dom mating. This is most likely at broad geographic scales,
especially at spatial scales that are greater than maximum
dispersal distances. At very large spatial scales, limitations
on dispersal distances ensure that mating will be nonran-
dom and more localized than if individuals in the large geo-
graphic population mated randomly.
When considered on a spatial scale, FILshould be nega-
tive at the level of social groups, since dispersal out of social
groups usually occurs by one sex or the other (Greenwood
1980; Dobson 1982). As the spatial scale increases, esti-
mates ofFISwill approach zero. WhenFISis not significantly
different from zero, mating may in fact be nonrandom. For
example, black-tailed prairie dogs have strongly negativeFIL
values, and FISat the colony (viz., subpopulation) level can
be close to zero (Dobson et al. 1997). But mating is probably
not random at either level of population structure, because
females in social groups (female kin-based families called
coteries) are most likely to mate with the dominant co-
terie male (i.e., females make nonrandom mating choices).
The fact that FISis zero occurs because at the colony level,
the inbreeding coefficient (F) and co-ancestry among coter-
ies (a) are equal. So the issue of whether there is truly a level
of population structure at which matings are random is a
genetic red herring. As the spatial scale becomes broader,
FITvalues are likely to be positive due to geographic isola-
tion of subpopulations (Chesser 1983; Daley 1992).
FLSand FSTmeasure genetic differentiation among social
groups, such as coteries, and subpopulations, such as colo-
nies, respectively. Theoretically, these values range between
zero and one. A value of zero would indicate no genetic
differences among social groups or subpopulations, and a
value of one would indicate complete genetic differences
(viz., each subpopulation fixed for alternative alleles at all
loci, an unlikely event). Note that equation (1) implies that
often uad. Basically, as the spatial scale increases, our
estimates of co-ancestry at each level of structure should de-
cline. The values of FLSand FSTdepend on the relative val-
ues of uand a, and aand d, respectively. In general, we
might expect FLSto often be somewhat greater than FST, as
Chesser (1983) found for coteries and colonies of black-
tailed prairie dogs. This indicates that more genetic differ-
entiation occurs at the level of the social group than among
geographically disparate subpopulations (but see Patton
and Smith 1990).
At genetic equilibrium, it is unlikely that FLSand FSTwill
often exceed 0.25, despite the theoretical maximum of 1.0.
FLScan be used to examine why this is so. As can be seen
from equation (2), the value of uis the limiting value of FLS,Gene Dynamics and Social Behavior 165Figure 14.1 Hierarchical gene dynamics. A regional population is depicted, with an expanded view of a local subpop-
ulation. Within the subpopulation, several social breeding groups are depicted. Single examples of genetic correlations
are shown: the inbreeding coefficient (F) for a female, co-ancestry (u) within a breeding group for two females, the
gene correlation between individuals in two different social groups (a), and the gene correlation between two subpopu-
lations (d).