a higher proportion of /tanimals surviving to sexual ma-
turity in the laboratory (Dunn et al. 1958). Also perplexing
are laboratory studies that found the fertility of /tmales
to be both higher (Dunn and Suckling 1955) andlower (Le-
vine et al. 1980) than /males.
Recognizing the powerful potential of sexual selection
to discriminate among genotypes, Sarah Lenington and her
colleagues have done considerable work evaluating the role
of mate choice in maintaining thaplotypes. The costly ef-
fects of homozygote sterility/lethality should favor individ-
uals who avoid t-bearing mates, especially if the choosy in-
dividual already carries one thaplotype. In accord with this
prediction, Lenington and her colleagues found evidence
of odor preferences in both sexes for /versus /tindi-
viduals (Lenington 1983; Egid and Lenington 1985; Len-
ington and Egid 1985). Since the major histocompatibility
complex is linked to the tcomplex, this obvious potential
source of odor differences was tested for its contribution to
t-associated odor discrimination. Odors from recombinant
mouse strains carrying similar MHC haplotypes, with or
without an associated tcomplex, were discriminated by
males but not by females, suggesting that both the MHC
complex and other genes within the tcomplex play a role
in mediating t-associated odors (Egid and Lenington 1985;
Lenington and Egid 1985). But despite t-associated odor
avoidance behavior, short-term experiments conducted in
indoor arenas revealed very little evidence for /biased
mating preferences (Franks and Lenington 1986), empha-
sizing the need to study mating preferences in an ecologi-
cal context where animals assess mates on a variety of com-
plex cues. In these short-term arenas, /tmales sired more
offspring (but not significantly more) than their wild-type
competitors, a result the authors attributed to higher fertil-
ity of /tmales (Dunn and Suckling 1955; Franks and Len-
ington 1986).
The competitive dynamics of /and /tmice were
further studied by Lenington et al. (1996) in large outdoor
seminatural enclosures. The results of their male dominance
and survivorship analysis appear to corroborate the pre-
vious short-term population studies: /tmales were more
aggressive and had slightly lower mortality than wild-type
competitors (Lenington et al. 1996). Unfortunately, inter-
pretation of these results is somewhat confounded by the
experimental design, which assayed male aggression during
staged encounters before release into the enclosures, and
again after retrapping. Aggressive behavior during staged
encounters could be influenced by the removal of impor-
tant ecological and motivational cues normally present dur-
ing territorial defense. Additionally, having spent part of
the preceding night in traps, aggression scores might have
reflected differential behavioral responses to stress between
/tand /males. Finally, since reproductive data were
not reported for this study, it is unknown whether aggres-
sion scores reflected actual fitness differences (Lenington
et al. 1996).
Despite these multifaceted approaches to measuring se-
lection on heterozygote tcarriers, empirical support for
the predicted heterozygote disadvantage remains murky.
Table5.1 outlines the major factors that have been studied
for their potential effects on tfrequencies. With so many
different relevant factors, it is nearly impossible to integrate
these data into a cohesive model that would successfully
predict the fitness of thaplotypes in nature. We therefore
decided to measure fitness itself.The tcomplex in seminatural enclosures
Measuring the fitness of theterozygotes requires a compet-
itive environment in which sexual selection can operate on
t-complex and wild-type genomes. Collecting long-term re-
productive data is equally important for allowing estimates
of lifetime fitness. The discovery of an unprecedented five-
fold fitness decline in inbred relative to outbred males when
measured across much of the adult lifespan (Meagher et al.
2000) clearly demonstrates the strength of large-scale sem-
inatural experiments to reveal hidden effects of gene func-
tion (discussed previously). One year after publishing the
inbreeding study, we made the fortuitous discovery of an
entire natural experiment submerged within the original
data set (Carroll et al. 2004). Some of the original mouse
founders from the inbreeding study (Meagher et al. 2000)
were found to harbor t-complex haplotypes, making it pos-
sible to test aspects of sexual selection that might serve to
limit the spread of this selfish genetic complex in the wild.
Because we were unaware of the presence of thaplotypes
during the two generations of breeding to create founders
for the inbreeding experiment, we had taken no steps to
control the spread of this meiotic drive complex in our lab-
oratory colony. The level of transmission distortion from t-
bearing males during laboratory breedings was 0.88 (88%
of all pups sired by /tmales inherited the t-complex ver-
sion of chromosome 17), representing significant deviation
from the 0.5 prediction of Mendelian inheritance. Conse-
quently, when founder mice for the inbreeding study were
selected from among the progeny of caged breeders, tfre-
quencies had increased from 9.7% to 15.3%, a jump of
58%, but still well within the range of reported estimates
from actual surveys. This fortuitous discovery allowed us to
study the population dynamics of the tcomplex at biologi-
cally relevant initial frequencies. Since house mice are short
lived and have relatively rapid generation times, ten rep-
licate experiments, running for ten months, allowed us toSexual Selection: Using Social Ecology to Determine Fitness Differences 65