Full text of DOI:10.1016/S0380-1330(01)70642-9

J. Great Lakes Res. 27(3):267–280
Internat. Assoc. Great Lakes Res., 2001
Genetic and Biogeographic Relationships of the Invasive Round
(Neogobius melanostomus) and Tubenose (Proterorhinus marmoratus)
Gobies in the Great Lakes Versus Eurasian Populations
Alison K. Dillon1 and Carol A. Stepien*
Great Lakes Environmental Genetics Laboratory
Center for Environmental Science, Technology and Policy
Cleveland State University
1899 East 22nd Street, MC 219
Cleveland, Ohio 44114-4435
ABSTRACT. Population genetic structure and systematic relationships were investigated for two exotic
fishes introduced to the Great Lakes in 1990, the round goby Neogobius melanostomus and the tubenose
goby Proterorhinus marmoratus, using DNA sequences from the left domain of the mitochondrial DNA
control region. Samples of round gobies were compared from different sites in the Great Lakes, an intro-
duced population from the Gulf of Gdansk in Poland, and a native population from the northern Black
Sea. The round goby was characterized by relatively high genetic variability, and 17 haplotypes were
identified from 64 individuals. Levels of genetic variation for the round goby were similar in the invasive
and native sampling sites, suggesting relatively large founding populations and lack of bottlenecks. The
northern Black Sea was eliminated as a probable founding source for both the Great Lakes and the exotic
population in Poland. Substitutions in the left domain of the control region revealed significant differ-
ences among samples from the Great Lakes and Eurasia, and between Lakes Erie and St. Clair, suggest-
ing non-random mating. No variation was detected in the tubenose goby population in the Great Lakes,
which has been less successful in terms of spread and population growth. A molecular clock calibration
suggested that the genera Neogobius and Proterorhinus diverged about 5.2 ± 1.0 million years ago,
apparently separating from a common ancestor shared with Gobius during the isolation of the Paratethys
basin from the Mediterranean Tethys Sea.
INDEX WORDS: DNA, exotic species, founder effect, genetics, Gobiidae, Great Lakes, Neogobius
melanostomus, Proterorhinus marmoratus.
INTRODUCTION
patterns, and founding sources for exotic species
may help in understanding the common factors reg-
ulating their relative successes.
The repeated successes of nonindigenous species
in the Great Lakes have had serious ecological im-
pacts, resulting in many permanent changes to the
ecosystem (Mills et al. 1993, Ricciardi and
MacIsaac 2000). Levels of genetic variability in
populations of exotic species may be positively cor-
related with invasive success (Williamson 1996),
which was examined in the present study and
through other investigations by our laboratory
(Stepien et al. 1998, Stepien and Dillon in press).
Comparisons of levels of genetic variability, spread
In 1990, the Eurasian round Neogobius melanos-
tomus and tubenose Proterorhinus marmoratus gob-
ies (Teleostei: Gobiidae) were discovered in the St.
Clair River (Fig. 1) of the Laurentian Great Lakes
(Crossman et al. 1992, Jude et al. 1992), where
they presumably had been introduced from foreign
ballast water discharge (Mills et al. 1993). The
round goby has spread to all five Great Lakes—
faster than any previously introduced fish and its
population sizes are large in the lower Great Lakes
(Fig. 1, Jude et al. 1992, Marsden et al. 1997). The
tubenose goby has not spread to other locations as
rapidly, and presently reaches northwestern Lake
Erie (Fig. 1, U.S. Geological Survey 2001). This
*Corresponding author and all reprint requests: E-mail: c.stepien@csuo-
hio.edu
1Present Address: Athersys, Inc., 3201 Carnegie Ave., Cleveland, Ohio
44115
267

268
Dillon and Stepien
FIG. 1. North American distribution and collection sites for Neogobius
melanostomus and Proterorhinus marmoratus. Sites for N. melanosto-
mus are 1 = Shiawassee River, near Linden, Michigan, 2 = Flint River.,
near Russelville, Michigan, 3 = Lake St. Clair, 5 = Lake Erie, off Avon
Point, Ohio, and 6 = Lake Erie, off the Chagrin River, Ohio. Proterorhi-
nus marmoratus was collected at sites 3 = Lake St. Clair and 4 = St. Clair
River. (Distribution information is from Biological Resources Division,
USGS 2001).
study primarily concentrated on the round goby,
since it now presents a greater ecological problem.
(Simonovic et al. 1996). The neogobiins have fused
pelvic fins and elongated dorsal and anal fins
(Miller 1986) and are distinguished from the At-
lantic and Mediterranean basin Gobius spp. by the
absence of a swimbladder and location of the up-
permost rays of the pectoral fins within the fin
membrane (Pinchuk 1991). The Neogobiinae is hy-
pothesized to have descended from an ancestor
Systematic and Biogeographic Relationships
The genera Neogobius and Proterorhinus com-
prise the subfamily Neogobiinae, which is endemic
to the Ponto-Caspian region (including the Mar-
mara, Black, Azov, Caspian, and Aral seas, Fig. 2)

Population Genetics of Invasive Gobies
269
FIG. 2. European collection sites for Neogobius melanostomus are numbered
7 and 8. 7 = Gulf of Gdansk, Poland, and 8 = Black Sea, near Varna, Bulgaria.
The Ponto-Caspian region includes the Marmara, Black, Azov, Caspian, and
Aral seas.
shared with Gobius, which either entered the Black
Sea basin during Quaternary interglacial connec-
tions with the Mediterranean Sea (Stevanovic 1982)
or earlier during the mid-Miocene Epoch (McKay
and Miller 1991). The monotypic Proterorhinus is
distinguished from Neogobius by its unique long
anterior nostrils (Miller 1986). The native distribu-
tion of P. marmoratus is limited to the Black and
Caspian seas, where it is threatened by anthro-
pogenic habitat disturbance (Lelek 1989).
A previous DNA study by Dougherty et al.
(1996) sequenced 365 base pairs (bp) of the mito-
chondrial (mt) DNA cytochrome b gene from 44
round gobies and two tubenose gobies finding 63
nucleotide differences between the species (pn =
0.173) and four intraspecific polymorphisms in the
former (pn = 0.011). Cytochrome b evolves rela-
tively slowly in fishes (Kocher et al. 1989, Stepien
and Kocher 1997) and the Dougherty et al. (1996)
study did not discern enough variation to evaluate
differences among sample sites. Dougherty et al.
(1996) found five haplotypes (one of which was
shared among sites), each differing by a single nu-
cleotide, for round gobies in the Black Sea (n = 15,
n = 3 haplotypes) and the St. Clair and Detroit
rivers (n = 29, n = 3 haplotypes, one shared with
the Black Sea). They concluded that more than one
female round goby type was introduced to the
Great Lakes basin and excluded the northern Black
Sea as the founding source (Dougherty et al.
1996). The Dougherty et al. (1996) samples were
made available in this study, allowing a direct com-
parison of the levels of resolution obtained from
the mtDNA control region and cytochrome b se-
quences.

270
Dillon and Stepien
Objectives
(site 4, 42° 45′ N, 82° 25′ W, n = 11) and Lake St.
Clair (site 3, n = 1) in the Great Lakes (Fig. 1).
Samples were stored frozen at -80°C or in 95%
ethanol at room temperature.
The purpose of this investigation was to geneti-
cally characterize the exotic round and tubenose
goby populations in the Great Lakes using se-
quences from the left domain of the mtDNA control
region. Specific objectives were to: (1) evaluate the
possible origin(s) of the founding population(s), (2)
compare levels of genetic variation in round gobies
from native versus introduced areas of their range,
(3) contrast the relative levels and distribution of
intra- versus interspecific genetic variation, and (4)
provide a data base for evaluating whether future
colonization sites are due to spread from nearby lo-
cations or new introductions from Eurasia.
Hypotheses for the above objectives (1 to 4)
were: (1, A) the exotic round goby population in
the North American Great Lakes did/did not origi-
nate from the northern Black Sea, (1, B) the exotic
population in the Gulf of Gdansk did/did not origi-
nate from the northern Black Sea, (1, C) the two in-
troduced populations did/did not share a common
origin, (2, A) levels of genetic variability were sim-
ilar/different in the exotic populations versus the
native population sample, (2, B) levels of genetic
variability were similar/different between the two
introduced populations, (3, A) levels of genetic
variability were similar/different between the round
and tubenose gobies, (3, B) significant differences
occurred/ did not occur among sampling locations
of the round goby in North America. Results of this
investigation will aid understanding of the relative
genetic variability and patterns of dispersal of the
gobies, in comparison with other exotic species.
DNA Extraction, Amplification,
and Sequencing
Methodology for extracting and purifying DNA
from the samples followed Stepien (1995). The left
domain of the control region, extending from the
tRNA-proline gene to the central conserved section
(Fig. 3), was amplified using the polymerase chain
reaction (PCR) with the oligonucleotide primers
L16007 (5′-CCC-AAA-GCT-AAA-ATT-CTA-A-3′,
Kocher et al. 1989) and H16498 (5′-CCT-GAA-
GTA-GGA-ACC-AGA-TG-3′, Meyer et al. 1990)
and then sequenced according to Stepien (1995).
Data Analysis
Proportions of polymorphic nucleotides (pn) were
calculated according to Nei (1987). Population ge-
netic analyses were conducted in two ways: (1) by
analyzing evolutionary relationships among haplo-
type sequences using genetic distance and cladistic
phylogenetic trees, and (2) by comparing levels of
genetic variability within and among sampling
sites, and testing for geographic heterogeneity (Fer-
ris and Berg 1987, Weir 1996).
The first approach included maximum parsimony
in PAUP* 4.0 (Phylogenetic Analysis Using Parsi-
mony, Swofford 2001) and neighbor-joining (NJ)
genetic distance analyses (Saitou and Nei 1987) for
651 bp (bases 90 to 740). Parsimony analyses em-
ployed the branch-and-bound algorithm (Hendy and
Penny 1982) and 50% majority rule consensus of
the most parsimonious trees (Margush and McMor-
ris 1981). Kimura’s (1980) two-parameter and pair-
wise (p-) genetic distances (Nei 1987), their
standard errors, and neighbor-joining (NJ) trees
were calculated with MEGA (Molecular Evolution-
ary Genetics Analysis, vers. 1.01, Kumar et al.
1993). Kimura’s (1980) two-parameter distances
were used to correct for the unequal rate of transi-
tional versus transversional substitutions that occur
in animal mtDNA sequences (Kocher and Carlton
1997). Pairwise distances were used to estimate
possible divergence times, using a hypothesized
rate of sequence divergence of 2% per million years
(reviewed by Avise 2000). This rate is slower than
that of mammals (Avise 2000) but average for
fishes, whose mtDNA evolves more slowly appar-
MATERIALS AND METHODS
Samples Collected
The left domain (~750 bp) of the mtDNA control
region was sequenced for N. melanostomus from
five sites in the Great Lakes (Fig. 1), including: two
tributaries of Saginaw Bay, Lake Huron; the Shi-
awassee River (site 1, 45° 15′ N, 84° 05′ W, n = 5)
and the Flint River (site 2, 43° 00′ N, 83° 45′ W, n
= 5); Lake St. Clair (site 3, 42° 22′ N, 82° 39′ W, n
= 10); and two sites in Lake Erie, Avon Point (site
5, 41° 50′ N, 82° 02′ W, n = 10) and the mouth of
the Chagrin River (site 6, 41° 44′ N, 81° 25′ W,
n = 10). Eurasian sampling locations (Fig. 2) in-
cluded the Gulf of Gdansk, Poland (site 7, 54° 20′
N, 18° 40′ E, n = 20) and the Black Sea, Bulgaria
(site 8, 43° 14′ N, 27° 58′ E, n = 15). Twelve P.
marmoratus were analyzed from the St. Clair River

Population Genetics of Invasive Gobies
271
FIG. 3. A schematic diagram of the left domain of the control region for (A)
Neogobius melanostomus and (B) Proterorhinus marmoratus. Locations of the
repeated sequences, short repeat (SR), termination associated sequences (TAS),
and the central conserved section (CCS) are shown. Vertical lines: A = intraspe-
cific substitutions, B = interspecific differences.
vealed similar results and to compare their relative
levels of phylogenetic resolution.
ently due to poikilothermy (Bernatchez et al. 1992,
Bernatchez and Danzmann 1993, Stepien and
Kocher 1997). Bootstrap support values were deter-
mined from 1,000 permutations for the nodes of the
distance and parsimony trees.
Frequency differences in the geographic distribu-
tions of the haplotypes also were analyzed with
modified chi-square tests, with a Monte Carlo simu-
lation approach with 10,000 randomizations (Roff
and Bentzen 1989) and the MONTE program in
REAP (McElroy et al. 1992). A Bonferroni correc-
tion that divided the P-value by the number of pair-
wise comparisons was used for the multiple
post-hoc tests (Fry et al. 1993).
The second approach compared the relative fre-
quencies of the haplotypes and their geographic di-
vergences, using haplotype diversity (h, which is
equivalent to heterozygosity, Nei and Tajima 1983)
and maximum-likelihood analysis within (nu-
∏
cleotide diversity— , Nei 1987) and among sam-
Hierarchical AMOVA (Analysis of MOlecular
VAriance, vers. 1.53, Excoffier et al. 1992, Ex-
coffier 1995) analyses, based on Euclidean dis-
tances among pairs of haplotypes and their relative
frequencies, tested the partitioning of variation be-
tween regions (North America and Europe), and
among sampling sites. Significance of the ϕST (FST
analog) values was determined with 1,000 random
permutations (Excoffier et al. 1992).
Tajima’s (1989) test of selective neutrality based
on the infinite-site model was performed using the
Arlequin software package (vers. 1.1, Schneider et
al. 1997). This test evaluated whether the observed
number of polymorphic sites and the average num-
pling locations (nucleotide divergence—dxy, Nei
and Tajima 1983) from the DA2 program in REAP
(the Restriction Enzyme Analysis Package, vers.
5.0, McElroy et al. 1992).
Phylogenetic utility of the repeated sequence ar-
rays versus the unique sequences (see Faber and
Stepien 1998) were compared through separate
analyses of: (1) the entire data set (17 haplotypes),
(2) the data subset of the repeated arrays (11 haplo-
types for nucleotide positions 090 to 595), and (3)
the data subset comprising the unique sequences
(11 haplotypes for nucleotide positions 000 to 089
and 596 to 733). The purpose was to test whether
the repeated and the unique sequence regions re-

272
Dillon and Stepien
ber of pairwise nucleotide differences deviated sig-
nificantly from theoretically predicted values
(Tajima 1989).
teen transitions (S), four transversions (V), and a
two-base insertion occurred within the four tandem
repeats (S:V ratio = 3.25, Table 1).
The unique sequences differed by eight transi-
tions (S), two transversions (V), and a single inser-
tion (S:V ratio = 4.0). Most of this variation (10 of
11 polymorphic sites) occurred in the Eurasian sam-
ples (Table 2). Two of the transitions (bases 575
and 588) occurred in single individuals sampled
from the Black Sea (Table 2). Nucleotide position
671 was fixed in both of the Eurasian populations
(G), and a transversion (53% T - 47% G) distin-
guished some North American samples.
The complete data set, including repeated and
non-repeated information, revealed 17 round goby
haplotypes and 43 polymorphic nucleotide sites
(pn = 0.07, S:V ratio = 13.0; Table 2, Fig. 3). Nu-
cleotide polymorphism within the four repeats (pn =
0.08) was higher than in the non-repetitive regions
(pn = 0.05). Intrasite polymorphisms were detected
in five of the seven sampling locations (the Shi-
awassee and Flint rivers of Lake Huron were
monomorphic). Nucleotide polymorphism (pn),
haplotype diversity (h), and nucleotide diversity
(Π) values (Table 3) were largest in the native pop-
ulation from the Black Sea (pn = 0.05, h = 0.96 ±
0.04) and the Lake St. Clair sampling site (Π =
0.010 ± 0.007). Haplotype diversity (h) averaged
0.46 ± 0.02 overall. Nucleotide divergences (dxy)
between pairs of sampling sites ranged from 0.0002
(between Poland and the Black Sea) to 0.0185 (be-
tween Poland and Lake Huron). Within the three
Great Lakes, dxy was smallest between Lakes
Huron and St. Clair (0.0009), intermediate between
Lakes Erie and St. Clair (0.0015), and larger be-
tween Lakes Erie and Huron (0.0060).
RESULTS
Structure and Substitutions in the Left Domain
of the mtDNA Control Region
A single haplotype discerned for P. marmoratus
and the five most common N. melanostomus haplo-
types (aa, cc, dd, ee, and gg) were reported in Gen-
Bank (accession numbers AF082969 to AF082974).
No intraspecific variation was found in P. mar-
moratus (n = 12).
Locations of the inter- and intraspecific substitu-
tions were mapped on a structural diagram of the
left domain for the mtDNA control region (Fig. 3).
Two putative Termination Associated Sequences
(TAS, Doda et al. 1981, Foran et al. 1988) were
identified in each species. The region immediately
flanking the 5′ tRNA-proline end (bases 1 to 89)
differed in sequence and length between the
species, and was omitted in all analyses due to
alignment uncertainty. The left domain of N.
melanostomus was approximately 400 bp longer
than that of P. marmoratus, resulting from four non-
identical tandemly repeated sequences (Table 1,
Fig. 3). Faber and Stepien 1998 discuss tandem re-
peats in the mtDNA control region of fishes. A fifth
shortened version (26 bp) of the repeated sequence
immediately followed, and also occurred in P. mar-
moratus (Fig. 3).
Interspecific Differences
The species N. melanostomus and P. marmoratus
differed by 26 transitional (S) and 23 transversional
(V) substitutions within 306 bp of aligned sequence
data (excluding the four repeats, S:V ratio = 1.13).
They also diverged by nine insertion/deletions en-
compassing a total of 50 nucleotide positions. Mean
genetic distances between the two species were
Kimura (1980) two parameter d = 0.112 ± 0.024
and pairwise p = 0.104 ± 0.020.
Variance components and associated ϕ statistics
showed significant hierarchial geographic partition-
ing of genetic variation (Table 4, Excoffier et al.
1995). Variation between the North American and
Eurasian regions was similar to that found among
sampling sites within regions. Modified chi-square
tests (Roff and Bentzen 1989, McElroy et al. 1992)
also revealed significant differences in the distribu-
tions of the haplotypes between North America and
Eurasia and among the five sampling sites overall
(Table 5). Pairwise comparisons showed a signifi-
cant difference between samples from Lakes Erie
and St. Clair.
Intraspecific Variation in the
Round Goby N. melanostomus
Round goby haplotypes varied in both the re-
peated array (Table 1) and the unique sequences of
the mtDNA control region (Table 2). Sequence sim-
ilarity among the repeated variants (lettered A
through K) alone ranged from 87% to 99%. Thir-
Tajima’s (1989) test supported selective neutral-
ity (D = –0.23, P = 0.43) when all samples of round
gobies were grouped as a single population. When
the sample sites were analyzed separately, selective

Population Genetics of Invasive Gobies
273

274
Dillon and Stepien

Population Genetics of Invasive Gobies
275
TABLE 3. Nucleotide polymorphism (pn), haplotype diversity (h ± SE), and
nucleotide diversity (Π ± SE) of Neogobius melanostomus within sampling sites
(REAP vers. 2.0, McElroy et al. 1992)
Sampling location
pn
h ± SE
Π ± SE
Avon Point
Chagrin River
Shiawassee River
Flint River
Lake St. Clair
Poland
0.018
0.002
0.000
0.000
0.025
0.014
0.046
0.644 ± 0.152
0.356 ± 0.159
0.000 ± 0.000
0.000 ± 0.000
0.844 ± 0.080
0.564 ± 0.134
0.962 ± 0.041
0.007 ± 0.005
0.001 ± 0.001
0.000 ± 0.000
0.000 ± 0.000
0.012 ± 0.007
0.004 ± 0.003
0.009 ± 0.004
Black Sea
TABLE 4. Hierarchical analyses of molecular variance among haplotypes of N. melanostomus
(AMOVA, Excoffier et al. 1992). * = significant. (A) Values based on the entire sequence data, (B) Values
using haplotypes from the repeated sequence region, (C) Values using haplotypes from the nonrepeated
sequences.
Variance component
Variance
% Total variance
Φ statistic
P
(A) Entire Data Set
2
Between regions
Among sampling sites
Within sampling sites
Φa2
1.04
0.55
0.96
40.7%
21.7%
37.6%
Φct = 0.41
< 0.001*
< 0.001*
< 0.001*
Φb2
Φc
Φsc = 0.37
Φst = 0.62
(B) Haplotypes from the tandem repeat region
2
Between regions
Among sampling sites
Within sampling sites
Φa2
Φb2
Φc
0.47
0.38
0.55
33.8%
27.2%
39.0%
Φct = 0.34
< 0.001*
< 0.001*
< 0.001*
Φsc = 0.41
Φst = 0.61
(C) Haplotypes from the unique sequences
2
Between regions
Among sampling sites
Within sampling sites
Φa2
0.56
0.10
0.51
48.0%
08.8%
43.2%
Φct = 0.48
< 0.001*
< 0.001*
< 0.001*
Φb2
Φc
Φ
sc = 0.17
Φst = 0.57
TABLE 5. Modified chi-square values from Monte Carlo tests (Roff and
Bentzen 1989) for the distribution of mtDNA control region sequence haplotypes
of N. melanostomus (REAP vers. 2.0, McElroy et al. 1992). * = significant,
using a Bonferroni correction for multiple post-hoc tests (Fry et al. 1993).
Locations Compared
χ2
P
Between regions
59.7
< 0.001*
(North America and Eurasia)
Among five sampling sites
159.9
< 0.001*
(two Eurasian populations and
Lakes Erie, Huron, and St. Clair)
Among Lakes Erie, Huron, and St. Clair
Between Lakes Erie and St. Clair
Between Lakes Erie and Huron
32.9
23.6
16.7
5.0
< 0.001*
< 0.001*
< 0.001*
0.198
Between Lakes St. Clair and Huron

276
Dillon and Stepien
FIG. 4. Neighbor-joining tree for haplotypes of Neogobius melanostomus using Kimura
(1980) two-parameter distances from the computer program MEGA 1.01 (Kumar et al.
1993). Genetic distances may be calculated by adding horizontal branch lengths between
pairs. Percentages indicate bootstrap support for nodes computed from 1,000 replicates.
The major types present in the Great Lakes are labeled 1–4, and the primary groups are
denoted as I and II (see Results). This tree was congruent with results from maximum
parsimony analysis using PAUP* 4.0 (Swofford 2001).
neutrality also was supported in four of the five lo-
cations (Lake Erie, D = –0.16, P = 0.46; Lake
Huron, D = 2.42, P = 0.99; Lake St. Clair, D = 1.45,
P = 0.91; and Poland, D = 0.27, P = 0.61). There
was a deviation in the sample from the Black Sea
(D = –1.80, P = 0.02).
Eurasian types (90% exclusively Eurasian haplo-
types and 92% of the Eurasian individuals, desig-
nated as II; Fig. 4). Group I contained three major
types, including two from Lakes Erie and St. Clair
(type 1, haplotypes aa and bb; type 3, haplotype ee)
and another found in Lakes Erie, Huron, and St.
Clair (type 2, haplotypes cc and dd). One Black Sea
haplotype (qq) clustered with the North American
type 3 haplotypes. Type 4 in Lake St. Clair also was
identified in the Black Sea, and was the sister taxon
to a cluster of exclusively Eurasian haplotypes.
Genetic distances among the haplotypes overall
ranged from d = 0.002 to 0.042, averaging 0.014 ±
0.005. Haplotypes in group I (Fig. 4) were sepa-
rated by an average genetic d = 0.009 ± 0.004, and
group II by d = 0.006 ± 0.003. The four types iden-
tified in the North American population (Fig. 4) di-
verged by a mean d = 0.011 ± 0.004. Maximum
parsimony analysis with PAUP* 4.0 (Swofford
Genetic Divergences
Kimura’s (1980) two-parameter and p-distances
among round goby haplotypes were similar, with
overlapping standard errors. The NJ tree (Fig. 4)
showed that haplotype ll from the Black Sea was
basal to all other round gobies, due to its unique re-
peat sequences (Tables 1 and 2). The remaining
haplotypes formed two groups, one containing the
majority of the North American haplotypes (83% of
the haplotypes and 97% of the individuals, desig-
nated as I), and the other containing mainly

Population Genetics of Invasive Gobies
277
2000) found 12 most parsimonious trees, having
497 steps and consistency indices (excluding unin-
formative characters) of 0.86. The 50% majority
rule consensus tree was congruent with the NJ tree
(Fig. 4).
or a less diverse source population. The Gulf of
Gdansk population was identified in 1990 (the same
year as the appearance in North America) and pre-
sumably was introduced through ballast water dis-
charge from ships originating in the Black or
Caspian Seas (Skora and Stolarski 1993).
Only one mtDNA haplotype was shared between
the samples from Lake St. Clair and the Black Sea,
indicating that the North American population did
not originate from this region of the northern Black
Sea. This corroborates the results of Dougherty et
al. (1996), who likewise found little similarity be-
tween samples from the Great Lakes and the north-
ern Black Sea. In this study, only a single haplotype
(not found in the Great Lakes) was shared between
the Black Sea and the Gulf of Gdansk population,
indicating that the Black Sea also was not its found-
ing source. The exotic populations in the Gulf of
Gdansk and the Great Lakes apparently were
founded by independent sources, since they shared
no haplotypes.
The presence of four divergent groupings of hap-
lotypes in the North American Great Lakes sug-
gests that the introduction was founded by a large
number of individuals from one or more unknown
source populations. Estimates suggest the haplotype
groups differentiated during the mid-
Pleistocene, presumably due to geographic isolation
in Eurasian glacial refugia (Hewitt 1996, 2000;
Stepien et al. 1998). Other freshwater fishes in Eu-
rope exhibit similar levels and patterns of genetic
divergences related to glacial refugia (Bernatchez
and Osinov 1995, Haenfling and Brandl 1998,
Stepien et al. 1998). Differences among sites in the
Great Lakes are consistent with the possibilities of
multiple founding sources, low gene flow among
sites, and differential lineage extinction.
All mtDNA haplotypes from Lakes Erie and
Huron also were present in the Lake St. Clair site,
except for one (haplotype bb, n = 3, Lake Erie). The
presence of a unique haplotype in Lake Erie may
have been the result of an independent introduction
or to sampling error. Five of the six Great Lakes
haplotypes were found in Lake St. Clair fish, sug-
gesting that populations in the other sites were the
result of intra- and inter-lake spread. Sampling lo-
cations in Lakes Erie and Huron (presumably areas
of spread) had fewer haplotypes and lower haplo-
typic diversity, possibly due to founder effects. The
two tributaries of Saginaw Bay, Lake Huron each
contained a single haplotype, suggesting that they
were founded by different genetic types or experi-
enced differential lineage extinction. It has been hy-
DISCUSSION
Interspecific Divergences and Biogeography
It has been hypothesized that the Gobius (repre-
sented by the outgroup in this study) and the neogo-
biin (Proterorhinus and Neogobius) lineages
diverged due to the isolation of the Paratethys from
the Mediterranean Tethys Sea basins, about 10 to
12 million years ago (Steininger and Rogl 1984).
Congruent with this hypothesis, genetic divergences
in this study suggest that the neogobiins diverged
from the Gobius lineage during the mid-Miocene,
and then speciated into Neogobius and Proterorhi-
nus during the late Miocene/early Pliocene Epochs
(an average sequence p-distance of 0.104 ± 0.020,
corresponding to an estimated 5.2 ± 1.0 million
years). These results refute the hypothesis that the
neogobiins diverged from Gobius later during Qua-
ternary interglacial connections between the basins
of the Black and Mediterranean seas (Stevanovic
1982).
Population Divergences and Invasion
Biogeography
Polymorphisms in the mtDNA control region of
the round goby fit a selective neutrality hypothesis.
A single deviation found in the Black Sea popula-
tion may be due to sampling error and small sample
size, lack of random mating, or selection (Tajima
1989). Levels of haplotypic and nucleotide diver-
sity appeared similar overall in samples from Lake
St. Clair (the North American founding population)
and a native population from the Black Sea, indicat-
ing that the round goby introduction may have been
relatively large and did not undergo a genetic bot-
tleneck (a marked reduction in population size).
Dougherty et al. (1996) also found that samples
from the Great Lakes and the Black Sea housed
similar levels of genetic diversity in the mtDNA cy-
tochrome b gene—although that gene was less vari-
able overall than the control region. In this study,
the Lake St. Clair sample was less variable in the
unique sequences and more diverse in the repeated
array, compared with the Black Sea population. In
contrast, the exotic population found in the Gulf of
Gdansk of the Baltic Sea, Poland was less geneti-
cally variable, suggesting a possible founder effect

278
Dillon and Stepien
pothesized that round gobies from Lake Huron trib-
utaries (Flint and Shiawasse rivers) may have origi-
nated from transport of bait fish from Lake St. Clair
(Jude 2000). The results of this study were consis-
tent with the hypothesis of separate bait-fish intro-
ductions (Jude 2000), as both genetic types were
found in Lake St. Clair.
additional introductions occur and evaluate their fu-
ture patterns of spread.
ACKNOWLEDGMENTS
This study was supported by the Ohio Sea Grant
Program development fund #RF-729753 (1997–98),
and the Lake Erie Protection Fund #SG 50-97
(1997–98) to C. Stepien. This project was part of
the Environmental Risk Analysis Program at Cleve-
land State University’s Center for Environmental
Science, Technology and Policy. Specimens were
provided by the Michigan Department of Natural
Resources (B. Haas and M. Thomas), the Ohio Di-
vision of Wildlife (C. Knight and K. Kayle), D.
Jude, J. Ram, S. Rudnicka, P. Simonovich, and M.
Sapoto.
In contrast to the cytochrome b study by
Dougherty et al. (1996), mtDNA control region se-
quences revealed greater variability and significant
differences among sampling locations of the round
goby. Both the repeated and nonrepeated mtDNA
control region sequences revealed phylogeographic
utility and yielded congruent relationships. This
study found significant differences among samples
from Lakes Erie and St. Clair, which may suggest
barriers to gene flow or non-random mating due to
localized spawning sites. Lack of a pelagic larval
stage (Marsden et al. 1997) may limit gene flow
among sites by limiting dispersal opportunities.
Male round gobies are aggressive, actively recruit
mates, and defend the nest (Jude 1997). Reports in-
dicate that males reproduce only once at a large size
(about age 3), do not feed during nest-guarding, and
then may die (Marsden et al. 1997). These life his-
tory features indicate highly selective mate choice
and substantial fitness investment per reproduction.
These characteristics also would tend to preserve
local population genetic differences and limit gene
flow over geographic distances.
REFERENCES
Avise, J.C. 2000. Phylogeography: The history and for-
mation of species. Cambridge: Harvard Univ. Press.
Bernatchez, L., and Danzmann, R.G. 1993. Intraspecific
DNA sequence variation of the mitochondrial DNA of
brook char (Salvelinus fontinalis Mitchill). Mol. Biol.
Evol. 10:1002–1014.
, and Osinov, A. 1995. Genetic diversity of trout
———
(genus Salmo) from its most eastern native range
based on mitochondrial DNA and nuclear gene varia-
tion. Mol. Ecol. 4:285–297.
, Guyomard, R., and Bonhomme, F. 1992. DNA
———
sequence variation of the mitochondrial control region
among geographically and morphologically remote
European brown trout Salmo trutta populations. Mol.
Ecol. 1:161–173.
All tubenose gobies surveyed from the St. Clair
region were a single haplotype, suggesting that the
source population had low genetic variability or the
founding population was small. Investigations of
native Eurasian tubenose goby populations are nec-
essary to address this question.
Crossman, E.J., Holm, E., Cholmondeley, R., and
Tuininga, K. 1992. First record for Canada of the
rudd, Scardinius erythrophthalmus, and notes on the
introduced round goby, Neogobius melanostomus. The
Canadian Field-Naturalist 106:206–209.
Doda, J.N., Wright, C.T., and Clayton, D.A. 1981. Elon-
gation of displacement-loop strands in human and
mouse mitochondrial DNA is arrested near specific
template sequences. Proc. Nat. Acad. Sci. USA
78:6116–6120.
Dougherty, J.D., Moore, W.S., and Ram, J.L. 1996.
Mitochondrial DNA analysis of round goby (Neogob-
ius melanostomus) and tubenose goby (Proterorhinus
marmoratus) in the Great Lakes basin. Can. J. Fish.
Aquat. Sci. 53:474–480.
Excoffier, L. 1995. AMOVA, Analysis of molecular vari-
ance, vers. 1.55. Geneva, Switzerland: Univ. of
Geneva.
The round goby has undergone a more rapid ex-
pansion in the Great Lakes and has a higher level of
genetic variability than the tubenose goby. Differ-
ences in relative success between the two species
also may be related to the larger size and more ag-
gressive nature of the round goby (Jude et al. 1995,
Marsden et al. 1997). Expansion of the round goby
population may be augmented by the abundance of
its zebra mussel prey, for which there are few com-
petitors (Jude 1997). The smaller tubenose goby
primarily consumes smaller prey, including aquatic
insects and benthic zooplankton, for which there are
more native competitors (Jude et al. 1995). This
study provides a data base that will allow popula-
tions of the round and tubenose gobies to be geneti-
cally monitored, in order to elucidate whether
, Smouse, P.E., and Quattro, J.M. 1992. Analysis
———
of molecular variance inferred from metric distances

Population Genetics of Invasive Gobies
279
among DNA haplotypes: application to human mito-
chondrial DNA data. Genetics 131:479–491.
Faber, J.E., and Stepien, C.A. 1998. Tandemly repeated
sequences in the mitochondrial DNA control region
and phylogeography of the pike-perches Stizostedion.
Mol. Phyl. Evol. 10:310–322.
Ferris, S.D., and Berg, W.J. 1987. The utility of mito-
chondrial DNA in fish genetics and fishery manage-
ment. In Population genetics and fishery
management, eds. N. Ryman and F. Utter, pp.
277–299. Seattle, WA: Univ. of Washington Press.
Foran, D.R., Hixson, J.E., and Brown, W.M. 1988. Com-
parison of ape and human sequences that regulate
mitochondrial DNA transcription and D-loop DNA
synthesis. Nucleic Acids Res. 16:5841–5861.
Paabo, S., Villablanca, F.X., and Wilson, A.C. 1989.
Dynamics of mitochondrial DNA evolution in ani-
mals: amplification and sequencing with conserved
primers. Proc. Nat. Acad. Sci. USA 86:6196–6200.
Kumar, S., Tamura, K., and Nei, M. 1993. MEGA: Mole-
cular Evolutionary Genetics Analysis, version 1.02.
University Park, PA: Institute of Molecular Evolu-
tionary Genetics, Pennsylvania State University.
Lelek, A. 1989. Threatened fishes of Europe, volume 9.
In The Freshwater Fishes of Europe, eds. Council of
Europe: European Committee for the conservation of
nature and natural resources. Wiesbaden: AULA Ver-
lag.
Margush, T., and McMorris, F.R. 1981. Consensus n-
trees. Bull. Math. Biol. 42:239–244.
Fry, J.C., Iles, T.C., Randerson, P.F., Bridge, P.D., Dun-
stan, F.D.J., Bowker, D.W., and Weingert, R.G. 1993.
Biological data analysis: a practical approach. New
York, NY: Oxford Univ. Press.
Haenfling, B., and Brandl, R. 1998. Genetic differentia-
tion of the bullhead Cottus gobio L. across watersheds
in Central Europe: evidence for two taxa. Heredity
80:110–117.
Hendy, M.D., and Penny, D. 1982. Branch and bound
algorithms to determine minimal evolutionary trees.
Math. Biosci. 59:277–290.
Hewitt, G.M. 1996. Some genetic consequences of Ice
Ages, and their role in divergence and speciation.
Biol. J. Linn. Soc. 58:247–27.
Marsden, J.E., Charlebois, P., Wolfe, K., Jude, D., and
Rudnicka, S. 1997. The round goby (Neogobius
melanostomus): a review of European and North
American literature with notes from the Round Goby
Conference, Chicago. Illinois Natural History Survey.
Special Pub. No. 20, Champaign, IL.
McElroy, D.P., Moran, P., Bermingham, P.E., and Korn-
field, I. 1992. REAP: An integrated environment for
the manipulation and phylogenetic analysis of restric-
tion data. J. Hered. 83:157–158.
McKay, S.I., and Miller, P.J. 1991. Isozyme criteria in
the testing of phyletic relationships between species
of Gobius and related eastern Atlantic-Mediterranean
genera (Teleostei: Gobiidae). J. Fish Biol. 39 (Suppl.
A):291–299.
Meyer, A., Kocher, T.D., Basatbwaki, P., and Wilson,
A.C. 1990. Monophyletic origin of Lake Victoria
cichlid fishes suggested by mitochondrial DNA
sequences. Nature 347:550–553.
Miller, P.J. 1986. Gobiidae, p. 1019–1095. In Fishes of
the Northeast Atlantic and Mediterranean, eds. P.J. P.
Whitehead, M.L. Bauchot, J.C. Huread, J. Nielsen,
and E. Tortonese. Paris: UNESCO.
Mills, E.L., Leach, J.H., Carlton, J.T., and Secor, C.L.
1993. Exotic species in the Great Lakes: a history of
biotic crises and anthropogenic introductions. J. Great
Lakes Res. 19:1–54.
. 2000. The genetic legacy of the Quaternary ice
ages. Nature 405:907–915.
———
Jude, D.J. 1997. Round gobies: Cyberfish of the third
millennium. Great Lakes Res. Rev. 3:27–34.
. 2000. Round and tubenose gobies: 10 years with
———
the latest Great Lakes phantom menace. Dreissena!
The Digest of the National Aquatic Nuisance Clear-
inghouse 11(4): 1–13.
, Reider, R.H., and Smith, R. 1992. Establishment
of Gobiidae in the Great Lakes basin. Can. J. Fish.
Aquat. Sci. 49:416–421.
———
, Janssen, J., and Crawford, G. 1995. Ecology,
———
distribution, and impact of the newly introduced
round and tubenose gobies in the biota of the St. Clair
and Detroit Rivers. In The Lake Huron Ecosystem:
Ecology, Fisheries and Management, eds. M.
Munawar, T. Edsall, and J. Leach, pp. 447–460. Ams-
terdam: SPB Academic Publishing.
Nei, M. 1987. Molecular Evolutionary Genetics. New
York, N.Y.: Columbia Univ. Press.
, and Tajima, F. 1983. Maximum likelihood esti-
mation of the number of nucleotide substitutions from
restriction-site data. Genetics 105:207–217.
———
Kimura, M. 1980. A simple method for estimating evolu-
tionary rate of base substitution through comparative
studies of nucleotide sequences. J. Mol. Evol.
16:111–120.
Pinchuk, V.I. 1991. Kvoprosy grupirovkakh vidov v pre-
delakh roda Neogobius (Perciformes) (About specia-
tion within the genus Neogobius (Perciformes)).
Voprosy Ikhtiologii 31:380–393.
Kocher, T. D., and Carlton, K. L. 1997. Base substitution
in fish mitochondrial DNA: patterns and rates. In
Molecular Systematics of Fishes, eds. T. D. Kocher
and C. A. Stepien, pp. 13–24. San Diego, CA: Acade-
mic Press.
Ricciardi, A., and H.J. MacIsaac. 2000. Recent mass
invasion of the North American Great Lakes by
Ponto-Caspian species. TREE 15:62–65.
Roff, D.A., and Bentzen, P. 1989. The statistical analysis
of mitochondrial DNA polymorphisms: X2
and the
problem of small samples. Mol. Biol. Evol. 6:539–545.
, Thomas, W.K., Meyer, A., Edwards, S.V.,
———

280
Dillon and Stepien
Saitou, N., and Nei, M. 1987. The neighbor joining
method: a new method for reconstructing phyloge-
netic trees. Mol. Biol. Evol. 4:406–425.
Schneider, S., Kueffer, J., Roessli, D., and Excoffier, L.
1997. Arlequin vers. 1.1: a software for population
genetic data analysis. Univ. of Geneva, Switzerland:
Genetics and Biometry Laboratory
ogy in studies of fish evolution,. In Molecular System-
atics of Fishes, eds. T.D. Kocher and C.A. Stepien,
pp. 1–11. San Diego, CA: Academic Press.
, Dillon, A.K., and Chandler, M.D. 1998. Genetic
———
identity, phylogeography, and systematics of ruffe
Gymnocephalus in the North American Great Lakes.
J. Great Lakes Res. 24:361–378.
Simonovic, P.D., Nikolic, V.P., and Skora, K.E. 1996.
Vertebral number in Ponto-Caspian gobies: phyloge-
netic relevance. J. Fish Biol. 49:1027–1029.
Stevanovic, P.M. 1982. Istorijska Geologija-Kenozoik
(Geological History-Cenozoic). Belgrade: Univ. of
Belgrade.
Swofford, D.L. 2001. Phylogenetic analysis using parsi-
mony (PAUP)* 4.0. vers. b5. Sunderland, Massachu-
setts: Sinauer Assoc., Inc.
Tajima, F. 1989. Statistical method for testing the neutral
mutation hypothesis by DNA polymorphism. Genetics
123:858–595.
U. S. Geological Survey (Biological Resources Divi-
sion). 2001. Online (range maps). http://nas.er.
usgs.gov/fishes/images/goby_map.gif (round
goby) http://nas.er.usgs.gov/fishes/images/pr_marmo.
gif (tubenose goby)
Weir, B.S. 1996. Genetic analysis II: methods for discrete
population genetic data. In Molecular Systematics, 2nd
ed., eds. D.M. Hillis, C. Moritz, and B.K. Mable, pp.
385–405. Sunderland, MA: Sinauer Assoc., Inc.
Williamson, M. 1996. Biological Invasions. New York,
NY: Chapman and Hall.
Skora, K.E., and Stolarski, J. 1993. New fish species in
the Gulf of Gdansk, Neogobius spp. [cf. Neogobius
melanostomus (Pallas 1811)]. Bull. Sea Fish. Inst. 1:83.
Steininger, F.F., and Rogl, R. 1984. Paleogeography and
palinspastic reconstruction of the Neogene of the
Mediterranean and Paratethys, In The Geological
Evolution of the eastern Mediterranean, eds. J.E.
Dixon and A.H.F. Robertson, pp. 659–686. Oxford:
Blackwell Sci. Pub., Inc.
Stepien, C.A. 1995. Population genetic divergence and
geographic patterns from DNA sequences: examples
from marine and freshwater fishes, In Evolution and
the Aquatic Ecosystem: Defining Unique Units in
Population Conservation, ed., J.L. Nielsen, pp.
263–287. Bethesda, MD: American Fisheries Society,
Symposium 17.
, and Dillon, A.K. 2001. Phylogeography of the
———
invasive ruffe Gymnocephalus cernuus (Percidae):
glacial refugia, the leading edge hypothesis, and
exotic introductions. Evolution in press.
Submitted: 16 January 2000
Accepted: 14 December 2000
Editorial handling: David J. Jude
, and Kocher, T.D. 1997. Molecules and morphol-
———