-
Genetic ecotoxicology II: population geneticstructure in
mosquito®sh exposed in situ toradionuclides
CHRISTOPHER W. THEODORAKIS 1 and LEE R. SHUGART2�
1Texas A&M University, College Station, TX77483, USA and 2LR
Shugart & Associates, Inc., PO Box 5564,
Oak Ridge, TN 37831-5564, USA
Received 1 July 1996; accepted 16 December 1996
In 1977, approximately 250 mosquito®sh (Gambusia af®nis) from a
relatively uncontaminated site
(Crystal Springs) were transplanted into a small pond on the
Department of Energy Oak Ridge
Reservation which is heavily contaminated with radionuclides
(Pond 3513). Starting in 1992, DNA
polymorphism was evaluated using the RAPD (Randomly Ampli®ed
Polymorphic DNA) and allozyme
genotype techniques to determine if genetic differentiation had
occurred between the two populations.
Fish from a second radionuclide-contaminated population (White
Oak Lake) and another unrelated
non-contaminated population (Wolf Creek) were also examined. For
the RAPD analyes, 15 RAPD
primers (from a total of 40) were found to produce polymorphic
banding patterns in at least two of
the four populations and subsequently were used to produce a
total of 142 bands. Data generated by
these RAPD primers indicated an increased genetic diversity in
radionuclide-contaminated sites
relative to reference sites. Furthermore, the patterns from six
RAPD primers produced a higher
average number of bands when using DNA from
radionuclide-contaminated populations than from
non-contaminated, and for three RAPD primers the average number
of bands from radionuclide-
contaminated populations was lower. In addition, 17 bands
occurred at a higher frequency in the
radionuclide-contaminated compared to the non-contaminated
populations. For the allozyme analyses,
it was found that there was a higher precentage of polymorphism
and heterozygosity in the
radionuclide-contaminated relative to non-contaminated sites.
These ®ndings contribute to our
understanding of the evolutionary effects of contaminant
exposure as well as to the development of
population-level biomarkers.
Keywords: population genetics; mosquito®sh; radiation; RAPD;
allozymes; evolutionary toxicology.
Introduction
There is inevitably some variation in measurable responses of
®sh or other organisms totoxic stress (Theodorakis et al., 1992).
This can be attributed in part to genetic variation.If this is the
case, then there is a possibility of natural selection in
contaminatedenvironments on genetic variants, which may lead to
observable changes in genefrequencies of genetic diversity (Lavie
and Nevo, 1982; Nevo et al., 1986; Ben-Schlomoand Nevo, 1988;
Strittholt et al., 1988; Gillespie and Guttman, 1989; Kopp et al.,
1992;Guttman, 1994; Anderson et al., 1994).
Ecotoxicology 6, 335±354 (1997)
0963±9292 # 1997 Chapman & Hall
�To whom correspondence should be addressed.
-
It has been suggested that changes in population genetic
structure may precedeecological damage that either (a) results from
declines in recruitment or survivorship(`ecological effects';
Benton and Guttman, 1992), or (b) results in decreased potentialfor
adaptation and increased probabilities of local extinctions
(`evolutionary effect';Bickham and Smolen, 1994). These types of
effects are of special concern because theycan manifest themselves
long after the source of contamination has been eliminated(Anderson
et al., 1994). Therefore, monitoring changes in population genetic
structurecan be a valuable component of ecological risk assessments
by providing a type ofpopulation-level biomarker of contaminant
exposure or effects (Suter, 1990).
In this study, the potential for toxic stress to affect
population genetic structure wasexamined in four populations of
mosquito®sh (Gambusia af®nis) living on or near theDepartment of
Energy Oak Ridge Reservation. These populations are particularly
suitedfor such studies because in 1977 ®sh from a non-contaminated
site (Crystal Springs)were transplanted to a pond heavily polluted
with radionuclides (Pond 3513). In aprevious study, Theodorakis et
al., (1996) examined the relationship between DNAdamage and
reproductive parameters in these same populations and evidence
waspresented which suggested that individuals with more DNA damage
(strand breakage)had lower reproductive success. If strand breakage
is a re¯ection of susceptibility togenotoxicants, then these
®ndings imply that ®sh which are less susceptible may have
aselective advantage in contaminated environments. This selection
may be manifested bydifferences in population genetic structure
between contaminated and non-contaminatedpopulations.
Population genetic structure was examined using RAPD (Randomly
Ampli®edPolymorphic DNA) and allozyme techniques. Although the
latter technique has beenused extensively for this purpose, the
former has received less attention. Nevertheless,the RAPD technique
has several advantages over other methods to assess polymorphismin
DNA. First, only minute amounts of DNA are needed. Second, an a
priori knowledgeof genomic sequences is not required. Third, the
technique has the potential to accessmany loci, thus leading to
identi®cation of loci that were not previously known to beunder
contaminant-selective pressure. Finally, RAPD primers amplify from
invertedrepeat sites within the DNA. This is important because
certain DNA sequences thatcontain inverted repeats (e.g.
transposons) have been shown to be responsive to DNAdamaging agents
(Doring and Starlinger, 1986).
Although the RAPD method may be appropriate for genotoxic study,
theexperimentalist should be aware of drawbacks. The technique is
sensitive to reactionconditions that sometimes result in the
production of non-reproducible bands (Williamset al., 1991; Hadrys
et al., 1992). If, however, reaction conditions are
rigidlystandardized and care is taken to identify non-reproducible
bands, the banding patternscan be highly reproducible (Dinesh et
al., 1995; personal observation). Also, thespeci®c locations and
identities of RAPD-ampli®cation sites within the DNA are notknown a
priori, which makes interpretation of the adaptive signi®cance of
such markersdif®cult.
The objective of this study was to determine if exposure to
radionuclides, a specialclass of genotoxicants, causes changes in
population genetic structure (e.g. geneticdiversity and=or genotype
frequencies). An ancillary goal was to evaluate theeffectiveness of
the RAPD and allozyme techniques to detect such changes. It
ishypothesized that genetic diversity will decrease in
genotoxicant-contaminated
336 Theodorakis and Shugart
-
populations. Futhermore, it is predicted that certain genotypes
will have a selectiveadvantage in contaminated populations (in
terms of fecundity), and that the frequencyof these genotypes will
be greater in the contaminated populations.
It should be noted that the long-term goals of this and other
studies by the authors(Shugart and Theodorakis, 1994, 1996;
Theodorakis et al., 1996) are (a) to developpopulation-level
biological markers (and indicators) of contaminant exposure
andeffects, and (b) to understand the evolutionary consequences of
exposure to environ-mental contamination, especially as they may
affect the process of natural selection.
Methods and materials
Chemicals
The following materials were used: Taq polymerase with assay
buffer, from PromegaCorp. (Madison, WI); DNA molecular length
markers (bacteriophage ÖX174 DNA,Hae III digest) from Promega or
Sigma Chemical Corp. (St Louis, MO); agarose, Tris-HCl, EDTA, boric
acid, phenol, and ethidium bromide from Gibco-BRL
Inc.(Gaithersburg, MD); deoxynucleotides (dATP, dCTP, dGTP, and
dTTP) from BoehringerManheim (Indianapolis, IN); bovine serum
albumin (BSA) from Idaho Technology(Idaho Falls, ID); chloroform
and ethanol from J.T. Baker (Phillipsburg, NJ); RAPDoligonucleotide
primers OPD1 through OPD20 from Operon Technologies (Alameda,CA)
and primers UBC1 through UBC20 from the Nucleic Acid and Protein
ServiceUnit, Biotechnology Laboratory, University of British
Columbia (Vancouver, BC). Thesources of reagents for allozyme
starch gel electrophoresis are reported in Kramer et al.(1989).
Study sites=specimen collection
The four populations examined in this study were from two
radionuclide-contaminatedsites situated on the US Department of
Energy reservation in Oak Ridge, TN: Pond 3513and White Oak Lake;
and two reference sites located off the reservation: Crystal
Springsand Wolf Creek. A detailed description of these sites as
well as the method for thecollection of adult, female mosquito®sh,
the preservation of samples and isolaton ofDNA from the blood, the
determination of brood size and relative fecundity are found
inTheodorakis et al. (1996).
RAPD analysis
RAPD reactions were carried out using either a DNA Thermal
Cycler 480 (Perkin-ElmerCetus Corp., Norwalk, CT) or an Air Thermo
Cycler 1605 (Idaho Technologies, IdahoFalls, ID). For each
particular RAPD primer, all populations were examined using thesame
apparatus. For the Perkin-Elmer apparatus, the reaction mixture
consisted of2.5 mM MgCl2, 100 ìM of each nucleotide triphosphate,
15 ng primer, 10 ng DNA, 1Unit Taq polymerase and a 1=10 volume of
the assay buffer supplied by Promega Corp.in glass double-distilled
water. Reactions were carried out in a voume of 25 ìl followingthe
ampli®cation protocol modi®ed from Williams et al. (1991). For the
Idaho apparatus,the reaction mixture consisted of 2.5 mM MgCl2, 10
mg ml
ÿ1 BSA, 100 ìM of eachnucleotide triphosphate, 12 ng primer, 20
ng DNA and 0.6 units Taq polymerase in glassdouble-distilled water.
The reactions were carried out in a volume of 10 ìl following
the
Population genetics in mosquito®sh 337
-
ampli®cation protocol supplied by the manufacturer (Skroch and
Nienhuis, 1992). All ofthe above procedures were carried out using
sterile technique.
Ampli®cation products were loaded onto a 1.5% agarose gel with
0.5 ìg mlÿ1ethidium bromide and subjected to electrophoresis at 2.5
V cmÿ1 for 3 h. An aliquot ofmolecular length standards
(bacteriophage ÖX174 DNA digested with Hae III) was runalongside
the samples for determination of realtive molecular length (Weir,
1990). Afterelectrophoresis, the gels were photographed under UV
light.
RAPD data analysis
The nomenclature for each primer and for each band was as
follows. The name of eachprimer consisted of the abbreviation of
the manufacturer followed by a number assignedby the manufacturer
to identify each primer. Operon Technologies produces several
setsof primers. For this experiment, set D was used, so the primers
in this set were identi®edas OPD. The Univeristy of British
Columbia primers were identi®ed as UBC. Each bandis indicated by
the primer which was used to amplify it, followed by its molecular
lengthin subscript. For example, band UBC2830 is a fragment 830
nucleotides long produced byprimer UBC2.
The data generated from bands on the gels included: (a) the
average number of bandsper individual; (b) the frequency of each
band (i.e. the proportion of individuals in apopulation for which
each band is produced); and (c) genetic diversity. Number ofbands
were scored because, for other types of DNA ®ngerprinting, this
parameter hasbeen associated with heterozygosity (Lynch, 1990).
Differences between populations interms of average number of bands
were tested using the Kruskall-Wallace test withmultiple
comparisons; and in terms of band frequencies using a ÷2 test.
The RAPD technique is relatively new and no one standard method
is routinely usedfor determining genetic diversity with this
procedure. Nevertheless, for comparison, twodifferent but accepted
methods were utilized. They were: (a) the similarity index
(S),which measures average similarity of banding patterns between
all possible pairs ofindividuals in the population (Lynch, 1990);
and (b) the nucleon diversity index (h: Nei,1987). Differences
between populations were tested by calculation of a standard
normaltest statistic using the variance of S and h (Lynch, 1990;
Nei, 1987). Because S isinversely proportional to genetic
diversity, genetic diversity is reported as thedissimilarity index
(D), where D 1 ÿ S (Lynch, 1990). The D and h indices rangebetween
0 (population completely homogeneous) and 1 (populations are
completelyheterogeneous).
Selection coef®cients were calculated for ®sh with or without
particular bands, wherethe selection coef®cient 1 ÿ w. The relative
®tness of ®sh without these bands (w)was calculated by dividing the
average fecundity of ®sh with these bands by theaverage fecundity
of ®sh without the bands. The correlations between
selectioncoef®cients and band frequency for ®sh from Pond 3513 were
calculated and examinedwith Kendall's test.
Genetic distances between populations were calculated using the
formula for Roger'sgenetic distance (Nei, 1987) utilizing band
frequencies instead of allele frequencies.This index has a value of
0 to 1 when the populations are completely identical andcompletely
different, respectively. Testing for differences between pair-wise
distancescan be problematic (Nei, 1987), therefore, the distances
for each locus were treated as a
338 Theodorakis and Shugart
-
sample, and the differences between pair-wise distances were
then tested using aStudent's t-test, as suggested by Nei
(1987).
Dendrograms were constructed employing the unweighted pair-group
method usingan arithmetic average (Weir, 1990).
Allozyme analysis
The methods for allozyme analysis were as described in Diamond
et al. (1989). Brie¯y,the whole carcass of each ®sh was homogenized
with a glass rod in approximately50±150 ìl of buffer (10 mM Tris, 1
mM EDTA, 50 mM NADP, pH 7.0) and centrifugedat 12 000 rpm in a
micro centrifuge for 45 s, loaded into a 12.5% starch gel
andsubjected to electrophoresis for at least 10 h. Conditions for
electrophoresis and stainingof each enzyme can be found in Harris
and Hopkinson (1976), Ayala et al. (1972) andSelander et al.
(1971). Allelic variants of each enzyme were scored as fast (F) or
slow(S), depending on their relative mobility. Homozygotes were
given the connotation FF orSS, while heterozygotes were depicted as
FS.
Genetic variation was measured as degree of heterozygosity or
polymorphism.Differences in heterozygosity were tested by
calculating the variance for the hetero-zygosity estimate (Weir,
1990) and using this variance to calculate a standard normaltest
statistic. Differences in polymorphism between populations were
tested using a ÷2
test. Genetic distances (D) between populations were calculated
usign Roger's distance(Nei, 1987). The differences between
distances were tested and dendrograms wereconstructed as above.
Results
Of the 40 different RAPD primers (OPD1 to OPD20 and UBC1 to
UBC20) used toscreen for polymorphic banding patterns, only 15 were
found to produce banding patternswhich were polymorphic in at least
two of the four ®sh populations (Table 1). The resultsusing these
15 RAPD primers are reported here.
Genetic diversity
Genetic diversity within the four ®sh populations, as revealed
by the RAPD and allozymeassays, is shown in Fig. 1 and is higher
within the radionuclide-contaminated sitesrelative to
non-contaminated sites.
RAPD assay. Data in Fig. 1A represent the dissimilarity index (D
1 ÿ S) where S isinversely proportional to genetic diversity. Data
from all RAPD primers were pooledtogether to calculate S (Lynch,
1990). When all RAPD primers were pooled, the nucleondiversity
index (h) was 1 (indicating each individual ®sh had a unique set of
RAPDprimer-speci®c banding patterns). Therefore, the h values were
calculated for each RAPDprimer separately and averaged
together.
Allozyme assay. The degree of heterozygosity and polymorphism
were observed to behigher in the two radionuclide-contaminated
populations relative to non-contaminatedpopulations (Fig. 1B).
Population genetics in mosquito®sh 339
-
Genetic distance
The genetic distances were calculated using both the RAPD and
allozyme techniques.Data from the RAPD assay and the allozyme NP
assay indicated that the tworadionuclide-contaminated populations
were more similar to each other than to the non-contaminated
populations (Fig. 2A and C). When data from all allozyme loci
wereincluded (Fig. 2B), the population in the
radionuclide-contaminated Pond 3513 were mostsimilar to the
reference population in Crystal Springs.
A
B
Dis
sim
ilarit
y In
dex
(D)
Het
eroz
ygos
ity
Pol
ymor
phis
mN
ucle
on D
iver
sity
Inde
x (h
)
0.4
0.3
0.2
0.1
0
1
0.8
0.6
0.4
0.2
0
0.8
0.4
0.2
0White Oak Lake
Pond 3513 CrystalSprings
WolfCreek
Heterozygosity
Polymorphism
DissimilarityNucleon Diversity
0.04
0.03
0.02
0.01
0
A B C D
A A B C
a b ca
a b c d
Fig. 1. Genetic diversity measurements in mosquito®sh
populations from radionuclide-
contamined (Pond 3513 and White Oak Lake) and non-contamined
(Crystal Springs and
Wolf Creek) sites using, A: the RAPD assay; or B: the allozyme
assay. Bars labelled with
the same letters are not signi®cantly different (P , 0.05).
340 Theodorakis and Shugart
-
A Pond 3513
White Oak Lake
Crystal Springs
Wolf Creek
Roger’s Genetic Distance0.30 0.25 0.20 0.15
B Pond 3513
White Oak Lake
Crystal Springs
Wolf Creek
Roger’s Genetic Distance0.40 0.30 0.20 0.10
Pond 3513
White Oak Lake
Crystal Springs
Wolf Creek
Roger’s Genetic Distance
C
0.08 0.07 0.06 0.05 0.04 0.03
Fig. 2. Dendrograms (unweighted pair-group method average)
representation of genetic
distance between mosquito®sh populations from
radionuclide-contamined (Pond 3513 and
White Oak Lake) and non-contamined (Crystal Springs and Wolf
Creek) sites. Data
generated using, A: the RAPD assay; B: the allozyme assay (12
loci); or C: the alloyzyme
assay with nucleoside phosphorylase locus only (C).
Population genetics in mosquito®sh 341
-
Genotype distributions
Three different metrics used to describe genotypes produced by
allozyme and RAPD datawere: (a) genotype frequencies, (b) RAPD band
frequencies, and (c) number of RAPDbands.
Genotype frequencies. Twelve loci were examined: malate
dehydrogenase (MDH, threeloci: MDH1, MDH2, MDH3); fumarase (FUM);
isocitrate dehydrogenase (IDH);nucleoside phosphorylase (NP);
hydroxybutarate dehydrogenase (HBDH); lactatedehydrogenase (LDH);
adenosine deaminase (ADA); phosphoglucomutase
(PGM);glucose-6-phosphate dehydrogenase (GPD); and lysyl-alanine
peptidase (PEP). Theresults for individual allozyme loci revealed
that the levels of heterozygosity werevery low. If a locus was
polymorphic, there were only two genotypes present: FSand SS (PEP)
or FS and FF (all other loci). The nucleoside phosphorylase locus
wasthe exception. The percentage of NP heterozygotes at this locus
was an order ofmagnitude higher than any of the other loci, in
addition, all three genotypes werepresent. Therefore the NP locus
was analysed separately. As demonstrated in Fig. 3, theNP
heterozygotes were present at a higher frequency and the SS
homozygote waspresent at a lower frequency in the two
radionuclide-contaminated sites (P , 0.05, ÷2
test).The only other locus which indicated a difference between
the radionuclide-
contaminated and non-contaminated sites was HBDH, with
heterozygosities in WhiteOak Lake and Pond 3513 of 0.04 and 0.067,
respectively (no HBDH heterozygoteswere observed in the other
sites). The other polymorphic loci found in populations fromPond
3513 were FUM and GPD, with heterozygosities of 0.085 and 0.067,
respectively.
Fast/Fast Fast/Slow Slow/Slow
Genotype
Freq
uenc
y
1.0
0.8
0.6
0.4
0.2
0.0
A,B A,B A A A BB CA BC D
Fig. 3. Nucleoside phosphorylase genotype frequencies in
mosquito®sh populations from
radionuclide-contamined (Pond 3513 and White Oak Lake) and
non-contamined (Crystal
Springs and Wolf Creek) sites. `Fast' and `slow' refer to
relative electrophoretic mobility of
alleles on starch gels. Bars labelled with the same letters are
not signi®cantly different (P ,0.05, ÷2 test). White Oak Lake ;
Pond 3513 ; Crystal Springs ; Wolf Creek .
342 Theodorakis and Shugart
-
In White Oak Lake, the polymorphic loci included MDH1 (0.082
heterozygotes),MDH2 (0.04), LDH (0.082), and ADA (0.061). In
Crystal Springs, NP was the onlypolymorphic locus. In Wolf Creek,
the polymorphic loci were MDH1 and LDH, withheterozygosities of
0.076 and 0.051, respectively. Genotypic distributions of all
lociwere in agreement with Hardy-Weinberg expectations.
RAPD band frequency. There were a total of 142 bands produced by
the 15 RAPDprimers used in this analysis. Seventeen of these bands
increased in frequency in both ofthe radionuclide-contaminated
populations relative to both of the non-contaminatedpopulations
(Table 2). For purposes of discussion, these bands will be referred
to as`contaminant-indicative bands'. Four of the bands decreased in
frequency in bothcontaminated sites relative to both
non-contaminated sites.
Number of RAPD bands. The average number of bands was greater in
radionuclide-contaminated populations than for non-contaminated
populations for six RAPD primers(Fig. 4A). Again, for purposes of
discussion, these RAPD primers will be referred to
as`contaminant-indicative primers'. The average number of bands was
lower in radio-nuclide-contaminated populations for three of the
RAPD primers (Fig. 4B). When thenumber of bands produced by all 15
of the RAPD primers were pooled, there were smallbut signi®cant
differences between sites, with radionuclide-contanimated sites
havinghigher numbers of bands than non-contaminated sites (Fig.
4C). In addition, when thenumber of RAPD bands present in
populations at the various sites was compared withallozyme data
(Fig. 5), it was observed that heterozygotes had more RAPD bands
thanhomozygotes (P , 0.05, Wicoxon rank-sum test).
Genotype and ®tness
NP heterozygotes had a higher fecundity than homozygotes for
both Pond 3513 andWhite Oak Lake, but not for either of the
reference sites (P , 0.05, Wilcoxon Rank-Sumtest; Fig. 6). There
was also a positive correlation between number of heterozygous
lociand fecundity for both of the radionuclide-contaminated sites
(Pond 3513: r2 0.395,P , 0.001; White Oak Lake: r2 0.384, P 0.05,
Kendall correlation test). In thereference sites no individual had
more than one heterozygous locus, so no suchcorrelations could be
performed.
In Pond 3513, ®sh which possessed eight of the
`contaminant-indicative bands' hadhigher fecundity than ®sh which
did not (Fig. 7A). For three other bands, there was ageneral trend
in this direction, but the differences were statistically
signi®cant only at
Table 1. List of RAPD primersa that produced polymorphic banding
patterns in the mosquito®sh
(Gabusia af®nis)
OD RAPD primers
OPD2 OPD7 OPD8 OPD11 OPD12 OPD13 OPD18 OPD20
UBC RAPD primers
UBC1 UBC2 UBC4 UBC6 UBC9 UBC12 UBC16
aSource of RAPD primers: Operon Technology (OD); Univeristy of
British Columbia (UBC).Sequences can be obtained from source.
Population genetics in mosquito®sh 343
-
P < 0.10. Similar results were found in White Oak Lake (Fig.
7B). However, for bothradionuclide-contaminated sites, individuals
with band UBC41610 had a lower fecunditythan did ®sh without this
band, even though this band was elevated in frequency at bothsites.
This is opposite of what was observed for other
`contaminant-indicative bands'. Inthe Crystal Springs population
(Fig. 7C) one band was noted for which fecundity wassigni®cantly
greater for ®sh with vs. without a `contaminant-indicative band'.
Fishwithout band UBC9920 on the other hand, had a higher fecundity
than did ®sh with thisband which is the opposite of what was found
at the radionuclide-contaminated sites.Similar results were
observed for bands UBC560 and UBC161020 in the Wolf Creekpopulation
(Fig. 7D).
The eight `contaminant-indicative bands' that correlated with
higher fecundity in the®sh populations of Pond 3515 and White Oak
Lake (Fig. 7A and B) were used tocalculate selection coef®cients. A
plot of the selection coef®cients versus the frequency
Table 2. Frequency of selected RAPDa bands for two
radionuclide-contaminated (Pond 3513 and
White Oak Lake) and two non-contaminated (Crystal Springs and
Wolf Creek) mosquito®sh
populationsb
Populationc
RAPD band Pond 3513 White Oak Lake Crystal Springs Wolf
Creek
A. Frequency in radionuclide-contaminated .
non-contaminatedOPD21586 1.00
1 0.971 0.062 0.573
OPD21060 1.001 1.001 0.092 0.003
OPD71721 0.311 0.181 0.002 0.022
OPD71390 0.621 0.362 0.213 0.203
OPD13428 0.961 0.742 0.593 0.004
OPD20605 0.201 0.672 0.063 0.113
OPD20433 0.781 0.482 0.093 0.004
UBC42055 0.911 0.742 0.523 0.104
UBC41610 0.711 0.771 0.312 0.272
UBC41407 0.471 0.531 0.342 0.133
UBC6575 0.441 0.642 0.213 0.313
UBC6455 0.201 0.231 0.002 0.002
UBC9921 0.771 0.522 0.453 0.034
UBC121269 0.571 0.482 0.003 0.003
UBC121158 0.721 0.192 0.003 0.003
UBC12861 0.131 0.292 0.003 0.023
UBC161016 0.761 0.781 0.362 0.432
B. Frequency in non-contaminated .
radionuclide-contaminedOPD8560 0.85
1 0.672 0.933 0.973
OPD13787 0.261 0.211 1.002 0.972
OPD13677 0.481 0.531 0.832 0.973
OPD13263 0.021 0.001 0.832 0.972
aBands which showed consistent differences in band frequency
between sites are reported here.bSample sizes for populations were:
Pond 3513, 40; White Oak Lake, 40; Crystal Springs, 38; Wolf Creek,
35.cValues with same numerical superscrips are not signi®cantly
different (P . 0.05, ÷2 test).
344 Theodorakis and Shugart
-
of these bands produced a positive correlation (Fig. 8). This
correlation was greater inPond 3513 (P 0.03; Fig. 8A) than in White
Oak Lake (P 0.10; Fig. 8B).
For many of the `contaminant-indicative primers' (i.e. RAPD
primers for whichthe average number of bands per individual was
greater in the two radionuclide-contaminated sites), there was a
positive correlation between fecundity and number ofbands (i.e.
individuals with more bands had a higher fecundity) at P <
0.05(Table 3). This was true for six RAPD primers in White Oak
Lake, four in Pond3513, and two in Crystal Springs. For RAPD
primers OPD2 and UBC9, the trendwas the opposite of that for at
least one of the radionuclide-contaminated sites. ForRAPD primer
OPD13, there was a negative correlation in White Oak Lake but
apositive correlation in Crystal Springs. Interestingly, for this
RAPD primer, thecontaminated populations showed on average fewer
bands per individual than at thereference sites.
A A A
A A B C
14
12
10
8
6
4
2
0
12
10
8
6
4
2
0
10
8
6
4
2
0
100
80
60
40
20
0OPD8 OPD12 OPD13 White Oak Lake
Pond 3513 CrystalSprings
Wolf Creek
OPD2 OPD7 OPD20 UBC4 UBC9 UBC12
Pond 3513White Oak LakeCrystal SpringsWolf Creek
A. Contaminated > Reference A. (cont')
B. Reference > Contaminated C. All Primers Pooled
Ave
rage
num
ber
of B
ands
per
Indi
vidu
al
A B C B C D A B B A A B C A A B C A A B C
A A B BA A
B B A B C C
Fig. 4. Average number of bands produced by nine different RAPD
primers for mosquito®sh
populations from radionuclide-contamined (Pond 3513 and White
Oak Lake) and non-contamined
(Crystal Springs and Wolf Creek) sites. RAPD primers selected on
demonstrated differences between
radionuclide-contaminated and non-contamined populations. Bars
and error bars represent medians and
quartiles, and similar letters indicate no signi®cant
differences. Data for RAPD primers, A: that
indicate a higher average number of bands in the
radionuclide-contaminated sites; B: that indicate a
lower average number of bands in the non-contamined sites; and
C: when all RAPD primers were
pooled together.
Population genetics in mosquito®sh 345
-
WhiteOak Lake
Pond 3513 Crystal Springs Wolf Creek
100
80
60
40
20
0
Num
ber
of R
AP
D B
ands
Heterozygote
Homozygote
[15] [32] [16] [31] [11] [29] [4] [36]
Fig. 5. Number of ampli®ed RAPD bands displayed by mosquito®sh
homozygous or
heterozygous for at least one locus. Bars and error bars are
medians and quartiles. Numbers
above bars are sample size.
Pond 3513 White Oak Lane
HeterozygoteHomozygote
Fec
undi
ty (
Bro
od S
ize/
Bod
y Le
ngth
)
2.5
2
1.5
1
0.5
0
Fig. 6. Fecundity of mosquito®sh from radionuclide-contamined
(Pond 3513 and White Oak
Lake) homozygous or heterozygous for the nucleoside
phosphorylase (NP) genotypes. Bars
and error bars indicate medians and quartiles. No signi®cant
differences were found in non-
contamined populations.
346 Theodorakis and Shugart
-
Fec
undi
ty (
Bro
od S
ize/
Bod
y Le
ngth
)
2
1.5
1
0.5
0
2.51.4
1.2
1
0.8
0.6
0.4
0.2
0
**D. Wolf Creek
[12] [28] [17] [23]
UBC6 UBC16560 1020
B. White Oak Lake
********
[19]
[29]
[17]
[31]
[41]
[7]
[36]
[12]
[32]
[16]
[34]
[14]
[37]
[11]
[23]
[25]
[9]
[39]
[14]
[34]
[37]
[11]
OP
D2
1590
OP
D7
1390
OP
D8
560
340
OP
D13
OP
D20 610
910
1610
UB
C2
UB
C4
UB
C12
1270
1160 860
1020
UB
C12
UB
C12
UB
C16
OP
D20 430
1610
UB
C4
UB
C4
440
580
460
UB
C6
UB
C6
UB
C9
920
1270
1160
1020
UB
C12
UB
C12
UB
C16
OP
D20
OP
D7
610
1150 OPD7
1150OPD13 510
UBC4 1210
UBC4 690
UBC9 1290
C. Crystal Springs[16][19]
[7][28]
[29][6] [29][6] [29][6]
**
A. Pond 3513 With Band Without Band
* * * * * ***
[26]
[14]
[8]
[32]
[31]
[9]
[34]
[14]
[39]
[9]
[17]
[23]
[11]
[37]
[30]
[9]
[22]
[18]
[28]
[12]
[30]
[10]
3
2.5
2
1.5
1
0.5
0
2
1.5
1
0.5
0
Fig. 7. Fecundity of mosquito®sh populations from
radionuclide-contamined (Pond 3513 and White Oak Lake) and
non-contamined (Crystal Springs
and Wolf Creek) sites with and without contaminated±indicative
RAPD bands. � indicates P < 0.05 (Ho: No difference between ®sh
with and withoutband; Wilcoxon Rank-sum test).
Popula
tion
gen
eticsin
mosq
uito
®sh
347
-
Discussion
Population genetic structure data derived from the RAPD and
allozyme assays revealdifferences between radionuclide-contaminated
and non-contaminated populations andsuggest that changes have
occurred since Pond 3513 was colonized from Crystal Springsin 1977.
In addition, the genetic similarity observed between Pond 3513 and
White OakLake suggest that these differences may be due to
radionuclide-induced selection.
It should be noted that the differences between
radionuclide-contaminated and non-contaminated populations may not
be due solely or even primarily to radionuclidecontamination. This
is a preliminary investigation and the number of populationsstudied
was small, therefore, other processes may very well be involved in
producingthe changes observed between the various populations. For
example, differencesbetween band frequencies could be due to random
drift, neutral mutation or selectiondue to some other environmental
variable not directly related to radiation exposure.Differences in
genetic diversity can also be driven by neutral processes. For
instance,relationships have been shown between genetic polymorphism
and habitat heterogeneity
0.7
0.6
0.5
0.4
0.3
0.2
0.1
00 0.2 0.4 0.6 0.8 1
Band Frequency
0 0.2 0.4 0.6 0.8 1
0.2
0.15
0.1
0.05
0
A. Pond 3513
r2 5 0.423P 5 0.423
B. White Oak Lake
r2 5 0.333P 5 0.10
Sel
ectio
n co
effic
ient
..
.
..
.
.
..
.
.
..
.
.
.
.
.
..
.
Fig. 8. Correlation between frequency of
`contaminated-indicative' RAPD bands vs.
selection coef®cient for RAPD bands. No signi®cant correlations
noted for either non-
contamined site.
348 Theodorakis and Shugart
-
(DeMeeus et al., 1993) and between genetic variability and
population density (Nei etal., 1975). Alternatively, the
mosquito®sh population in Pond 3513 could haveundergone a
bottleneck upon introduction from Crystal Springs. It has been
argued inthe past that bottlenecks should decrease variability (Nei
et al., 1975), but some recentevidence suggests the opposite (Wills
and Orr, 1993), both for single-locus (Leberg,1992) and
quantitative traits (Lopez-Fanjul and Villaverde, 1989).
Several lines of evidence which would argue against these
neutral hypotheses are:
(i) Measures of genetic diversity (Figs 1 and 2) indicate that
the contaminatedsites are more similar to each other than to the
non-contaminated sites, eventhough the Crystal Springs (reference)
site is ancestral to Pond 3513(radionuclide-contaminated).
(ii) Relationships observed between genotype and ®tness in the
contaminated sitesare not re¯ected in the non-contaminated sites.
These include differences infecundity between ®sh with and without
contaminant-indicative bands (Fig. 7),correlations between number
of bands and fecundity for `contaminant-indicativeprimers' (Table
3), and correlations between band frequency and
selectioncoef®cients (Fig. 8).
(iii) Recently published studies demonstrated a relationship
between genotype andDNA strand breakage (an effect of radiation
exposure) which parallels therelationship between fecundity and
genotype (Theodorakis et al., 1996).
(iv) Other studies utilizing the RAPD assay that indicate an
increase in geneticdiversity in organisms other than mosquito®sh at
contaminated sites on the OakRidge Reservation (K.L. Lee, Oak Ridge
National Laboratory, unpublisheddata; Nadig, 1996).
Table 3. Kendall correlation coef®cients r and probability
values (P) between number ofRAPD bands and fecundity for
mosquito®sh from two radionuclide-contaminated populations
(White Oak Lake and Pond 3513) and two non-contaminated (Crystal
Springs and Wolf
Creek) populations
Population
White Oak Lake Pond 3513 Crystal Springs Wolf Creek
Primer r Pa r P r P r P
OPD2 0.35 0.03 nsb ns ÿ0.44 ,0.01 ns nsOPD7 ns ns 0.34 0.03 ns
ns ns ns
OPD8 0.40 0.03 0.21 0.08 ns ns ns ns
OPD13 ÿ0.50 0.01 ns ns 0.35 0.05 ns nsOPD20 0.53 ,0.01 ns ns ns
ns ns nsUBC2 0.48 ,0.01 ns ns ns ns ns nsUBC4 0.32 0.07 ns ns 0.35
0.05 ns ns
UBC6 ns ns 0.32 0.01 ns ns ns ns
UBC9 ns ns 0.35 0.01 ns ns ÿ0.76 0.02UBC12 0.50 ,0.01 0.39 ,0.01
0.64 0.03 ns ns
aProbability (á) level for the test of Ho: r 0.bNot
signi®cant.
Population genetics in mosquito®sh 349
-
(v) Current studies indicate that when mosquito®sh from a
non-contaminated sourceare caged in Pond 3513, those which posses
`contaminant-indicative bands' have agreater chance of survival
than those that do not (unpublished data).
Taken together, these observations imply that radionuclide
exposure has an effect onpopulation genetic structure and genetic
diversity.
The ®nding of a higher genetic diversity in the
radionuclide-contaminated populationsis the reverse of what was
hypothesized initially (see Introduction) and what has beenreported
in other studies (Nevo et al., 1986; Strittholt et al., 1988; Kopp
et al., 1992).At present, de®nitive explanations for the occurrence
of higher diversity are unknown,but possibilities include: (a)
genomic rearrangements, which are known to be inducedby radiation
exposure (Doring and Starlinger, 1986) or other stressors (Cullis,
1990);and (b) differences in life history (Scribner et al., 1992),
reproductive (Chesser et al.,1984) or population
dynamic=demographic processes (Feder et al., 1984). Past
studies(Lavie and Nevo, 1982; Nevo et al., 1986; Ben-Schlomo and
Nevo, 1988; Strittholt etal., 1988; Gillespie and Guttman, 1989;
Kopp et al., 1992) have focused on structuralgenes (i.e.
allozymes), while many RAPD ampli®cation sites may be located in
non-coding regions. Thus discrepancies between the present study
and other studies may beexplained by differential gene structure
and function.
However in the present study, allozyme heterozygosity was also
higher in theradionuclide-contaminated populations, again in
contradiction to previous studies. Acontributing factor to this
apparent discrepancy may be the very low levels ofheterozygosity of
the Oak Ridge mosquito®sh populations compared to
populationsexamined in these other studies. An exception was the NP
locus, for which thepercentage heterozygosity was an order of
magnitude greater than any other loci. Forthis locus, there was
higher percentage heterozygosity in the
radionuclide-contaminatedsites (Fig. 3), and these heterozygotes
had higher fecundity than the homozygotes(Fig. 6). The nucleoside
phosphorylase enzyme is involved in nucleotide synthesis,
animportant biochemical pathway associated with DNA repair, and
nucleotide synthesisrates increase in response to gentoxicant
exposure (Kunz and Kohalm, 1991).Regardless of the reasons for
increased genetic diversity in contaminated habitats, thisparameter
may provide a population level biomarker of contaminant
exposure.
Other potential population-level biomarkers that re¯ect a
difference betweenreference and contaminated populations include
average number of bands per individualand frequency of certain
bands. In addition, there may be some relationship betweennumber of
RAPD bands and heterozygosity, because ®sh with at least one
heterozygouslocus had more bands than completely homozygous ®sh
(Fig. 5). Furthermore, for manybands and RAPD primers, there were
relationships between ®tness and genotype thatwere not re¯ected in
the reference populations. The nature of the relationship
betweengenotype and fecundity may vary between bands and RAPD
primers (Table 4).Interestingly, in some cases the relationships
between fecundity and genotype inreference sites were the opposite
of those in radionuclide-contaminated sites (see bandsUBC6560,
UBC91290, and UBC161020, Fig. 7; and RAPD primers OPD2, OPD13
andUCB9, Table 2). The speci®c relationship for each RAPD primer
and band may dependupon the structure and function of RAPD
ampli®cation sites within the DNA, as well asmolecular- and
population-level determinants of band presence or absence.
Overall, the frequency of such bands was correlated to the
selective coef®cient in
350 Theodorakis and Shugart
-
Pond 3513, and to a lesser extent in White Oak Lake, but in
neither reference site(Fig. 8). Furthermore, when all RAPD primers
were pooled it was found that theaverage number of bands was
greater in the radionuclide-contaminated populations(Fig. 4D), and
in these populations there was a positive correlation between
number ofbands and fecundity (Table 3). These ®ndings strongly
support the hypothesis ofcontaminant-induced selection as a
possible explanation for differences in bandfrequencies or number
of bands.
Another potential indicator of such selection is the genetic
distance betweenradionuclide-contaminated and non-contaminated
populations (Fig. 2). If no contami-nant-induced selection were
occurring, one would expect that the smallest distancewould be
between Pond 3513 and Crystal Springs, but RAPD data indicated
thesmallest distance was between Pond 3513 and White Oak Lake
(Figs. 2A and 2C).Deviations from expected genetic distance
relationships at contaminated sites have alsobeen found in other
studies (Guttman, 1994), and may be a re¯ection of population-level
responses to contamination.
Finally, another notable observation is that both the genetic
distribution and thegenotype=®tness relationships for Crystal
Springs is intermediate between theradionuclide-contaminated sites
and Wolf Creek. Although the relative levels ofcontamination in the
two reference sites (Crystal Springs and Wolf Creek) are
unknown,
Table 4. Summary of the relationship between genotype and
fecundity for various RAPD primers and
bands
Effect of genotype on ®tness in radionuclide-
contaminated populationsb
Genotypic distribution Selective advantage
Selective
disadvantage
A. RAPD bandsa
Band frequency greater in radionuclide-
contaminated populations
OPD1590, OPD71390, OPD71160,
OPD20610, UPD20430, VBC2910,
UBC4440, UBC6580, UBC6460,
UBC9920, UBC121270, UBC121160,
UBC12860, UBC161020
UBC41610
Band frequency less in radionuclide-
contaminated populations
OPD8560 none
B. RAPD Primersa
Average number of bands greater in
radionuclide-contaminated populations
OPD2, OPD7, OPD20, UBC2,
UBC4, UBC6, UBC9, UBC12
none
Average number of bands less in
radionuclide-contaminated populations
OPD8 OPD13
aOnly those bands and RAPD primers that displayed a difference
between radionuclide-contaminated and non-contaminated populations
are presented.bSelective advantage: Higher fecundity for ®sh
possessing a particular band or displaying more bands for
aparticular RAPD primer, in the radionuclide-contamined
populations, but not in either non-contamined population.Selective
disadvantage: Lower fecundity for ®sh possessing a particular band
or displaying more bands for aparticular RAPD primer, in the
radionuclide-contamined populations, but not in either
non-contamined population.
Population genetics in mosquito®sh 351
-
it should be noted that the amount of DNA strand breakage in the
Crystal Springspopulation was also intermediate beteen the
radionuclide-contaminated sites and WolfCreek (Theodorakis et al.,
1996). In this regard, radon at a concentration of 100 pCi perlitre
has been reported in water obtained from Crystal Springs (J.
Larson, Oak RidgeNational Laboratory, unpublished observation).
Conclusion
The results presented in this study provide evidence that
radionuclide contamination mayin¯uence evolution of exposed
populations through contaminant-induced selection.Furthermore,
genetic markers produced by the RAPD or allozyme assays can be
useful inidentifying evolutionary effects of contaminant exposure,
and may lead to developmentof population-level biomarkers of
contaminant exposure or effects. Speci®cally, the datareported here
suggest that contaminant-induced selection may have occurred after
the ®shfrom Crystal Springs were introduced into Pond 3513. Such
`evolutionary effects' are ofconcern because they have the
potential to negatively affect population dynamics andlocal
extinction rates (Bickham and Smolen, 1994). Also, adaptation to
speci®c pollutantsmay be disadvantageous, because individuals more
adapted to one class of pollutant maybe more susceptible to other
pollutants (Depledge, 1994) or environmental stressors(Trabalka and
Allen, 1977). In addition, if population or physiological responses
at acontaminated site change over time, it may be possible by
applying the approaches andprinciples of evolutionary toxicology to
determine the cause (i.e. changes in contaminantconcentrations vs.
selection) for resistant individuals. Finally, differential
susceptibility togenotoxins is present in the human population
(Rudiger, 1991), so investigations into themechanisms of
differential genotoxic susceptibility in animals can contribute to
humanhealth risk assessment.
Use of RAPD methodology as described here for biomonitoring must
be viewedcautiously at this point, because the genetic location or
the possible functions of theampli®cation sites from which the RAPD
bands are produced is not known withcertainty. Furthermore, effects
of random drift as a contributing factor have not beenruled out.
The relationship between genotype and fecundity and the fact that
bothradionuclide-contaminated sites show similar trends would argue
against this hypothesis,but until more studies are done, genetic
drift remains a viable alternative hypothesis.
The results presented in this study, although compelling, are
only preliminary andmore work needs to be done before the RAPD
technique can con®dently be used inroutine environmental monitoring
work.
Acknowledgements
I would sincerely like to thank B.G. Blaylock, K.L. Lee and R.
Clark for their assistancewith ®eld collections and G.G. McCraken,
J. Tuskan, and B.P. Bradley for their critiquesof this manuscript.
This project was sponsored in part by an appointment of CWT to
theUS Department of Energy Laboratory Cooperative Postgraduate
Research TrainingProgram administered by the Oak Ridge Associated
Universities and by the US ArmyBiomedical Research and Development
Laboratory (DOE project No. 1061-BO47-A1).The Oak Ridge National
Laboratory is managed by Lockheed Martin Energy ResearchCorp. for
the US Department of Energy under contract DE-AC05-96OR22464.
Portions
352 Theodorakis and Shugart
-
of this study were presented at the 14th and 15th Annual
Meetings of The Society ofEnvironmental Toxicology and
Chemistry.
References
Anderson, S., Sadinski, W., Shugart, L., Brussard, P., Depledge,
M., Ford, T., Hose, J., Stegeman, J.,
Suk, W., Wirgin, I. and Wogan, G. (1994) Genetic and molecular
ecotoxicology: A research
framework. Environ. Health Perspect. 102, 3±8.
Ayala, F.J., Powell, J.R., Tracey, M.L., Mourae, C.A. and
Perezsalas, S. (1972) Enzyme variability in
the Drosophila willistoni Group. III. Genetic variations in
natural populations of Drosophila
willistoni. Genetics 70, 113±39.
Ben-Schlomo, R. and Nevo, E. (1988) Isozyme polymorphism as
monitoring of marine environments:
the interactive effect of cadmium and mercury pollution on the
shrimp. Palaemon elegans. Mar.
Pollut. Bull. 19, 314±17.
Benton, M.J. and Guttman, S.I. (1992) Allozyme genotype and
differential resistance to mercury
pollution in the caddis¯y, Nectopsyche albida. II. Multilocus
genotypes. Can. J. Fish. Aquat. Sci.
49, 147±9.
Bickham, J.W. and Smolen, M.J. (1994) Somatic and heritable
effects of environmental genotoxins and
the emergence of evolutionary toxicology. Environ. Health
Perspect. 102, 25±8.
Chesser, R.K., Smith, M.W. and Smith, J.H. (1984) Biochemical
genetics of mosquito®sh: 3. Incidence
and signi®cance of multiple insemination. Genetica 64,
77±81.
Cullis, C.A. (1990) DNA rearrangements in response to
environmental stress. Advances in Genetics
28, 73±98.
DeMeeus, T., Michalakis, Y., Renaud, F. and Olivier, I. (1993)
Polymorphism in heterogeneous
environments, evolution of habitat selection and sympatric
speciation: Soft and hard selection
models. Evolutionary Ecology 7, 175±98.
Depledge, M.H. (1994) Genotypic toxicity: Implications for
individuals and populations. Environ.
Health Perspect. 102, 101±4.
Diamond, S.S., Newman, M.C., Mulvey, M., Dixon, P.M. and
Martinson, D. (1989) Alloyzme genotype
and time to death of mosquito®sh, Gambusia af®nis (Baird and
Girard), during acute exposure to
inorganic mercury. Environ. Toxicol. Chem. 8, 613±22.
Dinesh, K.R., Chan, W.K., Lim, T.M. and Phang, V.P.E. (1995)
RAPD markers in ®shes: An evaluation
of resolution and reproducibility. Asia Pac. J. Mol. Biol.
Biotechnol. 3, 112±18.
Doring, H.P. and Starlinger, J. (1986) Molecular genetics of
transposable elements in plants. Ann. Rev.
Genet. 20, 175±200.
Feder, J.L., Smith, M.H., Chesser, R.K., Godt, M.J.W. and
Asbury, K. (1984) Biochemical genetics of
mosquito®sh. 2. Demographic differentiation of populations in a
thermally altered reservoir.
Copeia 1984, 108±19.
Gillespie, R.B. and Guttman, S.I. (1989) Effects of contaminants
on the frequencies of allozymes in
populations of the central stoneroller. Environ. Toxicol. Chem.
8, 309±17.
Guttman, S.I. (1994) Population genetic structure and
ecotoxicology. Environ. Health Persepct. 102,
97±100.
Hadrys, H., Balick, M. and Schierwater, B. (1992) Applications
of random ampli®ed polymorphic
DNA (RAPD) in molecular ecology. Molecular Ecology 1, 55±63.
Harris, H. and Hopkinson, D.A. (1976) Handbook of enzyme
electrophoresis in human genetics.
Amsterdam, the Netherlands: North-Holland Publishing
Company.
Kopp, R.L., Guttman, S.I. and Wissing, T.E. (1992) Genetic
indicators of environmental stress in
central mudminnow (Umbra limi) populations exposed to acid
deposition in the Adirondack
Mountains. Environ. Toxicol. Chem. 11, 665±76.
Kramer, J., Newman, M.C., Mulvey, M. and Ultsch, G.R. (1989)
Glycolysis and Krebs cycle
Population genetics in mosquito®sh 353
-
metabolites in mosquito®sh, Gambusia holbrooki, Girard 1852,
exposed to mercuric chloride:
allozyme genotype effects. Environ. Toxicol. Chem. 11,
357±64.
Kunz, B.A. and Kohalm, S.E. (1991) Modulation of mutagenesis by
deoxyribonucleotide levels. Ann.
Rev. Genet. 25, 339±59.
Lavie, B. and Nevo, E. (1982) Heavy metal selection of
phosphoglucose isomerase allozymes in
marine gastropods. Marine Biol. 71, 17±22.
Leberg, P.L. (1992) Effects of population bottlenecks on genetic
diversity as measured by allozyme
electrophoresis. Evolution 46, 477±94.
Lopez-Fanjul, C. and Villaverde, A. (1989) Inbreeding increases
genetic variation for viability in
Drosophila melanogaster. Evolution 43, 1800±4.
Lynch, M. (1990) The similarity index and DNA ®ngerprinting.
Mol. Biol. Evol. 7, 478±89.
Nadig, S. (1996) Genetic diversity of redbreast sun®sh in East
Fork Poplar Creek. Master's Thesis.
University of Tennessee: Knoxville, TN, USA.
Nei, M. (1987) Molecular Evolutionary Genetics. New York:
Columbia University Press.
Nei, M., Maruyama, T. and Chakraborty, R. (1975) The bottleneck
effect and genetic variability in
populations. Evolution 29, 1±10.
Nevo, E., Noy, R., Lavie, B., Bieles, A. and Muchtar, S. (1986)
Genetic diversity and resistance to
marine pollution. Biol. J. Linn. Soc. 29, 139±44.
Rudiger, H.W. (1991) Genetic susceptibility to DNA methylating
agents. In Ecogentics (P. Grandjean,
ed.) pp. 205±15. London: Chapman & Hall.
Scribner, K.T., Wooten, M.C., Smith, M.H., Kennedy, P.K. and
Rhodes, Jr O.E. (1992) Variation in life
history and genetic traits of Hawaiian mosquito®sh populations.
J. Evol. Biol. 5, 267±88.
Selander, R.K., Smith, M.H., Yanng, S.Y., Johnson, W.E. and
Gentry, J.B. (1971) Biochemical
polymorphism and systematics in the genus Peromyscus. I.
Variation in the old ®eld mouse
(Peromyscus polionotus). Studies in Genetics VI. Univ. Texas
Publ. 7130, 42±120.
Shugart, L.R. and Theodorakis, C.W. (1994) Environmental
genotoxicity: Probing the underlying
mechanisms. Environ. Health Perspect. 102, 13±17.
Shugart, L.R. and Theodorakis, C.W. (1996) The genotypic
diversity approach. J. Comp. Physiol.
Biochem. 113C, 273±6.
Skroch, P.W. and Nienhuis, J. (1992) RAPD protocol for the air
thermo-cycler. The Rapid Cyclist 1, 9.
Strittholt, J.R., Guttman, S.I. and Wissing, T.E. (1988) Genetic
variability of yellow perch (Perca
¯avescens) in Lake Erie and selected impoundments. In The
Biogeography of the Island Region
of Western Lake Erie (J.F. Downhower, ed.) pp. 246±57. Columbus,
OH: Ohio State Univ. Press.
Suter, G.W. (1990) Use of biomarkers in ecological risk
assessment. In Biomarkers of Environmental
contamination (J.F. McCarthy and L.R. Shugart, eds) pp. 419±28,
Boca Raton, FL: Lewis
Publishers.
Theodorakis, C.W., D'Surney, S.J., Bickham, J.W., Lyne, T.B.,
Bradley, B.P., Hawkins, W.E., Farkas,
W.L., McCarthy, J.F. and Shugart, L.R. (1992) Sequential
expression of biomarkers in bluegill
sun®sh exposed to contaminated sediment. Ecotoxicology 1,
45±73.
Theodorakis, C.W., Blaylock, B.G. and Shugart, L.R. (1996)
Genetic exotoxicology: I. DNA integrity
and reproduction in mosquito®sh exposed in situ to
radionuclides. Ecotoxicology 6, 205±18.
Trabalka, J.R. and Allen, C.P. (1977) Aspects of ®tness of a
mosquito®sh Gambusia af®nis exposed to
chronic low-level environmental radiation. Radiat. Res. 70,
198±211.
Weir, B.S. (1990) Genetic Data Analysis. Sutherland, MA: Sinauer
Associates.
Williams, J.G.K., Kubelik, A.R., Livak, K.J., Rafaski, J.A. and
Tingey, S.V. (1991) DNA
polymorphisms ampli®ed by arbitrary primers are useful as
genetic markers. Nucleic Acids
Res. 18, 6531±5.
Wills, J.H. and Orr, H.A. (1993) Increased heritable variation
following populations bottlenecks: The
role of dominance. Evolution 47, 949±57.
354 Theodorakis and Shugart
IntroductionMethods and materialsChemicalsStudy sites/specimen
collectionRAPD analysisRAPD data analysisAllozyme
analysisResultsGenetic diversityGenetic distanceGenotype
distributionsGenotype and
fitnessDiscussionConclusionAcknowledgementsReferences