QhE),ohonf S9(10), 2005, pp 218 2199
BREEDING SYSTEM, COLONY STRUCTURE, AND GENETIC DIFFERENTIATION IN
THE CAMPONOTUS FESTINATUS SPECIES COMPLEX OF CARPENTER ANTS
MICHAEL A. D. GOODISMAN1'2'3'4 AND DANIEL A. HAHN1'5'6
1Center for Insect Science, Universitv ofArizona, Tucson, Arizona 85721
Department of Biochemistry and Molecular Biophysics, University of Arizona, Tucson, Arizona 85721
3School of ', The Georgia Institute of Technology, Atlanta, Georgia 30332
5hnterdisciplinary Graduate Program in Insect Science, University of Arizona, Tucson, Arizona 85721
6Department of Entomology and Nematology, University of Florida, Gainesville, Florida 32608
Abstract.-All social insects live in highly organized societies. However, different social insect species display striking
variation in social structure. This variation can significantly affect the genetic structure within populations and,
consequently, the divergence between species. The purpose of this study was to determine if variation in social
structure was associated with species diversification in the Camponotus festinatus desert carpenter ant species complex.
We used polymorphic DNA microsatellite markers to dissect the breeding system of these ants and to determine if
distinct C. festinatus forms hybridized in their natural range. Our analysis of single-queen colonies established in the
laboratory revealed that queens i :., ,i: mated with only a single male. The genotypes of workers sampled from a
field population suggested that multiple, related queens occasionally reproduced within colonies and that colonies
inhabited multiple nests. Camponotusfestinatus workers derived from colonies of the same form originating at different
locales were strongly differentiated, suggesting that gene flow was ..:.1- .: I restricted. Overall, our data indicate
that C. festinatus populations are highly structured. Distinct C. festinatus forms possess similar social systems but are
genetically isolated. Consequently, our data suggest that diversification in the C. Jfstinatus species complex is not
necessarily associated with a shift in social structure.
Key words.-Formicidae, genetic structure, microsatellites, polyandry, polygyny, relatedness, social insects, speciation.
Received November 3, 2004. Accepted July 12, 2005.
Social insects are remarkable because they form highly
cooperative societies (Wilson 1971; Oster and Wilson 1978).
However, social insect species differ in many important as-
pects of their social biology. For example, species-level var-
iation occurs in the number of reproductive within colonies,
the number of nests that single colonies inhabit, and the re-
lationships of colonymates (Holldobler and Wilson 1990;
Bourke and Franks 1995; Crozier and Pamilo 1996; Pamilo
et al. 1997). Ai!h ..:,. h such variation in social structure is
well documented, surprisingly little is known about how var-
iation in social system H.:hi. i.. speciation. This is unfor-
tunate because changes in social structure alter the genetic
structure of populations, which may affect species diversi-
fication (Endler 1977; Barton and Clark 1990; Frank 1998;
Avise 2004). In addition, variation in colony- and population-
genetic structure may alter the strength of kin selection and
kin conflict, potentially leading to changes in the helping
behavior of workers, the skew in progeny production among
reproductive, and the sex ratio produced by colonies (Bourke
and Franks 1995; Crozier and Pamilo 1996; Frank 1998;
Keller and Reeve 1999).
One of the key determinants of genetic structure within
social insect populations is the breeding system of social
groups (Ross 2001). The breeding system incorporates sev-
eral aspects of the social system, including the number of
breeders within groups, their relatedness, and how they par-
tition reproduction. Colonies of many hymenopteran social
insects are headed by a single queen (monogyne colonies)
that has mated only once. However, in more complex sys-
tems, the number of female reproductive in the colony and
the number of males with which queens mate may vary. The
presence of multiple queens polygynyy) decreases the relat-
edness of individuals within colonies and creates within-col-
ony genetic substructure. In addition, variation in queen num-
ber is associated with a number of important life-history traits
that likely influence speciation, including aggression toward
nonnestmates and queen dispersal ability (Keller 1993; Bour-
ke and Franks 1995; Ross and Keller 1995; Crozier and Pam-
The number of nests that a single colony inhabits is fre-
quently associated with colony queen number and may have
additional consequences for the evolution of life-history traits
related to speciation (Bourke and Franks 1995; Crozier and
Pamilo 1996). Members of single colonies in monogyne spe-
cies frequently reside within a single nest (monodomy), and
the spatial distribution of these nests is thought to affect
behaviors such as territoriality, brood raiding, and intraspe-
cific parasitism (Bourke and Franks 1995; Gadau et al. 2003).
In contrast, the functional colony units of many polygyne
species are separated into multiple nests (polydomy; Bourke
and Franks 1995; Crozier and Pamilo 1996). Spreading the
colony among multiple nests allows polydomous species to
control larger areas (Pamilo and Rosengren 1984; Herbers
1990, 1993). In addition, polydomy can ,:1 genetic struc-
ture and alter kin conflict within colonies (Herbers 1990;
Pamilo 1990), which may, consequently, affect speciation.
The number of males with which a queen mates may also
influence evolutionary diversification. Like colony queen
number, multiple mating by queens (polyandry) generally
leads to a decrease in nestmate relatedness relative to single
mating monandryy) and potentially lowers inclusive fitness
benefits received by workers who help raise colonymates.
C 2005 The Society for the Study of Evolution. All rights reserved.
M. A. D. GOODISMAN AND D. A. HAHN
However, polyandry can be advantageous if it increases the
variability of the workforce or reduces conflict within the
colony (Boomsma and Ratnieks 1996; Crozier and Fjerdings-
tad 2001: Strassmann 2001; Brown and Schmid-Hempel
Between-species hybridization can also substantially affect
population genetic structure among social groups. Hybrid-
ization yields a mosaic of novel genotypes, which can pro-
duce phenotypes that may differ in fitness from either parental
species (Barton and Hewitt 1985, 1989; Barton and Gale
1993; Harrison 1993; Arnold 1997). In social insects, hy-
bridization is often associated with remarkable changes in
social system. For example, inter- or intraspecific hybridiza-
tion has been implicated as an important factor in ant caste
determination (Julian et al. 2002; Volney and Gordon 2002;
Cahan and Keller 2003; Cahan and Vinson 2003), as a caus-
ative agent of changes in bee behavior (Clarke et al. 2002;
Schneider et al. 2003, 2004), and as a contributor to unusual
speciation or gene flow dynamics within populations (Good-
isman and Asmussen 1997; .I. ... et al. 1998, 2000;
Hochberg et al. 2003). These studies highlight the importance
of understanding multiple levels of genetic structure between
closely related species, particularly in areas where these spe-
cies occur in sympatry.
To understand how social structure influences speciation,
it is first necessary to have a good working knowledge of the
evolutionary relationships among the social taxa of interest.
However, such information in social insects is often severely
limited by our current ability to positively identify closely
related species and infer their phylogenetic relationships
(Jenkins et al. 2001; Arevalo et al. 2004; Hunt and Carpenter
2004; Ye et al. 2004). This problem is particularly apparent
in the ants (Ilymenoptera: Formicidae). Ecologically impor-
tant in almost all terrestrial ecosystems, ants possess an as-
tounding range of life-history diversity, including :.:i. i, i.. ,
in queen number, queen mating status, colony founding
mechanisms, mating strategies, reproductive skew, and caste
ratios (IIolldobler and Wilson 1990; Bourke and Franks 1995;
Crozier and Pamilo 1996). In fact, there are numerous ex-
amples of species that differ significantly in their ecology or
life histories but that cannot be clearly differentiated by mor-
phology alone (Creighton 1950; Umphrey 1996; MacKay and
MacKay 1997; Feldharr et al. 2003; Goropashnaya et al.
2004; Kronauer et al. 2004b).
In this study, we attempted to understand the factors con-
tributing to the population-genetic structure of a single spe-
cies complex within the ant genus Camponotus and determine
if diversification in the complex is associated with changes
in social structure. Camponotus is the second most diverse
genus within the ants and has been shown to be ecologically
important in numerous locales worldwide (Holldobler and
Wilson 1990; Bolton 1995; Brady et al. 2000). However, the
systematics of several groups within Camponotus is unclear
(Creighton 1950; Snelling 1968, 1970, 2000; MacKay and
MacKay 1997, 2000). Indeed, significant genetic structuring
has been found within some groups that are morphologically
indistinguishable, suggesting that molecular genetic data are
a necessary complement to morphological data for under-
standing the evolutionary relationships within Camnponotus
and other ant genera (Bolton 1995; Umphrey 1996; Brady et
al. 2000; Feldharr et al. 2003; Goropashnaya et al. ** I).
Our study focuses on the Camponotus.istinatus (Buckley)
species complex, whose members are distributed throughout
the southwestern United States and northern Mexico (Creigh-
ton 1950; '.. ...._ 1968). The morphological variation pre-
sent in this group has long been recognized and has led sev-
eral authors to partition this variance into consistent varieties,
subspecies, or forms (Emory 1893; Wheeler 1902, 1910).
This paper focuses on the colony and population-genetic
structure of three of the C. festinatus fomis that live in south-
The two desert forms, C. nr. Jistinatus desert dark (here-
after desert dark) and C. nr. festinatus desert light (hereafter
desert light), occur in the deserts of southern Arizona ranging
from approximately 700 to 1500 m in elevation. The two
forms differ morphologically in that the cuticle of the desert
dark form is mostly brown, while that of the desert light form
tends to yellow. These colors do not reflect the maturation
process of the ants, but rather represent relatively discrete
differences in adult pigmentation. Indeed. the desert dark and
light forms can also be reliably distinguished by other mor-
phological features, such as the length of their antenna
scapes, and have been shown to differ in their life histories
with respect to allocation to fat storage (A. N. Lazarus, D.
A. Hahn, S. P. Cover, and J. J. Wernegreen, unpubl. data;
). A. Hahn, unpubl. data). Moreover, the distinct forms have
never been found occupying the same nest.
The third form included in this study, C. nrfestinatus mid-
elevation, is typically found in the oak-grassland transition
zone in southern Arizona. The midelevation form generally
occurs at slightly higher elevations (1200 to 2500 m) than
the desert forms, although there is a zone of overlap where
the desert and midelevation forms occur in sympathy. Mid-
elevation individuals are distinguished from the two desert
forms by the head shape of their minor workers, which tapers
more drastically behind the eyes, and the presence of fully
erect hairs on their antennal scapes, rather than the decumbent
hairs characteristic of the desert forms. The occurrence of
the morphological and life-history differences among forms
suggests that they could be separate species.
We investigated the 1 .... n. and social structure of the
desert dark, desert light, and midelevation forms of C. fes-
tinatus. Specifically, we analyzed the genotypes of workers
reared in laboratory colonies to assess whether queens mated
with multiple males. We then examined worker genotypes in
field-collected colonies to determine if multiple queens in-
habited single colonies. We were particularly interested in
determining if the 1... i,.) system of the C. feslinatus forms
differed, a result that would indicate that selection might have
operated differently in the i -i 1 forms. Next, we developed
a new method of spatial analysis that facilitated the identi-
fication of colony boundaries and genetic structure among
colonies within a field site. We also investigated whether the
forms of C. festinatus were genetically distinct and whether
they formed hybrids in a natural population. Finally, we pro-
vide general suggestions for untangling the relationships of
species complexes in other social insect groups.
GENETIC STRUCTURE IN CA IfPONOTUS ANTS
FIG. 1. Locations of 31 field-sampled Camponotusfestinatus nests. Nests containing desert dark and midelevation form ants are denoted
by filled circles with black lettering and empty circles with gray lettering, respectively. Circled pairs of nests contain workers belong
to single colonies.
MATERIALS AND METHODS
We analyzed worker genotypes in both laboratory and field
colonies to examine the breeding system of C. festinatus.
Single-queen colonies were established in the laboratory by
introducing newly mated queens, captured in the vicinity of
Tucson, Arizona, into trays. The desert dark and desert light
forms of C. festinatus were chosen for laboratory analysis
because queens of these forms are readily captured after their
mating flights and reproduce under laboratory conditions.
Colonies were allowed to develop for 3-5 years in the lab
at 28-30C and ambient humidity. Ants were fed a diet of
frozen Nauphoeta cinerea or 0Manduca sexta larvae three
times a week, and 50% honey solution with 50 mg each of
Vanderzants vitamins (Sigma, St. Louis, MO) and Wesson's
Salt Mixture per 100 ml (Sigma). Approximately 15 adult
workers were sampled from each colony. In addition, the
reproductive queens from three colonies were sacrificed for
Camponotus festinatus workers of the desert dark and mid-
elevation forms were obtained from field colonies sampled
from a single site. We selected these forms for study, because
they are known to occur sympatrically in relatively high den-
sities (Fig. 1). The collection site was located opposite Syc-
amore Canyon in the Atascosa Mountains in Santa Cruz
County near Nogales, Arizona (3126.29'N, 11111.00'W),
at an elevation of 1255 m above sea level. The field site was
60-80 km from the location where the laboratory-reared
queens were captured. Between two and 27 adult worker ants
were collected from nests located under small rocks through-
out an area of 7500 m2. Worker larvae were also obtained
from two nests. All samples were placed in 95% ethanol for
preservation and subsequent genetic analysis.
Genomic DNA was extracted from ants using a modifi-
cation of the Chelex protocol (Walsh et al. 1991) as described
by Crozier et al. (1999). Polymerase chain reaction (PCR)
amplification was used to determine the genotypes of indi-
vidual ants at the four polymorphic microsatellite loci,
Cconl2, Ccon20, Ccon42, and Ccon70, which were origi-
nally cloned from the congener Camponotus consubrinus
(Crozier et al. 1999). In addition, genotypes were assayed at
the polymorphic microsatellite locus Cfesl, which was de-
tected in an EST screen of C. festinatus (Genbank accession
CK656570; forward amplification primer 5'-GAGA-
AGTCTAGAGAAAAGT-3'; reverse amplification primer 5'
TTTGTCAA1TGTTTAATAAGTG-3'). The PCR products
were -N i I by end-labeling the forward primers of Cfes1,
Cconl2, Ccon20, Ccon42. and Ccon70 with the -.. -....
dyes 6-FAM, HEX, HEX, 6-FAM, and NED, respectively.
PCRs were conducted in a final volume of 10 pl containing
102 1 4
0 o 113
10o0 0110 _
121 50 meters
136.0o01 0 N
1340 0 1
US 1"0 140
oj1 50 meters
M. A. D. GOODISMAN AND D. A. HAHN
1 pl genomic DNA and 0.5 U Taq DNA polymerase (New
England Biolabs, Ipswich, MA), and a final concentration of
200 gM dNTPs, 0.5 gM of each of the forward and reverse
PCR primers, and IX New England Bioloabs PCR buffer.
The PCR cycling profiles for the five markers began with
an initial denaturation at 94C for 2 min, and then proceeded
with 40 cycles of 94C for 30 sec, 45C for 45 sec, and 72C
for 45 sec, followed by a final extension of 72C for 10 min.
PCR products from all loci were combined in a 10 pl cocktail
containing 1.0-p[, 2.0 pl, 2.0 L, 2.0 p1, and 3.0 p1 of Cfesl,
Cconl2, Ccon20, Ccon42, and Ccon70 PCR product, re-
spectively. Two .... i. :.1. of this cocktail were combined
with a labeled size standard, electrophoresed, and scored on
an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems,
Foster City, CA).
Statistical A analyses
We calculated the genetic diversity at each microsatellite
locus for the laboratory and field colonies independently.
Allele frequencies were estimated using the program GE-
NEPOP 3.2 (Raymond and Rousset 1995), weighting esti-
mates from each colony equally. The variability at each locus
was quantified by Nei's (1987) estimate of gene diversity.
We directly examined the genotypes of workers sampled
from the monogyne laboratory colonies to determine the min-
imum number of males with which each queen mated. In
general, male mate number is readily determined in Hyme-
noptera, because males are haploid and females are diploid.
However, multiply mated queens can be overlooked if in-
sufficient numbers of worker offspring are sampled and if
male mates contribute unequally to offspring production. We
estimated the probability of missing male mates to determine
if such errors were likely under our sampling scheme. In
addition, we calculated the probability that we failed to detect
male mates of the queen, because the queen's putative mates
possessed identical genotypes. This likelihood is quantified
by the nondetection error, which is the :. ii that two
males will share the same multilocus genotype. In ,: ...,-
ploid taxa such as the Hymenoptera, this probability is the
product of the sum of the squared allele frequencies at each
microsatellite locus (Boomsma and Ratnieks 1996).
Workers within social insect colonies are typically related
and, therefore, should not be considered independently of
each other in population analyses. To avoid the potential
problems caused by this genetic nonindependence, we used
a resampling technique that yielded unbiased measures of
population genetic structure. Specifically, we created 10 new
datasets, each of which contained genetic information from
a single worker randomly selected from each colony. We used
these reduced datasets for most analyses, and took the arith-
metic mean from the analyses of the 10 datasets as our best
estimates for population parameters and associated proba-
We examined the laboratory-reared workers for evidence
of genetic differentiation between the desert dark and desert
light forms. We used probability or exact tests, as imple-
mented by the program GENEPOP 3.2, to determine if there
was significant allelic or genotypic differentiation between
the color forms (Raymond and Rousset 1995). We also used
GENEPOP 3.2 to calculate the magnitude of differentiation
between the color forms through the estimation of the sta-
tistics FsT and Psi, the latter being an estimator of genetic
differentiation that considers variation in allele size as an
informative measure of divergence.
We examined the genotypic structure of workers from each
form to determine if the samples were derived from a single,
randomly mating population. If samples were derived from
distinct populations, then an increase in homozygosity rel-
ative to Hardy-Weinberg expectations might arise through
the Wahlund effect (Hartl and Clark 1989). We used both
probability tests, where the probability of the observed sam-
ple of genotypes defines the rejection region, and score tests,
where the alternative hypotheses are defined as either excess
or deficiency of heterozygotes, to estimate the significance
of deviations from Hardy-Weinberg expectations ,i' ..,..
and Raymond 1995). The magnitude of disequilibrium was
quantified by the statistic Pis.
'1he relatedness of laboratory-reared workers was esti-
mated using the program Relatedness 5.0 (Queller and Good-
night 1989). We expected the relatedness of workers to equal
0.75 if workers were full sisters produced by a single, once-
mated queen. Lower estimates would be obtained if queens
were frequently polyandrous. Standard errors for estimates
were obtained by jackknifing over colonies, and t-tests were
used to determine if calculated relatedness estimates differed
from specified values (i.e., 0.75) or from each other. We also
used the estimate of worker relatedness from laboratory col-
onies to calculate the effective mating frequency of queens,
which incorporates information on both the number of times
a queen mates and the unequal contribution of her male mates
to progeny production (Starr 1984).
Camponotus festinatus workers collected from distinct
field-sampled nests may belong to single colonies (polydo-
my). To determine if polydomy was common, we tested for
genetic structure among nests in our field site in two ways.
First, we examined the relationship between pairwise geo-
graphic distance between nests and genetic distance, F, as
estimated by Weir and Cockerham's (1984) method using
GENEPOP 3.2. The magnitude of the correlation was mea-
sured by Spearman's correlation coefficient, rs, and the sig-
nificance of the correlation was determined by a Mantel test.
The Mantel test, which was carried out by GENEPOP, com-
pares the matrices of genetic and geographic distances to
determine if the observed correlation differs from correlations
obtained by permuting the observed data. If polydomy was
common and the effects of polydomy ranged over relatively
large distances, we expected a significant, positive relation-
ship between pairwise genetic distances and geographic dis-
Second, we examined the data for evidence of genetic
structure over varying geographic scales. The motivation for
this technique is derived from the possibility that genetic
structure may be limited to certain spatial scales within sites.
For example, if new colonies within sites are founded by
GENETIC STRUCTURE IN CAfJPONOTUS ANTS
queens partaking in extensive nuptial flights, and colonies
inhabit multiple nests, then genetic similarity of individuals
from nests may only be detectable over the distance that a
single colony occupies (i.e., a few meters). That is, the re-
lationship between genetic similarity and geographic distance
may be limited, because workers from distinct colonies will
not be related regardless of their proximity. In this case,
extensive queen dispersal coupled with limited colony range
would lead to genetic structure on a microgeographic scale
only, and a widespread relationship between geographic and
genetic distance, as predicted under standard models of iso-
lation by distance, would not be detectable (Rousset 1997).
To test for this possibility, we compared estimates of ge-
netic differentiation (F) for pairs of nests that were separated
by a specific range of .:-.: ,.-.. (L)1 to L)2 meters, with L)2
> Di) to pairs of nests that were separated by distances
greater than that range (D)2 to meters; Goodisman and Hahn
2 -' i). We first calculated estimates of F for all pairs of nests
in the field site. We then calculated (1) the mean F-value for
all pairs of nests that were separated by at least Di meters
but no more than D2 meters and (2) the mean F-value for all
pairs of nests separated by more than D2 meters. The dif-
ference between (1) and (2), defined as dif-D, is expected to
be negative if individuals from nest pairs separated by Di to
D2 meters are more similar to each other than individuals
from nests separated by distances greater than L)2 meters. For
example, if nests occupied by a single colony generally lie
within 5 m of one another, then nests separated by Di = 0
to D2 = 5 m should show lower values of genetic differen-
tiation (F) than nests separated by more than D2 = 5 m.
The significance of the test statistic, dif-D, was calculated
using a resampling protocol whereby nest pairs were ran-
domly assigned values of F from the original dataset with
replacement. We then calculated the new test statistic rdif-
D, which represented the difference in the mean values of F
for the randomly constructed dataset. This procedure was
repeated 10,000 times. We considered there to be evidence
of genetic structure if the observed value of dif-D was less
than 5% of the calculated rdif-D values.
If evidence of polydomy was apparent from either of the
statistical tests above, we attempted to group nests into true
colony units. To determine which nests belonged to distinct
colonies, we compared the genotypic frequencies of workers
from all pairs of nests by means of an exact test using the
program GENEPOP 3.2 (Goodisman and Crozier 2002). As
was the case with the laboratory workers, the mean estimate
obtained from 10 datasets containing a single worker per
colony was used for calculations. Significant differentiation,
after correcting for the I.1..; 1. tests. ,. i .1 (Rice 1989),
suggested that workers sampled from distinct nests did not
belong to the same colony. Nonsignificant differentiation was
generally taken as evidence that workers were derived from
the same colony. However, because few workers were sam-
pled from some nests (e.g., nests 102, 104, 106, 108, 111,
117, and 139), leading to a substantial loss in power, we used
partially subjective means to determine colony boundaries.
Specifically, nests that contained relatively few workers and
were separated by considerable distances were not necessarily
viewed as belonging to the same colony unit, even if the
workers from the nest pair were not significantly differen-
tiated. Any error of misidentifying colony boundaries (i.e.,
considering individuals from distinct nests as belonging to
distinct colonies, when, in fact, they belonged to the same
colony) would be conservative in that it would decrease our
estimates of nestmate relatedness relative to their true values.
After grouping workers from distinct nests into colony
units, we attempted to determine the social structure of C.
festinatus field colonies. We examined the genotypes of work-
ers to determine if one or many queens contributed to worker
production in C. festinatus colonies. Single queens were con-
sidered to head colonies if the workers therein possessed a
maximum of three alleles at any locus and their genotypes
conformed to those expected under Mendelian segregation of
alleles from a singly mated diploid female and haploid male.
This method could result in missing some instances of mul-
tiple matrilines within nests if few workers were sampled
from nests or if reproducing queens were closely related.
As was the case with the laboratory colonies, we use 10
reduced datasets, each of which consisted of a single ran-
domly selected worker from each field colony, for population
analyses. We took the mean of the 10 datasets as our best
estimates for population parameters. We tested for Itardy-
Weinberg disequilibrium within forms using GENEPOP 3.2
to determine if workers from each form were part of the same
population. We expected to find disequlibrium if colonies of
different forms were not part of the same randomly mating
We next examined the field samples to determine if the
desert dark and midelevation forms were genetically differ-
entiated. In addition, allele and genotype frequencies were
compared between laboratory and field-sampled ants to detect
differentiation within and between C. festinatus forms. The
significance of allelic and genotypic differentiation between
the forms was estimated using exact tests, and the magnitude
of differentiation, as measured by the statistics FST and pST,
was also determined.
The program RELATEDNESS 5.0 was used to estimate
the relatedness of workers from field-sampled C. festinatus
colonies. Standard errors for estimates were obtained by j ack-
knifing over colonies. We used t-tests to determine if relat-
edness estimates differed from specified values or if the re-
latedness of desert dark and midelevation form workers dif-
The method of Pritchard et al. (2000) was used to establish
if field-sampled desert dark and midelevation form workers
originated from more than a single interbreeding group. This
algorithm uses a Bayesian clustering method to assign in-
dividuals to K distinct populations while simultaneously es-
timating allele frequencies in those populations. The method
assumes that alleles from distinct populations will be in Har-
dy- ..M.. and linkage equilibrium; deviations from equi-
librium within a sample signal the presence of population
structure. We were specifically interested in determining if
sampled ants could be i, i.',ii' *:..1I according to form and
-*.. I.. -! Ill- distinguishing the presence of hybrids.
Two distinct models of population structure were consid-
ered within the clustering framework. In the first model, we
considered the probability of observing our genotypic data
given K = 1-10 populations and no admixture between pop-
ulations. The second model allowed for the possibility that
M. A. D. GOODISMAN AND D. A. HAHN
TABLE 1. Nei's gene diversity (and number of alleles) at five microsatellite loci in four samples of Camponotus festinatus ants.
Laboratory desert Laboratory desert Field desert Field
Locus dark light dark midelevation
Cfesl 0.86 (8) 0.93 (14) 0.58 (3) 0.83 (10)
Cconl2 0.90 (11) 0.94 (14) 0.22 (2) 0.87 (10)
Ccon20 0.53 (4) 0.48 (4) 0.37 (4) 0.14 (5)
Ccon42 0.38 (8) 0.86 (9) 0 (1) 0.62 (6)
Ccon70 0.97 (23) 0.92 (17) 0.62 (5) 0.35 (5)
Mean 0.73 (10.8) 0.83 (11.6) 0.36 (3.0) 0.56 (7.2)
individuals resulted from admixture from the K = 1-10 the-
oretical populations. In both cases, the program Structure 2.1
(available via ],ii: /pritch.bsd. uchicago.edu/structure.html)
was used to derive the most likely value of K. In addition,
estimates of the probability of population membership, for
the no-admixture model, or fractional membership of each
individual into each of the K theoretical populations, for the
admixture model, were also obtained. Simulations assumed
correlated allele frequencies within populations, and burning
and simulation length were set to 10,000.
AMicrosatellite Locus Quality'
A concern when applying microsatellite markers to pop-
ulation genetic data is the occurrence of false signals of evo-
lutionary processes that may originate from errors in the gen-
eration of the data themselves. For example, a deficit of het-
erozygotes, which typically signals the presence of selection,
population structure, or inbreeding, can also arise through
errors in the PCR amplification of alleles. Two such errors
include null alleles and allele dropout. A null allele at mi-
crosatellite locus arises when an allele fails to PCR-amplify
because it carries a mutation in the region of primer binding.
In contrast, allele dropout takes place when particular pairs
of alleles occur in a single heterozygous individual. If the
alleles differ substantially in i 1I i1i then the shorter allele
may outreplicate the longer allele during PCR, ultimately
....~,i to the longer allele remaining undetected. In either
case, individuals that are heterozygous for distinct alleles
would incorrectly be scored as homozygous, thereby inflating
estimates of homozygosity (Jones and Ardren 2003).
There is no evidence that null alleles or allele dropout
influenced our study of genetic variation in C. festinatus.
High-frequency null alleles would have been evident in our
analysis of parentage in laboratory colonies. If null alleles
were present at high ,! ...-,...'. then some proportion of mat-
ings would occur between queens heterozygous for normal
alleles and males carrying the null allele (e.g., AB queen X
0 male). Such matings would result in progeny with a 1:1
ratio of one of the two maternal alleles and the null paternal
allele (worker genotypes AO and i *). The genotypes of these
individuals would be naively interpreted as homozygous for
the maternal alleles (AA and BB). However, a 1:1 ratio of
homozygous genotypes among progeny is not possible in a
haplodiploid genetic system. A singly mated queen cannot
produce this genotypic ratio, and a multiply mated queen
would be expected to produce some heterozygous genotypes.
Consequently, the presence of two, distinct homozygote ge-
notypes at a given locus within a family would immediately
be suspect. However, our analysis of family groups revealed
no instances of such anomalous genotypic ratios. Therefore,
we have no evidence that null alleles are present in our da-
Our data also did not show evidence of allele dropout. If
allele dropout occurred frequently at our loci, then we would
expect many putative homozygotes to consist of short alleles.
Such a pattern would result from .':i!i!.- .. .I. bias in het-
erozygotes where the allele of shorter size would outreplicate
the longer allele during PCR amplification. We tested whether
the smallest allele present in putative homozygotes was sig-
nificantly smaller than the smallest allele present in hetero-
zygotes, as expected .: I ..!. :, frequently resulted from
allele dropout of larger alleles. We considered all five loci
for all four samples ( I.. : : ..i..i desert/midelevation),
resulting in a total of 18 statistical tests (some loci in some
populations were monomorphic). We found that the smallest
allele in homozygous genotypes was significantly greater
than that smallest allele in heterozygous genotypes in 14
cases, in direct contrast to the effect produced by allele drop-
out. In three cases there was no significant difference between
homozygous and heterozygous allele size, and in one case
the smallest allele for homozygotes was significantly smaller
than that in heterozygotes. Consequently, these calculations
indicate that allele dropout did not occur in our dataset. Over-
all, our analyses -..-..-I that our data are robust and not
affected by errors arising from PCR amplification of micro-
The multilocus genotypes of 510 workers were assayed
from 37 laboratory colonies (13.78 1.53 workers per col-
ony, x SD) established by single queens. The genotypes
of the three desert dark form queens were also determined.
Workers from 26 and 11 of these laboratory colonies were
designated as desert dark and desert light, respectively, based
on morphological analysis. In addition, 327 workers from 31
nests were sampled from our field site (13.55 + 7.51 workers
per nest). Workers from eight of these nests were categorized
as desert dark, while those from the remaining 23 nests were
categorized as midelevation (Fig. 1). Identification was based
on phenotypic (i.e., color, morphology) information. No nest
contained workers of more than one form.
The five microsatellite markers used in this study proved
to be highly variable, :'! ....h there were substantial dif-
ferences in variability among samples (Table 1). In total, the
GENETIC STRUCTURE IN CAfJPONOTUS ANTS
TABLE 2. Genetic differentiation between Camponotus festinatus
forms. Values above and below the diagonal are estimates of FST
and PST, respectively. All forms differed significantly from each
other (P < 0.0001), as determined by allelic and genotypic prob-
Laboratory desert dark
Laboratory desert light
Field desert dark
loci Cfesl, Cconl2, Ccon20, Ccon42, and Ccon70 possessed
22, 22, 9, 16, and 36 alleles, respectively. The field samples
displayed lower diversity than the laboratory samples, with
the field desert dark forms having the lowest variation as a
group. The distribution of allele frequencies among the four
samples also varied substantially (see Appendix available
online only at http://dx.doi.org/10.1554/04-672.1.sl).
Levels of ''.'.., ',.' faating by Queens
Direct analysis of the genotypes of workers from the 37
laboratory colonies allowed us to estimate the number of
males with which queens mated. We first considered the prob-
ability that we would underestimate male mate number be-
cause of the finite variation of our markers. Considering the
desert dark and desert light laboratory samples indepen-
dently, the I-.l-..i.l; that two males would have the same
haploid genotype given the variability of our markers was
0.00047 and 0.00013, respectively. Consequently, it is un-
likely that we failed to detect male mates because of the
insufficient variability of our markers. Another potential
source of error could arise due to the process of sampling.
That is, we may have failed to detect male mates by chance
because our worker sample size was finite. However, we
found that this probability was also relatively low. For ex-
ample, if both males contributed equally to progeny produc-
tion, then the probability of sampling 14 workers (the mean
number sampled per colony) from the same patriline is 2 X
0.514 = 0.00012. Skew in male reproductive success would
need to be substantial, with one male .1 i!.,i',. approxi-
mately 75% of '..' -.. ,-_ for the probability of missing a
male mate through sampling effects to exceed 0.05. There-
fore, we do not expect sampling biases to contribute major
error to our estimate of male mate number.
We found that most queens from our laboratory colonies
mated with a single male. In total, the genotypes of workers
from 32 of the 37 nests were consistent with having been
produced by a singly mated queen. Four of the 37 nests
contained workers that would have required double mating
by the queen, and one nest contained workers that would
have required triple mating by the queen. Therefore, the mean
number of mates per queen from these data comes to 1.16.
However, upon closer inspection, we found that the worker
genotypes from the colonies headed by three of the putatively
doubly mated queens and the triply mated queen could more
parsimoniously be explained by mutation events or by a sin-
gle contaminating worker sample. Under these assumptions,
only one of the 37 queens was multiply (doubly) mated, and
our best estimate of mean number of mates per queen falls
Genetic Structure of Labtoratoy Samples
We examined the laboratory samples for evidence of ge-
netic differentiation between the two desert forms. We found
evidence for highly significant allelic and genotypic differ-
entiation between desert dark and desert light C. festinatus
(P < 0.0001). The magnitude of dissimilarity was also fairly
substantial, with our estimates of differentiation equaling Fs'T
= 0.09 and Ps'r = 0.39 (Table 2). The relatively large estimate
of PST was caused, in part, by the locus-specific estimates at
Cfesl and Ccon70, where the two color forms differed not
only in the actual alleles present but also in the size ranges
of alleles (online Appendix).
We next tested for evidence of deviations from Hardy-
Weinberg equilibrium within laboratory desert color forms.
The applied probability test revealed no significant disequi-
librium in the desert dark form (FIs = -0.01, P > 0.40) and
weak but not statistically significant evidence for disequilib-
rium in the desert light form (Fis = 0.16, P < 0.10). Because
of the marginally significant result, we conducted further
analysis using the statistically more :: .: one-sided tests
with predefined rejection regions of heterozygous excess or
deficit. We found that the desert light form possessed a mod-
estly significant deficit of heterozygotes (heterozygote ex-
cess: desert light, P > 0.9; desert dark, P :> 0.1; heterozygote
deficit: desert light, P < 0.05; desert dark, P > 0.8).
The average relatedness of workers sampled from the lab-
oratory colonies was 0.74 + 0.01 (r + SEM) when differences
in allele frequencies between the color forms were considered
(Table 3). This estimate of nestmate relatedness did not differ
significantly from the value of 0.75 expected if all colonies
were headed by a single, once-mated queen (t36 = 0.86, P
> 0.35), and yielded an estimate of effective mating fre-
quency of 1.02. In addition, worker relatedness of desert dark
TABLE 3. Relatedness of Camponotus festinatus workers from laboratory and field colonies.
Laboratory Laboratory All laboratory Field desert Field All field
Locus desert dark desert light samples dark midelevation samples
Cfesl 0.72 0.81 0.76 0.79 0.70 0.72
Cconl2 0.71 0.79 0.73 0.83 0.62 0.63
Ccon20 0.78 0.54 0.72 0.75 0.89 0.83
Ccon42 0.73 0.82 0.78 1.00 0.64 0.64
Ccon70 0.73 0.66 0.71 0.75 0.69 0.71
Mean SEM 0.73 + 0.03 0.75 + 0.02 0.74 + 0.01 0.77 + 0.06 0.68 0.11 0.69 0.09
M. A. D. GOODISMAN AND D. A. HAHN
FIG. 2. Relationship between pairwise estimates of genetic distance and geographic distance for Camponotusfestinatus nests.
and desert light C. festinatus did not differ significantly (t34
= 0.60, P > 0.5). As expected, the relatedness estimates for
the one colony that was clearly headed by a doubly mated
queen (r = 0.49) was lower than the value of 0.75 and quite
close to the estimate of 0.5 expected if the doubly mated
queen used the sperm of each male equally. Nevertheless,
consideration of the entire dataset supports the hypothesis
that C. festinatus queens almost always mate once.
Social Structure of Field Samples
We examined the relationship between pairwise estimates
of genetic differentiation for ants collected from distinct nests
and the distance between nests to determine if polydomy was
common in field populations of C. festinatus. The correlation
between nest distance and genetic similarity of workers with-
in nests (rs = 0.179) differed significantly from zero as
judged by a Mantel test (P = 0.014; Fig 2).
To better understand the nature of the significant corre-
lation, we compared the pairwise estimates of differentiation
for nests separated by DZ to D2 meters to the differentiation
of nests separated by distances greater than D2 meters. As
expected under a simple model of polydomy, significant ge-
netic structure was observed over relatively short distances
(< 10 m) in both the desert dark and midelevation forms as
well as both forms combined (Table 4). The signal grew
weaker at moderate distances (25-50 m) but then reemerged
as the distances between nests increased. This unusual pattern
was observed in both the desert dark and midelevation forms,
suggesting that it may reflect some real aspect of colony
movement or colonization.
Because our analysis of genetic structure suggested that
C. festinatus sometimes formed polydomous colonies, we at-
tempted to group individuals sampled from distinct nests into
true colonies of related individuals. We used the significance
of genotypic differentiation between nests in conjunction
with the proximity of nests to determine if individuals be-
longed to the same colony. Our analyses revealed four in-
stances of polydomy. Specifically, we determined that work-
ers from the following pairs of nests belonged to the same
colonies: 105 and 106, 107 and 108, 111 and 112, 124 and
130 (Fig. 1).
We examined the genotypes of field-sampled workers to
determine the breeding system of natural C. festinatus col-
onies. We assumed that field-sampled desert dark and mid-
elevation C. festinatus queens mated with only a single male.
as indicated by our study of laboratory colonies. This as-
sumption is justified, even in the midelevation form, as queen
mate number (i.e., whether queens mate multiply vs. singly)
is not an evolutionarily labile trait. Typically, all species
within a hymenopteran social insect genus are either mo-
nandrous or polyandrous (although the extent of : .. 1iili,
mating does vary among species within polyandrous genera;
see Strassmann 2001).
We found that the genotypes of workers from 20 of the
27 colonies conformed to the simplest type of social system
found in hymenopteran social insects; these nests appeared
to be headed by a single, once-mated queen. However, the
genetic structure in the remaining seven colonies (101, 103,
118, 121, 124/130, 141, and 142, where "/" indicates that
the two members of the two nests belong to the same colony)
was more complex. The genotype of workers from all these
colonies required either reproduction by multiple queens or
the mating of a single queen with at least three males. Because
only one of the 37 laboratory-reared queens was found to
mate multiply (see above), the latter possibility is unlikely.
Consequently, it appears that C. festinatus colonies some-
times contain multiple matrilines.
Another notable outcome of our analysis was that all seven
of the colonies that contained i1: I ..,_ from multiple ma-
trilines were midelevation form. This result initially sug-
gested that the desert dark and midelevation forms differed
in social structure. However, because the majority (21 of 27)
of the colonies sampled were midelevation in our field pop-
ulation, the probability that all seven multiple-matriline col-
GENETIC STRUCTURE IN CA IXPONOTUS ANTS
TABLE 4. Genetic structure of Camponotusfestinatus nests over varying spatial scales. In this analysis, the mean genetic differentiation
of pairs of nests separated by distances of D1 to D2 meters was compared against the mean genetic differentiation of pairs of nests
separated by distances greater than D meters. The resulting difference in means ( .' .D) was compared against the distribution of differences
in means for random nest pairs at various .; ,,;. ... limits i ... 0). The number of nest pairs separated by distances of D1 to D2 meters
and distances greater than D2 meters are given. Significant differences are given in bold.
Number of nest
pairs separated by
D2 D1 to D, meters
Number of nest
pairs separated by
more than D, meters
onies would be midelevation by chance is not particularly
small (calculated as the probability that all polygyne nests
would be midelevation given that seven nests were polygyne;
P = 21/27 X 20/26 ... 15/21 = 0.13). In addition, the
variability of the desert dark field samples was substantially
lower than that of the midelevation samples (Table 1), a factor
that would diminish our ability to detect multiple matrilines
in desert dark ants. Most importantly, we note that the re-
latedness of desert dark and midelevation form ants did not
differ significantly (Table 3). Therefore, there is no con-
vincing evidence that the breeding system of C. festinatus
desert dark and midelevation form ants differs.
Genetic Structure of Field Samples
We analyzed the field-sampled workers' genotypes for ev-
idence of genetic structure within and between the desert dark
and midelevation forms. The genotype frequencies were in-
consistent with Hardy-Weinberg equilibrium in the midele-
vation form (P < 0.005, midelevation; P > 0.7, desert dark).
Subsequent analysis revealed that this departure from equi-
librium was due to a heterozygote deficit (P < 0.0001, mid-
elevation; P > 0.7, desert dark) and not a heterozygote excess
(P > 0.9, midelevation; P > 0.3, desert dark). These signif-
icant deviations were accompanied by a relatively poor match
between the expected and observed frequency of heterozy-
gotes under Hardy-Weinberg equilibrium in the two forms
(FTS = 0.16, midelevation; FTs = -0.17, desert dark).
Tests for genetic differentiation between the forms were
unambiguous (Table 2). The allele and genotype frequencies
differed significantly between all forms sampled (allelic and
genotypic tests for all pairwise comparisons; P < 0.0001).
This significant differentiation was consistent with the large
magnitude of differentiation estimated by PST and FST be-
tween samples (Table 2). Of interest, we discovered that the
desert dark field samples were strongly differentiated from
the desert dark laboratory samples. We also note that the
estimates of PST were substantially greater than the estimates
of F'sr, a result that reflects that the sizes of the alleles, in
addition to the frequencies of the alleles, differed between
the samples (online Appendix).
The relatedness of C. festinatus field sampled ants was
relatively high (Table 3). The overall estimate of relatedness
of workers (0.69 0.03) was not significantly different from
the value of 0.75 expected if all colonies were headed by a
single, once-mated queen (t26 = 0.66, P > 0.5). However,
the average relatedness estimate for colonies deemed to con-
tain multiple matrilines (0.44) was substantially lower than
the overall mean estimate. We found that the mean related-
ness of desert dark form ants (0.77) and midelevation form
ants (0.68) did not differ significantly (t4 = 0.70, P > 0.5).
M. A. D. GOODISMAN AND D. A. HAHN
Dark form o Mid-elevation form
N .V / /\ O\ o
FIG. 3. Probability of membership of desert dark and midelevation
Camponotus festinatus ants into populations 1, 2, and 3 of K = 3
theoretical populations under a model of no-admixture.
This further substantiated the hypothesis that social structure
of desert dark and midelevation C. Jiestinalus did not differ.
Analyses of the field data using a model-based clustering
algorithm yielded additional insights into the nature of dif-
ferentiation of the forms. Using the no-admixture model, the
field-sampled C. festinatus ants were clustered into K = 3
populations with high : 1 .1 :1;! (P > 0.999, Fig. 3). The
probability of the number of populations equaling some other
value of K was very low (P < 0.001 for all other probabilities
combined). The no-admixture clustering-algorithm succeed-
ed in sorting the ants into their color forms based on genotype
(Fig. 3). The desert dark form ants were strongly differen-
tiated from the midelevation form ants. In addition, all the
desert dark form ants were determined to be part of a single
theoretical population (population 2), as they all lie in one
of three corners of the ternary diagram in Figure 3. In con-
trast, the midelevation form ants showed more complex char-
acteristics. Although midelevation form ants were :,s ...
from desert dark form ants, midelevation form ants were
apparently comprised of mixtures of genes from two popu-
lations, theoretical population 1 and population 3 (Fig. 3).
The results of the clustering algorithm with admixture lent
some support to two different estimates of K. The ."- -:- ,l-i..
that the field samples were derived from K = 2 populations
under the no-admixture model was fairly high (P 0.84).
In addition, the K = 2 population model successfully clas-
sified desert dark and midelevation form C. festinatus into
distinct groups, with the exception of colony 116, which was
misclassified (Fig. 4). However, the admixture model also
provided some support to a model of K = 3 populations (P
0.18; the probability of K = I or 4 < K < 10 was P <
0 1 1 1 1 1 1 I I I I I F---
0t -0 d- M^ r '0 00 1 0 0 ff) 10 oo 01 O
Dark form Mid-elevation form
FIG. 4. -' :.* .:;.. membership of desert dark and midelevation form Camponotus festinatus ants into population 1 ofK =2 theoretical
populations under a model of admixture. The proportional membership of each individual into population 2 equals one minus the
proportional membership in population 1 (1 P1). Based on morphological analysis, ants from the first seven and last 20 colonies listed
were originally classified as desert dark and midelevation C. festinatus, respectively.
GENETIC STRUCTURE IN CAIIUPONOTUS ANTS
Population genetic structure .1:.., I the evolution of life-
history traits, which may subsequently affect species diver-
sification in social organisms (Endler 1977; Barton and Clark
1990; Bourke and Franks 1995; Crozier and Pamillo 1996;
Frank 1998; Avise 2004). We used a combination of labo-
ratory and field studies to explore the genetic structure of a
social insect taxon at a variety of levels including within
colonies, among colonies in a single field site, between spa-
tially separated populations, and across morphologically dis-
tinguishable forms. Using highly polymorphic DNA micro-
satellite markers, we successfully resolved questions con-
cerning the i-......i '-, structure, nesting biology, spatial struc-
ture, and species status within this complex group. We also
gained a greater understanding of whether changes in social
system were associated with diversification.
Our analysis of queen mate number in both desert dark
and desert light C. festinatus strongly indicated that queens
typically mated only once. The genotypes of workers sampled
from 36 of the 37 monogyne laboratory colonies were par-
simoniously explained by queen monandry. Under such a
scenario, worker relatedness would be expected to be close
to 0.75. Indeed, our estimates of worker relatedness (0.73
and 0.75 for desert dark and desert light, respectively) were
very close or equal to this value.
Single mating by desert dark and desert light C. festinatus
queens is consistent with results from other studies in Cam-
ponotus (Gadau et al. 1996, 1998; Seppa and Gertsch 1996;
Satoh et al. 1997; Crozier et al. 1999; Goodisman and Hahn
2004) and most other '- :..:..:- .. .. social insects (Bourke
and Franks 1995; Boomsma and Ratnieks 1996; Crozier and
Pamilo 1996; Strassman 2001). Single-mating appears to be
an ancestral trait in the social Hymenoptera, and substantial
polyandry has only evolved in the ant genera Atta, Acro-
myrmex, Dorylus, and Pogonomyrmex, the bee genus Apis,
and wasp genus Vespula (Schmid-Hempel 1995; Boomsma
and Ratnieks 1996; Crozier and Fjerdingstad 2001; Strass-
mann 2001; Brown and Schmid-Hempel 2003; Kronauer et
al. 2004a). Presumably, multiple mating incurs considerable
costs in time and energy and increases the risk of predation
or contraction of I or parasites. Consequently, selec-
tion for multiple mating likely occurs only when ecological
conditions dictate that it is important to acquire additional
sperm for reproduction, produce a genetically variable worker
force, or achieve a reduction in intracolony conflict (Booms-
ma and Ratnieks 1996; Strassmann 2001; Brown and Schmid-
Another important feature of the breeding structure of in-
sect societies is the number of queens cohabiting within col-
onies (Keller 1993). If we assume that desert dark and mid-
elevation form queens from our field populations were singly
mated, as was the case in the desert dark and desert light
laboratory colonies, then analysis of worker genotypes in-
dicates that workers from approximately 25% of C. :. ... ..
field-sampled colonies were derived from multiple matrilines.
This suggests that multiple queens inhabited colonies, the
primary reproductive queen within colonies changed over
time, or brood raiding occurred in this species.
Ants in the genus Camponotus generally display life-his-
tory characteristics associated with monogyny, such as strong
aggression toward nonnestmates, widespread dispersal
through extensive nuptial flights, worker caste polymor-
phism, and independent colony foundation. In contrast, spe-
cies whose colonies are normally headed by multiple queens
often exhibit low aggression toward nonnestmates, relatively
weak dispersal, an absence of worker caste polymorphism,
and reproduction through colony budding (H11lldobler and
Wilson 1990; Bourke and Franks 1995; Crozier and Pamilo
1996). However, recent studies, primarily conducted with
molecular markers, have uncovered several cases of polygyny
or queen replacement in Camponotus (Satoh 1989; Carlin et
al. 1993; Akre et al. 1994; Gertsch et al. 1995; Gadau et al.
1998, 1999; Fraseret al. 2000). These ,. !,, show that social
structure may be more complex than previously appreciated
in some Camponotus species, and the association between
colony queen number and other life-history traits likely oc-
curs as a continuum rather than the dichotomous suites of
traits drawn above.
i- ..! xI averages for nestmate relatedness in both desert
dark and midelevation field-sampled C. jestinatus colonies
were high (Table 3), despite the evidence for multiple ma-
trilines within colonies. Moreover, relatedness within colo-
nies containing workers from multiple matrilines was also
fairly high (r = 0.44), suggesting that queens within nests
were related. Indeed, queens cohabiting within nests are re-
lated in most social insects (Crozier and Pamilo 1996). The
exceptions tend to occur in nonequilibrium situations, such
as when a social insect has recently invaded a new environ-
ment or occupies a :. :, :...I habitat (HIlldobler and Wilson
1990; Bourke and Franks 1995; Crozier and Pamilo 1996).
S.. 'Patterns of f: Genetic :
in the Field
Single social insect colonies sometimes inhabit multiple
nests, a syndrome known as polydomy. Polydomy is common
in ants and may occur if nest space is limiting, nests are
frequently disrupted, moving into multiple nesting sites in-
creases foraging efficiency, or if intracolony -. .:1. 1 can be
reduced by expanding into multiple nests (IIHlldobler and
Wilson 1977, 1990; Pamilo and Rosengren 1984; Banschbach
and Herbers 1996a,b; Cerda et al. 2002). In natural popu-
lations, a signature of polydomous colonies is the discovery
of genetic structuring over short geographic distances.
We detected genetic structure within our field site and were
able to group workers from distinct nests into single colonies
based on their genotypes (Fig. 1). Our analysis of genetic
patterns within our field site showed that pairs of nests that
were separated by less than 10 m were significantly more
similar than pairs separated by more than 10 m (Table 4).
This pattern of structure fell off as the distances between nest
pairs increased (10-50 m). This drop-off in signal is expected
in polydomous ants if colonies inhabit multiple nests that
remain in proximity (C .i .-.. r., and Crozier 2002). Poly-
domy may be facilitated by the nesting biology of C. festin-
atus, which resides in loose soil under rocks or fallen trees,
M. A. D. GOODISMAN AND D. A. HAHN
rather than forming conspicuous mounds (Cokendolpher
1990). In addition, this result is consistent with the obser-
vation that polydomy is frequently associated with polygyny
in ants (Bourke and Franks 1995; Crozier and Pamilo 1996).
Our analysis of genetic structure also uncovered additional
patterns within the field site. Unexpectedly, workers from
pairs of nests separated by relatively large .i:-i ,,..-.- (e.g..
50-150 m in the midelevation form) were significantly more
similar than expected by chance alone. This pattern was ev-
ident in both desert dark and midelevation forms, suggesting
that it is not simply an artifact of one of the datasets. It is
unlikely that these patterns are signatures of extensive po-
lydomy or colony budding in this species for two reasons.
First, C. festinatus colonies are relatively small, making po-
lydomy or budding an unlikely source for spatial structure
at this scale. Second, polydomy or budding would be ex-
pected to leave patterns of genetic structure over intermediate
distances (i.e., 10-50 m) if it left such patterns over greater
It is difficult to reconcile the patterns of genetic structuring
with the known biology of this species. A possible expla-
nation for these unusual patterns is that they -1:.. -the way
this site was colonized over a period of years. This site was
located on a slope in the foothills of the Atascosa Mountains
in the high desert grassland-to-oak transition zone near a dirt
road and a cattle loading pen. This area could have been
affected by both natural events such as fire or flooding, and
,..i .-.. =_..- factors such as grazing or pesticide applica-
tion. Colonizing queens from any one year may have been
derived from relatively few parent colonies and, therefore,
would have formed a related group of individuals when com-
pared to colonizing queens from other years. In addition, this
site may have been colonized in patches of approximately
100 m2, such that one patch within the site was colonized
per year. Under such a scenario, workers sampled from col-
onies within patches would be more similar than workers
sampled from colonies between patches, and genetic simi-
larity of nests over fairly large distances could result. Testing
this complex scenario, however, would require monitoring
the spatial distribution and demography of colonies in both
intentionally disturbed and undisturbed sites over a period
of many years.
Genetic Differentiation within Forms
We discovered strong genetic differences between the lab-
oratory- and field-sampled desert dark form. Indeed, the mag-
nitude of differentiation between the field and laboratory de-
sert dark form exceeded some estimates of differentiation
between forms (Table 2). This suggests that spatial or tem-
poral variation in allele frequencies exists within forms of
the C. festinatus species complex, as the samples were col-
lected at distinct sites in different years. Additional evidence
of genetic differentiation within forms arises from the fact
that the genetic diversity of laboratory desert dark samples
was higher than that of the field desert dark samples (Table
1). Moreover, laboratory desert light ants displayed nonran-
dom association of alleles at the microsatellite loci, sug-
gesting a deficit of heterozygotes relative to Hardy-Weinberg
equilibrium. Such patterns could result from the Wahlund
effect (Hartl and Clark 1989), whereby ants were sampled
from multiple distinct populations resulting in a heterozygote
deficit. Overall, these data suggest that forms within this
species group may still be poorly defined, and more data from
other populations may be required to gain a full understand-
ing of the magnitude of variability in this species.
Genetic D .:-- : .. .... between Forms
The three forms of C. festinatus ants possessed distinct
complements of alleles (Table 2). Moreover, the field- and
laboratory-sampled desert dark ants, which were collected
from distinct sites, also differed significantly in allele fre-
quencies. Measures of differentiation between the forms that
incorporated information on allele size (psT) were substan-
tially larger than those that considered only allele identity
(FsT). This result indicates that gene flow between the forms
has been restricted for a substantial period of time (Hardy et
al. 2003). In addition, we were able to group the field-sampled
ants into discrete theoretical populations that largely coin-
cided with form, and our analyses did not uncover evidence
of hybridization between the desert dark and midelevation
forms. Therefore, based on the combined lines of evidence,
we consider the desert dark, desert light, and midelevation
forms of C. festinatus to be reproductively isolated. Signif-
icant genetic structure has similarly been shown among pop-
ulations separated by comparable distances in another tax-
onomically controversial group, the Florida carpenter ant, C.
floridanus (Deyrup et al. 1988; Gadau et al. 1996).
Our analyses suggested that the field-collected desert dark
form ants belonged to a single .,:.i -i._._;.:.,- group (Figs. 3,
4). The midelevation form ants, however, displayed some-
what more variation. There was some evidence from the ge-
netic data that individuals classified as midelevation form
originated from multiple populations. This is distinctly pos-
sible, given the current difficulty in morphologically distin-
guishing the different forms within this species complex and
the observed differentiation between field- and laboratory-
sampled desert dark form ants. Workers from one midele-
vation form, field-sampled colony (116, Fig. 4) were deemed
to be genetically distinct from other midelevation form ants.
Further inspection of this colony indicated a very high es-
timate of colony-specific relatedness relative to other mid-
elevation form colonies (r = 0.97). However, ants from this
nest were also Ii n1. I from desert dark form ants; the small-
est estimate of differentiation between workers from the focal
nest and workers from any other field-sampled nests was F
= 0.68. These data indicate that individuals from this colony
were strongly differentiated from the other C. festinatus sam-
ples. Consequently, it appears that workers from this nest
may originate from a breeding group separate from desert
dark, desert light, and midelevation C. festinatus.
In summary, our results indicate that substantial restric-
tions in gene flow occur within and among C. festinatus
forms. These findings suggest that population -.,=. .... .
may be more common in ants than previously recognized. In
addition, such patterns of genetic structure may have im-
portant impacts on speciation, life-history evolution, and the
distribution morphological variation within ants.
GENETIC STRUCTURE IN CA tPONOTUS ANTS
One of the aims of this study was to determine if species
diversification in social insect taxa was associated with
changes in social structure. We found that distinct forms
within C. festinatus were genetically differentiated, suggest-
ing incipient or complete speciation. However, we found no
evidence that the forms i1 1 .. ...I in their breeding or social
structure. Consequently, our data indicate that speciation in
ants can occur without significant changes in social structure.
In addition, by documenting the .".'- ,1. structure, nest
structure, and spatial genetic structure within C. festinatus,
we have taken the first steps toward understanding the evo-
lution of this species complex. Current research on C. fes-
tinatus is focused on understanding the I1, 1 .-, :: 1, of
the group by using a combination of morphological and mo-
lecular genetic analyses of the ants and their intracellular
endosymbiotic bacteria (A. N. Lazarus, D. A. Hahn, S. P.
Cover, and J. J. Wernegreen, unpubl. data) and characterizing
the life-history differences displayed by the different forms
(D. A. Hahn, unpubl. data). These studies will allow for the
reconstruction of the evolutionary history of the C. ." ..... ..
species complex, which is critical for understanding the ob-
served variation within this group. The large geographic
range covered by this species complex combined with the
apparent morphological and molecular divergences make C.
festinatus a promising model for studying speciation and se-
lective forces associated with speciation events in social in-
Using our experiences with the C. festinalus species com-
plex as a guide, we make the following recommendations for
others who want to understand evolutionary relationships
within complex social insect taxa. A i.. ;. .1 i morphological
analysis should be the first step in characterizing any group
suspected to contain multiple distinct lineages. This will re-
quire careful examination of multiple castes from many dis-
tinct nests and geographic locations. Statistical approaches
to morphological analyses have been used successfully to
separate taxa within ants in the past, and new analytical tools
derived from studies of functional morphology and devel-
opmental allometry should facilitate further progress in using
continuous characters in systematic studies (Snelling 1968;
Umphrey 1996; Zelditch et al. 1 I).
In addition to morphological analysis, molecular genetic
tools should be used to understand patterns of gene flow
within taxa. Numerous markers, including variable DNA mi-
crosatellites, AFLPs, and single nucleotide polymorphisms,
can be useful in this arena. The type of marker, in addition
to the number of markers and samples necessary, will vary
depending on the group of interest and the level of analysis.
However, we make the following conservative recommen-
dations. Microsatellite markers are particularly useful be-
cause they are codominant and tend to possess many alleles,
properties that lead to substantial power in determining col-
ony boundaries and population differentiation. A thorough
molecular genetic study will include analysis of the breeding
system, spatial analyses of colony and population structure,
and phylogenetic analyses and biogeography. To thoroughly
analyze -'i.-...::i system and distinguish colony boundaries,
we recommend sampling 30 individuals per nest unit and 30
nest units per site. When considering multiple taxa, samples
should be compared from multiple areas of both allopatry
and sympathy. In addition, molecular data derived from sym-
biotic organisms should also be included for comparison
when possible (Clark et al. 2000; Abbot and Moran 2002).
We acknowledge that gathering evidence from all of the
above sources is a daunting task, but such complex studies
may be required to understand the factors contributing to
evolutionary diversification of social taxa.
We are grateful to M. A. Wells and D. E. Wheeler for
providing the laboratory facilities and resources necessary
for this study. We thank A. Cutter, B. J. Haeck, J. K. IHatt,
J. Krenz, and C. Schmidt for help collecting ants; N. Buck
for caring of ants; A. Lazarus and S. Cover for identification
of the three forms; M. Deyrup, B. Johnson, C. Johnson, and
J. Hunt for useful discussion; and A. Lazarus, U. Mueller, J.
Weregreen, and four anonymous reviewers for helpful com-
ments on the manuscript. This research was supported in part
by a National Institutes of Health Postdoctoral Excellence in
Research and Teaching fellowship administered through the
Ilniversity of Arizona Center for Insect Science (GM00708-
Abbot, P., and N. A. Moran. 2002. Extremely low levels of genetic
polymorphism in endosymbionts (Buchnera) of aphids (Pem-
phigus). Mol. Ecol. 11:2649-2660.
Akre, R. D., L. D. Hansen, and E. A. Myhre. 1994. Colony size
and polygyny in carpenter ants (Hymenoptera: Formicidae). J.
Kans. Entomol. Soc. Am. 67:1-9.
Arevalo, E., Y. Zhu, J. M. Carpenter, and J. E. Strassmann. 2004.
The phylogeny of the social wasp subfamily Polistinae: evidence
from microsatelite flanking sequences, mitochondrial COI se-
quence, and morphological characters. BMC Evol. Biol. 4:8.
Arnold, M. L. 1997. Natural hybridization and evolution. Oxford
Univ. Press, Oxford, U.K.
Avise, J. C. 2004. Molecular markers, natural history, and evolution.
Sinauer, Sunderland, NMA.
Banschbach, V. S., and J. M. Herbers. 1996a. Complex colony
structure in social insects. I. Ecological determinants and genetic
consequences. Evolution 50:285 297.
---. 1996b. Complex colony structure in social insects. II. Re-
production, queen-worker conflict, and levels of selection. Evo-
lution 50:298 307.
Barton, N. II., and A. G. Clark. 1990. Population structure and
processes in evolution. Pp. 115-173 in K. Wohrmann and S. K.
Jain, eds. Population biology: ecological and evolutionary view-
points. ..;. _... Berlin.
Barton, N. I., and K. S. Gale. 1993. Genetic analysis of hybrid
zones. Pp. 13 45 in R. G. Harrison, ed. Hybrid zones and the
evolutionary process. Oxford Univ. Press. New York.
Barton, N. II., and G. M. Hewitt. 1985. Analysis of hybrid zones.
Annu. Rev. Ecol. Syst. 16:113-148.
1989. Adaptation, speciation and hybrid zones. Nature 341:
Bolton, B. 1995. A new general catalogue of the ants of the world.
Belknap Press, Cambridge, MA.
Boomsma, J. J., and F. L. W. Ratnieks. 1996. Paternity in eusocial
Hymenoptera. Philos. Trans. R. Soc. Lond. 351:947-975.
Bourke, A. F. G., and N. R. Franks. 1995. Social evolution in ants.
Princeton Univ. Press, Princeton, NJ.
Brady, S. G., J. Gadau, and P. S. Ward. 2000. Systematics of the
ant genus Camponotus: a preliminary analysis using data from
the mitochondrial gene cytochrome oxidase I. Pp. 131-139 in
M. A. D. GOODISMAN AND D. A. HAHN
A. D. Austin and M. Downtown, eds. Hymenoptera: evolution,
biodiversity, and biological control. CSIRO Press, Collings-
Brown, M. J. F., and P. Schmid-Hempel. 2003. The evolution of
female multiple mating in social Hymenoptera. Evolution 57:
Cahan, S. II.. and L. Keller. 2003. Complex hybrid origin of genetic
caste determination in harvester ants. Nature 424:306-309.
Cahan, S. H., and S. B. Vinson. 2003. Reproductive division of
labor between hybrid and nonhybrid offspring in a fire ant hybrid
zone. Evolution 57:1562-1570.
Carlin,N. F., H. K. Reeve, and S. P. Cover. 1993. Kin discrimination
and division of labour among matrilines in the polygynous car-
penter ant, Camponotus planatus. Pp. 362-401 in L. Keller, ed.
Queen number and sociality in insects. Oxford Univ. Press, Ox-
Cerda, X., A. Dahbi, and J. Retana. 2002. Spatial patterns, temporal
variability, and the role of multi-nest colonies in a monogynous
Spanish desert ant. Ecol. Entomol. 27:7-15.
Clark, M. A., N. A. Moran, P. Baumann, and J. J. Wernegreen.
2000. Cospeciation between bacterial endosymbionts (Buchnera)
and a recent radiation of aphids (Uroleucon) and the pitfalls of
testing for phylogenetic congruence. Evolution 54:517-525.
Clarke, K. E., T. E. Rinderer, P. Franck, J. G. Quezada-Euan, and
B. P. Oldroyd. 2002. The Africanization of honeybees (Apis
L.) of the Yucatan: a study of a massive hybridization
event across time. Evolution 56:1462 1474.
Cokendolpher, J. C. 1990. The ants (Hymenoptera, Formicidae) of
western Texas. II. Subfamilies Ecitoninae, Ponerinae, Pseudo-
myrmecinae, Dolichoderinae, and Formicinae. Texas Tech Univ.
Press, Lubbock, TX.
Creighton, W. S. 1950. Ants of North America. Harvard Univ. Press,
Crozier, R. H., and E. J. Fjerdingstad. 2001. Polyandry in social
Hymenoptera: Disunity in diversity? Ann. Zool. Fenn. 38:
Crozier, R. H., and P. Pamilo. 1996. Evolution of social insect
colonies: sex allocation and kin selection. Oxford Univ. Press,
Crozier, R. H., B. Kaufmann, M. E. Carew, and Y. C. Crozier. 1999.
Mutability of microsatellites developed for the ant Camponotus
consobrinus. Mol. Ecol. 8:271-276.
Deyrup, M. A., N. Carlin, J. Trager, and G. Umphrey. 1988. A
review of the ants of the Florida Keys. Fla. Entomol. 71:
Emory, C. 1893. Beitroge zur Kentniss der nordamsrikanischen
Ameisen-fauna. Zool. Jahrb. Syst. 7:663-682.
Endler, J. A. 1977. Geographic variation, speciation, and dines.
Princeton Univ. Press, Princeton, NJ.
Feldhaar, H., B. Fiala, J. Gadau, M. Mohamed, and U. Maschwitz.
2003. Molecular phylogeny of Crematogaster subgenus Deca-
crema ants and the colonization of Iacarange trees. Mol. Phy-
logenet. Evol. 27:441 452.
Frank, S. A. 1998. Foundations of social evolution. Princeton Univ.
Press, Princeton, NJ.
Fraser, V. S., B. Kaufmann, B. P. Oldroyd, and R. H. Crozier. 2000.
Genetic influence on caste in the ant Camponotus consobrinus.
Behav. Ecol. Sociobiol. 47:188 194.
Gadau, J., J. Heinze, B. H6lldobler, and M. Schmid. 1996. Popu-
lation and colony structure of the carpenter ant Camponotus
floridanus. Mol. Ecol. 5:785-792.
Gadau, J., P. J. Gertsch, J. Heinze, P. Pamilo, and B. Holldobler.
1998. Oligogyny by unrelated queens in the carpenter ant, Cam-
ponotus ligniperdus. Behav. Ecol. Sociobiol. 44:23 33.
Gadau, J.. S. Brady, and P. S. Ward. 1999. Systematics. distribution,
and ecology of an endemic California Camponotus quercicola
(Hymenoptera: Formicidae). Ann. Entomol. Soc. Am. 92:
Gadau, J. C.-P. Strehl, J. Oettler, and B. H6lldobler. 2003. Deter-
minants of intracolonial relatedness in Pogonomyrmex rugosus:
mating frequency and brood raids. Mol. Ecol. 12:1931-1938.
Gertsch, P., P. Pamilo, and S.-L. Varvio. 1995. Microsatellites re-
veal high genetic diversity within colonies of Camponotus ants.
Mol. Ecol. 4:257 260.
Goodisman, M. A. D., and M. A. Asmussen. 1997. Cytonuclear
theory for haplodiploid species and X-linked genes. I. Hardy-
Weinberg dynamics and continent-island model. Genetics 147:
Goodisman, M. A. D., and R. H. Crozier. 2002. Population and
colony genetic structure of the primitive termite MIastotermes
darwiniensis. Evolution 56:70-83.
Goodisman, M. A. D., and D. A. Hahn. 2004. Colony genetic struc-
ture of the ant Camponotus ocreatus (Hymenoptera: Formicidae).
Goodisman, M. A. D., D. D. Shoemaker, and M. A. Asmussen.
1998. Cytonuclear theory for haplodiploid species and X-linked
genes. II. A .,. "-, .. models of gene flow and application
to a fire ant hybrid zone. Evolution 52:1423-1440.
Goodisman, M. A. D., K. G. Ross, and M. A. Asmussen. 2000. A
formal assessment of the effects of gene flow and selection in
the fire ant Solenopsis invicta. Evolution 54:606-616.
Goropashnaya, A. V., V. B. Federov, and P. Pamillo. 2004. Recent
speciation in the Formica rufa group ants: inference from mi-
tochondrial DNA phylogeny. Mol. Phylogenet. Evol. 32:
Hardy, O. J., N. Carbonnel, H. Freville, and M. Heuertz. 2003.
Microsatellite allele sizes: a simple test to assess their ; .::
chance on genetic differentiation. Genetics 163:1467-1482.
Harrison, R. G. 1993. Hybrid zones and the evolutionary process.
Cornell Univ. Press, Ithaca, NY.
Hart, D. IT., and A. G. Clark. 1989. 7'.4;... of population ge-
netics. Sinauer, Sunderland, MA.
Herbers, J. M. 1990. Reproductive investment and allocation ratios
for the ant Leptothorax longispinosus: sorting out the variation.
Am. Nat. 136:178-208.
1993. Ecological determinants of queen number in ants.
Pp. 262 293 in L. Keller, ed. Queen number and sociality in
insects. Oxford Univ. Press, Oxford, U.K.
Hochberg, M. E., B. Sinervo, and S. P. Brown. 2003. Socially me-
diated speciation. Evolution 57:154-158.
HI11dobler, B., and E. O. Wilson. 1977. The number of queens: an
important trait in ant evolution. Naturwissenschaften 64:8 15.
1990. The ants. The .. .. i,. Press of Harvard Univ. Press,
Hunt, J. H., and J. M. Carpenter. 2004. Intra-specific nest form
variation in some Neotropical swarm-founding wasps of the ge-
nus Parachartergus. J. Kans. Entomol. Soc. 77:448 456.
Jenkins, T. M., R. E. Dean, R. Verkerk, and B. T. Forschler. 2001.
Phylogenetic analyses of two mitochondrial genes and one nu-
clear intron region illuminate European subterranean gene flow,
taxonomy, and introduction dynamics. Mol. Phylogenet. Evol.
Jones, A. G., and W. R. Ardren. 2003. Methods of parentage anal-
ysis in natural populations. Mol. Ecol. 12:2511-2523.
Julian, G. E., J. 11. Fewell, J. Gadau, R. A. Johnson, andD. Larrabee.
2002. Genetic determination of the queen caste in an ant hybrid
zone. Proc. Natl. Acad. Sci. USA 99:8157-8160.
Keller, L. 1993. Queen number and sociality in insects. Oxford
Univ. Press, Oxford, U.K.
Keller, L., and H. K. Reeve. 1999. Dynamics of conflicts within
insect societies. Pp. 153 175 in L. Keller, ed. Levels of selection
in evolution. Princeton Univ. Press, Princeton, NJ.
Kronauer, D. J. C., C. .i, ,.;.-_ J. S. Pedersen, J. J. Boomsma,
and J. Gadau. 2004a. Extreme queen-mating frequency and col-
ony fission in African army ants. Mol. Ecol. 13:2381 2388.
Kronauer, D. J. C., B. Holldobler, and J. Gadau. 2004b. Phyloge-
netics of the new world honey ants (genus Myrmecocystus) es-
timated from mitochondrial DNA sequences. Mol. i. -._.
MacKay, W. P., and E. E. MacKay. 1997. A revision of the Neo-
tropical ants of the montivagus species complex, genus Cam-
ponotus, subgenus Myrmentoma. Sociobiology 30:319 334.
S2000. A note in response to Mr. Roy Snelling's review of
the I .... .. .. complex. Sociobiology 36:613-615.
GENETIC STRUCTURE IN CA/1IPONOTUS ANTS
Nei, M. 1987. Molecular evolutionary genetics. Columbia Univ.
Press, New York.
Oster, G. F., and E. O. Wilson. 1978. Caste and ecology in the
social insects. Princeton Univ. Press, Princeton, NJ.
Panilo, P. 1990. Sex allocation and queen-worker conflict in po-
lygynous ants. Behav. Ecol. Sociobiol. 27:31 36.
Pamilo, P., and R. Rosengren. 1984. Evolution of nesting strategies
of ants: genetic evidence from different population types of For-
mica ants. Biol. J. Linn. Soc. 21:331 348.
Pamilo, P., P. Gertsch, P. Thoren, and P. Seppa. 1997. Molecular
population genetics of social insects. Annu. Rev. Ecol. Syst. 28:
Pritchard, J. K., M. Stephens, and P. Donnelly. 2000. Inference of
population structure using multilocus genotype data. Genetics
Queller, D. C., and K. F. Goodnight. 1989. Estimating relatedness
using genetic markers. Evolution 43:258 275.
Raymond, M., and F. Rousset. 1995. GENEPOP (version 1.2): pop-
ulation genetics software for exact tests and ecumenicism. J.
Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution
Ross, K. G. 2001. Molecular ecology of social behaviour: analyses
of breeding systems and genetic structure. Mol. Ecol. 10:
Ross, K. G., and L. Keller. 1995. Ecology and evolution of social
organization: insights from fire ants and other highly eusocial
insects. Annu. Rev. Ecol. Syst. 26:631-656.
Rousset, F. 1997. Genetic differentiation and estimation of gene
flow from F-statistics under isolation by distance. Genetics 145:
Rousset, F., and M. Raymond. 1995. Testing heterozygote excess
and deficiency. Genetics 140:1413 1419.
Satoh, T. 1989. Comparisons between two apparently distinct forms
of Camponotus nawai Ito (Hymenoptera: Formicidae). Insectes
Satoh, T., K. Masuko, and T. Matsumoto. 1997. Colony genetic
structure in the mono- and polygynous sibling species of the
ants Camponotus nawai and Camponotus yamaokai: DNA fin-
gerprinting analysis. Ecol. Res. 12:71-76.
Schmid-Hempel, P. 1995. Parasites and social insects. Apidologie
Schneider, S. S., L. J. Leamy, L. A. Lewis, and G. DeGrandi-
Hoffman. 2003. The influence of hybridization between African
and European honeybeesApis -.. -..:.* on asymmetries in wing
size and shape. Evolution 57:2350 2364.
Schneider, S. S., G. DeGrandi-Hoffman, and D. R. Smith. 2004.
The African honey bee: factors contributing to a successful bi-
ological invasion. Annu. Rev. Entomol. 49:351 376.
Seppa, P., and P. Gertsch. 1996. Genetic relatedness in the ant
Camponotus herculeansus: a comparison of estimates from al-
lozyme and DNA microsatellite markers. Insectes Soc. 43:
:1... R. R. 1968. Studies on California ants. 4. Two species of
Camponotus. Proc. Entomol. Soc. Wash. 70:350-358.
1970. Studies on California ants. 5. Revisionary notes on
some species of Camponotus subgenus Tanemyrmex. Proc. En-
tomol. Soc. Wash. 72:390-397.
2000. A review of the Camponotus montivagus complex.
Starr, C. K. 1984. Sperm competition, kinship, and sociality in the
aculeate Hymenoptera. Pp. 427-464 in R. L. Smith, ed. Sperm
competition and the evolution of animal mating systems. Aca-
demic Press, New York.
Strassmann, J. 2001. The rarity of multiple mating by females in
the social Hymenoptera. Insectes Soc. 48:1-13.
Umphrey, G. J. 1996. Morphometric discrimination among sibling
species in the fulva-rudis-texana complex of the ant genus
Aphaenogaster. Can. J. Zool. 74:528 559.
Volny, V. P., and D. M. Gordon. 2002. Genetic basis for queen-
worker ::,1. i: h ,in a social insect. Proc. Natl. Acad. Sci.
USA 99:6108 6111.
Walsh, P. S., D. A. Metzger, and R. IIiguchi. 1991. Chelex 100 as
a medium for simple extraction of DNA for PCR-based typing
from forensic material. Biotechniques 10:506-513.
Weir, B. S., and C. C. Cockerham. 1984. Estimating F-statistics for
the analysis of population structure. Evolution 38:1358 1370.
Wheeler, W. M. 1902. A new agricultural ant from Texas with
remarks on the known North American species. Am. Nat. 36:
1910. The North American ants of the genus Camponotus.
Ann. NY Acad. Sci. 20:295-354.
Wilson, E. 0. 1971. The insect societies. Harvard Univ. Press, Cam-
Ye, W., C.-Y. Lee. R. H1. Scheffrahn, J. M. Aleong, N.-Y. Su, G.
W. Bennett, and M. E. Scharf. 2004. Phylogenetic relationships
of nearctic Reliculitermes species with particular reference to
Reticulitermes arenincola Goellner. Mol. Phylogenet. Evol. 30:
Zelditch, M. L., D. L. Swiderski, H. D. Sheets, and W. L. Fink.
2004. Geometric morphometrics for biologists: a primer. Elsev-
Corresponding Editor: U. Mueller