DISPERSAL PAïTERNS OF DREISSENA BUGENSIS IN THE
LAURENTIAN GREAT LAKES AS INFERRED FROM HIGHLY
POLYMORPHIC MICROSATELLITE MARKERS
A Thesis
Presented to
The Faculty of Graduate Studies
of
The University of Guelph
by
ANTHONY B. WILSON
In partial fulfillment of requirements
for the degree of
Master of Science
September, 1998
O Anthony B. Wilson, 1 998
National Library Bibliothèque nationale du Canada
Acquisitions and Acquisitions et BiMiographic Services services bibliographiques
395 Wellington Street 395, rue Wellington Ottawa ON K1A ON4 OttawaON K 1 A W Canada Canada
The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distriiute or seli copies of this thesis in microform, paper or electronic formats.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts f?om it may be printed or othenivise reproduced without the author's permission.
L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfichelnlm, de reproduction sur papier ou sur format électronique.
L'auteur conseNe la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.
Abstract
DISPERSAL PATTERNS OF DREISSEA BUGENSIS IN THE LAURENTIAN GREAT LAKES AS INFERRED FROM HIGHLY
POLYMORPHIC MICROSATELLITE MARKERS
Anthony B. Wilson University of Guelph. 1998
Advisor: Professor E. G. Boulding
The freshwater Dreissenidae are well suited for rapid dispersal.
retaining a primitive f o m of planktotrophic developrnent and a capacity to
produœ byssal threadç, both of which may have accelerated their spread
through North America. In contrast to the radiation of Dreissena polymorpha,
which has been characterized by several jump dispersal events, D. bugensis
has exhibited a more gradua1 diffusion from its point of introduction in Lake
Erie through the lower Laurentian Great Lakes. In this thesis, I develop six
highly polymorphic species-specific microsatellite markers and use these
markers to examine the presentday population genetic structure of D.
bugensis in North America. Data suggest that, in contrast to census data
suggesting diffusive dispersal of D. bugensis, high levels of gene fiow have
been maintained between geographically disjunct populations. presumably
due to boater-mediated transport of adult mussels. At the same time, more
proximate populations are significantly different, indicating non-random
settlement of veliger larvae. These results suggest that divergent dispersal
strategies have played an important role in the success of the quagga mussel
invasion.
I
Acknowledgements
I would especially like to thank Dr. Kerry Naish for her assistance
throughout the course of the project. In addition to a crucial role during the
development of the microsatellite markers, Kerry has been the source of a
great deal of productive consternation (you know what I mean), wine, and has
greatly broadened my views on science and the world in general. Thanks for
everything Keny !
I would also like to thank both rny thesis and defense cornmittees for
their insight and feedback which greatly improved on earlier drafk of my work
and helped to keep me focussed on the task at hand. Thanks to Liz, Paul,
Gerry, Terry, and Steve Scadding for their time and patience.
This work would not have been possible without a great deal of
assistance from dreissenid mussel researchers in North America and abroad.
I would especially like to thank Trevor Claxton, the Canadian Coast Guard,
Gary Sprules, Steve Smith and the crew of the research vesse1 Limnos, Bob
OIGorrnan, and Yves De Lafontaine and Brigitte Cusson at Environment
Canada for their help with sample collection and insight into patterns of
dreissenid mussel population structure in the Laurentian Great Lakes.
Finally. thanks to my family for trying to understand the pressures which
govern the life of a graduate student and for putting up with my continued
absence from the family homestead. Thanks for everything guys!
Table of Contents
Acknowledgements .......................... .. ........................................ i Table of Contents ....................................................................... ii List of Tables ...................................................................... iii List of Figures ........................................................................ iv
Chapter 1: General Introduction ................................................... General l ntroduction ................................................... Theoretical Basis ............................................................. Introduction & S p read of Dreissena bugensis
in North America .............................................................. Dispersal Mechanisrns ................................................... Previous Genetic Studies ...................................................
............................. Species-Level Investigations ........................................ Population-Level Study
.............................................................. Microsatellites ....................... Microsatellites: Potential Advantages
Microsatellite Mutation ........................................ ............... Microsatellite Analyses: Statistical Models
Thesis Outline ............................................................. Bibliography ........................................................................
Chapter 2: Development of novel tri- and tetra-nucleotide microsatellite loci in the invasive mollusc.
.................................................... Dreissena bugensis 22 Introduction ................... .. .................................................. 23 Materials and Methods ............................................. 24 Results and Discussion ...................................... ,. .......... 27 Conclusions ................... .... ........................................... 30 Bibliography ....................................................................... 30
Chapter 3: Dispersal patterns of Dreissena bugensis in the Laurentian Great Lakes as inferred from hig hly polymorp hic microsatellite markers ..............................
Abstract ......................................................................... Introduction ......................................................................... Materials and Methods ................................................... Results .........................................................................
......................................................................... Discussion .............................................. ..................... Conclusions ...
................................................ ..................... Bibliography .. Appendix 3.1 : Gel photographs ........................................ Appendix 3.2. Allelic frequency data .............................
iii
List of Tables
Table 2.2. Pairwise linearized FST and RST estimates for six study
populations - Impact of nuIl allele at Dbu@ ............................... 33
Table 3.1. Dreissena bugensis sampling site locations ............................... 65
Table 3.2. Fisher's exact test for genotypic differentiation - Pairwise
population cornparisons .............................................................. 66
Table 3.3. Pairwise linearized FST and Rm estimates for eight study
populations ............ .. ................................................................. 67
Table 3.4. Analysis of Molecular Variance (AMOVA) ................................... 68
iv
List of Figures
Fig. 1.1. Current North American distribution of dreissenid mussels .... 21
Fig. 2.1. Plot of average number of alleles present I population at
variable sample sizes for six Dreissena bugensis
microsatellite loci.. ...,.... .... ..... . ..... ... . . - - . - . . . . .. . . . . . 34
Fig. 3.1. Map of Dreissena bugensis sampling sites ............................ 69
Fig. 3.2. Average mussel length / population .................. .. ............... 70
Fig. 3.3. Allele frequency histograms for quagga mussel
microsatellite loci.. . . . . .. . . . . .. . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . ., . . .. . . . 71
Fig. 3.4. Hardy-Weinberg test for each locus in each population ... . .. .. . 72
Fig. 3.5. Test of isolation-by-distance mode1 ......................... - ........ ... . 73
CHAPTER ONE
General Introduction
"The Dreissena is perhaps better fitted for dissemination by man and subsequent establishment than any other fresh-water s hell; tenacity of life, unusually rapid propagation, the faculty of becoming attached by string byssus to extraneous substances,
and the power of adapting itself to strange and altogether artificial surroundings have combined to make it one of the
most successful molluscan colonists in the world."
- Kew (1 893)
General lntroduction
The rate of dispersal of living organisms has grown rapidly with increased
anthropogenic activity and has lead to the widespread dissemination of species
across broad geographic boundaries. As many as ten percent of introduced
species may have substantial economic and ecological impacts on native
resourœs (Mills et al. 1993b; Williamson and Fritter 1996). Consequentfy,
developing rnodels to predict the sucœss and impact of invasive species has an
important role in the protection of natural ecosystems (Pimrn 1989; Lodge 1993;
Williamson and Fritter 1996). In addition to an important predictive and
conservational role, modeling the success or failure of an introduced species can
help c lam the relative importance of life history adaptations that govem
invasions and the dispersal success of native species.
The quagga mussel. Dreissena bugensis. was introduced to North
Arnerica frorn the Ukraine in 1989 and rapidly spread throughout the lower
Laurentian Great Lakes (Mills et al. 1 993a). While the quagga mussel has
colonized the lower Laurentian Great Lakes and St Lawrence River, its present
day North Arnerican distribution is much more restricted #an that of an earlier
dreissenid invader, D. polymorpha (Fig. 1.1). 80th of these dreissenid mussel
species have several important life history characteristics hat rnay contribute to
their success as invaders. Passive dispersal via a planktotrophic larval stage
and potentially longdistanœ dispersal via byssal thread attachment to ships
rnay have accelerated the spread of dreissenid mussels in North Arnerican
waterways.
In this study, I develop and apply highly polymophic species-specific
microsatellite markers to an investigation of the presentday population genetic
structure of the quagga musse1 in North Amerim. These genetic data can be
used as an indirect estimator of mussel dispersal and may help illuminate
historical colonization and radiation events which have characterized the spread
of D. bugensis throug h the iaurentian Great Lakes. Genetic data may help
clam the relative significance that dispersal mechanisms play in the colonization
and spread of invasive aquatic species.
Theoretical basis
Modes of dispersal and gene fiow Vary substantially between terrestrial
and aquatic systems (Valentine and Jablonski 1983), which may lead to
significant differenœs in colonization patterns. High dispersal ability is a feature
cornmon to both manne organisrns (Palumbi 1995) and many freshwater
species (Banarescu 1990). Although the expectation is that high dispersal
potential leads to high levels of gene flow and low levels of population
structuring, study on manne species suggests rnany exceptions to this paradigm
(Palumbi 1996).
Theoretical ecologists have developed a number of models in an effort to
clam the relationship between dernographic parameters and the rate of range
expansion of introduced species (Hastings 1996). Dispersal over short
distances, or neighbourhood diffusion. resuits in the spread of individuals to
adjacent areas (Hengeveld 1989). While the magnitude of this radiation is
direcüy related to the eficiency of dispersal, neighbourhood diffusion generally
progresses along a closed front (Hengeveld 1989). In contrast, longdistance
jump dispersal may lead to the disjunct distribution of a species over its range.
High levels of jump dispersal may result in increased gene flow between disjunct
populations, effectively confounding geographic patterns of genetic stnicturing
(Rousset 1997). Between these two extrernes are a continuum of intemediate
hierarchiwl diffusion models. where both neighbourhood diffusion and jump
dispersal play a role in range expansion (Hengeveld 1989). While ernpirical
verfication of models of diffusion is incornplete. research suggests that rare,
longdistance dispersal events may play an important role in the spread of recent
invaders (Hastings 1996).
Introduction and Spread of D. bugensis in North America
Dreissenid mussels are native to the Dneiper river drainage system in the
Ukraine (Mills et al. 1996) and were first identifieci in Nom America in 1988, with
the discovery of a population of D. polymopha in Lake St. Clair (Hebert et al.
1989). Based on the size of these mussels, Hebert et al. (1 989) suggest that
0. polymopha has been present in Lake St. Clair since 1986. Following its
establishment the zebra mussel spread rapidly and by 1990, populations of
D. polymorpha were found in al1 five of the Laurentian Great Lake (see Mackie
and Schloesser 1996 for review). Presently. zebra mussel populations have
been identified outside the Laurentian Great Lakes in the Mississippi drainage
system, the St. Lawrence River and in several lakes which are not connecteci to
infected waterbodies (Fig 1 -1 ; Mackie and Schloesser 1996).
In 1989, the first specirnen of a morphologically distinct fom of dreissenid
was found in eastem Lake Erie (Mills et al. 1993a), putatively called the quagga
mussel (May and Marsden 1 992). Through an extensive survey program, Mills
et al. (1 993a) chronicled the spread of the quagga mussel frorn Lake Erie into
Lake Ontario and. by 1991. populations of D. bugensis were found in the St.
Lawrence River.
While survey data suggests that the quagga mussel was introduced to the
Laurentian Great Lakes only three years after the introduction of D. polymopha
(Mills et al. 1993a), its present day North American distribution is limited to the
lower Laurentian Great Lakes and St. Lawrence River (Fig. 1.1). While the
present-day distribution of the zebra mussel suggests that long distance jump
dispersal events have played a large role in its colonization and spread, sunrey
data chronicling the diffusive radiation of D. bugensis suggest veliger-madiatecl
dispersal from its founding population in Lake Erie. Molecular data will help to
clanfy the apparent variability in colonization and spread which exists between
these two dreissenid rnussel species.
Dispersal Mechanisrns
The radiation of both invasive dreissenid species has been accelerated
by a naturally high dispersal capability. While high dispersai ability is wmmon
among freshwater molluscan species. freshwater dispenal via a planktotrophic
larval stage is unique to the Dreissenidae (Ackerman et al. 1994). The
development of rnost other freshwater molluscan species involves some
combination of direct development (eg. Sphaeriidae) or obligate parasitism (eg.
Unionidae) (McMahon 1991 ). afthough juvenile Cohicula fluminea may also be
passively dispersed by water currents (McMahon 1991). Levin and Bridges
(1 995) suggest that planktonic larvae may be ineffective for dispersal in
freshwater due to unidirectional currents in streams and rivers, which may
explain the rarity of a veliger stage in freshwater molluscan development.
Nonetheless, the retention of a planktotrophic lawal stage in the Dreissenidae
greatly increases downstream dispersal potential. Dispersal via a passively
floating larval stage should best fit a model of neighbourhood diffusion. where
the genetic distance between individuals is proportional to the geographic
distance separating thern. Rousset (1997) has proposed a model of isolation-
bydistance which tests the correlation between hierarchical measures of
population subdivision and geographic distance. A system which fitç the
neighbourhood diffusion mode1 should exhibit a strong positive correlation
between estimates of population subdivision and geography .
Byssal adhesion is a common feature of larval and postrnetamorphic
bivalves which evolved to fundion during metamorphosis from larva to post-
veliger (Yonge 1962). Both species of Dreissena present in North Arnerica have
the ability to produœ byssal threads throughout their life cycles (Mills et al.
1996), which may have a signficant impact on molluscan dispersal patterns
(Johnson and Carlton 1 996). In contrast to dfisive dispersal via the
planktotrophic larval stage, human-mediated dispersal of mussels via byssal
thread attachment to boats may result in disjund dispersal over large distances.
Viable dreissenid mussels have also been found on boats transported overfand
between watersheds (Cariton 1993; Ricciardi et al. 1 995), suggesting that rapid
dispersal of dreissenid mussels may be enhanced by boater rnovement patterns
(Padilla et al. 1996). If jurnp dispersal mechanisrns have played a large role in
the spread of the quagga mussel, deviations from the isolation-bydistance
mode1 are expected.
Previous Genetic Studies
Genetic analyses have played a signifimnt role in our understanding of
the dreissenid mussel invasion of North Amencan wateways. m i l e the zebra
mussel had been predicted to invade North America based on its spread through
Europe (Marsden et al. 1 W6), the discovery of a second, rnorphologically distinct
form of dreissenid mussel has lead to conœms regarding the initial allozyme
surveys of O. polymorpha. which assumed that only a single species of
Dreissena was present in North Amerka.
Species-Level Investigations
May and Marsden (1 992) completed the first allozyme study on a single
population of the second f o n of Dreissena, calling this distinct morphotype the
quagga mussel. Using 12 allozyme loci, May and Marsden (1 992) provided
strong molecular evidence ta suggest that the quagga mussel is a separate
species from D. polymorpha.
While May and Marsden (1 992) provided evidenœ supporting a species-
level designation for the quagga mussel, a clear alignment of the North American
quagga mussel with its Ukrainian counterpart is important to establish the
environmental limits and potential ramifications of this new invader. Spidle et al.
(1994) were the first to confimi the identity of the quagga mussel as D. bugensis
using molecular data. Through an analysis of 1 1 allozyme loci in Ukrainian
specimens of D. bugensis and the North Arnerican quagga musel, Spidle et al.
(1994) supported the conclusions of Rosenberg and Ludyanskiy's (1994)
nomenclatural review of Dreissena spp., which identifid the North American
quagga mussel as D. bugensis on the basis of morphological characteristics.
While natural interbreeding between D. bugensis and D. polymorpha has
not been demonstrated in the wild, laboratory manipulation of zebra and quagga
mussel gametes has been used to produœ putative hybrid ofipring (Nichols
and Black 1993). Although the offspring produced from these crosses were not
successfully reared to the settling stage, the co-occurrenœ of the two dreissenid
mussel species at sites in the Laurentian Great Lakes has lead to concern over
the potential ramifications of naturally occumng hybrid mussels with distinct
physiological characteristics. Spidle et al. (1 995) used two proteincoding loci
believed to differentiate between the two species to investigate a cross-section
of mussels from sites in the lower Laurentian Great Lakes. Based on a sample
of 750 individual dreissenid mussels, Spidle et a' (1 995) found no evidenœ of
naturally occumng hybrids.
Denott and Munawar (1993) found a third unique phenotype of quagga
mussel inhabiting the profundal zone of Lake Erie during a 1992 survey of the
lake. Baldwin et al. (1 996) examined the mitochondrial cytochrome c oxidase
subunit I gene (mtCOI) from two representative individuals of the profunda
mussel as part of a larger survey and found four nucleotide difFerences (out of
61 9bp = 0.65% sequenœ divergence) between the profunda mussel and
D. bugensis. A more detailed study on twenty profunda mussels found no
nucleotide differences in this same gene fragment, suggesting that the profunda
mussel is a deep water phenotype of D. bugensis. and not a separate species
(Claxton et al. in press). While molecular tools have provided important
taxanornic information on the dreissenid invasion, the conflicting resufts of these
two mtCOl studies suggest a need for standardization of protocols and larger
sample sizes in taxonomie studies of Dreissena.
Population-Level Study
Given the extent of the invasion of Nom American waterways by
D. bugensis, the lack of any substantive population-level study of the quagga
mussel has lefi a large gap in Our understanding of its colonization and spread
through North Arnerica. As discussed above. Spidle et al. (1994) analyzed three
North American and one Ukrainian population of the quagga mussel at 11
allozyme loci. Based on 7 polymorphic loci, Spidle et al. (1994) found low levels
of genetic differentiation between North Arneriwn and European populations,
with overall heterozygosity estimates ranging from 0.1 09 to 0.1 31, approximately
I l3 of those reported for D. polymorpha (Marsden et al. 1 996). While study of
allozyme variation in the quagga mussel is limited, it will form an important
dataset for comparative analysis with a more extensive molecular survey.
Microsatellites
For the present study of quagga mussel population genetic structure, I
have developed and employed hig h ly polymorp hic microsatellite markers, as
they offer several bene- over more traditional genetic tools. Microsatellites,
short tandem repeats (STRs), or simple sequenœ repeats (SSRs). are stretches
of short (2-6 bp) repetitive sequences widely distributed in eukaryotic genomes.
Microsatellites are often highly polymorphic due to variation in the number of
repeat units, and mutation rates at microsatellite loci have been estimated ta
Vary between lo4 and 10" (Bniford and Wayne 1993). Microsatellite alleles are
inherited in simple Mendelian fashion and are likely selectively neutral (Ashley
and Dow 1994). As they are widespread in the genome and highiy variable.
microsatellites are excellent markers for genome mapping (Routman and
Cheverud 1994). studies of kinship (Queller et al. 1993) and in population-level
investigations (Bruford and Wayne 1 993).
Microsatellites: Potential Advantages over Allozymes
While the identification of microsatellite loci from a genomic library may be
time-cansuming and expensive, the operating costs associated with
microsatellite work following development are comparable to those for allozyme
markers (Parker et al. 1998). As rnicrosateliiie loci exhibl a higher relative level
of variation, they should be more sensitive than allozymes to changes in
population breeding size, structure, and patterns of dispersal (Estoup et al.
1 998). Recent comparative analysis of these two molecular techniques supports
this hypothesis (Hughes and Queller 1 993; Estoup et al. 1998).
In a cornparison of microsatellite and allozyme data. Hughes and Queller
(1 993) observed mean microsatellite heterozygosity of 0.62 in a species of social
wasp with little allozyme polyrnorphisrn (H,=0.035), suggesting that
microsatellites may offer an excellent alternative to allozymes with low variability.
Estoup et al. (1998) also found a higher level of polymorphism at microsatellite
loci than for allozymes, resulting in higher statistical power in tests of population
differentiation and linkage disequilibrium. The high nurnber of alleles present at
some microsatellite loci may also facilitate the detection of private alteles in
population-genetic studies, clanfying rates of gene flow between divergent
populations (Estoup et al. 1998). Although these studies suggest that
microsatellites are more powerful for studies of population genetics, it is
important to establish the goals of a population-level study in order to employ the
most appropriate molecular marker (Parker et a/. 1998).
Microsatellite Mutation
While microsatellite loci have been an important tool in population-genetic
investigations across a wide variety of taxa (e.g. Garza et al. 1995; Estoup et al.
1995; Angers et a/. 1995), our understanding of the mutational proœsses
goveming the evolution of microsatellites is still incomplete. Levinson and
Gutrnan (1 987) have suggested that the basic evolution of microsatellites fits a
slipped-strand mispairing model, where repeat units are either added or deleted
by replication slippage during DNA transcription.
The results of detailed empirical study on microsatellite variation in
humans and primates suggests that the dynamics of microsatellite mutation may
Vary between species and that a slipped-strand mispairing model of mutation
may be insufficient to explain microsatellite variation between closely related
taxa (Garza et al. 1995). Through an examination of eight microsatellite loci,
Garza et al. (1 995) found a bias towards an increase in microsatellite allele size
with constraints on average allele size, suggesting that the maximum number of
microsatellite repeats may be limited and that size homoplasy may be an
important consideration in the analysis of microsatellite data (Estoup et ai. 1995).
Microsatellite Analyses: Statistical Models
The statistical analysis of microsatellite data is contingent on basic
assurnptions regarding the mutational proœsses regulaüng microsatellite
variation. Three mutationai models have been applied to the analysis of
microsatellite data. The infinite alleles model (IAM) (Kimura and Crow 1964)
negates homoplasy, assuming an equal probability of any mutation in allele
length and that any allele arising by mutation is different from those previously
present in the species (Estoup et al. 1995). While the IAM has been shown to
provide an adequate fit to minisatellite data (Shriver et al. 1993). this mode1 fails
to consider the most up-todate information on microsatellite mutation.
Consequently, caution should be used in interpreting statistical analyses which
are based solely on this model.
In contrast. the stepwise mutation model (SMM) assumes that a
microsatellite mutation eÏther increases or decreases the number of alleles by a
single repeat unit (Kimura and Ohta 1978). An important consequenœ of the
SMM is that similar alleles may be not be identical by desœnt (homoplasy)
which may confound the analysis of microsatellites from widely divergent taxa
(Estoup et al. 1995). The SMM reflects the importance of the slipped-strand
mispairing process in microsatellite evolution, but fails to consider alternative
mechanisms which may lead to disjunct allelic distributions. While Shriver et al.
(1 993) have demonstrated that the evolution of human microsatellites with 3-5
bp repeat units is well described by the SMM, recent empirical tests of
invertebrate microsatellite markers have indicated significant deviations from this
model (Estoup et al. 1995).
The ho-phase rnodel has reœntly been proposed as an alternative to the
IAM and SMM (Di Rienzo et al. 1994). This model assumes that while rnost
mutations are single-step (SMM). a srnall number are mutti-step, which rnay
result in a non-normal distribution of alleles. In a study of ten dinucleotide
human microsatellites, Di Rienzo et al. (1994) found that data best fit a twa-
phase model. While the two-phase model rnay be more appropriate for the
analysis of some microsatellites, the frequency of a rnulti-step mutation rnay not
be consistent across different loci and rnay not be easily detemined given the
scope of most population-level studies. Due to these complications, most
methods for the analysis of microsatellite data continue to be based on either the
IAM or SMM. As additional empirïcal data are collected and theoretical work
continues. a more complete picture of microsatellite mutation rnay be
established.
Thesis Outline
In this thesis, I investigate patterns of colonization and dispersal which
have characterized the spread of D. bugensis in the Laurentian Great Lakes
through an investigation of the population structure of the species eight years
after the initial introduction. As allozyme studies rnay have been confounded
due to the presence of two species of dreissenid mussel (Marsden et al. 1996).
s pecies-specifc rn icrosatellite markers are most appropriate for a b road survey
of quagga musse1 population genetic structure. The use of microsatellite
markers should also result in more robust estimates of population differentiation
(Estoup et al. 1998). Chapter two outlines the development and optimization of
microsatellite marken in D. bugensis.
In chapter three, I use six microsatellite loci to investigate the role that
larval dispersal and byssal thread attachment have played in the spread of the
quagga mussel. Passive diffusion via a planktonic veliger stage should result in
a cleariy defined population structure which fits an isolation-bydistanœ model
over a broad geographic scale. In contrast. human-mediated transport of veliger
larvae and adult mussels via boat traffic could result in a disjunct distribution of
mussel populations and high levels of gene flow between populations,
eliminating the potential for genetic differentiation on the basis of distance. I
present data on the presentday population structure of the quagga mussel in
the lower Laurenüan Great Lakes and use these data to infer patterns of
colonization and spread which have played a significant role in the success of
the quagga mussel invasion of North Arnetica.
Bi bliography
Ackeman, J.D., Sim, B., Nichols, S.J., and R. Claudi. 1994. A review of the early life histoiy of zebra mussels (Dreissena bugensis): Cornparisons with marine bivalves. Can. J. Zool. 72: 1 169-1 179.
Angers, B., Bematchez, L., Angers, A, and L, Desgroseillerç. 1995. Specfic microsatellite loci for brook chan reveal strong population subdivision on a microgeographic =le. J. Fish. Bol. 47(Supp. A): l ï ï- l85.
Ashley, M.V., and B.D. Dow. 1994. The use of microsatellite analysis in population biology: Background. methods, and potential applications. In Molecular ecology and evolution: Approaches and applications. Edited by B. Schierwater, B. Streit, G.P. Wagner, and R. Desalle. Birkhauser Veriag, Basel. Switzeriand. pp. l85-2OI.
Baldwin, B.S., Black, M., Sanjur, O., Gustafson, R., Lm, R.A, and R.C. Vrijenhoek. .tg96 A diagnostic molecular marker for zebra mussels (Dmissena poIymopha) and potentially m-occumng bivalves: rnitochondrial COI. Mol. Mar. Biol. Biotech. 5: 9-14.
Banarescu, P. 1990. Zoogeography of f'resh waters -Vol. 1: General distribution and dispersal of freshwater animals. AU LA-Veriag, Wiesbaden.
Bruford, M.W., and R.K Wayne. 1993. Microsatellites and their application to population genetic studies. Cuni Opin. Gene. Dev. 3: 939-943.
Carlton, J.T. 1993. Dispersal mechanisrns of the zebra mussel (Dreissena polymorpha). In Zebra mussels: Biology, impacts, and contml. E d M &y T.F. Nalepa and 0. Schloesser. Lewis Publishers, Boca Raton, Florida. pp. 677-697.
Claxton, W.T., Wilson, AB., Mackie, G.L., and E.G. Boulding. 1998. A genetic and morphological cornparison of shallow and deep water populations of the introduced dreissenid bivalve, DreÏssena bugensis. Can. J. Zool. (in press).
Dermott, R., and M. Munawar. 1993. Invasion of Lake Erie offshore sediments by Dreissena, and its ecological implications. Can. J. Fish. Aquat Sci. 50: 2298-2304.
Di Rienzo, A, Peterson, AC., Gam, J.C., Valdes, AM., Slatkin, M., and N.B. Freimer. 1994. Mutational processes of simple-sequence repeat loci in human populations. Proc. Natl. Acad. Sci. USA 91: 3166-3170.
Estoup, A, Tailliez, C., Comuet, J-M, and M. Solignac. 1995. Size hornoplasy and mutational processes of intempted microsatellites of two bee species, Apis meltifera and Bombus temsfns (Apidae). Mol, Biol. Evol. 12: 1074-1 084.
Estoup, A, Rousset, F., Michalakis, Y., Comuet J-M, Adriamanga, M., and R. Guyomard. 1998. Comparative analysis of microsatellite and allozyme markers: A case study investigating microgeographic differentiation in brown trout (Salmo butta). Mol. Ecol. 7: 339-353.
Garza, J-C., Slatkin, M., and N.B. Freimer. 1995. Microsatellite allele frequencies in humans and chimpanzees with implications for constraints on allele size. Mol. Biol. Evol. 12: 594-603.
Hastings, A 1996. Models of spatial spread: Is the theory complete? Ecology 77: 1675-1679.
Hebert, P.D.N., Muncaster, B.W., and G.L. Mackie. 1989. Ecological and genetic studies on Dmissena polymorpha (Pallas): A new mollusc in the Great Lakes region. Can. J. Fish. Aquat- Sci. 46: 1587-1 591.
Hengeveld, R. 1989. Dynamics of biological invasions. Chapman and Hall Ltd., London.
Hughes, CR., and DG. Queller. 1993. Detection of highly polymorphic microsatellite loci in a species with litüe allozyme polymorphisrn. Mol. €col. 2: 131 -1 37.
Johnson, L.E., and J.T. Cariton. 1996. Post establishment spread in large-sale invasions: Dis persal rnechanisrns of the zebra mussel Dreissena polymorpha. Ecology TI: 1686-1 690.
Kew, H.W. 1833. The dispersal of shells. Kegan Paul, Trench, Trubner & Co. Ltd.. London.
Kimura, M., and J.F. Crow. 1964. The number of alleles that can be maintained in a finite population. Genetics 49: 725-738.
Kimura, M., and T. Ohta. 1978. Stepwise mutation mode1 and distribution of allelic muencies in a finite population. Proc. Natl. Acad. Sci. USA 75: 2868-2872.
Levin, LA, and T.S. Bridges. 1995. Pattern and divenity in reproduction and development in Ecology of manne invertebrate larme. Edited by LR. McEdward. CRC Press, Boca Raton, Florida. pp. 1-48.
Lodge, D.M. 1993. Biological invasions: Lessons for ecology. Trends Ecol. Evol. 8: 133-1 37.
Mackie, G.L., and D.W. Schloeser. 1996. Comparative biology of zebra mussels in Europe and North Arnerica: An overview. Am. Zool. 36: 244-258.
Marsden, J.E., Spidle, AP., and B. May. 1996. Review of genetic studies on Dreissena spp. Am. ZWI. 36: 259-270.
May, B. and J.E. Marsden. 1992. Genetic identification and implications of a second invasive species of dreissenid musse1 in the Great Lakes. Can. J. Fish. Aquat Sci. 49: 1501- 1506.
McMahon, R.F. 1991. Mollusca: Sivalvia. In Ecology and classification of North American freshwater invertebrates. Edited by J.H. Thorp and AP. Covich. Academic Press, San Diego, California. pp. 31 5-399.
Mills, E L , Demott, R.M., Roseman, E.F., Dustin, D., Mellina, E., Conn D.B., and A.P. Spidle. 1993a. Colonkation, ecology, and population structure of the 'quagga" musse1 (Bivalvia: Dreissenidae) in the lower Great Lakes. Can. J. Fish. Aquat Sci. 50: 2305- 2314.
Mills, E.L., Leach, J.H., Carlton, J.T., and C.L. Secor. 1993b. Exotic species in the Great Lakes: A history of biotic crises and anthropogenic introductions. J. Great Lakes Res. 19: 1-54.
Mills, E.L., Rosenberg, G., Spidle, AP., Ludyanskiy, M., Pligin, Y., and B. May. 1996. A review of the biology and ecology of the quagga mussel (Dreisena bugensis), a second species of freshwater dreissenid introduced to North America. Am. 2001. 36: 271-286.
Padilla, D.K., Chotkowski, M.A, L-AJ. Buchan 1996. Predicting the spread of zebra mussels (Dreissena polymorpha) to inland lakes using boater movement patterns. Glob. Ewl. Biogeog- Let 5: 353-359.
Palumbi, S.R. 1995. Using genetics as an indirect estirnator of larval dispersal. ln Ecology of manne invertebrate lawae. Edited by L.R. McEdward. CRC Press, Boca Raton, Florida. pp. 369-387.
Palurnbi, S.R. 1996. Macrospatial genetic structure and speciation in manne taxa with high dispersal abilies. In Molecular zoology: Advances, strategies, and procedures. Edited by J.D. Ferraris and SR, Palurnbi, Wiley-Lis, Inc., New York pp. f 01 -1 17.
Parker, P.G., Snow, AA, Schug, M.D., Booton, G.C., and P.A Fuerst 1998. What molecules can tell us about populations: Choosing and using a molecular marker. Ecalogy 79: 361- 382.
Pimm, S.L* 1 989. fheories of predicting success and impact of introduced species. In Biological invasions: A global perspective. Edited by J.A Drake and H A Mmney. John Wiley & Sons Ltd.. London.
Pligin, Y.V. 1984. Extension of the distribution of Drer'ssena bugensis. Malac. Rev. 17: 143-144.
Queller, D.C., Strassmann, J.E., and C.R. Hughes. 1993. Microsatellites and kinship. Trends E d . Evo~. 8: 285-288.
Ricciardi, A, Serrouya, R., F.G. Whoriskey. 1995. Aerial exposure tolerance of zebra and quagga mussels (Bivalvia: Dreissenidae): Implications for overland dispersal. Can. J. Fish. Aquat Sci. 52: 470-477.
Rosenberg, G., and M.L. Ludyansk iy. 1994. A nomenclatural review of Dreissena (Bivalvia: Dreissenidae), wiai identification of the quagga mussel as D ~ i s e n a bugensis. Can. J . Fish. Aquat Sci. 51: 1474-1484.
Rousset, F. 1997. Genetic difkrentiation and estimation of gene fiow from F-statistics under isolation by distance. Genetics 145: 121 9-1 228.
Routrnan, E., and J.M. Cheverud. 1994. Individual genes underiying quantitative traits: Molecular and analytical methods. In Molecular ecology and evolution: Approaches and applications. Edited by B. Schienrvater, B. Streit, G.P. Wagner, and R. Desalle. Birkhauser Verlag, Basel, Switzeriand. pp. 593-606.
Shriver, M.D., Jin, L., Chakraborty, R., and E. Boerwinkfe. 1993. VNTR allele frequency distributions under the stepwise mutation rnodel: A cornputer simulation approach. Genetics 134: 983-993.
Spidle, AP., J.E. Marsden, and B. May. 1994. Identification of the Great Lakes quagga mussel as Dreissena bugensis from the Dneiper River, Ukraine, on the basis of allozyme variation. Can. J. Fish, Aquat Sci. 51: 1485-1489.
Spidle, AP., E.L. Mills, and B. May. 1995. Absence of naturally occumng hybridization between the quagga mussel (Dreissena bugensis) and the zebra mussel (D. polymorpha) in the lower Great Lakes. Can. J. 2001.73: 400-403.
United States Geological Survey. 1997. National zebra mussel and aquatic nuisance species clearinghouse: United States distribution maps. http:lhwvw.nfrcg.gov/zebra.mussel.
Williamson, M., and A Fritter. 1996. The varying success of invaders. Ecology ï7: 1661-1666.
Yonge. C.M. 1962. On the primitive significance of the byssus in the Bivaivia and its effect on evofution. J. Mar. Biol. Assoc. U.K. 42: 261-271.
Dreissena b ugensis
Fia 1 .l. Current North American distribution of dreissenid mussels. Stars indicate putative points of introduction. From Mills et al. (1 996); USGS (1997).
CHAPTER 2
Development of Novel Tri- and Tetra-Nucleotide Microsatellite
Loci in the Invasive Mollusc, Dreissena bugensis.
Introduction:
The quagga mussel, Dreissena bugensis, is native to the Dneiper River
drainage system in the Ukraine (Mills et al. 1996). In 1989, a population of
D. bugensis was identified in eastern Lake Erie, Canada (Mills et al. 1993).
Following its introduction, the quagga mussel spread rapidly within the lower
Laurentian Great Lakes and populations have now been identified in Lake
Erie, Lake Ontario, the St. Lawrence River (Mills et al. 1996). It is
hypothesized that dreissenid mussels were introduced to the Laurentian Great
Lakes following transoceanic transport of quagga mussel individuals in the
ballast water of one or more vessels (Hebert et al. 1989).
A recent study investigated the introduction of the quagga mussel using
allozymes. but failed to d istingu ish any sig nificant population-level structuring
of D. bugensis in the Laurentian Great Lakes (Spidle et al. 1994). The
development of highly variable microsatellite markers will provide a powerful
new tool for the investigation of quagga mussel populations and may help
clarify patterns of colonization and subsequent diffusion that have
characterized the rapid spread of dreissenid mussels in North Amen'ca. In this
study, I describe the identification of six highly polymorphic tri- and tetra-
nucleotide microsatellite markers in D. bugensis which can be visualized using
a non-radioactive silver-staining protocol.
Materials & Methods:
DNA Isolation 8 Preparation for Cloning:
Genomic DNA was isolated from 18 adult D. bugensis using a modified
dreissenid extraction protocol, w lh DNA precipitation in one volume of 100%
isopropanol (Claxton et al. 1997). Extracted DNA was pooled and digested
with Rsal and Haelll. The restricted DNA was electrophoresed through a 1%
agarose gel and fragments of 300-1000bp were excised and purified using a
glass rnilk purification method (Sephaglas BandPrep kit, Phanacia). A size-
selected Iibrary was constructed by ligating the DNA into the vector pZErO
(Invitrogen) (restricted with EcoRV) following manufacturer's
recommendations, and transfecting the modified vector into electrowmpetent
E. coli Top 1 OF' cells by electroporation (E. coli pulser, Bio-Rad).
The pZErO plasmid (Invitrogen) contains the lethal gene cedB in its
multiple-cloning site. This gene produces a protein whose expression results
in a degradation of the host cell chromosome, leading to cell death. If the
plasmid contains an insert, the ccW gene is interrupted and cells grow
nomally and if not, the hast cell dies. In this way, pZErO functions as a
positive-selection system, optimizing the number of insert-containing clones.
Colony Screening and Microsatellite Development:
Colonies were lifted ont0 nylon membrane (Zeta-probe GT Membrane,
Bio-Rad) for 5 min in 10% SDS, denatured in 0.5M NaOH and 1.5M NaCl for 5
min, and neutralized for 5 min in a 1 .OM Tris-HCI (pH 7.5) 1 1.5M NaCl buffer.
DNA was fixed to the membrane by baking for one hour in a 80°C incubator.
Pnor to hybridization, membranes were washed for 30 min at 50°C in a
solution containing 5X SSC, 0.5% SOS and 1 mM EDTA (pH 8.0) to remove
cellular debris. Following this wash, membranes were transferred to a 5X
SSC, 1X Denhart's solution and 0.1% SDS hybridization solution and
prehybridized for 2 hours in a hybridization oven (Hybridiser HB-1. Techne). A
total of 50 000 colonies were screened using the oligonucleotides AAATe and
AATto (Hybridization T = 52%) and GATA7, M C 9 , and M G e (Hybridization T
= 55°C) end-labelled with PP-ATP. y 3 3 ~ - ~ ~ ~ - l a b e l l e d probe was added to
fresh hybridization solution and membranes were hybridized overnight at
hybrid ization tempe rature (see above). After hybridization, membranes were
washed twice for 15 min in 5X SSC and 0.1 % SDS at 50°C and exposed to
autoradiographic film (Biomax MR, Kodak) ovemight.
Fifty-two positive clones were sequenced using the AB1 PRlSM dye
teminator cycle sequencing-ready reaction kit (following manufacturer's
recommendations) and analyzed with an AB1 PRlSM 377 automated
sequencer. Pdmer sets were developed using PRIMER V0.5 (Lincoln and
Daly 1991) for ten of the f@-two sequences with long, unintempted repeats
and adequate single-copy fianking DNA for primer design.
Polymerase Chain Reaction of Primer Sets:
?CR was perfomed using a PTC-100 themocycler (MJ Research) as
follows: 5 min at 95°C; followed by 28-31 cycles of 1 min at 95OC, 1 min at
annealing temperature (see Table l ) , 1 min extension at 72°C. Ten microlitre
reactions contained 20 ng of total genomic DNA, 0.2 mM each of dATP, dCTP.
dGTP, and dlTP (Pharmacia), 2.0 mM MgC12 (Gibco-BRL), 20 mM Tris-HCI
(pH 8.4), 50 mM KCI, 0.2 FM of each primer, and 0.5 units of Taq DNA
polymerase. The PCR products were electrophoresed through a 5% non-
denaturing l9:l acrylamide:bisacrylarnide gel and visualized using a silver-
staining protocol. Gels were washed for six minutes in a 10% ethanol and 5%
acetic acid buffer, transferred to a 0.1% silver nitrate solution for ten minutes,
washed twice in distilled water and stained for a further ten minutes in a buffer
containing 1 3% NaOH, 0.1 % NaBH4, and 0.1 5% formaldehyde. This staining
protocol effectively reduces gel visualization to 0.5 hr and enables the use of
tri- and tetra-nucleotide primer sets in laboratories that lack facilities for
radioactive work.
Results & Discussion:
Of the ten primer sets developed, six amplified repeatably and were
polymorphic within a single population of 36 individuals collected from Long
Point Bay, Lake Erie. Of the four loci that failed to yield an acceptable
product, Iwo loci were monomorphic (1 AAAT,, 1 MCn) and primer sets for
two AAT, repeats yielded non-specific banding patterns. The six acceptable
loci were used in a more detailed within- and cross-species investigation.
Test for CrossSpecies Amplification:
All six primers failed to amplify samples of the zebra mussel.
D. polymorpha, over a 4863°C temperature range. The failure of cross-
species amplification is not unexpected, as D. polymorpha has been reported
to be 47% divergent from O. bugensis at a 710-bp fragment of the
mitochondrial cytochrorne c oxidase subunit I gene (Baldwin et al. 1996).
suggesting a long period since divergence of the two species. An
independent microsatellite library has been developed for the zebra mussel
(KA Naish, University of Guelph, Ontario, Canada, unpublished data).
Amplification of D. bugensis:
The two tetra- and four tri-nucleotide polymorphic loci were fumer
analyzed on a group of six populations from the Laurentian Great Lakes
comprising 324 individuals (Table 2.1 ). Levels of intraspecific variability were
high, with allele numbers at each locus ranging from 4 to 44 and observed
heterozygosity ranging between 0.108 and 0.81 1.
Null allele at Dbug6:
Of the six microsatellite loci analyzed, the original primer pair developed
for Dbug6 yielded a lower success rate of amplification and a significantly
lower heterozygosity when compared to other loci. These two factors
suggested the presence of a nuIl allele. where mutation in one or both priming
sites results in non-priming of samples. In an attempt to verify the presence of
a nuIl allele at Dbug6, a new set of primers was developed for this locus.
Upon redesign of primers for Dbug6,100% of sarnples arnplified and
heterozygosity levels at this locus increased from 0.108 to 0.389 (Table 1).
confining that one or more mutations in the region targeted by the initial
primer pair had resulted in non-amplification of samples.
The collection of data for Dbug6 both with a nul1 allele and upon
redesign of primers offers an unique opportunity to examine the impact of a
nuIl allele-biased data set on comrnon statistical estimators. The hierarchical
statistics, FST and RsT were calculated for al1 loci (Wieir and Cockerham 1984)
and linearized (Slatkin 1995) using ARLEQUIN VI -1 (Schneider et al. 1997)
(Table 2.2). Overall significance of 20% of Fsr estimates (3/15) and 13.3% of
IQT estimates (211 5) was affected by the presence of a nuIl allele (Permutation
test (4 0,000 permutations; pe0.05)). As demonstrated here, the presence of a
nul1 allele at a single microsatellite locus may significantly bias overall
measures of genetic distance, potentially confounding conclusions regarding
the presenœ or absence of population structuring in natural populations.
Caution should be taken to eliminate any potential sources of error in the
collection of microsatelfite data.
Allelic Data: Assessrnent of Sample Sire:
Ruzzante (1998) examined the effect of sample size on several
measures of genetic distance and population structure for highiy variable
microsatellite loci and suggested that sample sizes of between 50 and 100 are
generally necessary for precise estimation of genetic distances. In an
investigation of the sample size necessary for adequate allelic representation
of microsatellite loci in D. bugensis. sample sets within each of eig ht
populations (n=54) were randomized (to avoid entry bias) and groups of five
individuals were selected without replacement from each population. Average
allele number and associated standard error were plotted versus sample size
(Fig. 2.1) in an effort to obtain a descriptive estimate of the sample size
required to adequately sample each microsatellite locus. Data for al1 six loci
suggest that at samples of greater than 50, the number of new alleles
sarnpled approaches a plateau. Based on these descriptive data, 1 appears
that while sample sizes of n r 50 are generally required to adequately sample
Dbugl-5, a sample size of 35 would be sufficient to sample the allelic
distribution in al1 populations for Dbug6, due to the low number of alleles
present at this locus.
Conclusion:
The high variability and species-specificity of these microsatellite
markers will make them excellent tools for the elucidation of gene fiow
between introduced populations of D. bugensis. While the conclusions of
earlier allozyme studies on the quagga mussel may have been confounded
due to misidentification of D. polymorpha individuals as D. bugensis (Marsden
et al. 1996), the high specificity of these new microsatellite markers will help to
minimize the potential error associated with such complications.
Bi bliography:
Baldwin, B.S., Black, M., Sanjur, O., Gustafson, R., Lutz, R A , and R.C. Vrijenhoek. 1996. A diagnostic molecular marker for zebra mussels (Dreissena polymorpha) and potentially co-occumng bivalves, mitochondrial COI. Mol. Mar. Biol. Biotech. 5: 9-14.
Claxton W.T., Martel A., Dermott, R.M., and E.G. Boulding. 1997. Discrimination of field- collected juveniles of two introduced dreissenids (Dreissena polymorpha and Dmissena bugensis) using mitochondrial DNA and shell morphology. Can. J. Fish. Aquat Sci. 54: 1280-1288.
Hebert P.D.N., Munmster, B.W., and G.L. Mackie. 1989. Ecological and genetic studies on Dmr'ssena polymorpha (Pallas): a new mollusc in the Great Lakes. Can. J. Fish. Aquat Sci. 46: 1587-1 591.
Lincoln S., and M. Daly. 1991. Primer, version 0.5. Whitehead lnstitute for Biomedical Research, Cambridge, Mass.
Marsden J.E., Spidle A.P., and B. May. 1996. Review of genetic studies of Dreissena spp. Am. Zool. 36: 259-270.
Mills, E-L, Dermott, R.M., Roseman, E.F., Dustin, D., Mellina, E., Conn D.B., and A P. Spidle. 1993. Colonïzation, ecology, and population structure of the "quagga" mussel (Bivalvia: Dreissenidae) in the lower Great Lakes. Can. J. Fsh. Aquat Sci. 50: 2305-2314.
Mills E-L., Rosenberg G., Spidle AP., Ludyanskiy M., Pligin Y., and B. May. 1996. A review of the biology and ecology of the quagga mussel (Drdsena bugensis), a second species of freshwater dreissenid introduced to North America. Am. Zool. 36: 271-286.
Ruzzante, D.E. 1998. A comparison of several measures of genetic distance and population structure with microsatellite data: Bias and sampling variance. Can. J. Fish. Aquat Sci. 55: 1-14.
Schneider, S., Kueffer, J-MI Roessli, O., and L. Excoffier. 1997. Arlequin VI. 1 : A software for population genetic analysis. Genetics and Biometry Laboratory, University of Geneva, Switzerland.
Slatkin, M. 1995. A measure of population subdivision based on microsatellite allele frequencîes. Genetics 139: 457-462.
Spidle A R , Marsden JE. , and B. May. 1994. Identification of the Great Lakes quagga mussel as Dreissena bugensis from the Dneiper River, Ukraine, on the basis of aIlozyme variation. Can. J. Fish. Aquat Sci. 51: 1485-1489.
Weir, B.S., and C.C. Cockerham. 1984. Estirnating F-statistics for the analysis of population structure. Evolution 38: 1358-1 370.
Table 2.1 Characterization of Dreissena bugensis microsatellite loci. All loci were analyzed using 54 individuals from each of six populations (n=324; see Chapter 3) in the Laurentian Great Lakes. Ho = observeci heterozygosity. Note the redesign of Dbug6 reverse primer.
Locus* Ho # of Alleles Primer Sequence Sequenced Size of Annealing (5'-3') Repeat Motif Sequenced (Oc)
lnsert
D b ~ g i 0.589 44 GTAGTlTATCAGGCTTGmGGAC M A T 4 269 56 GTACCGGTAAAGTCAAATCGTCCC
Db~g3 0.811 30 CTGCATATGCTTCGTGTTllATGC M T i 4 264 AAGCTGATTACGTTCGCTCTAACC
Db~g5 0.328 27 TTCCGCCATATCATCAGACTTTTC TAC5 M T z 7 298 GGTCGCATTCGTTGATGTCTG
Dbug6 0.108 4 TCTCCTGGGAGCTCATGTTAGC AAc7 205 WI Null AGTATGATTTATTGCACGGACATCG
Dbug6 0.389 4 TCTCCTGGGAGCTCATGTTAGC AAc7 205 WIO Null CTTGCCGCCCACCTAAAGT
Genbank Accession Numbers: XXX-XX
Table 2.2. Paiiwise linearized FST (upper right) and RsT (lower left) estimates (Slatkin 1995) for six study populations (see Table 3.1) calculated by ARLEQUIN V I . l (Schneider et al. 1997) - Impact of nuIl allele. (a) With null allele at Dbug6; (b) Without nuIl allele at Dbug6. * = Values are significantly different from zero (Permutation test; pg0.05).
{a) With nul1 allele: (b) Without nul1 allele: Popn CE LPI EE WO EO SF Popn CE LPI EE WO EO SF CE - 0.0164' 0.0050 0.0904' 0.01 59" 0.0000 CE - 0.0059 0.0033 0.0096* 0.0230' 0.01 13'
l
O O
LPI 0.0000 - 0,0147' 0.0147'0.0422' 0.0220' LPI 0.0000 - 0.01 50' 0.01 86' 0.0345' 0.0305~ I
Sample Size - 0 10 20 30 40 50
Sompie Size
Fig. 2.1. Plot of average number of alleles presentlpopulation at variable sam ple sizes for six Dreissena bugensis microsatellite loci. Sample sets were randomized prior to calculation. Nu= Total number of alleles scored.
30
25
$ 20 - a l . 3 15 C
O * 10
5 -
- 5
* DbugS N-47 4
- i T T -
,_.r- - - - --1 $ 3 - L I - r * s 1. - 1 O +
*
- = 1 T L
?t 2 -
I : 1 -
- Dbug6
- 1 ? -;- -; ;- - P
1 1. '
. O r - ' - * - 1 9 I 0.t - s 1
O 10 20 30 40 50 60 O 10 20 30 40 50 60 Sarnple Site Sample Ske
CHAPTER 3 Dis persal Patterns of Dreissena bugensis In The Lau rentian
Great Lakes As lnferred F rom Highly Polymorphic
Microsatellite Markers
Abstract:
While an idealized model of dispersai predictç that high dispersal potential
leads to high levels of gene fiow and low levels of population stfuctunng,
empirical study suggests many exceptions to this paradigrn. The recent invasion
of Dreissena polymorpha and D. bugensis to the Laurentian Great Lakes
presents a unique opportunity to study the spread of two closely related species
with high dispersal potential. While the dispersal of D. polymopha has been
charaderized by several jump dispersal events, survey data suggest a gradua1
diffusion of D. bugensis from its point of introduction through the lower
Laurentian Great Lakes. In this study, I use six highly polymorphic species-
specific microsatellite markers to investigate the present-day population genetic
structure of D. bugensis in North Amerka in an effort to clanfy the role of
divergent dispersal strategies in its colonization sucœss. In contrast to survey
data, which would suggest a fit to the isolation-bydistanœ modei, the significant
differences observed between 24 of 28 population pairs do not correlate with any
simple geographic relationship. At the same time, several disjunct populations
are not significantly different, suggesting that boater-mediated jump dispersai of
mussels may be having a significant impact on the population genetic structure
of D. bugensis.
Introduction:
Biological invasions are characterized by two distinct phases: the
establishment of a species at a single location, and its subsequent radiation in
spaœ (Hastings 1996). Both the establishment of an initial population and
sucœssful naturalkation of the species in its new habitat are crucial to the
survival of an invading species. Theoretical ecologists have developed a
number of models in an effort to c l am the relationship between dernographic
parameters and the rate of range expansion of introduced species (Hastings
1996). While empirical verfication of these rnodels is incomplete, existing data
suggest that rare, longdistanœ jump dispersal events may play an important
role in the spread of ment invaders (Hastings 1996).
Until reœntly, theoretical estimates of species spread have been limited
to ecological and laboratory studies of relevant Me-history parameters and
annual survey data documenting the sucœssful colonization and spread of an
invasive species (Hastings 1996). Wioi the advent of highly polymorphic
molecular markers, a greater understanding of the patterns of spread of a natural
population may be achieved, providing data on rates of gene flow between
populations and clanfying secondary introductions which may have a significant
impact on the population structure of an invasive species.
The quagga mussel, D. bugensis, is native to the Dneiper drainage
system in the Ukraine (Mills et al. 1996). In 1989, D. bugensis individuals were
identified in eastem Lake Erie, Canada (Mills et al. 1993; Figure 3.1). The
quagga mussel spread rapidly following its initial colonization and by 1996, it had
spread through Lake Erie to Lake Ontario and the St. Lawrence River (Mills et ai.
1996). Researchers have hypothesized that dreissenid mussels were
introduœd to the Laurentian Great Lakes following transoceanic transport of
mussels in the ballast water of ships (Hebert et al. 1989)
Studies on temporal variation in distribution of dreissenid mussels have
benefited from an extensive survey program (USGS 1997) which was
established following the identification of the zebra mussel, 0. polymorpha, in
Lake St. Clair, Ontario in 1986 (Hebert et al. 1989). Data suggest that in
contrast to the spread of the zebra mussel, which has been characterized by
several longdistance jump dispersal events within the Laurentian Great Lakes,
the quagga mussei has exhibited a more gradua1 diffusion from Ïts initial
introduction in Lake Erie eastvvard into Lake Ontario and the St. Lawrence River
(Mills et al. 1993).
Despite their differential spread, the radiation of both D. bugensis and
D. polymorpha has been exceedingly rapid following their initial colonization.
The quagga and zebra mussel have two notable Me-history characteristics which
may have accelerated their colonization and spread through the Laurentian
Great Lakes (Johnson and Cariton 1996). Firstly, both D. bugensis and D.
polymorpha have a planktonic veliger stage, whereby free-living larvae can
remain suspended in the water column for up to several weeks (Carkon 1993).
During this time, larvae passively diffuse in lacustrine currents and, upon
settlernent may found new populations.
Dreissenid mussels are also byssate, facilitating attachrnent to substrate
via production of byssal threads (Mills et al. 1996). 60th larvae and adults have
been identifid attached to the hulls of ships (Johnson and Carkon 1996).
Movernents of boats both within lakes (Johnson and Cariton 1996) and overland
(Ricciardi et al. 1995) may resuit in the formation of new mussel populations in
geographically disparate regions. Dispersal via boat trafic, and to a lesser
degree, birds and fish, may explain the jump dispersal which characterized the
rapid spread of the zebra mussel to upstream sites in Lake Superior, Lake
Michigan and Lake Huron from founding populations in Lake St. Clair (Carlton
'l993).
In this study, I use highly polymorphic microsatellite markers in an effort to
clarify the presentilay population genetic structure of the quagga mussel in
North Arnerica. As the quagga mussel has only been present in the Laurentian
Great Lakes since i 989, these data can serve as an indirect estimator of the
dispersal of D. bugensis through North Arnerîcan watennmys. As noted above,
survey data suggest a gradual diffusion of 0. bugensis from its initial introduction
in Lake Erie. Such a gradual diffusion should exhibit a strong fit to the isolation-
bydistance mode1 (Rousset 1997). In wntrast. if jump dispersal is playing a role
in mussel spread, genetic data should exhibit a departure from this model.
Microsatellite markers will help to quantify this pattern of spread. dariQing the
role that hurnannediated dispersal vedors have played in the spread of the
quagga mussel.
Materials and Methods:
Sample Collection:
Adult D. bugensis were collected from a combination of near-shore and
off-shore regions representing the extent of the species range in the lower
Laurentian Great Lakes (Figure 3.1 ; Table 3.1). Upon collection, samples were
extracted immediately or frozen at -90°C until further use. Shell length data was
collected for al1 extracted specimens and compared with shells from a single
year class collected from a Canadian Coast Guard buoy in Long Point Bay, Lake
Erie. Shells were preserved for further studies of correlation between genotypic
and phenotypic variation within the species.
Extraction and PCR Protocol:
Genomic D M was isolated from individual D. bugensis using a modified
dreissenid extraction protocol. with DNA precipitation in one volume of
isopropanol (Claxton et al. 1997). DNA samples were diluted 1 :20 in distilled
water prior to PCR amplification.
Samples were PCR-amplified at six polymorphic microsatellite loci Fable
2.1). These microsatellites have been tested on both species of dreissenid
mussel found in the Laurentian Great Lakes and are specific to D. bugensis (see
Chapter 2). Ten microlitre PCR readions were performed in a PTC-100
thermocycler (MJ Research) using 20 ng of total genomic DNA. 0.2 mM each of
ÇIATP, dCTP, dGTP, and dlTP (Phanacia), 2.0 mM MgCh (Gibco-BRL), 20
mM Tris-HCI (pH 8.4), 50 mM KCI, 0.2 uM of each primer and 0.5 units of Taq
DNA polymerase. The following ?CR profile was used: Denaturation for 5 min at
95OC; followed by 28-31 cycles of 1 min at 95OC. 1 min at annealing temperature
(54-63OC; see Chapter 2), 1 min extension at 72°C. The PCR products were
electrophoresed on a 5% nondenaturing 19:l acrylamide: bisacrylamide gel and
visualized using a silver staining protocol (see Chapter 2).
For each of the six loci, allelic ladders were constructed using a
combination of PCR product from individual mussels. These allelic ladders were
run every five lanes in each gel and helped to facilitate swring of gels and to
maintain consistency of scoring between difterent gel runs (Appendix 3.1).
Statistical Analyses:
Linkage Disequilibnum:
Loci data were analyzed for independenœ by an analysis of genotypic
disequilibrium (Garnier-Gere and Dillmann 1992) using GENEPOP V3.1 a
(Raymond and Rousset 1995a). GENEPOP generates R X C contingency
tables for each pair of loci within each population and performs a probability test
on each table using a Markov chain method. Given the nul1 hypothesis Ho =
'Genotypes at one locus are independent from genotypes at the other locusn, the
probability of the observed table is:
P k
Il (Ni.!) . ll (N.,!) F1 j=1
n = P k
(N..!) n il (Ni!) i=I j=i
where Ni. = number of sarnples of genotype i at locus A, N.j = number of samples
of genotype j at locus B. Ni = number of samples which have both genotype i (at
locus A) and j (at locus B), and N.. = sum of al1 cells (Raymond and Rousset
1995a). The typesne error probability of rejecting Ho is calculated by summing
the probabilities of al1 contingency tables with the same or lower probabilities and
with the same row and column sums (Raymond and Rousset 1995b). As the
Fisher exact test is impossible to perfom for most data sets (due to
computational constraints), an unbiased estirnate of the P-value can be obtained
using a Markov chain method. The Markov chain method explores the spaœ of
all possible contingency tables with the same row and column sums. The
proportion of penuted tables with a lower or equal probability than the observed
is an unbiased estimate of the P-value (Raymond and Rousset 1995b). For the
linkage disequilibrium analysis. a 10,000 batch, 1 ,000 iteration Markov chain
analysis was preceded by a 1 0.000-step dememorization proœss.
Hardy- Weinberg Equilibrium:
Data were tested for departure from Hardy-Weinberg expectations using
GENEPOP V3.l a (Raymond and Rousset 1995a). GENEPOP prepares R X C
contingency tables of observed data and theoretical expectations. For the nuil
hypothesis Ho = 'Random union of gametes". the probability of the observed
table is:
where S = sample population, N = number of individuals. k = nurnber of alleles,
nq = number of genotypes 4 (isj) (Rousset and Raymond 1995). As for the
linkage disequilibrium test, the type-one error probability of rejecting Ho is
calculated by summing the probabilities of al1 contingency tables with the same
or lower probabilities and with the same row and column sums (Raymond and
Rousset 199513). For loci with fewer than five alleles, an exact P-value was
calculated by a wmplete enurneration method and for loci with a greater number
of alleles, a 10,000 step, 1,000 iteration Markov chain method (1 0,000
demernorization steps) was used to calculate an unbiased estimate of the P-
value.
Popu/ation Dfle~ntiation:
Tests for genotypic differentiation of populations are less powerful than
those which assume random sampling of alleles, but more appropriate when
individuals exhibit non-random rnating (Goudet et al. 1996). Global tests for
genotypic differentiation yielded highly significant differences (P<0.00001)
among the eight study populations. In an effort to clanfy the pattern of
interpopulation differentiation, pairwise tests of genotypic differentiation were
performed using GENEPOP V3.1 a (Raymond and Rousset 1995a) under the
nuIl hypothesis: "the genotypic distribution is identical across populationsn. An
unbiased estimate of the significance of the probability test was calculated
through a 10,000 step, 1,000 iteration Markov chain permutation (1 0,000
dememorization steps) of a R X C contingency table of genotypic distribution
for each population. The significanœ of the P-values across the six loci was
determined using Fisher's probability combination test (Raymond and Rousset
1996). Fisher's method uses a chi-square test based on the following
equation:
where r = number of loci; Pi = P-value for the j-th locus. This statistic Ws a chi-
square distribution with 2 r degrees of freedom (Sokal and Rohlf 1991).
The parameters FS, FIT. and Fis were introduced by Wright (1 951 ) as a
convenient means of summarizing population structure. These estimators do
not take into account allele size and henœ are more conservative than those
which reflect the step-wise mutation mode1 (See Chapter 1). Weir and
Cockerham (1 984) have developed estimators of F-parameters applicable to the
analysis of multiple alleles and loci:
where a = variance between populations; b = variation between individuals
within populations; and c = variation behveen gametes within individuals (Weir
and Cockerham 1984). ARLEQUIN VI .1 (Schneider et al. 1997) uses these
estimators to construct paiiwise estimates of population structure. Genetic
distance between populations was measured by mlculating Fm and linearized
with population divergence time via a transformation (Linearized Fsr= FsT/(l-
FS); Slatkin 1995). The significanœ of these FsT estimates was tested under
the nuIl hypothesis Ho = "No difference between populations" by permuting
genotypes between populations (1 0.000 iterations). The P-value of the test is
the proportion of permutations leading to a Fm value larger or equal to that
observed (Schneider et al. 1997).
In addition to Fsr estimates, population subdivision was also estimated
through a calculation of &, a Fsranalog which takes into account distance
between alleles (Slatkin 1995). Rsr is defined as:
&=X- s w (3.7) - S
where S= average squared differenœ in allele size between pairs of genes
within population; Sw = average squared differenœ in allele size between pairs
of genes taken from a colledion of P populations (Slatkin 1995). Genetic
distance between populations was measured by caiculating Rsr and transfomed
as for Fm (Slatkin 1995). The significance of k~estimates was tested under the
nuIl hypothesis Ho = "No difference between populationsn by permuting
genotypes between populations (10,000 iterations). The P-value of the test is
the proportion of permutations leading to a Rsr value larger or equal to that
observed (Schneider et al. 1997).
Isolation by distancer
Models of isolation by distanœ offer an empirical means to test dispersal
patterns from measures of population subdivision. Natural populations which fit
a two-dimensional diffusive stepping stone model should exhibit a strong fit to
the isolation by distance model (Rousset 1997). Through simulation study.
Rousset (1997) has found a linear relationship between Fsr/(l-Fsr) and
logarithm of distance for two-dimensional habitats. Linearized pairwise FsT
estimates (ARLEQUIN VI .1 - Schneider et al. 1997; see above) were plotted
versus the natural logarithm of the linear distanœ between populations. In
addition to a computation of the linear regression between these two
parameters, a Mante1 pemutation test (10,000 permutations) was used to test
the nuIl hypothesis Ho = "Genetic distance and geographic distance are
independent" using a rank correlation coefficient calculated from genotype 1
distance matrices.
Anaiysis of Molecular Variance (AMOVA):
In a fumer analysis of the correlation between genetic distance and
topographic characterÎstics of the lower Laurentian Great Lakes, a hierarchical
analysis of molecular variance (AMOVA; Excoffier et al. 1992) was used to
clanfy the parttioning of molecular variance between three levels: variation
among lakes, among populations within a lake. and arnong individuals within
populations. Excoffier et al. (1 992) have developed O-statistics, F-statistic
analogs which reflect the contribution to genetic diversity from each of these
three levels of genetic subdivision. ARLEQU l N VI. 1 (Schneider et al. 1 997)
perfoms an AMOVA, testing the significanœ of the variance cornponents and 0-
statistics via a Mante1 permutation test (10,000 permutations) under the nuIl
hypothesis Ho = "Samples are drawn from a global populationn.
Results:
Mussel Length Data:
In an effort to quantify variation in shell length between sites, length of al1
extracted specimens was rneasured and cornpareci with a sample of a single
year class taken from Canadian Coast Guard buoy ET2 in Long Point Bay (Fig.
3.2). Average mussel length was signifimntly larger than ET2 in 4 out of 8
sampled populations (2-test; p<0.01).
Genetic Data:
Descriptive statistics:
A total of 167 alleles were observed for the six loci over eight populations
(432 individuals). ranging from a low of 4 alleles at Dbug6 to a high of 44 at
Dbugl. Overall allelic frequency data (Appendix 3.2) was plotted as a size-
ranked histogram to clarify the sampling space of the allelic distribution (Fig. 3.3).
The microsatellite markers exhibited a great deal of variation in allelic distribution
with Dbugl and Dbug4 exhibiting a bimodal distribution of alleles, Dbug5
exhibiting an approximately normal distri-bution. and Dbug2.3. and 6 suggesting
a unifom distribution of alleles.
Linkage Diseguilibrium:
Aithough exact tests for genotypic disequilibrium between microsatellite
loci within populations gave 5 significant P-values (X2 test; Pe0.05) out of 130
pairs of loci (3.85%). 7 significant values would be expeded on the basis of type
I error. Global tests of linkage disequilibrium calculated from withinpopulation
data were not significant at the 5% level (2 test) indicating that allelic variation at
al1 six loci is in linkage equilibrium.
Hardy- Weinberg Equilibrium:
Departure from Hardy-Weinberg expectations was measured via a
Markov chain permutation procedure for Dbugl-5 and a direct enurneration
approach for Dbug6. Ail six loci exhibit significant heterozygote deficiency (X2-
test; Pc0.05) in al1 populations. The inbreeding coefficient (Fis) was plotted by
locus for each population (Fig. 3.4), graphically illustrating the departure from
Ha rdy-Wein berg expectations.
Population DMerentiation:
Global X2 tests of population differentiation indicated signifimnt
heterogeneity in gene frequencies among the 8 populations (P<0.000010). In
an effort to further partition these data. paiMse estimates of genotypic
differentiation were calculated (Table 3.2). Twenty-four of twenty-eight pairwise
comparisons (85.7%) were significantly different (X2 test: Pe0.05). Wm the
exception of two comparisons between Lake Erie populations and SF which
suggested that populations were not significantly different, al1 non-signficant
comparisons were among neighbouring populations in Lake Erie.
Pairwise Fsr and Rsr estimates showed similarly high levels of genetic
differentiation between populations (Table 3.3). Twenty-seven out of twenty-
eight population pairs (96.4%) were significantly differentiated on the basis of
FST, while Ru estimates were signfcant in 7 out of 28 (25.0%) pairs. The single
population-pair which did not have a significant Fsr value (CE-EE) was also
undifferentiated by the exact test for population differentiation (P=0.2094; Table
3.2). The higher level of differentiation detected by Fçr compared to RST may
refiect the departure of data from the stepwise mutation model (See Chapter 1).
isolation by distance:
M i l e Fs1 and exact tests of population difïerentiation both suggest
signifiant differenœs behnreen populations, these differenœs are not correlated
with the geographic distanœ between populations (? = 0.0366; F ig. 3.5). A one-
tailed Mantel Test of the correlation between linearized Fçr and geographical
distance was also non-significant (P=0.32061). These results suggest that
diffusive dispersal of larvae may not be the sole mechanism of gene flow
between populations.
Analysis of Molecular Vanance:
Populations were partitioned into two groups on the basis of geographical
relation to the putative point of introduction in eastem Lake Erie (see Fig. 3.1) in
an effort to detenine if the presentday genetic structure has been defined by
historical wlonization events. While sig nacant (P=0.05), only 0.39% of the total
genetic variation can be explained on the basis of these groupings. These data
are consistent with the isolation-by-distance test, suggesting that while significant
difterences do exist, the current population genetic structure of D. bugensis in
North America is independent of its presumed point of introduction.
Discussion:
Previous survey data suggested that D. bugensis experienced a gradua1
diffusion from its initial point of introduction in Lake Erie (Mills et al. 1993). In
contrast, the results of the present population geneüc study indicate that long
distance jump dispersal may be playing a large role in maintaining gene flow
between geographically disjunct populations. Tests of population subdivision
and genotypic differentiation suggest that geographic distance is not the sole
factor goveming population subdivision of the quagga mussel in the Laurentian
Great Lakes.
Mussel length data - Existence of two year classes?
Four out of eight study populations varied significantly from a sample
wllected from a Canadian Coast Guard buoy in Long Point Bay (ET2). As these
buoys are removed every fall and replaced with span to prevent ice damage,
the mussels collected from the buoy represent a single year of settlement.
Mackie and Schloesser (1 996) have found that the relationship between age and
growth of mussels varies considerably based on the quality and quantity of food.
temperature, and body size, making accurate age assessrnent difficult in adult
mussels. Although wave action and food availability presumably dfler
significantly between a benthic i profundal habitat and in the region surrounding
a subsurface float. these ciifferences rnay not entirely explain the observed
variation in mussel size distribution. While study suggests that North Amencan
D. polymorpha individuals have a Iife span of 1.5 to 2 years (Mackie 1991) which
may be truncated by warmer water temperatures (Mackie et al. 1989), similar
study has not been conducteci on D. bugensis. If we assume that quagga
mussel longevity is similar to the zebra mussel, we cannot negate the possibility
that some of the collected specimens of D. bugensis were from the previous
year of settlement, which may have an impact on the inferences based on
microsatellite data.
All of the Western Ontario (WO) and Lake St. Francis (SF) mussels were
significantly larger than those present at the other six sites, with no mussels
found in the size range found on the Coast Guard buoy (8.5 - 14 mm). This
suggests either a much greater resource availability at these sites or a lack of
settlement in the previous year. These resuits are consistent with those of Mills
et al. (1 993), who found larger quagga mussels at sites in Lake Ontario than in
Lake Erie. In wntrast. while the majority of mussels wllected at Long Point Bay
I (LPI) and Eastern Erie (EE) were of similar length. the datasets were skewed
by several large mussels (>20mm), suggesting the presence of two year classes
at these sites.
Genetic Data:
The high variability observed at 5 of 6 microsatellite loci (T= 27.8 I 7.5
alleles) results in a higher power of statistical tests for population
differentiation than is possible using allozyme markers (Estoup et ai. 1998).
However, the overall low frequency of some of these alleles may confound
analyses of population structure which place a hig her significance on rare
alleles (such as the private alleles method; see Slatkin 1985). While 14
putative private alleles were identified (see Appendix 3.2). the low frequency at
which they occur suggests that they may be present in one or more additional
populations at a similar low frequency. In an effort to minimize bias associated
with statistical estimators. Fisher's exact test for population differentiation and
F- and R-statistics were chosen to estimate population structure. Fisher's
exact test and pairwise F- and Restimators weight al1 alleles equally.
Although conclusions bas& on these statistics rnay be conservative. they
minimize the potential bias associated with sampling error. While fve of six
loci have a large number of alleles. Dbug6 has significantly fewer alleles. The
low number of alleles present at Dbug6 (an AAC, repeat locus) is consistent
with the observation of a monomorphic locus at the only other MC,-repeat
locus developed (see Chapter 2). These data suggest a lower allelic diversity
at AAC, loci in O. bugensis.
Qualitative interpretation of the allelic distribution at these six loci
suggests variable mutational modes exist across microsatellites in D. bugensis.
As the quagga mussel has only been present in the Laurentian Great Lakes for
eight years, the likelihood that de-novo mutations have become established in
North Arnerican populations seems extremely unlikely. Although the high level
of genetic variation at 5 of 6 microsatellite loci suggests that the mussel
introduction involved a large number of individuals, the allelic distribution at these
six microsatellite loci rnay not refiect the full spectrum of alleles present in native
Ukrainian populations. This is moût evident at Dbugl and Dbug4. whose
bimodal distribution of allele frequencies suggests a deviation frorn the stepwise
mutation model (SMM). As the presentday allelic distribution rnay be biased by
historical founder effects, statistical bias rnay be introduced when using tests
based on the SMM. In this study, the use of Fm, a more conservative estimator
based on the infinite alleles model (Siatkin 1995). rnay be more appropriate.
Al1 six loci exhibit a significant heterozygote deficiency, suggesting that
non-random mating rnay be occumng in North Arnerican populations of D.
bugensis. Two hypotheses rnay explain this deviation from Hardy-Weinberg
equilibrium. The first hypothesis suggests that an excess of homozygous
individuals rnay refiect the Wahlund effect (Hartl and Clark 1 989). in which a
population which is composed of a mixture of individuals frorn several
subpopulations exhibits a heterozygote deficiency. The deviations from Hardy-
Weinberg equilibrium in this study rnay refiect the present nonequilibrium state
of quagga mussel populations in North Ameriw, as passively dispersed
planktonic larvae continually re-colonize sites based on predominant current
patterns. Even if larvae themselves are the progeny of randomly mating
populations. the settlement of multiple cohorts at a single site rnay lead to
significant departures fram Hardy-Weinberg expectaüons.
Haag and Garton (1995) have suggested an alternative hypothesis for the
heterozygote deficiency observed in dreissenid mussels. Manne invertebrates
with planktonic larvae often exhibl sirnilar low frequencies of heterozygotes
(Singh and Green 1 984; Beaumont 1 991 ). Sing h and Green (1 984) sugçest Lh t
heterozygous larvae may sufter greater mortality due to increased food
requirements associated with higher growth rates, which may lead to an exœss
of homozygotes upon settlernent. Haag and Garton (1995) tested this
hypothesis in D. polymoqha. examining variation in allelic ffequencies at a
single allozyme locus at different life-history stages. Aithough Haag and Garton
(1 995) detected a differenœ between genotype frequencies in larvae and adults,
they found no significant variation in heterozygote frequency over life-history
stages. While the results of Haag and Garton's (1 995) study were limited to a
small number of individuals at a single allozyme locus, their work suggests that
the high heterozygote deficiency I observed at D. bugensis microsatellite loci
may be more reasonably attribiited to admixture of populations than to
differential mortality of larvae.
FsT values for D. bugensis ranged from 0.0033 to 0.0363
(SD =0.0084; Table 3.3). While population genetic data is Iimited for
planktonic dispersers in freshwater systems, allozyme study suggests
comparable estimates of Fm (in the range of 0.004 to 0.015) over a similar
geographic range for marine species with extended planktotrophic
development (Watts et al. 1990; Sarver and Folk 1993, Holbom et al. 1994).
Both FST estimates and tests of population differentiation consistently
demonstrate significant differenœs between 24 of 28 painnrise population
comparisons. While RST estimates indicated that only 7 of 28 population-pairs
were significantly different, these estimates may be biased due to violations of
the stepwise mutational model (see above). This high level of population
differentiation does not correlate with any simple geographic relationship
between populations, suggesting a complex pattern of gene fiow between
populations.
While the isolation-bydistance model is an excellent tool for illuminating
simple relationships between population structure and geography,
1 is limited in that it may be invalidated by as few as a single pairwise
cornparison which deviates from expectaüons (Rousset 1997). In this study, the
two most geographically separated populations (CE & SF) are not statistically
differentiated by the exact test and differenœs as estimated by Fsr are only
marginally significant. This major deviation from the isolation-by-distance model
contributes to the poor overall fd of the genetic data to geography. Although the
AMOVA is more conseivative than the isolation-bydistanœ test, it too suggests
that the majority of the observed genetic variation cannot be explained on the
basis of geography. The resuits of both of these tests indicate a more complex
model of mussel dispersal than the diffusive radiation suggested by survey data.
Although the two most distant populations have a similar allelic
distribution and low Fu, the two Long Point Bay populations (separated by 11
km) are significantly different. These data suggest that while boater
movement patterns may have had a significant impact on the population
structure of the quagga mussel in North Arnerica. non-random dispersal of
larvae via cunents may lead to significant genetic differences over a small
scale. Unfortunately. ecological study of dreissenid dispersal has been
hampered by limitations in Our ability to directly track larval dispersal. It is
unclear how far mussel larvae may disperse in lacustrine environments and if
dispersing larvae aggregate, as has been demonstrated in oysters (Andrews
1979) and crabs (Jillett and Zeldis 1985). A recent study investigated larval
dispersal in a riverine metapopulation of D. polymorpha (Stoeckel et al. 1997).
Based on changes in mean size of larvae as the cohort drif€ed downriver,
Stoeckel et al. (1997) estimated that larvae probably travel more than 300 km
before settlement. While current flows Vary significantly between riverine and
lacustrine environments, these data may help to explain the low levels of gene
flow observed between the two most proximate populations in the present
study, as larvae may travel large distances before settlement. A more detailed
understanding of larval dynamics in lakes is essential to clarify patterns of
settlement which govern the spread of dreissenid species.
Reœnt mussel surveys suggest an increased incidence of quagga
rnossels outside their established North American range. Although aduit quagga
mussels have been found in the Mississippi (Mills et al. 1996), Ohio (Brenœ and
Miller 1994), and Rideau Rivers (A. Martel, Canadian Museum of Nature.
Ottawa, Ontario, personal communication), 1 is unclear if these mussels are the
resuit of local recniitment or 'stray aduiW from boat trafîïc. In seven yearç of
intensive study of the Rideau River, A. Martel (personal communication) has
found only a single quagga mussel near Smith Falls, Ontario (in 1996).
Regardless of whether mussels in these disjunct locations were locally recruited
or introduced as aduits by boat traffic. the increased number of sightings of the
quagga mussel outside the lower Laurentian Great Lakes suggests an
increasing dissemination of D. bugensis.
Dreissena bugensis was first identifed in the offshore zone of Lake Erie
(Mills et al. 1993) and became established quickly. presumably due to a lack of
competition from the zebra mussel. While populations of D. bugensis grew to
dominate the offshore zone (Mills et al. 1993), mussels from these areas were
much less likely to corne into contact with boat trafiic than individuals from zebra
mussel populations in the near-shore zone. Boater-mediated transport of D.
po/ymorpha may help to explain the more widespread distribution of the zebra
mussel in North America.
Claxton (1 998) and Mills et al. (1 998) suggest that D. bugensis may
currentiy be displacing neanhore zebra mussels in Eastern Lake Erie and on the
south shore of Lake Ontario, perhaps due to a faster growth rate of the quagga
musse1 under resourcdimited conditions (Mills et al. 1998). This pattern of
competitive exclusion has also been described in eastem Europe, where D.
bugensis has corne to dominate populations of Dreissena at nearshore sites in
several U krainian reservoirs (Pligin 1984; Mills et al. 1996). Wfi the movement
of quagga mussel populations into the neanhore zone, they are now more Iikely
to be dispersed by boat traffic. This hypothesis is consistent with reœnt mussel
sightings (see above) and the results of the present study which suggests high
levels of gene flow between disjunct populations. Although the quagga mussel
may be limited by environmental factors or competitive exclusion by established
zebra mussel populations. my results suggest that widespread dissemination of
D. bugensis is not only possible. but is presently occumng on a large scale.
Conclusions:
The resuits of this study suggest that jump dispersal, probably mediated
by boater movement patterns, is having a significant impact on quagga musse1
dispersal patterns. M i l e deviations from Hardy-Weinberg equilibrium suggest
that this system has not yet reached an equilibrium state, the increased number
of sightings of quagga mussels outside their established range suggests that,
with the movement of D. bugensis into the nearshore zone, increased transport
by boat traffic may result in large scale mixing of quagga müssels, elininâtkg
population stnicturing of D. bugensis in the lower Laurentian Great Lakes and
increasing the potential for the establishment of new populations outside the
main range. In contrast, significant differenœs were found between populations
separated by short distances, suggesting that non-random dispersal of larvae
may play a role in limiong gene fiow between populations.
Concurrent molecular study on O. poiymorpha (K.A. Naish, unpubfished
data) may provide an "eye to the futuren, illuminating what might be expected
from the quagga mussel in the next 5-10 years. A cornparison of quagga and
zebra mussel data may help clan@ large scale patterns which have lead to the
success of these molluscan invaders.
Bi bliography:
hdrews, J.D. 1979. Pelecypoda: Osteridae. In Reproduction of marine invertebrates. Edited by AC. Giese and J.S. Pearce. Academic Press Inc., New York.
Beaumont, AR. 1991. Genetic studies of laboratory reared mussels, Myüius dulis: Heterozygote deficiencies, heterozygosity, and growth. Biol. J. Linn. Soc. 44: 273-285.
Brence, A, and M. Miller. 1994. Quagga found in the Ohio River. ln Zebra mussel update #21. Edifed by C. Kraft Gopherr //gopher.adp.wisc.edu:80/.bro~~e/~METASGIZW .data/.00000137.
Carlton, J.T. 1993. Dispersal mechanisms of the zebra mussel (Dreissena polymorpha), ln Zebra mussels: Biology, impacts, and control. Edited by T.F. Nalepa and D. Schloesser. Lewis Publishers, Boca Raton, Florida. pp. 677-697.
Claxton. W.C.. Martel. A, Demott. R.M., and E.G. Boulding. 1997. Discrimination of field- collecteci juveniles of two in troduced d reissenids (Drer'ssena polymorpha and Dreissena bugensis) using rnitochondrial DNA and shell morphology. Can. J. Fish. Aquat Sci. 54: 1280-1288.
Claxton, W.T. 1998. Molecular sy~ te rn~cs and ecology of deep and shallow water populations of Dreisena polymorpha and Dreisena bugensis. Ph.D. thesis, University of Guelph, Guelph, Ont
Estoup, A, Rousset, F., Michalakis, Y., Comuet J-M, Adriamanga, M., and R. Guyomard. 1998. Comparative analysis of microsatellite and allozyme markers: A case study investigating microgeographic differentation in brown trout (Salmo tmtta). Mol. Ewl. 7: 339-353.
Excoffier, L., Smouse, P.E., and J.M. Quattro. 1992. Analysis of molecular variance inferred from metric distances among DNA haplotypes: Application to human m itochondrial DNA restriction data. Genetics 131 : 479-491.
Gamier-Gere, P., and C. Dillmann. 1992. A cornputer program for testing pairwise linkage disequilibria in subdivided populations. J. Heredity. 83: 239.
Goudet J., Raymond, M., de Meeus. T., and F. Rousset 1996. Testing differentiation in diploid populations. Genetics 144: 1933-1 940.
Haag, W.R., and D.W. Garton. 1995. Variation in genotype frequencies during the life cycle of the bivalve, Dreissena polymorpha. Evolution 49: 1284-1 288.
Hartl, D L . and AG. Clark 1989. Principles of population genetics. Sinauer Associates Inc., Sunderland, Mas.
Hastings, A 1996. Models of spatial spread: Is the aieory complete? Ecology ï7: 1675-1 679.
Hebert, P.D.N., Muncaster, B.W., and G.L Mackie. 1989. Ecological and genetic studies on Dreissena polymorpha (Pallas): A new mollusc in the Great Lakes reg ion. Can. J. Fish. Aquat Sci. 46: 1587-1 591.
Holbom, K, Johnson, M.S., and R. Black. 1994. Population genetics of the corallivorous gastropod Drupella cornus at Ningaloo Reef, Western Australia. Coral Reefs 13: 33-39.
Jillett, J.B. and J.R. Zeldis. 1985. Aerial observations of surface patchiness of a planktonic crustacean. Bull. Mar. Sci. 37: 609-619.
Johnson, L.E., and J.T. Cariton. 1996. Post establishment spread in large-scale invasions: Dis persal mechanisrns of the zebra mussel Dreissena polymorpha. Ecology 77: 1 686- 1690.
Mackie, G.L. 1991. Biology of the exotic zebra mussel, Dreissena pulymorpha, in relation to native bivalves and its potential impact in Lake St Clair- In Environmental assessrnent and habitat evaluation of the upper Great Lakes connedng channels. Edited by M. Munawar and T. Edsall. Hydrobiologia 219: 251-268,
Mackie, G.L., Gibbons, W-N., Muncaster, B.W., and I.M. Gray. 1989. The zebta mussel, Dmkena polymorpha: A synthesis of European experiences and a preview for North America. Report prepared for Water Resources Branch, Great Lakes Section. Available tom Queen's Printer for Ontario, ISBN 0-7729-5647-2.
Mackie, G.L., and D.W. Schloesser. 19%. Comparative biology of zebra mussels in Europe and North America: An overview. Am. Zool. 36: 244-258.
Mills, E-L, Derrnott, R.M., Roseman, E.F., Dustin, O., Mellina, E., Conn D.B., and AP. Spidle. 1993. Colonization, ecology, and population structure of the "quagga" mussel (Bivalvia: Dreissenidae) in the lower Great Lakes. Can. J. Fish. Aquat Sci. 50: 2305-2314.
Mills, E L , Rosenberg, G., Spidle, AP., Ludyanskiy, M., Pligin, Y., and 8. May. 1996. A review of the biology and ecology of the quagga mussel (Dreisena bugensis), a second species of freshwater dreissenid introduced to North Arnerica. Am. 2001. 36: 271-286.
Mills. E.L., Chrisman, J.R., Baldwin, B.S., Owens, R.W., OIGonnan, R., Howell, T., Roseman, E.F., and M.K Raths. 1998. Recent changes in the dreissenid community and a shift toward the quagga mussel in Lake Ontario. 41st Conference of the International Assocation of Great Lakes Research, Hamilton, Ont., May 1998. p. 1 18 (Abstract only).
Pligin, Y.V. 1984. Extension of the distribution of Dreissena bugensis. Malacol. Rev. 17: 143-144.
Raymond, M., and F. Rousset 1995a. GENEPOP (version 1.2): Population genetics software for exact tests and ecumenicism. J. Heredtty 83: 239.
Raymond, M., and F. Rousset: 1995b. An exact test for population differentiation. Evolution 49: 1280-1283.
Rousset, F. 1997. Genetic differentiation and estimation of gene flow frorn F-statistics under isolation by distance. Genetics 145: 121 9-1 228.
Rousset, F., and M. Raymond. 1995. Testing heterozygote excess and deficiency. Genetics i40: 141 3-141 9.
Ricciardi, A, Serrouya, R., F.G. Whonskey. 1995. Aerial exposure tolerance of zebra and quagga mussels (Bivalvia: Dreissenidae): Implications for overland dispersal. Can. J. Fish, Aquat Sci. 52: 470-477.
Sarver, S.K., and D.W. Foltz. 1993. Genetic population structure of a species' wmplex of blue mussels (Mytilus spp.). Mar. Biol. Ili: 105-1 12.
Schneider, S., Kueffer, J-Ml Roessli, D., and L. Excoffier. 1997. Ariequin V I -1 : A software for population genetic analysis. Genetics and Biometry Laboratory, University of Geneva, Switzerland.
Singh, S.M., and R.H. Green. 1984. Excess of allozyme homozygosity in marine molluscs and its possible biological significance. MaIacologia 25: 569-581.
Slatkin, M. 1985. Rare alleles as indicators of gene flow. Evoiution 39: 53-65.
Slatkin. M. 1995. A measure of population subdivision based on microsatellite allele frequencies. Genetics i 39: 457-462.
Sokal, R.R. and F.J. Rohlf. 1981. Biometry, 2d ed. W.H. Freeman & Co., San Francisco.
Stoeckel, J.A., Schneider, D.W., Soeken, L.A, Blodgett, KD., and R.E. Sparks. 1997. L a ~ a l dynamics of a riverine metapopulation: Implications for zebra mussel recruitrnent. dispersal, and control in a large-river system. J. NA. Benth. Soc. 16: 586-601.
Watts, R.J., Johnson, M.S., and R. Black. 1990. Effects of recruitment on genetic patchiness in the urch in Echinometra mafhaei in Western Australia. Mar. Biol. 105: 1 45-1 5 1 .
Weir, B.S., and C.C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38: 1358-1 370.
Wright, S. 1951. The genetical structure of populations. Ann. Eugen. 15: 323-354.
Table 3.1. Dreissena bugensis sampling site locations.
Sampling Site Coordinates Depth(m) Sampling Technique Date
Central Erie (CE) 41°57.000' N 10.0 Ponar Grab (2) 1 0/07/97 81°37.470' W
Long Point Bay 1 (LPI) 42O42.960' N 8.4 Ponar Grab (3) 06/24/97 80°15.084' W
Long Point Bay Il (LPII) 42O46.290' N 8.4 Ponar Grab (6) 06122197 80°08.460' W
Eastern Erie (EE) 42O30.394' N 63.0 Ponar Grab (2) 06/25/97 79O53.798' W
Western Ontario (WO) 43O13.471' N 14.0 Ponar Grab (3) 0611 8197 79O16.356' W
Rochester (RC) 43O24.500' N 1 .O Shoreline 12/07/97 77O53.000' W
Eastern Ontario (EO) 44O09.020' N 22.0 Ponar Grab (3) IO129197 76O38.000' W
Lake St. Francis (SF) 45O07.140' N 15.0 SCUBA 74°21.010' W
Table 3.2. Fisher's exact test for genotypic dïfferentiation - Pairwise population comparisons. Locus by locus pairwise ?-w!iies w r e computed by GENEPOP V3.1 a (Raymond and Rousset 1995a) and pooled using Fisher's chi-square method for combining probabilities (Sokal and Rohlf 1981). Global tests indicated mat al1 loci were highly significantly differentiated among the eight populations (P~0.0001+0.0000).
Population - Population X2 P-Value
CE-LPI CE-LPII CE-EE CE-WO CE-RC CE-EO CE-SF LPI-LPII LPI-EE LPI-WO LPI-RC LPI-EO LPI-SF LPI 1-EE LPI 1-WO LPI 1-RC LPI 1-wo LPI 1-SF EE-WO EE-RC EE-€0 EE-SF WO-RC WO-EO WO-SF RC-EO RC-SF EO-SF 36.600 0.0003* ' = Allelic distributions are significantly different ( X 2 test Pe0.05; df = 12)
Table 3.3. Painvise linearized Fst (upper right) and Rst (lower left) estimates (Slatkin 1995) for eight study populations calculated by ARLEQUIN VI . 1 (Schneider et al. 1997).
Pop"
CE
LPI
LPll
EE
WO
RC
€0
SF
LPI
0.0059*
LPll
0.0089*
0.01 le*
- 0.0000
0.01 85*
0.0000
0.0162*
0.0037
= Populations are significantly different (Permutation test; ~ ~ 0 . 0 5 ) .
Table 3.4. Analysis of molecular variance (AMOVA) calculated by ARLEQUIN V I .1 (Schneider et al. 1997). Populations were divided into two groups on the basis of location in relation to putative point of introduction (Mills et al. 1993).
Source of Variation d.f. Variance Components
Percentage of Variation
Among groups 1 0.00950 Va 0.39*
Among populations 6 0.02526 V b
within groups
Among individuals 424 0,86084 VC within populations
Within individuals 432 1,55556 Vd 63.46*
Group 1: Central Erie, Long Point Bay 1, Long Point Bay II, Eastern Erie Group 2: Western Ontario, Rochester, Eastern Ontario, Lake St. Francis
* = Variance component is significant (Permutation test; P<=0.05)
Fis 1
Fis 1 . -
CE LPI LPll EE WO RC EO SF Population
Fis I
CE LPI LPll EE WO RC EO SF Population
Fis 1
CE LPI LPI1 EE WO RC €0 SF Population
Fis 1
CE LPI LPll EE WO RC EO SF Population
Fis 1
CE LPI LPll EE WO RC €0 SF CE LPI LPll EE WO RC EO SF
Population Population
Fig. 3.4. Hardy-Weinberg test for each locus in each population calculated by GENEPOP V3.l a (Raymond and Rousset 1995a). * indicates a significant heterozygote deficiency (Pc0.05). All loci exhibit a significant departure from H-W equilibrium (P<0.00001).
Ln (distance) (km) Fig. 3.5. Test of isolation-by-distance rnodel. Linearized Fst calculated by Arlequin VI . 1 (Schneider et al. 1997) vs.natural logarithm of linear distance (km).
LPll LPll LPII LPll L 27 26 25 24
Appendix l(a): Dbugl - Gel photograph and idealized schematic diagram. Allelic state indicated for each individual. Arrow indicates sequenced plasmid band (wl allele number), L = allelic ladder.
LPI I LPll LPll L
Appendix l(c): Dbug3 - Gel photograph and idealized schematic diagram. Allelic state indicated for each individual. Arrow indicates sequenced plasrnid band (wl allele number), L = allelic ladder.
0204 - - LPI 50
LPI LPI 49 48
LPI L 47
Appendix l(d): Dbug4 - Gel photograph and idealized schematic diagram. Allelic state indicated for each individual. Arrow indicates sequenced plasmid band (wl allele number), L = allelic ladder.
~ 0 & 0 0 0 0 ~ 0 ( Y o m 0 0 0 - 0 s s a s s s s q
0 0 0 0 0 0 0 0
e 4 O O N ~ 0 < 3 0 0
" g q q s s g g g 0 0 0 0 0 0 0 0
" f f g % J g % 3 0 0 0 0 0 0 0 0
O O O O O O r O t N O C Y O q q q g S = g g
0 0 0 0 0 0 0 0
0 0 0 - m m 0 0 r F & g 2 X m o " "
9 9 9 9 0 0 9 0 0 0 0 0
= r o - o r < 3 0 l -
x s s s s g 9 5 0 0 0 0 0 0 0 0
S - O N h 0 0 0 = xsoz%ss, 0 0 0 0 0 0 0 0
œ o 0 0 o C ) O o C Y r w g a o m o m m 8e ig%%X%8 i D 0 0 0 0 0 0 0 0 r - 0 0 0 0 0 0 yx?Xgxx * 0 0 0 u 3 0 D r o r r - 0 0 q o o g 3 6 8 8
0 0 0 0 0 0 0 0
2 N 0 8 = = z % = q8999qq9 0 0 0 0 0 0 0 0
N 0 0 - 0 ( Y V ) 0 0 r r O I N 0 m o q q q q q 8 6
0 0 0 0 0 0 0 0
r - o o r m m o c u r u 3 0 4 0 m qqq*$eq;
0 0 0 0 0 0 0 0
o m 0 ( 3 m ( D m o r rwr- - -Cymu3m
.9*rrrq? 0 0 0 0 0 0 0 0
O - O F O * > I O O ~ h O - - O m O I C > 9 r 9 9 9 r 9 7 0 0 0 0 0 0 0 0
W < 3 O ( P N F ( D O m 0 < 3 ( 3 = r w m 0 - . r r c u - ' ? ' ? 0 0 0 0 0 0 0 0
b588880" f Z 9 9 9 9 4 9 4 9 0 0 0 0 0 0 0 0
< D m O O Q Q u ) O O I
4?4$$?88 0 0 0 0 0 0 0 0
- o o g s n z s s 8 8 , - 2 8 8 2 8 4 0 0 r m r i n o r
8Ô636:88 0 0 0 0 0 0 0 0
( 3 0 0 1 0 N C 3 e ) O F N - h O m 0 t - m 9 9 9 r 9 9 9 9 0 0 0 0 0 0 0 0
N - O O . & 8 g O o
a~3~~g~ m p r o o r o - N x = p 5 9 9 9 9 0 9 9
o o o o o e o o
i a a f f w q O w w o - o o u r o w ~ o i ~ ~ w u r
IMAGE EVALUATlON TEST TARGET (QA-3)
APPLIED IMAGE. lnc 1653 East Main Street - -. , , Rochester, NY 14609 USA -- --= Phone: 71 6/42-0300 -- --= Fax: 71 6/288-5989
O 1993. Applied Image. Inc. AU Rlghts fbsarvetl