inbreeding being two such stressors. Although the relation-
ship between FA and fitness is unclear, greater susceptibil-
ity to such developmental perturbations will generally be
costly (Møller and Swaddle 1997), especially if there is a
relationship between morphology and performance. If it
can be shown that FA affects any performance measure in-
volved in male-male competition, then male territoriality
and ultimately fitness can be greatly compromised by in-
heriting a single thaplotype.
Combining the enclosure study discussed earlier with
prior work from other labs, it is clear that no single factor
is responsible for limiting the invasion and spread of t-
complex haplotypes in wild populations. Rather, there are
a variety of selective factors, operating on both males and
females to balance the extreme t-haplotype transmission
bias from t-bearing males. But with so many selective com-
ponents operating against the tcomplex, both in homozy-
gote and heterozygote carriers, it is unclear why this “dele-
terious gene” persists, even at low levels. One solution to
this apparent paradox is if sexual selection is less efficient
in small populations. Deterministic models, such as those
used to predict gene frequencies in the study by Carroll
et al. (2004), assume random mating within infinitely large
populations. However, Muspopulation sizes vary greatly,
from small, single-male founder territorial units (Selander
1970), to populations containing hundreds of individuals
(Ardlie and Silver 1998). Without competition, t-bearing
animals are quite prolific — as was evident in our labora-
tory breeding colony. Small populations founded by one or
a few t-bearing males are predicted to favor rapid increases
in tfrequencies by virtue of meiotic drive. Such populations
may serve as the primary sources of thaplotypes, which can
then infect neighboring populations as t-bearing individu-
als emigrate. In contrast, larger populations with extensive
competition will be more effective at selecting against indi-
viduals bearing thaplotypes, driving down the frequencies
of this selfish genetic complex. Frequency data from natu-
ral populations appear to fit such a model, where small and
medium-sized populations (60 individuals) have a high
prevalence of thaplotypes, and tend toward fluctuations in
tfrequency (Ardlie and Silver 1998). By contrast,tfrequen-
cies in large populations (60 individuals) tend to be much
lower (avg. 3%) or are completely absent. Although this
prediction of density-dependent selection has not yet been
tested experimentally, seminatural populations promise a
fertile approach for such studies.Summary and ConclusionsFunction is fitness! The three case examples presented here
all involve major measurable fitness differentials result-
ing from naturally occurring genetic variation among wild
(or wild-derived) house mice. However, these major fit-
ness differences often went undetected using laboratory ap-
proaches. Recent theory and empirical work, including a
meta-analysis encompassing 14 years worth of selection
studies, suggest that the dominant forms of selection-driven
evolution may prove to be intra-male competition and fe-
male choice (Hoekstra et al. 2001). This does not mean that
parasites, predators, and other agents of natural selection
are not important, but that in many species, sexual selec-
tion may be a good proxy for natural selection, because it
screens for those who will command the best resources and
will have the best mating success. Consequently, it becomes
a good predictor of who will survive and ultimately repro-
duce. Seminatural competitive populations of Muslargely
capture the social ecology of this species, and thus present
a very tractable experimental system for measuring fitness,
which is required for answering many questions in biology.
Fortunately, using the social ecology approaches reviewed
previously can help reveal many phenotypes that are cryp-
tic under laboratory conditions, serving the functional ge-
nomics community and the molecular research program in
general.Sexual Selection: Using Social Ecology to Determine Fitness Differences 67