Population genetics and phylogeography of freshwater mussels in North America, Elliptio dilatata and Actinonaias ligamentina (Bivalvia: Unionidae).

Abstract

Extrinsic and intrinsic forces combined shape the population structure of every species differently. Freshwater mussels are obligate parasites to a host fish during a juvenile stage (glochidia). Elliptio dilatata (ED) and Actinonaias ligamentina (AL) are co-occurring freshwater mussel taxa with similar North American distribution and share some potential host fish. Using mitochondrial DNA, we determined the genotypes of 190 + individuals from collection sites in at least two tributaries in the Lake Erie and Ohio River watersheds, along with the Ouachita and Strawberry rivers in the southeast. Both species had followed a stepping-stone model of dispersal, with greater pairwise genetic structure among collection sites of ED. Also, phylogeographical analysis for ED found significant geographical structuring of haplotype diversity. Overall, within-population variation increased significantly from north to south, with low genetic diversity in the Strawberry River. We calculated significant among-population structure for both species (ED: Phi(ST) = 0.62, P < 0.001; AL: Phi(ST) = 0.16, P < 0.001). Genetic analysis identified the Ouachita River as an area of significant reproductive isolation for both species. Results for AL indicated dispersal into northern areas from two genetically distinct glacial refugia, where results for ED indicated dispersal followed by low gene flow in northern areas. The conservation strategies for mussels that co-occur in the same 'bed' could be species specific. Species such as ED have management units on the population scale, where AL has a more homogeneous genetic structure across its range.

Full text

Page 1

Molecular Ecology (2008) 17, 2149–2163doi: 10.1111/j.1365-294X.2008.03745.x
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Blackwell Publishing Ltd
Population genetics and phylogeography of freshwater
mussels in North America, Elliptio dilatata and Actinonaias
ligamentina (Bivalvia: Unionidae)
CURT L. ELDERKIN,* ALAN D. CHRISTIAN,* § JANICE L. METCALFE-SMITH† and DAVID J. BERG‡
*Department of Zoology, Miami University, Oxford, OH 45056, USA, †National Water Research Institute, Burlington, Ontario,
Canada L7R 4A6, ‡Department of Zoology, Miami University, Hamilton, OH 45011, USA
Abstract
Extrinsic and intrinsic forces combined shape the population structure of every species
differently. Freshwater mussels are obligate parasites to a host fish during a juvenile stage
(glochidia). Elliptio dilatata (ED) and Actinonaias ligamentina (AL) are co-occurring freshwater
mussel taxa with similar North American distribution and share some potential host fish.
Using mitochondrial DNA, we determined the genotypes of 190 + individuals from collection
sites in at least two tributaries in the Lake Erie and Ohio River watersheds, along with the
Ouachita and Strawberry rivers in the southeast. Both species had followed a stepping-stone
model of dispersal, with greater pairwise genetic structure among collection sites of ED.
Also, phylogeographical analysis for ED found significant geographical structuring of
haplotype diversity. Overall, within-population variation increased significantly from north
to south, with low genetic diversity in the Strawberry River. We calculated significant among-
population structure for both species (ED: ΦST= 0.62, P < 0.001; AL: ΦST= 0.16, P < 0.001).
Genetic analysis identified the Ouachita River as an area of significant reproductive isolation
for both species. Results for AL indicated dispersal into northern areas from two genetically
distinct glacial refugia, where results for ED indicated dispersal followed by low gene flow
in northern areas. The conservation strategies for mussels that co-occur in the same ‘bed’
could be species specific. Species such as ED have management units on the population
scale, where AL has a more homogeneous genetic structure across its range.
Keywords: COI, genetic diversity, mtDNA, phylogeography, population genetics, postglacial
dispersal, Unionidae
Received 2 November 2007; revision received 6 February 2008; accepted 11 February 2008
Introduction
Relationships among populations of freshwater organisms
have been formed by major geological changes in aquatic
systems (Bernatchez & Wilson 1998). In North America, this
relationship of genetic and geological changes has been
explored extensively on fish species. The major event which
has affected genetic diversity and structure among aquatic
organisms was the climate change during the Pleistocene
(Near et al. 2001). During this time, before 15 000–8000 bp,
several ice sheets proceeded south, rendering northern areas
uninhabitable. Widespread aquatic species would maintain
populations in southern areas; however, northern taxa was
forced south (Bernatchez & Wilson 1998). From previous work
on fish biogeography, we know that fish followed multiple
dispersal routes into the formerly glaciated northern areas
following the global warming at the end of the Pleistocene
(Mandrak & Crossman 1992). Interestingly, evidence from
aquatic organisms indicates that these climate changing
events may have facilitated the dispersal following the
receding ice sheet in Europe (Freeland et al. 2004). Also, in
North American species, climate change during the Pleis-
tocene is hypothesized to be responsible for rapid genetic
change in upland fish taxa in eastern and central North
Correspondence: Curt L. Elderkin, Department of Biology, The
College of New Jersey, Ewing, NJ 08628, USA. Fax: 609-637-5118;
E-mail: elderkin@tcnj.edu
§Present address: Department of Biological Sciences, Arkansas
State University, State University, AR 72467, USA.

Page 2

2150 C. L. ELDERKIN ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
America (Mayden 1988; Near et al. 2001; Berendzen et al.
2003; Near & Keck 2005; Ray et al. 2006). However, species
that inhabit bottom land habitat appear to have the opposite
effects from this event, where little population structure has
been found across the entire Mississippi River watershed
(Nedbal & Philipp 1994; Avise 2004).
Genetic relationships within and among populations
are not only formed by historic geological events, but also
affected by intrinsic effects (Wares & Turner 2003), including
habitat preferences and dispersal abilities among habitats
(Bohonak 1999; Bilton et al. 2001; Berg et al. 2007). Dispersal
ability appears to be a major determinant of among-
population genetic structure of freshwater organisms
(Bilton et al. 2001; Miller et al. 2002). Certain fish, particularly
those typically found in large bottomland rivers, may
disperse on the order of tens or hundreds of kilometres
(Lyons & Kanehl 2002; Vokoun & Rabeni 2005) where upland
fishes, such as darters may disperse less than 1 km or on the
order of metres (Scalet 1973; Freeman 1995). There is no
other organism whose dispersal or evolutionary history is
more closely allied to freshwater fish than the bivalve family
Unionidae. Most members of this family rely on a host fish
for dispersal during a parasitic juvenile stage known as a
glochidia (McMahon 1991).
Mussel biologists have primarily used mitochondrial DNA
(mtDNA) to determine taxonomic status among mussels
before their extinction (Lydeard et al. 1996; Mulvey et al. 1997;
Lydeard & Roe 1998; Roe & Lydeard 1998; Serb et al. 2003;
Grobler et al. 2006) and population numbers are declining
for most species (Williams et al. 1993). In addition, mtDNA
has been used to determine the genetic structure of several
endangered species (King et al. 1999; Roe et al. 2001; Jones
et al. 2006; Zannatta & Murphy 2007), yet very few non-
endangered species have been studied (Turner et al. 2000;
Elderkin et al. 2007). Before using mtDNA, studies using
allozymes were successful for determining genetic structure
of other mussel taxa (Mulvey et al. 1997; Berg et al. 1998;
Curole et al. 2004). Similar to fish studies, mussel taxa have
demonstrated genetic structure congruent with the historic
isolation/connectivity among the drainages where they
were collected (Mulvey et al. 1997; Roe et al. 2001). Conversely,
other studies have found little genetic structure on large
geographical scales in a formerly widespread endangered
species (Monroe et al. 2007). Most recently, studies have
found relatively low genetic structure among populations
of mussels native to the Mississippi and Great Lakes water-
sheds (Elderkin etal. 2007) and for common species in Europe
(Curole et al. 2004). Based on these recent studies, mussel
taxa may have very different genetic structure, yet they
often co-occur (mussel habituate in multispecies ‘beds’) in
the same habitats.
Fish phylogeography shows that climate change affected
genetic structure; however, intrinsic effects such as dispersal
and habitat preferences also have immense impacts. Among
freshwater mussels, little attention has been focused on the
effect of climate changes during the Pleistocene (Elderkin
et al. 2007) or specific dispersal abilities and how these affect
population genetic structure (Berg et al. 2007). It appears
from preliminary evidence that although mussels share
similar drainage distributions and geological history, these
events combined with intraspecific behaviour may yield very
different patterns of genetic diversity among co-occurring
mussel taxa. Specifically, these data may enhance our under-
standing of mussel vagility, historic distribution, and
population densities.
The current study involves Elliptio dilatata (ED), the
spike, and Actinonaias ligamentina (AL), the mucket, which
are both common, widespread species in the Mississippi
and Great Lakes watersheds (Williams et al. 1993). These
two species often co-occur in rivers and streams throughout
North America (Cummings & Mayer 1992; Hoggarth
1992; Hoggarth 1995–96; Matthews 2007); however, many
times ED occurs at much smaller densities (Hoggarth
1995–96; Matthews 2007). Both species are obligate gill
parasites, and have similar potential host fish (Cummings
& Watters 2002). However, the genetic diversity of these
species has never been studied over a large portion of
their range, encompassing habitat both above and below
the proposed glacial maximum. We determine the amount
of genetic variation and phylogeographical relationships
among populations of these two widespread mussel taxa
to establish if these codistributed species have similar
genetic diversity. We quantify genetic structure across
their range to determine if extrinsic forces (i.e. glaciation
of the upper Mississippi River and Great Lake watersheds)
and intrinsic effects, such as dispersal ability/population
size, had a similar effect on two species of mussel taxa.
We conclude by briefly considering the implications of
our findings for the development of future conservation
strategies.
Methods
Locations — ED
Mussels were collected from rivers in three distinct drainages
(Fig. 1). From the Lake Erie drainage, two populations
each were collected in the Sydenham, Grand, and Speed
Rivers of Ontario, Canada (Metcalfe-Smith et al. 1998) and
two populations on the St. Joseph River. Ohio River drainage
samples were taken from two tributaries, two populations
each from the Big Darby and the Little Miami Rivers.
Southern Mississippi River collection sites were located in
two major tributaries, the Red River of the South and the
White River. Two separate collection sites were in the
Ouachita River, which flows into the Red River. In addition,
two separate collection sites were sampled in the Strawberry
River, which flows into the White River.

Page 3

POPULATION GENETICS AND PHYLOGEOGRAPHY OF FRESHWATER MUSSELS 2151
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Locations — AL
Mussels were collected from a similar distribution across
three drainages (Fig. 2). From the Lake Erie drainage, two
populations each were sampled in the Sydenham and
Thames rivers of Ontario, Canada (Metcalfe-Smith et al.
1998) and the Sandusky River. Ohio River samples were
collected from two tributaries, two populations each from
the Walhonding and the Licking rivers. From the Southern
Mississippi River drainage, collections were taken from
one tributary in the Red River, two populations on the
Ouachita River.
Sample collection
Individuals were collected by sight or feel, with the aid of
snorkel or SCUBA. At each site, mussels were found 1–2
miles (up or down stream) from the access point. Mussels
were collected regardless of age or size; they were only
selected by species. At each site, mussels were carefully
opened and the mantle was nondestructively sampled using
a mantle biopsy (Berg et al. 1995). Tissue was flash frozen in
liquid nitrogen and each individual was returned to the
location collected. Tissue was shipped to Oxford, Ohio and
stored at – 80 °C until processed. We used global positioning
system (GPS) latitude and longitude coordinates to mark
the geographical position of each sampling location (Table 1).
Sequence data
Individuals from both species were treated similarly for
the DNA analysis. A fragment of the cytochrome oxidase
subunit I (COI) was amplified using polymerase chain
reaction (PCR). Amplification, sequencing, and alignment
followed procedures in Elderkin et al. (2007).
Genetic diversity/phylogeographical analysis
Geographical distances were calculated using natural
dispersal routes along river channels. For each location,
GPS coordinates were entered into arcview digital mapping
system (Environmental Systems Research Institute). Along-
river distances were calculated by tracing the river courses
connecting sample sites. Most collection sites in this study
were connected through the main channel of the Mississippi
River. A theoretical connection was established to the
Canadian populations from the Wabash River, overland
into the St. Joseph River (in the vicinity of Fort Wayne,
Indiana), the Maumee River, and into Lake Erie. This route
is based on historic information outlined in Graf (2002),
where geological evidence suggests that the Wabash River
and Lake Erie were briefly connected after the Pleistocene
glaciers receded.
We used tcs 1.21 to obtain a minimum-spanning tree
of all individuals for both species (Clement et al. 2000).
Fig. 1 Map of the central USA and Canada
with collection sites of Elliptio dilatata sampled
for this study. Population abbreviations
follow those in Table 1. (1) Ouachita River
(OUA 1–2); (2) Strawberry River (STR 1–2);
(3) Little Miami River (LMI 1–2); (4) Big
Darby Creek (BDC 1–2); (5) St. Joseph River
(STJ 1–2); (6) Sydenham River (SYD 1–2);
(7) Grand River (GRR 1–2). Pie charts
represent haplotype group frequencies
within each sampling location; group colour
is consistent with Fig. 3.

Page 4

2152 C. L. ELDERKIN ET AL.
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Table 1 Descriptive statistics by collection site for both freshwater mussel species. Collection sites follow Figs 1 and 2. Bold indicates significance at P < 0.05.
SpeciesSite Name ID no.LatittudeLongitude
n
No. of
haplotypes
Location spec.
haplotypes
Mean no. of
base-pair difference
Nucleotide
diversity (π) Tajima’s D
Mismatch
no. of Modes
Mismatch
variance
Elliptio dilatata
Sydenham 2
Sydenham
Grand
Speed
St. Joeseph
St. Joeseph 2
Darby
Darby 2
Little Miami
EF Little Miami
Strawberry 2
Strawberry
Ouachita 2
Ouachita
SYD2
SYD1
GRR1
GRR2
STJ1
STJ2
BDC1
BDC2
LMI1
LMI2
STR2
STR1
OUA2
OUA1
42.9750
42.6041
43.6305
43.4541
41.7533
41.5201
40.0746
39.9507
39.7817
39.0596
36.1490
36.0909
34.3525
34.2252
−81.9710
−82.0720
−80.4780
−80.2950
−84.6649
−84.6930
−83.4036
−83.8083
−83.8738
−84.1778
−91.7003
−91.5548
−92.8763
−92.9700
16
19
13
11
10
17
6
14
13
15
10
18
15
15
6
5
5
4
3
7
2
4
2
5
2
2
2
2
2
1
1
3
0
1
1
4
0
0
9
4
1.9
1.3
0.7
0.5
0.7
0.8
0.3
0.8
0
1
0.4
0.4
3.8
6.5
1.96
1.31
0.74
0.55
0.67
1.09
0.33
0.78
0.15
0.99
0.36
0.37
4.20
6.50
−1.55
0.43
−1.44
−1.60
−0.18
−1.30
−0.93
−0.53
−1.15
−0.62
0.01
0.49
−0.97
0.60
2
1
1
1
1
1
1
1
1
1
1
1
2
2
6.81
0.79
0.43
0.36
0.36
0.64
0.24
0.42
0.13
0.49
0.23
0.23
17.50
23.92
11
7
Actinonaias ligamentina
Sydenham
Sydemham 2
Thames 2
Thames 6
Sandusky
Sandusky 2
Walhonding
Walhonding 2
Licking
Licking 2
Ouachita 1
Ouachita 2
SYD1
SYD2
THM2
THM6
SAN1
SAN2
WAL1
WAL2
LIK1
LIK2
OUA1
OUA2
42.6040
42.8500
42.7080
42.9870
41.2649
41.2385
40.3500
40.3300
38.7913
38.1801
34.3090
37.1966
−82.0720
−81.8250
−81.6170
−81.0710
−83.1682
−83.1439
−81.9085
−82.1446
−82.4034
−83.6064
−92.9125
−93.0331
17
15
14
14
15
17
21
16
32
15
17
16
62
7
2
1
5
4
3
4
5
6
3
3.5
2.2
2.8
2.1
1.7
2.6
4.1
3.4
4.1
4.3
6.1
2.97
3.53
2.84
2.18
2.15
1.71
2.57
4.12
3.30
4.12
4.29
6.11
−1.80
−1.31
−2.19
−1.46
−1.40
−1.77
−1.40
−0.58
−0.38
−1.20
−1.36
−0.77
2
2
3
2
2
2
2
2
2
2
1
2
10.49
6.62
9.07
5.21
6.25
4.36
8.69
8.73
7.15
15.13
5.71
13.59
11
9
6
6
8
8
7
12
9
15
14
11
10

Page 5

POPULATION GENETICS AND PHYLOGEOGRAPHY OF FRESHWATER MUSSELS 2153
© 2008 The Authors
Journal compilation © 2008 Blackwell Publishing Ltd
Ambiguity in the network (loops or circular pathways) was
resolved using a coalescence model as outlined in Fetzner
& Crandall (2003). For both species, we calculated a 95%
root probability. Nodes in the network were compared to
collection locations to determine if some haplotypes were
clustered in certain geographical locations. The tree was
labelled by regional groups based on where the 95% majority
of haplotypes within a group were found geographically.
Next, we calculated the number of haplotypes within and
the number of haplotypes unique to a particular collection
site and tested for correlation to latitude using linear
regression analysis.
Population genetics
We used arlequin 3.1 (Schneider et al. 2000) for analysis of
molecular variation. This software takes into account both
haplotype frequency and molecular sequence divergence.
We calculated mean nucleotide diversity (π) within a collection
site. We also calculated mismatch distributions, and Tajima’s
D for each collection site (1000 replicates). These estimates
were tested for a significant correlation with latitude using
linear regression analysis. Mismatch distributions also
indicate rapid demographic expansion, where we would
expect mismatch diversity to follow a Poisson distribution
(Slatkin & Hudson 1991; Rogers & Harpending 1992). We
calculated genetic variation partitioned into a maximum
of three hierarchical levels [always with collection site
(population) as the smallest group] using analysis of
molecular variance (amova, Excoffier et al. 1992). First, we
tested for structure among rivers: within populations (ΦST),
among populations within rivers (ΦSR), and among rivers
(ΦRT). We then tested for structure among drainages: within
populations (ΦST), among populations within drainages
(ΦSD), and among drainages (ΦDT). Significance of the Φ
statistics was tested using 1000 nonparametric permutations
(Weir & Cockerham 1984). We also calculated the pairwise
FST values for each combination of populations. The
significance of this estimate was calculated using 1000
permutations of the data and was considered significant at
P = 0.05. We compared the pairwise population structure
estimates to a matrix of geographical distance using a
Mantel test.
We used additional procedures to determine major
patterns of genetic structure across the ranges sampled for
both species. First, we applied samova 1.0 (Dupanloup
et al. 2002) to identify groups of collection sites that were
maximally differentiated at COI. We performed this test
using k = 2 through k = 8 partitions of the sampling area
with 100 simulated annealing steps within each analysis.
We then examined the number of partitions that maximized
genetic differentiation among groups (ΦCT), using the pre-
diction that collection sites in reproductively isolated regions
would be assigned into the same group and maximize ΦCT.
Fig. 2 Map of the central USA and Canada
with collection sites of Actinonaias ligamentina
sampled for this study. Population abbrevia-
tions follow those in the text and Table 1.
(1) Ouachita River (OUA 1–2); (2) Licking
River (LIK 1–2); (3) Walhonding River
(WAL 1–2); (4) Sandusky River (SAN 1–2);
(5) Sydenham River (SYD 1–2); (6) Thames
River (THM 2, 6). Pie charts represent
haplotype group frequencies within each
sampling location; group colour is consistent
with Fig. 4.