. The fate of paternalmitochondrial DNA in developing female mussels, Mytilus edulis: implicationsfor the mechanism of doubly uniparental inheritance of mitochondrial DNA.Genetics 148, 341–347.

Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: molecular evolutionarygenetics analysis (MEGA) software version 4.0. Mol. Biol. Evol. 24, 1596–1599.

Templeton, A.R., Crandall, K.A., Sing, C.F., 1992. A cladistic analysis of phenotypicassociations with haplotypes inferred from restriction endonuclease mappingand DNA sequence data. III. Cladogram estimation. Genetics 132, 619–633.

Turgeon, D.D., Bogan, A.E., Coan, E.V., Emerson, W.K., Lyons, W.G., Pratt, W.L., Roper,C.F.E., Scheltema, A., Thompson, F.G., Williams, J.D., 1988. Common andscientific names of aquatic invertebrates from the United States and Canada:Mollusks. American Fisheries Society, Bethesda, Maryland.

Turgeon, D.D., Quinn, J.F., Bogan, A.E., Coan, E.V., Hochberg, F.G., Lyons, W.G.,Mikkelsen, P.M., Neves, R.J., Roper, C.F.E., Rosenberg, G., Roth, B., Scheltema, A.,Thompson, F.G., Vecchione, M., Williams, J.D., 1998. Common and scientificnames of aquatic invertebrates from the United States and Canada: Mollusks.American Fisheries Society, Bethesda, Maryland.

Veillette, J.J., 1994. Evolution and paleohydrology of glacial Lakes Barlow andOjibway. Quat. Sci. Rev. 13, 945–971.

Walker, J.M., Curole, J.P., Wade, D.E., Chapman, E.G., Bogan, A.E., Watters, G.T., Hoeh,W.R., 2006. Taxonomic distribution and phylogenetic utility of gender-associated mitochondrial genomes in the Unionoida (bivalvia). Malacologia48, 265–282.

Wiens, J.J., Penkrot, T.A., 2002. Delimiting species using DNA and morphologicalvariation and discordant species limits in spiny lizards (Sceloporus). Syst. Biol.51, 69–91.

Williams, J.D., Warren, M.L., Cummings, K.S., Harris, J.L., Neves, R.J., 1993.Conservation status of fresh-water mussels of the United-States and Canada.Fisheries 18, 6–22.

Zouros, E., 2000. The exceptional mitochondrial DNA system of the mussel familyMytilidae. Genes Genet. Syst. 75, 313–318.

Zwickl, D., 2006. Genetic Algorithm Approaches for the Phylogenetic Analysis ofLarge Biological Sequence Datasets Under the Maximum Likelihood Criterion.Ph.D. Thesis. The University of Texas at Austin.