Morphological and genetic variability in an alien invasive mussel across an

environmental gradient in South America

Esteban M. Paolucci,1,2,* Paula Sardina,2,3 Francisco Sylvester,4 Pablo V. Perepelizin,1,4 Aibin Zhan,1,5

Sara Ghabooli,1 Melania E. Cristescu,1,6 Marcia D. Oliveira,7 and Hugh J. MacIsaac 1

1 Great Lakes Institute for Environmental Research, University of Windsor, Windsor, Ontario, Canada2 Museo Argentino de Ciencias Naturales ‘‘Bernardino Rivadavia’’ and Consejo Nacional de Investigaciones Cientıficas y Tecnicas,

Buenos Aires, Argentina3 School of Biological Sciences, Monash University, Australia4 Universidad de Buenos Aires, Departamento de Ecologıa, Genetica y Evolucion, and Consejo Nacional de Investigaciones Cientıficas y

Tecnicas, Buenos Aires, Argentina5 Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Haidian District, Beijing, China6 Biology Department, McGill University, Stewart Biology Building, Montreal, Quebec, Canada7 Embrapa Pantanal, Corumba, Mato Grosso do Sul, Brazil

Abstract

Adaptation is an essential step in the establishment and spread of alien species in new environments, withphenotypic plasticity or genetic variability often contributing to this success. The golden mussel Limnopernafortunei is a biofouling mollusc native to Southeast Asia that was introduced to South America near the Rıo de laPlata estuary, Argentina, though the species has subsequently spread more than 2000 km upstream. We analyzedmorphological and genetic variation in 24 introduced populations of L. fortunei across its South American range.Relative gill area and shell morphology differed significantly, even among geographically proximate populations.Differences in relative gill area were especially marked across the species’ range and were negatively correlatedwith total suspended solids. Whereas mean gill cilia length, filament width, and interfilamental ciliary junctiondistance did not differ significantly among populations, mean gill cilia density was significantly lower inpopulations from areas with high suspended solids. Conversely, morphological differences were not related to thenumber of haplotypes, haplotype diversity, or nucleotide diversity, based upon analyses of the mitochondrialcytochrome c oxidase subunit 1 gene. Our results indicate that introduced populations of golden mussels in SouthAmerica exhibit pronounced morphological variation in shell and gill metrics that appear to result fromdevelopmental plasticity in relation to total suspended sediments, as has been observed in other mussel species.These adaptations may have facilitated spread of this species to a wide range of habitats.

The establishment of a viable population is an essentialstep in successful biological invasions (Blackburn et al.2011), and adaptation to the new environment is a keyfactor contributing to establishment and spread of nonin-digenous species (Daehler 2003). Adaptation itself mayresult from phenomena such as developmental or pheno-typic plasticity, genetic variability, adaptive homeostasis,and maternal effects, among others (Ghalambor et al. 2007;Hulme 2008). In spatially or temporally variable environ-ments, rapid adaptation to changing conditions could bedecisive to the success of colonizing populations (Ghalam-bor et al. 2007). Rapid adaptation may be particularlyimportant in river floodplains that experience pronouncedspatial and temporal variation in flow rate and otherenvironmental conditions (Wells and Pigliucci 2000).

Phenotypic plasticity is one of the most importantmechanisms for rapid adaptation to changing environmen-tal conditions (Schlichting and Pigliucci 1998). Significantphenotypic plasticity has been reported in a number ofinvasive bivalves, including the Asian clam Corbiculafluminea, zebra mussel Dreissena polymorpha, and quaggamussel Dreissena rostriformis bugensis (Dreissena bugensis,Marsden et al. 1996; Sousa et al. 2007; Peyer et al. 2010). In

general, phenotypic plasticity in these species involveschanges in shell morphology, mainly in width : length andwidth : height ratios. Variation in gill structure in responseto changing environmental conditions also has beendescribed in bivalves, especially in lamellibranchs; thesemolluscs possess high filtering abilities and respond to foodtype and quantity, suspended sediment, and other factors(Payne et al. 1995; Lei et al. 1996). However, in some casesmorphological variation is related to genetic differencesamong populations rather than to phenotypic plasticity(Marsden et al. 1996).

The golden mussel Limnoperna fortunei was introducedto South America near the Rio de la Plata River,Argentina, around 1990 (Pastorino et al. 1993). The specieshas since colonized an array of natural and artificialenvironments (Darrigran 2002; Boltovskoy et al. 2006). L.fortunei has spread approximately 2000 km across almostthe entire Rio de la Plata basin, as well as to inland lakesand reservoirs, at rates as high as 250 km per year(Boltovskoy et al. 2006). The extensive spread across thebasin has been associated with downstream recruitment ofsettling, postmetamorphic larvae (Boltovskoy et al. 2006)and with upstream colonization by biofouling individualson local vessels (Karatayev et al. 2007). However, factorsunderlying this species’ remarkable adaptation to new* Corresponding author: estebanmpaolucci@gmail.com

Limnol. Oceanogr., 59(2), 2014, 400–412

E 2014, by the Association for the Sciences of Limnology and Oceanography, Inc.doi:10.4319/lo.2014.59.2.0400

400

environments have not been established. For example, arelatively high filtration activity was recorded for adults(Sylvester et al. 2005) and veligers (Gazulha 2010) inintroduced areas, but the morphological, developmental, orphysiological determinants associated with these observa-tions have not yet been explored.

Evidence exists for genetic variation in L. fortuneipopulations across its South American range. For example,L. fortunei populations exhibited strong genetic structurebased on nuclear microsatellite markers (Zhan et al. 2012),implying a possible genetic basis for its invasion success.On the other hand, morphological variation in thesepopulations has not been addressed. Better understandingof the roles of genetic and phenotypic variation in thesepopulations may help explain the spread of this highlysuccessful invader.

In this study, we assess the morphological variation in 24populations of L. fortunei throughout the entire invadedrange in South America and relate findings to both geneticstructure and environmental factors. We utilize conven-tional morphological measures, including shell morphologyand gill macro- and ultrastructure, to compare populationsliving under different environmental conditions across theSouth American range. Additionally, we relate morpho-logical variation to genetic diversity based upon mitochon-drial cytochrome c oxidase subunit 1 gene (COI) sequencesof individuals from different populations across the species’range.

Methods

Study area and sample collection and analysis—Samplingwas designed to cover most of the invaded range of L.fortunei in South America (Fig. 1; Table 1). Samples werecollected between 2007 and 2010 from the lower ParanaRiver (populations: Del Este Channel, EC; CarapachayRiver, CR; Tigre, TI; and San Fernando, SF) and the Rıode la Plata estuary (populations: Buenos Aires, BA;Quilmes, QU; Punta Lara, PL; Santa Lucıa River, SL;and Magdalena, MA), representing the species’ southern-most range in South America. The lower Parana River andthe Rıo de la Plata estuary are strongly influenced byanthropogenic activities, including sewage discharge andindustrial installations along the banks of the river and itstributaries. The invasion pathway was followed, samplingup to the northern limit of the species’ distribution, over2000 km upstream, in the upper Parana River (populations:Rio Baıa, RB; Itaipu Hydroelectric Power reservoir, IT;Yabebiry River, YR; Yacireta Dam, YD; Parana River,PR; Cayasta-1, CA-1; Cayasta-2, CA-2; and SetubalLagoon, SA) and upper Paraguay River (population:Corumba, CO) in Brazil. One population from the westernrange limit was also collected in the Rio Tercero reservoir(RT), Argentina. Samples were also collected along theUruguay River and Salto Grande Lake in Argentina, to thenorthern distributional limit (populations: Federacion, FE;Salto Grande Dam, SG; and Puerto Luis, PU). Finally, asample was collected from the Guaiba basin on the coastalplain of Rio Grande do Sul, Brazil (population SaoGoncalo Channel, SO). In total, 24 populations distributed

along the Rıo de la Plata and Guaiba basins were sampledin four countries: Argentina, Brazil, Paraguay, and Uruguay(Fig. 1; Table 1). All mussels were preserved in 95% ethanoland were later stored at 4uC prior to laboratory analyses.

Macrostructure analysis—A subsample of 15 musselswas randomly selected and analyzed from each populationfor morphometric analysis. Shell length, shell width, andshell height were measured to the nearest 0.1 mm usingcalipers (Electronic Digital) to estimate maximum an-terior–posterior, dorsal–ventral, and lateral dimensions,respectively.

The right valve of each mussel was removed, and shelland lamella areas (Fig. 2) were delineated and calculatedusing a dissecting microscope (Leica S8APO) equippedwith a digitizer and image processing software ‘‘QCapture’’version 6.0 and AutoCAD 2000. Shell area and lamellaarea for the inner ascending demibranchs were measuredto the nearest 0.1 mm2 using a digital image at 163magnification (Fig. 2). The average number of filamentsper millimeter (filament density, FD) was calculated in eachmussel (15 per population) from three randomly chosen gillsections from a 1283 magnification image of the gill(Fig. 2). Shell-free dry weight (dry wt) for each mussel wasmeasured after individuals were dried to constant weight at60uC. Population FE (Salto Grande Lake, Argentina) wasexcluded from macrostructure and ultrastructure analysis(see below), although not from genetic analyses, owing topoor preservation of gill structures.

To avoid size-based differences among mussels (Rey-ment et al. 1984), similarly sized mussels were used andshell distances and gill : shell areas data were standardizedusing the following ratios: lamella area : shell area (relativegill area, RLS), shell width : shell length (RWL), shellwidth : shell height (RWH), and shell height : shell length(RHL). Ratio data was, in turn, log-transformed prior tostatistical analysis. Analysis of variance (ANOVA) andTukey’s multiple comparison tests were applied to testdifferences in log-transformed ratios (dependent variables)among populations from different areas (independentvariables). Additionally, these ratios and the remainingmorphological variables, such as FD and shell-free dry wt,were used for multivariate analysis (see below).

Ultrastructure analysis—Individual mussels were exam-ined using scanning electron microscopy (SEM) to evaluatewhether differences in gill area (macrostructure patterns)were related to the gill ultrastructure. Previous work withD. polymorpha demonstrated that the contraction ofinterfilamental ciliary junctions modified the interfilamen-tal distance and, in consequence, the total gill area (Medlerand Silverman 1997; Medler et al. 1999). We measured theultrastructure variables filament width, interfilamentalciliary junction distance, and cilia length and density insections of tissue dissected from the tips of the gill from thesame mussels (15 per population) used for morphologicalanalysis (Fig. 3). Gill tissue was rehydrated from 70%ethanol to tap water and was mounted on small steelplatforms with a drop of water. Both surfaces of the innerascending lamellae were examined with a Hitachi S2500

Alien invasive mussel diversity 401

scanning electron microscope operated at 10 kV and underlow vacuum conditions (130 Pa). We excluded mussels forwhich we were not able to measure at least one of theaforementioned variables, and repeated measures on singleindividuals were averaged to avoid pseudoreplication.Total gill area was estimated as lamella area 3 8. Totalshell length, gill area, and ultrastructure variables from thepopulation group with the significantly highest RLS basedon ANOVA were compared to the other populations usinga t-test.

Genetic analysis—Genomic deoxyribonucleic acid (DNA)was extracted from the posterior adductor muscle of anotherrandomly selected subsample of more than 21 individualsfrom each population, following Elphinstone et al. (2003).Polymerase chain reaction (PCR) amplification, PCRproduct cleaning, and sequencing of the mitochondrialCOI gene were performed according to Zhan et al. (2012).Sequences were checked and then aligned using CodonCodeAligner software version 2.0.6 (CodonCode Corp.). Subse-quently, alignment was visually inspected and manually

Fig. 1. Sampling sites for the invasive golden mussel L. fortunei in South America. Site names as per Table 1.

402 Paolucci et al.

Table1.

En

vir

on

men

tal

cha

ract

eris

tics

(an

nu

al

mea

n)

of

sam

pli

ng

site

sfo

rL

imn

op

ern

afo

rtu

nei

.F

ull

sou

rce

refe

ren

ces

are

giv

enin

Web

Ap

pen

dix

,T

ab

leA

2.

WT

,w

ate

rte

mp

era

ture

;C

hl

a,

chlo

rop

hy

lla;

DO

,d

isso

lved

ox

yg

en;

TS

S,

tota

lsu

spen

ded

soli

ds.

Sa

mp

lesi

teR

egio

no

rst

ate

,co

un

try

La

titu

de

(u)

Lo

ng

itu

de

(u)

Co

llec

tio

ny

ear

WT

(uC

)p

HC

alc

ium

(mg

L2

1)

Ch

la

(mg

L2

1)

DO

(mg

L2

1)

TS

S(m

gL

21)

So

urc

e

CO

Co

rum

ba

,B

razi

l2

18

.99

72

57

.65

42

00

72

8.1

6.4

3.5

0.7

84

.13

9.8

Oli

vei

raet

al.

20

11

;O

liv

eira

an

dC

alh

eiro

s2

00

0R

BB

aıa

Riv

er,

up

per

Pa

ran

aR

iver

,B

razi

l2

22

.68

62

53

.25

22

00

72

5.4

6.9

4.0

3.3

88

.11

.9V

elh

oet

al.

20

01

;R

od

rig

ues

eta

l.2

00

9IT

Ita

ipu

Hy

dro

elec

tric

Po

wer

rese

rvo

ir,

Bra

zil

22

5.4

07

25

4.5

89

20

09

24

.37

.6—

3.8

08

.26

.9F

ilh

o2

00

6

YR

Ya

beb

iry

Riv

er,

Mis

ion

es,

Arg

enti

na

22

7.2

96

25

5.5

43

20

08

24

.47

.24

.10

.24

8.3

3.5

Nei

ffet

al.

20

00

YD

Ya

cire

taD

am

,B

razi

l,P

ara

gu

ay

an

dA

rgen

tin

a2

27

.47

12

56

.70

42

00

82

4.1

7.4

4.3

0.7

08

.21

.2

PR

Pa

ran

aR

iver

,S

an

taT

ecla

,C

orr

ien

tes,

Arg

enti

na

22

7.6

05

25

6.3

84

20

08

24

.17

.44

.00

.70

8.2

1.2

CA

-1C

ay

ast

a,

Sa

nta

Fe,

Arg

enti

na

23

1.1

83

26

0.0

80

20

08

21

.07

.21

1.0

10

.00

7.8

—M

ain

eet

al.

20

04

CA

-2C

ay

ast

a,

Sa

nta

Fe,

Arg

enti

na

23

1.1

86

26

0.0

33

20

08

21

.07

.21

1.0

10

.00

7.8

—F

EF

eder

aci

on

,S

alt

oG

ran

de

La

ke,

Arg

enti

na

23

0.9

92

25

7.9

15

20

10

27

.07

.72

7.9

16

.67

8.1

34

.0D

eL

eon

an

dG

uil

lerm

o2

00

3

SG

Sa

lto

Gra

nd

eD

am

,U

rug

ua

y2

31

.24

92

57

.95

22

00

82

7.7

7.9

5.8

16

.35

8.6

17

.0P

UP

uer

toL

uis

,S

alt

oG

ran

de

La

ke,

Arg

enti

na

23

1.2

49

25

7.9

51

20

10

27

.77

.95

.81

6.3

58

.61

7.0

SA

Set

ub

al

La

go

on

,S

an

taF

e,A

rgen

tin

a2

31

.63

52

60

.68

12

00

72

0.2

7.2

27

.95

.40

7.7

—D

ever

cell

ia

nd

Per

uch

et2

00

8

SO

Sa

oG

on

calo

Ch

an

nel

,B

razi

l2

31

.81

12

52

.38

82

00

71

8.6

6.9

49

.05

.19

9.4

71

.5Y

un

eset

al.

19

96

;P

aiv

aet

al.

20

08

;D

aR

och

aet

al.

20

09

UR

Uru

gu

ay

Riv

er,

Co

lon

,A

rgen

tin

a2

32

.15

22

58

.18

92

01

01

9.9

7.2

5.8

2.5

07

.91

7.0

L.

J.Ja

nio

tu

np

ub

l.

RT

Rıo

Ter

cero

Da

m,

Co

rdo

ba

,A

rgen

tin

a2

32

.21

32

64

.47

32

00

81

9.4

8.2

14

.56

.35

9.0

4.2

Ma

cD

on

ag

het

al.

20

08

;B

olt

ov

sko

yet

al.

20

09

EC

Del

Est

eC

ha

nn

el,

Bu

eno

sA

ires

,A

rgen

tin

a2

34

.34

62

58

.51

92

01

02

0.9

7.1

10

.82

.78

—4

1.9

Go

mez

eta

l.1

99

8;

Ca

tald

oet

al.

20

01

CR

Ca

rap

ach

ay

Riv

er,

Bu

eno

sA

ires

,A

rgen

tin

a2

34

.39

82

58

.59

42

00

81

9.0

7.1

8.9

4.3

56

.66

9.5

Piz

arr

oa

nd

Ale

ma

nn

i2

00

5;

Piz

arr

oet

al.

20

07

TI

Lu

jan

Riv

er,

Tig

re,

Bu

eno

sA

ires

,A

rgen

tin

a2

34

.41

62

58

.57

82

01

01

9.0

7.1

8.9

4.3

56

.66

9.5

Ca

tald

oet

al.

20

01

SF

Lu

jan

Riv

er,

Sa

nF

ern

an

do

,A

rgen

tin

a2

34

.42

82

58

.55

22

00

81

9.0

7.1

8.9

4.3

56

.64

9.7

BA

Bu

eno

sA

ires

city

,A

rgen

tin

a2

34

.60

62

58

.34

52

00

82

3.1

7.3

23

.54

.09

5.7

70

.4N

ag

yet

al.

20

02

;C

all

iari

eta

l.2

00

5Q

UQ

uil

mes

,B

uen

os

Air

es,

Arg

enti

na

23

4.7

16

25

8.2

14

20

10

23

.18

.42

3.5

4.0

91

0.7

70

.4C

all

iari

eta

l.2

00

5

PL

Pu

nta

La

ra,

Bu

eno

sA

ires

,A

rgen

tin

a2

34

.78

22

58

.01

12

01

02

3.1

8.4

23

.54

.09

9.4

70

.4

SL

Sa

nta

Lu

cıa

Riv

er,

Ca

nel

on

es,

Uru

gu

ay

23

4.8

10

25

6.4

31

20

07

20

.87

.8—

6.1

18

.22

9.2

Na

gy

eta

l.2

00

2

MA

Ma

gd

ale

na

,B

uen

os

Air

es,

Arg

enti

na

23

5.0

13

25

7.5

36

20

10

23

.18

.22

3.5

4.7

08

.75

8.2

Ca

llia

riet

al.

20

05

Alien invasive mussel diversity 403

adjusted. Individual haplotypes were identified using DNASequence Polymorphism version 5 (Rozas et al. 2003).

Evolutionary relationships among haplotypes wereexamined using Bayesian inference (BI) and neighbor-joining (NJ) phylogeny reconstructions, with the brownmussel Perna perna used as an outgroup. All molecularevolution model parameters were estimated using MrMo-deltest version 3.7 (Posada and Crandall 1998) with theAkaike Information Criterion. Bayesian analysis was con-ducted with MrBayes version 3.2 (Ronquist and Huelsen-beck 2003). Trees were sampled every 100 generations fortwo million generations, and the first 25% of all the treessampled before convergence were discarded as burn-in. NJphylogenetic analyses were performed using MolecularEvolutionary Genetics Analysis version 4.0 (Tamura et al.2007), based on nucleotide distances corrected using theTamura–Nei model (Tamura and Nei 1993). Clade supportwas estimated using bootstrap analysis with 1000 replicates.Additionally, relationships among haplotypes were furtherexamined using a statistical parsimony haplotype networkgenerated at the 95% connection limit, using softwaredeveloped by Templeton, Crandall, and Sing (TCS) version1.21 (Clement et al. 2000).

Intrapopulation genetic diversity was measured by thenumber of haplotypes per population (Nh), haplotypediversity (h), and nucleotide diversity (p), using DNASequence Polymorphism version 5.00.07 (Rozas et al.2003). As with the analysis applied to ultrastructure varia-bles, t-tests were used to examine differences betweengroups with the most salient significant differences inmacrostructure characteristics with respect to Nh, h, and p.A nonparametric Mann–Whitney U-test (Statistica 6.0software) was used to determine differences in individualhaplotypes frequencies (which represent a frequency of agiven haplotype in a given population) among populationgroups with the most salient significant differences inmacrostructure characteristics.

Morphological and environmental variables—Mean an-nual values of key environmental parameters, includingwater temperature, dissolved oxygen concentration(mg L21), total suspended solids (TSS, mg L21), pH, andchlorophyll a (Chl a) concentration (mg L21), wereobtained from the published literature (Table 1). Ifenvironmental information was not available for a givensampling site, we looked for the closest available locationto our site. Canonical correlation analysis (CCA) was usedto evaluate the relationship between log-transformedmorphometric data (dependent variables: dry wt, FD,RWL, RWH, and RLS) and environmental parameters(independent variables). CCA allows the assessment ofthe relationship between two sets of variables and includesa chi-square test on successive canonical roots to assesshow many roots should be considered. We used Statistica6.0 to compute CCA.

Spearman rank correlation coefficients were calculatedto explore relationships between the first significantcanonical root (independent variable) and log-trans-formed morphological ratios (macrostructure) and envi-ronmental and genetic variables (dependent variables).Environmental and morphological variables with asignificant Spearman rank correlation were comparedusing a t-test. Comparison groups were defined usingpopulations separated by ANOVA.

Results

Macrostructure—We observed significant differencesamong populations with respect to a number of gilland shell morphology attributes, including RLS, RWL,and RWH. Because these two last variables exhibited lesspronounced variation among populations than RLS, weused groups defined by the latter to carry out most of theremaining comparisons. ANOVA analysis of the mostsalient morphological difference, RLS, found three popula-tion groups (Fig. 4). Fourteen populations formed thebiggest group and had a significantly higher RLS thanpopulations in the other two groups (F21, 323 5 31.3, p ,0.001; ANOVA and Tukey’s test; Fig. 4). Most of theabovementioned populations also had significantly lowerRWL ratios (F21, 323 5 28.1, p , 0.001; ANOVA and Tukey’stest) and RWH ratios (F21, 323 5 14.5, p , 0.001; ANOVAand Tukey’s test) as compared to other populations (Fig. 4).

Fig. 2. Anteroposterior and lateral view of the L. fortuneigills. Inner demibranch (ID), outer demibranch (OD), descendinglamella (DL), ascending lamella (AL). Detail of the gill sectionused to calculate the number of filaments per millimeter (lowerleft corner).

404 Paolucci et al.

For the other nine populations, which exhibited low RLS

ratios, two overlapped groups were observed. QU displayedthe lowest value for RLS, but it was not significantly differentfrom any remaining populations. In addition, QU was notsignificantly different from other populations with respect toRWL and RWH ratios; therefore, we defined two comparisongroups: populations with low RLS (QU, MA, RB, PL, BA,SO, CR, CO, and TI) and high RLS (UR, YR, SA, RT, EC,SL, CA1, CA2, SF, SG, YD, PR, IT, and PU).

Ultrastructure—Strong morphological differences ob-served among populations with respect to RLS allowed usto group the populations into two main groups and testultrastructure differences. Mean cilia length, filamentwidth, and interfilamental ciliary junction distance didnot differ significantly among populations (t-tests, p .0.05; Table 2). However, populations with significantlylower RLS (QU, MA, RB, PL, BA, SO, CR, CO, and TI)exhibited significantly lower mean cilia density and average

Fig. 3. Scanning electron micrographs of L. fortunei inner descending lamellae showing (A–C) outer surface with details of meancilia length (cl) and mean cilia density (cd) measures: (A) 10003, (B) 16003, (C) 10003; (D–F) inner surface with details ofinterfilamental ciliary junction distance (ifc) and filament width (fw): (D) 8003, (E) 12003, (F) 4003.

Alien invasive mussel diversity 405

total gill area as compared to the remaining populations (t-tests, p , 0.001 and p , 0.021, respectively; Table 2).

Genetic analysis—The 510 base pair COI alignmentcontained 24 polymorphic sites, only 10 of which wereparsimony informative. Across all samples analyzed, weidentified 18 distinct haplotypes (GenBank accessionnumbers: HQ843794–HQ843808). The Nh varied fromfour to eight, and h ranged from 0.387 to 0.726 (Table 3).The most abundant haplotype (Lfm03) was detected athigh frequency in all populations (60.5%). Haplotypeswith frequencies higher than 2% were also shared amongmost populations. Frequencies of all population-specifichaplotypes were extremely low (0.1–0.3%). The TCSnetwork was highlighted by a star-shaped pattern with adominant haplotype (Lfm03) at the center. The otherhaplotypes were connected to the dominant one (Lfm03)by several mutation steps (Fig. 5a). BI and NJ phyloge-netic analyses resulted in very shallow phylogenies(Fig. 5b). Genetic variables, including Nh, h, and p, didnot differ significantly between the two macrostructuremorphological groups defined by RLS (t-tests, t 5 20.261,degrees of freedom [df] 5 20, p 5 0.796; t 5 20.827, df 520, p 5 0.418; and t 5 0.366, df 5 20, p 5 0.718,

respectively). Additionally, these two groups did not differsignificantly in haplotype frequencies (Mann–Whitney U-test, p . 0.05; Web Appendix, Table A1, www.aslo.org/lo/toc/vol_59/issue_2/0400a.html).

Morphological and environmental variables—pH anddissolved oxygen concentration exhibited little variationthroughout the study area (see Table 1). In general, waterwas alkaline and well oxygenated. The lowest oxygen valueswere recorded in QU, which is close to a petrochemicalcomplex. Low oxygen values (less than 6.6 mg L21) werealso recorded on the upper Paraguay River (Corumba,Brazil), in the Lujan and Carapachay Rivers, and on the Riode la Plata River near Buenos Aires. Water temperaturefollowed a latitudinal gradient, with values between 19.0uCin the Rıo de la Plata River in the south and 28.1uC inCorumba, Brazil, in the north. TSS and Chl a concentrationexhibited the greatest variation among sampling sites. Chl avalues indicate oligotrophic conditions in the upper Para-guay and Parana Rivers and eutrophic conditions in SaltoGrande reservoir. Other sampling sites, such as the MiddleParana River, Parana delta, and Rio de la Plata estuary, hadintermediate Chl a levels. TSS ranged from very high values(e.g., 70 mg L21) in the lower Parana and Rio de la Plata

Fig. 4. Morphometric ratios of L. fortunei populations studied: RWL, RWH, RHL, and RLS. Populations were ordered to pool thosewith nonsignificant differences. The grid on top of the figures shows the statistical results (black cells indicate populations that were notsignificantly different based on Tukey’s multiple comparisons test, a 5 0.05). Black diamonds indicate total suspended solids (TSS) foreach population in mg L21. Bar colors represent groups of similar gill morphology (on all panels).

406 Paolucci et al.

Rivers, to intermediate values (40 mg L21) in the upperParaguay River, to very low values (1–2 mg L21) in theupper Parana River and Rio Tercero reservoir.

CCA performed using biological and environmentalparameters revealed clear factor structure and significantcorrelations with the first canonical root (chi-square test, 0.01). The first canonical root was correlated with RLS

and TSS concentration, followed by shell metrics (i.e., RWL

and RWH shell ratios). This component explained 24.7% oftotal variance (first canonical root; Table 4), with mainloadings strongly positive with respect to TSS, RWH, andRWL and strongly negative with respect to RLS (Table 4).

Spearman rank correlation analysis demonstrated thatthe scores of the first canonical root were positivelycorrelated with TSS, RWH, and RWL (Spearman correlation:r 5 0.694, p , 0.01; r 5 0.573, p , 0.01; and r 5 0.635, p ,0.01, respectively; Fig. 6), were strongly negatively correlat-ed with RLS (Spearman r 5 20.754, p , 0.01; Fig. 6), andwere not correlated with any measure of genetic diversity(i.e., Nh, h, or p). In addition to significantly lower RLS,populations QU, MA, RB, PL, BA, SO, CR, CO, and TIhad a significantly higher mean RWL, RWH, and TSS (t-tests,t 5 4.43, df 5 20, p , 0.001; t 5 4.68, df 5 20, p , 0.001; andt 5 2.70, df 5 18, p 5 0.015, respectively).

Discussion

This is the first study to explore morphometric variationin L. fortunei throughout the species’ introduced range inSouth America. Our analysis revealed clear differences withrespect to gill structure and shell morphometry amongpopulations, which corresponded to environmental factorsrather than genetic composition. We observed significantvariation among populations with respect to gill propor-tions and shell morphometry (including width : length andwidth : height ratios). Similar findings have been reportedin other bivalve species (Drent et al. 2004), including theinvasive bivalves C. fluminea, Crassostrea gigas, D. poly-morpha, and D. rostriformis bugensis (Payne et al. 1995;Dutertre et al. 2009; Peyer et al. 2010). Morphometricvariation thus appears to allow invasive molluscs tosuccessfully colonize, become established in, and spreadfrom novel environments, including those exhibitingsubstantial environmental gradients.

One of the most important differences that we observedamong populations was with respect to RLS, which wassignificantly lower in some populations, especially thoseexposed to a high concentration of TSS (i.e., QU, MA, RB,BA, PL, SO, TI, CO, and CR; Table 1; Fig. 4). Our resultssupport a general tendency established in previous studies:bivalve suspension feeders, such as D. polymorpha and C.gigas, have smaller gill surfaces in turbid habitats as anadaptive mechanism for optimizing food intake andavoiding gill damage associated with elevated mineralturbidity (Payne et al. 1995; Schneider et al. 1998; Dutertreet al. 2009). Although other factors, such as low dissolvedoxygen concentration and high contamination levels, mayalso play an important role (as can be observed on thesignificantly lower gill area of mussels found downstreamfrom a petrochemical complex in QU) our analyses suggest

Table

2.

Ult

rast

ruct

ure

an

dto

tal

susp

end

edso

lid

sv

alu

es(m

ean

6st

an

da

rdd

evia

tio

n[S

D])

for

two

gro

up

so

fL

imn

op

ern

afo

rtu

nei

po

pu

lati

on

sd

efin

eda

cco

rdin

gto

dif

fere

nce

sin

gil

lm

acr

ost

ruct

ure

(t-t

est

resu

lts)

.T

SS

,to

tal

susp

end

edso

lid

s;n

,to

tal

nu

mb

ero

fv

ali

do

bse

rva

tio

ns

ad

din

gth

etw

op

op

ula

tio

ng

rou

ps

(nu

mb

ero

fo

bse

rva

tio

ns

inea

chg

rou

p);

df,

deg

rees

of

free

do

m;

p,

p-v

alu

es;

an

dt,

tv

alu

efo

rt-

test

an

aly

sis.

Po

pu

lati

on

gro

up

sM

ean

cili

ale

ng

th6

SD

(mm

)M

ean

cili

ad

ensi

ty6

SD

(nu

mb

erp

erm

m)

Inte

rfil

am

enta

lci

lia

ryju

nct

ion

sd

ista

nce

6S

D(m

m)

Fil

am

ent

wid

th6

SD

(mm

)

Av

era

ge

gil

lsu

rfa

ce6

SD

(mm

2)

To

tal

shel

lle

ng

th6

SD

(mm

)T

SS

6S

D(m

gL

21)

QU

,M

A,

RB

,B

A,

PL

,S

O,

TI,

CO

,C

R1

1.8

61

.31

6.7

63

.27

.36

4.6

19

.36

51

31

.36

24

.81

5.0

61

.25

7.9

62

3.4

UR

,Y

R,

SA

,R

T,

EC

,S

L,

CA

1,

CA

2,

SF

,S

G,

YD

,P

R,

IT,

PU

11

.56

0.4

25

.06

2.6

9.0

61

.41

6.7

63

.71

89

.06

65

.91

5.6

62

.61

7.2

61

6.7

Sta

tist

ica

la

na

lysi

s

n6

8(2

9+3

9)

54

(35

+19

)3

3(8

+25

)5

7(1

9+3

8)

23

(9+1

4)

23

(9+1

4)

20

(9+1

1)

df

66

52

31

55

21

21

18

t2

0.8

62

4.5

70

.99

0.9

26

.24

0.3

22

.7p

0.3

95

,0

.00

10

.32

80

.36

40

.02

10

.57

60

.01

4

Alien invasive mussel diversity 407

that both total gill area and RLS of L. fortunei were stronglyinfluenced by the amount of TSS (Fig. 4).

Previous studies have shown physiologically based,temporal variation in gill structure in D. polymorpha as aresult of increasing concentrations of Ca+2 and K+. Thisresulted in muscle contraction, decreasing the interfilamentdistance and, consequently, the gill area (Medler andSilverman 1997; Medler et al. 1999). However, a lack ofsignificant difference in interfilament spaces in gills amongLimnoperna populations supports the idea that gill varia-tion is not simply an ephemeral reaction to the environmentat that moment, but rather appears to be a permanentfeature developed during growth in different environments.Previous studies demonstrated that the relationship be-tween pallial organs and turbidity is more complex andmay involve other environmental and morphologicalfactors, including availability and quality of food, ordimension of the labial palp (Payne et al. 1995; Drentet al. 2004; Dutertre et al. 2009). In general, a higherconcentration of food does not cause a change in gill area;rather it effects a relative increase in the ratio of labial palparea to gill area, which results in mussels having a higherfood-sorting capacity (Drent et al. 2004). It is also possiblethat the relative decrease in gill area was accompanied by arelative increase in palp area, as has been observed inpreliminary data from the Uruguay River (E. M. Paolucciunpubl.).

Mussels in our study could be separated into two shellshape groups, which, as with gill structure, were correlatedwith TSS concentration. The first grouping had higherwidth : length and width : height shell ratios, whereas the

second had a more elongate shell shape. Differences in shellmorphology as an adaptation to new environments havebeen recorded in a number of alien bivalve species, such asD. polymorpha, D. bugensis, and C. fluminea (Marsden etal. 1996; Sousa et al. 2007; Peyer et al. 2010). Shell shapes,particularly the width : length and width : height ratios, mayrepresent an adaptive response to different environmentalconditions, such as population density, predation pressure,depth, substrate type, and food availability, among others.Despite the apparent complexity of the interaction betweenshell shape and environment, a recent experimentalapproach observed significant developmental plasticity inshell morphology (i.e., width : length and width : heightratios) of D. rostriformis bugensis (D. bugensis) exposed todifferent temperatures, regardless of food quantity or levelof water motion (Peyer et al. 2010). The authors found thatdifferences in temperature affect growth rate and result inproduction of two different morphotypes—shallow anddeep quagga mussels—which differ with respect to bothwidth : length and width : height ratios. Although weobserved that TSS was better correlated than temperaturewith mussel morphology, the former factor may also affectmussel growth rate and change shell morphology inconsequence.

Although mussels displayed mixtures of gill and shellcharacteristics in some populations (MA, CO, and SO), ourresults revealed an association between shell shape andRLS, with more elongated organisms having higher RLSsand shorter ones lower RLSs (Fig. 2). This associationappears to be related to developmental plasticity (Drentet al. 2004; Peyer et al. 2010). Some environmental factors

Table 3. Collection details: haplotype number per population (Nh); haplotype diversity (h); and nucleotide diversity (p) forLimnoperna fortunei.

Samplesite Region or state, country Sample size Nh Haplotype code(s) h p

CO Corumba, Brazil 29 5 Lfm01–05 0.416 0.0021RB Rio Baıa, Alto Rio Parana, Brazil 27 5 Lfm01–05 0.724 0.0059IT Itaipu Hydroelectric Power reservoir, Brazil 32 6 Lfm01–06 0.625 0.0033YR Yabebiry River, Misiones, Argentina 27 5 Lfm01–05 0.704 0.0037YD Yacireta Dam, Brazil, Paraguay and Argentina 34 4 Lfm01, Lfm03–05 0.677 0.0029PR Parana River, Santa Tecla, Corrientes, Argentina 26 6 Lfm01–06 0.609 0.0033FE Federacion, Salto Grande Lake, Argentina 24 6 Lfm01, Lfm03–05, Lfm07–08 0.638 0.0031CA-1 Cayasta, Santa Fe, Argentina 27 6 Lfm01–06 0.726 0.0051CA-2 Cayasta, Santa Fe, Argentina 32 4 Lfm01, Lfm03–05 0.567 0.0022SG Salto Grande Dam, Uruguay 46 8 Lfm01, Lfm03–05, Lfm09–12 0.570 0.0031PU Puerto Luis, Salto Grande Lake, Argentina 24 6 Lfm01–05, Lfm11, Lfm18 0.630 0.0064SA Setubal Lagoon, Santa Fe, Argentina 30 5 Lfm01–05 0.618 0.0042SO Sao Goncalo Channel, Brazil 34 5 Lfm03–06, Lfm10 0.631 0.0034UR Uruguay River, Colon, Argentina 23 4 Lfm02–03, Lfm05, Lfm17 0.387 0.0025RT Rıo Tercero Dam, Cordoba, Argentina 59 6 Lfm01, Lfm03–06, Lfm13 0.546 0.0022EC Del Este Channel, Buenos Aires, Argentina 24 6 Lfm01–03, Lfm05, Lfm07, Lfm14 0.594 0.0041CR Carapachay River, Buenos Aires, Argentina 29 7 Lfm01–07 0.700 0.0035TI Lujan River, Tigre, Buenos Aires, Argentina 24 5 Lfm01–05 0.540 0.0068SF Lujan River, San Fernando, Argentina 30 5 Lfm01–05 0.575 0.0064BA Buenos Aires city, Argentina 52 6 Lfm01–06 0.483 0.0056QU Quilmes, Buenos Aires, Argentina 22 4 Lfm03, Lfm05–07 0.541 0.0028PL Punta Lara, Buenos Aires, Argentina 21 4 Lfm03–05, Lfm15 0.414 0.0027SL Santa Lucıa River, Canelones, Uruguay 26 5 Lfm01, Lfm03–05, Lfm10 0.634 0.0038MA Magdalena, Buenos Aires, Argentina 22 7 Lfm01–06, Lfm16 0.688 0.0036

408 Paolucci et al.

can have a strong influence on growth rates and,consequently, on shell shape. The same factors that affectthe development of gills in this study (e.g., TSS) may alsoaffect growth and, consequently, modify shell proportions.

Our results indicate that, although h was relatively high,genetic structure was very low and could not account formorphological variability among populations. Similarfindings have been reported for other bivalves (Sousa etal. 2007). A previous study observed substantial microsat-ellite variation among Limnoperna populations (Zhan et al.2012), yet even these patterns do not coincide with ourobserved patterns of morphological variation. Rather, our

combined results support the hypothesis that morpholog-ical differences result from phenotypic plasticity. However,this is the first study comparing morphological and geneticvariation in L. fortunei, and further studies may benecessary in order to assess the importance of the geneticcomponent on shell and gill morphology for this species.

Morphological variation in key characters, such as gillstructure, may play an important role in the successfulspread of L. fortunei and other invasive bivalves (Payne etal. 1995; Sousa et al. 2007; Dutertre et al. 2009). Althoughit is not clear whether developmental plasticity conferscompetitive advantage on these invasive molluscs relative

Fig. 5. (A) Parsimony haplotype network and (B) NJ phylogenetic tree based on mitochondrial COI haplotypes for L. fortunei inSouth America. Percentages indicate relative number of individuals carrying the haplotype. Dark circles indicate unsampled haplotypes.The green mussel Perna perna was used as an outgroup. Colors indicate sampled locations as per Table 1. Haplotype names as perTable 3.

Alien invasive mussel diversity 409

to native or even other introduced species (Peyer et al. 2010),a comparison of morphologic characteristics and phenotypicplasticity of different invasive bivalves can help illuminatefactors contributing to invasion success and the impactsof these species in invaded areas. L. fortunei in our studyaveraged 166.4 6 60.2 mm2 in RLS for animals that averaged15.4 6 2.1 mm shell length (Table 2). These values areslightly higher than those reported for bivalves such as D.polymorpha (139.0 6 76.0 mm2 for individuals 14.6 6 3.3 mmin length; Payne et al. 1995; Lei et al. 1996). This differencewas even higher when only populations with high lamella :shell area ratios were considered (average 189.0 6 65.9 mm2,shell length 15.6 6 2.6 mm). The higher filtering surface areaof L. fortunei as compared to D. polymorpha was describedearlier by Morton (1973) and may explain the higherfiltration rate recorded by the former (Sylvester et al.2005). Such plasticity would appear particularly advanta-geous in very dynamic environments, such as the Rio de laPlata basin, which, with its extensive wetland areas, canexperience dramatic changes in environmental conditions.Given the dramatic changes in ecosystem characteristicscaused by introduced D. polymorpha populations in invaded

Table 4. The canonical loadings of the dependent andindependent variables and eigenvalues on the first threecanonical roots. TSS, total suspended solids; RWH, width : heightshell ratio; RWL, width : length shell ratio; WT, water temperature;Chl a, chlorophyll a; FD, filament density; DO, dissolved oxygen;dry wt, dry weight; RLS, lamella area : shell area ratio.

Variable

Component

Root 1 Root 2 Root 3

TSS 0.769 0.331 20.094RWH 0.510 0.476 0.631RWL 0.347 0.358 0.859pH 0.137 0.413 20.215WT 0.121 20.887 20.023Chl a 0.116 0.270 20.573FD 20.067 20.469 20.222DO 20.080 0.398 20.659Dry wt 20.221 0.757 0.510RLS 20.831 20.249 0.041Eigenvalues 0.247 0.174 0.079

Fig. 6. Spearman correlation between the first canonical root and morphological, environmental, and genetic variables. Lamellaarea:shell area (RLS), shell width:shell length (RWL), shell width:shell height (RWH), total suspended solids (TSS), haplotype number (Nh),haplotype diversity (h), and nucleotide diversity (p).

410 Paolucci et al.

systems across North America and parts of Europe,colonization and spread of an even more potent ecologicalengineer—the golden mussel L. fortunei—must be prevented(Karatayev et al. 2007; Boltovskoy et al. 2009).

AcknowledgmentsWe thank Dr. H. M. Valett and two anonymous reviewers for

their helpful comments. We are grateful to L. Melo, Z. Zhou, M.Barnes, R. Capitoli, S. Chao, H. Chih-Wei, V. Leytes, D. Nakano,D. Nagashio, J. Seo, A. Takeda, L. Zhang, C. Canzi, C.Cantarini, D. Giberto, F. R. Molina, G. Vite, J. Langone, andM. Echem for providing samples. We thank S. Lackie forassistance with SEM. E.M.P. and P.V.P. gratefully acknowledgescholarship support from the Emerging Leaders in the AmericasProgram from the Government of Canada. This study wassupported by Natural Sciences and Engineering Research Council(NSERC, Canada) Discovery Grants to M.E.C. and H.J.M., andby an NSERC Discovery Accelerator Supplement to H.J.M.

References

BLACKBURN, T. M., AND OTHERS. 2011. A proposed unifiedframework for biological invasions. Trends Ecol. Evol. 26:333–339, doi:10.1016/j.tree.2011.03.023

BOLTOVSKOY, D., N. CORREA, D. CATALDO, AND F. SYLVESTER. 2006.Dispersion and ecological impact of the invasive freshwaterbivalve Limnoperna fortunei in the Rıo de la Plata watershed andbeyond. Biol. Invasions 8: 947–963, doi:10.1007/s10530-005-5107-z

———, A. KARATAYEV, L. BURLAKOVA, D. CATALDO, V. KAR-

ATAYEV, F. SYLVESTER, AND A. MARINELARENA. 2009. Signif-icant ecosystem-wide effects of the swiftly spreading invasivefreshwater bivalve Limnoperna fortunei. Hydrobiologia 636:271–284, doi:10.1007/s10750-009-9956-9

CLEMENT, M., D. POSADA, AND K. CRANDALL. 2000. TCS: acomputer program to estimate gene genealogies. Mol. Ecol. 9:1657–1659, doi:10.1046/j.1365-294x.2000.01020.x

DAEHLER, C. C. 2003. Performance comparisons of co-occurringnative and alien invasive plants: Implications for conservationand restoration. Annu. Rev. Ecol. Evol. Syst. 34: 183–211,doi:10.1146/annurev.ecolsys.34.011802.132403

DARRIGRAN, G. 2002. Potential impact of filter-feeding invaderson temperate inland freshwater environments. Biol. Invasions4: 145–156, doi:10.1023/A:1020521811416

DRENT, J., P. C. LUTTIKHUIZEN, AND T. PIERSMA. 2004. Morpho-logical dynamics in the foraging apparatus of a deposit feedingmarine bivalve: Phenotypic plasticity and heritable effects.Funct. Ecol. 18: 349–356, doi:10.1111/j.0269-8463.2004.00839.x

DUTERTRE, M., L. BARILLE, P. G. BENINGER, P. ROSA, AND Y.GRUET. 2009. Variations in the pallial organ sizes of theinvasive oyster, Crassostrea gigas, along an extreme turbiditygradient. Estuar. Coast. Shelf. Sci. 85: 431–436, doi:10.1016/j.ecss.2009.09.007

ELPHINSTONE, M., G. HINTEN, M. ANDERSON, AND C. NOCK. 2003.An inexpensive and high throughput procedure to extract andpurify total genomic DNA for population studies. Mol. Ecol.Notes 3: 317–320, doi:10.1046/j.1471-8286.2003.00397.x

GAZULHA, V. 2010. The golden mussel Limnoperna fortunei(Dunker, 1857) in presence of cyanobacteria: Rates offiltration, feeding behavior and survival. Ph.D. thesis. FederalUniversity of Rio Grande do Sul.

GHALAMBOR, C. K., J. K. MCKAY, S. P. CARROLL, AND D. N.REZNICK. 2007. Adaptive versus non-adaptive phenotypicplasticity and the potential for contemporary adaptation innew environments. Funct. Ecol. 21: 394–407, doi:10.1111/j.1365-2435.2007.01283.x

HULME, P. E. 2008. Phenotypic plasticity and plant invasions: Is itall Jack? Funct. Ecol. 22: 3–7, doi:10.1111/j.1365-2435.2007.01369.x

KARATAYEV, A. Y., D. BOLTOVSKOY, D. K. PADILLA, AND L. E.BURLAKOVA. 2007. The invasive bivalves Dreissena polymor-pha and Limnoperna fortunei: Parallels, contrasts, potentialspread and invasion impacts. J. Shellfish Res. 26: 205–213,doi:10.2983/0730-8000(2007)26[205:TIBDPA]2.0.CO;2

LEI, J., B. S. PAYNE, AND S. Y. WANG. 1996. Filtration dynamics ofthe zebra mussel, Dreissena polymorpha. Can. J. Fish. Aquat.Sci. 53: 29–37, doi:10.1139/f95-164

MARSDEN, E. J., A. P. SPIDLE, AND B. MAY. 1996. Review ofgenetic studies of Dreissena spp. Am. Zool. 36: 259–270.

MEDLER, S., AND H. SILVERMAN. 1997. Functional organization ofintrinsic gill muscles in zebra mussels, Dreissena polymorpha(Mollusca: Bivalvia), and response to transmitters in vitro.Invertebr. Biol. 116: 200–212, doi:10.2307/3226897

———, C. C. THOMPSON, T. H. DIETZ, AND H. SILVERMAN. 1999.Ionic effects on intrinsic gill muscles in the freshwater bivalve,Dreissena polymorpha. Comp. Biochem. Physiol. A 122:163–172, doi:10.1016/S1095-6433(98)10161-7

MORTON, B. 1973. Some aspects of the biology and functionalmorphology of the organs of feeding and digestion ofLimnoperna fortunei (Dunker) (Bivalvia: Mytilacea). Malaco-logia 12: 265–281.

PASTORINO, G., G. DARRIGRAN, S. M. MARTIN, AND L. LUNASCHI.1993. Limnoperna fortunei (Dunker, 1857) (Mytilidae), nuevobivalvo invasor en aguas del Rıo de la Plata. Neotropica 39:101–102. [Limnoperna fortunei (Dunker, 1857) (Mytilidae),new invasive bivalve in the Rio de la Plata.]

PAYNE, B. S., J. LEI, A. C. MILLER, AND E. D. HUBERTZ. 1995.Adaptative variation in palp and gill size in zebra mussels(Dreissena polymorpha) and Asian clam (Corbicula fluminea).Can. J. Fish. Aquat. Sci. 52: 1130–1134, doi:10.1139/f95-109

PEYER, S. M., J. C. HERMANSON, AND C. E. LEE. 2010.Developmental plasticity of shell morphology of quaggamussels from shallow and deep-water habitats of the GreatLakes. J. Exp. Biol. 213: 2602–2609, doi:10.1242/jeb.042549

POSADA, D., AND K. CRANDALL. 1998. MODELTEST: Testing themodel of DNA substitution. Bioinformatics 14: 817–818,doi:10.1093/bioinformatics/14.9.817

REYMENT, R. A., BLACKITH, R. E., AND N. A. CAMPBELL. 1984.Multivariate morphometrics, 2nd ed. Academic Press.

RONQUIST, F., AND J. HUELSENBECK. 2003. MrBayes 3: Bayesianphylogenetic inference under mixed models. Bioinformatics19: 1572–1574, doi:10.1093/bioinformatics/btg180

ROZAS, J., J. SANCHEZ-DELBARRIO, X. MESSEGUER, AND R. ROZAS.2003. DnaSP, DNA polymorphism analyses by the coalescentand other methods. Bioinformatics 19: 2496–2497, doi:10.1093/bioinformatics/btg359

SCHLICHTING, C., AND M. PIGLIUCCI. 1998. Phenotypicevolution: A reaction norm perspective. Sinauer.

SCHNEIDER, D. W., S. P. MADON, J. A. STOECKEL, AND R. E.SPARKS. 1998. Seston quality controls zebra mussel (Dreissenapolymorpha) energetics in turbid rivers. Oecologia 117:331–341, doi:10.1007/s004420050666

SOUSA, R., R. FREIRE, M. RUFINO, J. MENDEZ, M. GASPAR, C.ANTUNES, AND L. GUILHERMINO. 2007. Genetic and shellmorphological variability of the invasive bivalve Corbiculafluminea (Muller, 1774) in two Portuguese estuaries. Estuar.Coast. Shelf. Sci. 74: 166–174, doi:10.1016/j.ecss.2007.04.011

SYLVESTER, F., J. DORADO, D. BOLTOVSKOY, A. JUAREZ, AND D.CATALDO. 2005. Filtration rates of the invasive pest bivalveLimnoperna fortunei as a function of size and temperature.Hydrobiologia 534: 71–80, doi:10.1007/s10750-004-1322-3

Alien invasive mussel diversity 411

TAMURA, K., J. DUDLEY, M. NEI, AND S. KUMAR. 2007. MEGA4:Molecular evolutionary genetics analysis (MEGA) softwareversion 4.0. Mol. Biol. Evol. 24: 1596–1599, doi:10.1093/molbev/msm092

———, AND M. NEI. 1993. Estimation of the number ofnucleotide substitutions in the control region of mitochon-drial DNA in humans and chimpanzees. Mol. Biol. Evol. 10:512–526.

WELLS, C. L., AND M. PIGLIUCCI. 2000. Adaptive phenotypicplasticity: The case of heterophylly in aquatic plants. Perspect.Plant Ecol. Evol. Syst. 3: 1–18, doi:10.1078/1433-8319-00001

ZHAN, A., AND OTHERS. 2012. Scale-dependent post-establishmentspread and genetic diversity in an invading mollusc in SouthAmerica. Divers. Distrib. 18: 1042–1055, doi:10.1111/j.1472-4642.2012.00894.x

Associate editor: H. Maurice Valett

Received: 19 December 2012Amended: 31 October 2013

Accepted: 26 November 2013

412 Paolucci et al.