Genetic Determination of The Queen Caste in An Ant Hybrid Zone
Genetic Determination of The Queen Caste in An Ant Hybrid Zone
Genetic Determination of The Queen Caste in An Ant Hybrid Zone
hybrid zone
Glennis E. Julian*†, Jennifer H. Fewell*, Jürgen Gadau‡, Robert A. Johnson*, and Debbie Larrabee*
*Department of Biology, Arizona State University, Tempe, AZ 85287; and ‡Department of Behavioral Physiology and Sociobiology, Biozentrum,
University of Würzburg, Würzburg D-97074, Germany
Communicated by Bert Hölldobler, University of Würzburg, Würzburg, Germany, April 12, 2002 (received for review February 24, 2002)
The question of how reproductives and sterile workers differen- ants, Pogonomyrmex rugosus and Pogonomyrmex barbatus, which
tiate within eusocial groups has long been a core issue in socio- seems to be associated with hybridization. These two sister
biology because it requires the loss of individual direct fitness in species are widespread in deserts and grasslands of the south-
favor of indirect or group-level fitness gains. The evolution of central and southwestern United States and northern Mexico
social behavior requires that differentiation between workers and (15). Of the two species, P. barbatus has a more eastern
female reproductives be environmentally determined, because distribution, occurring from western Louisiana to central Ari-
genetically determined sterility would be quickly eliminated. Nev- zona, whereas P. rugosus occurs from central Texas to California.
ertheless, we report clear evidence of genetic caste determination The two species have broadly overlapping geographic ranges
in populations of two seed harvester ant species common to the from central and southeastern Arizona to western Texas, but
southwestern USA, Pogonomyrmex rugosus and Pogonomyrmex both species also inhabit large areas of allopatry (15).
barbatus. The genetic differentiation between workers and queens We used randomly amplified polymorphic DNA (RAPD)
is found only in areas of sympatry of the two species, and thus genetic markers to examine genotypic patterns among sibling
appears to arisen from hybridization. Our data suggest that this workers, alate queens (alates), and males from colonies of P.
hybridization has had a profound historical effect on the caste rugosus and P. barbatus in areas of sympatry and allopatry.
determination systems and mating patterns of each of these Morphological data suggest that these two ant species hybridize
species. in areas of overlap (16), and both species often possess mtDNA
of the sister species in these areas (17). Mating aggregations of
both species within contact zones have been observed to contain
A n overriding principle of social insect biology is that the
determination of reproductive (queens) versus sterile
(workers) individuals within a social group is primarily environ-
low numbers of congeneric reproductives, supporting the pos-
sibility of hybridization (18).
mentally governed (1–3). A sterile caste can evolve under kin Methods
selection only if the genes for sterility are expressed condition-
We collected workers, alates, and males from colonies of P.
ally, because any allele that invariably caused sterility could
rugosus and P. barbatus during the mating flight season (July
not be passed on and would be quickly eliminated from the
through September) in areas of sympatry and allopatry. Ants
population (4, 5).
were immediately frozen at ⫺80°C or placed in 100% ethanol.
Numerous studies have accumulated evidence that differences
Both P. barbatus and P. rugosus have one reproductive queen per
in larval environment, particularly larval nutrition, determine
colony, and queens of both species mate with multiple males
whether a female egg will develop into a fully capable repro- (18). Thus, all individuals collected from a given colony were
EVOLUTION
ductive or a small worker female (6, 7). When a nutritional or offspring of the same queen.
other physiological threshold is reached during larval develop- Our sympatic site was in southeastern Arizona (Cochise
ment, neurosecretory changes result in elevated juvenile hor- County) and southwestern New Mexico (Hidalgo County),
mone levels, triggering development of reproductive potential where we collected five colonies of P. rugosus and nine colonies
(8). Recent research has focused on identifying genes involved in of P. barbatus. Our sites for allopatric samples were in Texas and
the developmental cascade underlying this polyphenism (9). Arizona, where we collected 17 colonies of P. rugosus and seven
However, the hypothesis that the underlying mechanism for colonies of P. barbatus. Colonies of P. rugosus were collected in
queen-worker caste differentiation could itself become geneti- Maricopa County (six colonies), Pinal County (five colonies),
cally based lacks support. and Navajo County (six colonies), Arizona, whereas samples of
There have been rare reported exceptions of genetic influence P. barbatus were collected in Yavapai County (three colonies),
on caste determination; however, these have involved the gen- Arizona, and Tarrant County (three colonies) and Wichita
eration of different queen types in ants with polymorphic County (one colony), Texas.
queens, rather than the fundamental differentiation of queens DNA was extracted from each individual according to Landry
and workers. Allelic differences between regular, winged queens et al. (19) or Gadau (20); abdomens of workers and alates were
and an intermediate, wingless queen have been described in the removed before extraction. The RAPD PCR was performed
slave maker ant, Harpegoxenus sublaevis (10), and a genetically according to Williams et al. (21). We screened an initial colony
based queen polymorphism also has been demonstrated in a of workers, alates, and males of each species with 45 different
species of an Australian ant, Monomorium (11). A third case, 10-base random primers. We then chose a 10-bp primer OPC9
using allozymes as genetic markers, has demonstrated that a (Operon Technologies, Alameda, CA) showing a fragment-
second queen morph in Acanthomyops is actually a genotype length polymorphism that was useful for identifying heterozy-
resulting from hybridization (12). In Melipona bees, it has been gotes (22, 23). All colonies were then screened with the C9
suggested that queen determination is in part genetically con- primer. Heterozygous individuals displayed three different
trolled by double heterozygosity at two independent loci (13). bands, one at 510 kb, one at 550 kb, and a heteroduplex band at
However, this example is highly controversial (14), and the 590 kb. We verified that the marker was a codominant fragment-
environmental influence on caste remains strong because all
genotypes become workers if given insufficient food.
This article presents clear evidence for genetic queen-worker Abbreviation: RAPD, randomly amplified polymorphic DNA.
caste determination in populations of two species of harvester †To whom reprint requests should be addressed. E-mail: [email protected].
P. barbatus (Colony 5)
Primer (size) 1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5
C9 (600) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
R11 (300) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
R11 (400) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
D18 (490) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
S19 (1000) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
G4 (650) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
Q9 (850) 0 0 0 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
R11 (600) 0 0 0 1 0 0 1 1 0 0 0 0 1 1 0 1 0 0 0 1
Q9 (450) 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 1 0 1
P. rugosus (Colony D)
Primer (size) 1 2 3 4 5 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5
C9 (600) 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
P13 (630) 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
S8 (1300) 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
T20 (750) 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
N11 (650) 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
K4 (300) 1 0 1 0 0 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
D18 (495) 0 0 0 0 1 0 1 0 0 1 1 0 0 0 0 0 1 0 0 1
A1 (800) — 0 0 1 0 0 0 1 0 — 1 0 1 0 0 1 0 0 1 1
length polymorphism by mixing PCR-amplified DNA products locus, in which genotypic frequencies were similar for alates and
from males of each alternative allele, heating the samples, and workers (Table 2). As a result, the proportion of heterozygotes
running them on a gel, which allows reannealing of the hetero- differed significantly between the sympatric and allopatric
duplex. All three bands appeared, indistinguishable from a groups (Fig. 1). In P. barbatus within the contact zone, heterozy-
heterozygous individual. gosity was completely associated with caste, whereas in the
allopatric population heterozygosity was not different between
Results alates and workers (Fig. 1). Heterozygosity levels in allopatric P.
We found distinct intra-colonial differences between the geno- rugosus were low for both alates and workers, because the 550
types of workers and their reproductive siblings in colonies allele was rare. However, heterozygosity was not associated with
within the contact zone. In a sample colony of each species, 13 caste (Table 2), and some colonies contained both a heterozy-
primers revealed 16 variable markers; 12 displayed complete gous queen and heterozygous daughter alates.
linkage with the worker caste (Table 1). Specifically, workers
displayed bands that were absent in their alate queen sisters or
in the males. The most parsimonious explanation for these
genotype differences is that the workers and queens come from Table 2. Genotypes revealed by RAPD genetic marker C9 for
different patrilines. Because Hymenoptera are haplo-diploid, alate queens, workers, and males from sympatric and allopatric
workers from the same father share all markers specific to that populations of P. rugosus and P. barbatus
patriline. Males are produced parthenogenetically, and their
Homozygotes Heterozygotes Homozygotes
genotypes reveal the queen’s genotype. Therefore, markers
Caste 510兾510 510兾550 550兾550 Total
absent in males but present in workers are inherited patrilinially.
Not all markers showed this distinct caste difference; four P. rugosus (sympatric) (n ⫽ 5 colonies)
markers showed variation independent of caste and sex (Table Queens 15 4 5 21
1). Most likely the queens are heterozygous (present兾absent) for Workers 0 33 0 33
these markers and the genotypes of their mates are absent, Males 17 — 12 29
because males—representing the queens genotype—also segre- P. rugosus (allopatric) (n ⫽ 17 colonies)
gated for all of these markers. Queens 41 2 0 43
In addition, we found distinct differences between alates and Workers 56 5 0 61
workers by using a fragment-length polymorphism (C9–550兾 Males 46 — 1 46
510) for colonies within the contact zone. All workers in the P. barbatus (sympatric) (n ⫽ 9 colonies)
sympatric populations of both P. rugosus and P. barbatus were Queens 29 0 38 67
heterozygous for the C9 marker (Table 2, Fig. 1). In contrast, Workers 0 60 0 60
their alate sisters were significantly more likely to be homozy- Males 37 — 30 67
gous for one of the two alleles. All P. barbatus alates (n ⫽ 67) and P. barbatus (allopatric) (n ⫽ 7 colonies)
80% of P. rugosus alates (n ⫽ 21) were homozygous. All males Queens 4 6 5 15
within a given colony shared the same allele, indicating that the Workers 12 11 13 36
queens of colonies in the sympatric zone are all homozygous. Males 4 — 5 9
This allele matched the one present in their sibling alates. Genotypes were assigned based on the fragment-length polymorphic
The allopatric populations of both species showed a more marker that revealed two alleles of different lengths, one at 510 bp, the other
typical pattern of within-colony genetic variation for the C9 at 550 bp. Males are hemizygous and therefore possess only one allele.
EVOLUTION
were homozygous. An additional 14 RAPD primers also indi-
cated genotypic differentiation between alates and workers ity may not be unique to Pogonomyrmex. Hung and Vinson (27)
(Table 1). These data collectively suggest that the C-9–550兾510 presented allozyme evidence consistent with genetic caste de-
locus is a marker for a large genomic linkage group affecting termination in fire ants, Solenopsis geminata and Solenopsis
caste determination. xyloni. Because they did not have DNA-based markers, they
How can a system of genetic caste determination arise? We interpreted their results to suggest differential enzyme expres-
suggest that caste determination in P. rugosus and P. barbatus is sion between reproductive queens and workers.
related (at least historically) to hybridization. Our results show The ability to shunt nonconspecific offspring into sterile castes
an extreme genetic effect on caste within areas of sympatry, but offers eusocial species a unique mechanism for countering the
we found no similar pattern in allopatric populations of either negative fitness consequences of hybridization (24). Because
species. We hypothesize that ants in the sympatric zone have workers are valuable to colony success, a differential caste
responded to introgression with a unique form of colony-level or trajectory for conspecific versus nonconspecific sperm would
social hybridogenesis (24). This hybridogenesis is similar on a offset negative fitness consequences of interspecies mating.
colony level to that found on an organismal level in certain Shunting heterospecific sperm into workers would also limit the
species of guppies (Poeciliopsis) (25). These Poeciliopsis females movement of hybrid genomes into F2 generations, because
hybridize with males of another species, but during meiosis of the reproductives are almost universally the product of conspecific
offspring only the maternal genes remain in viable gametes (26). matings. If so, eusociality may paradoxically contribute to hybrid
Thus the male’s genome is not transferred beyond the F1 colony success, but in turn limit the extent of hybridization
generation, which is the same in our system. beyond the F1 generation.
How could hybridogenesis in Pogonomyrmex work? Our data
indicate two separate multilocus genotypes present within both We thank T. Polony for assistance in collection of samples; S. Poloner
species, which are revealed by the 510 and 550 markers and the for assistance with PCR; and S. Cahan, T. Dowling, J. Harrison, J.
Parker, D. Wheeler, and members of the Social Insect Research Group
additional RAPD markers (Tables 1 and 2). To survive and
for discussion. We also thank three anonymous referees whose com-
reproduce, homozygous queens must mate with a male of ments helped to improve the manuscript. D.L. was supported by a
matching type to produce reproductively capable daughters. To Howard Hughes grant for undergraduate research. Research was funded
produce sterile workers they must also mate with a male of the by National Institute of Mental Health Grant R29 MH51329 to J.H.F.
opposite type. However, because workers do not reproduce, the This work was also funded by grants from the Deutsche Forschungsge-
genetic information of nonmatching males is not transferred meinschaft to J.G. (Ga-661兾1 and SFB 554兾TPC5).