Rodent Societies: An Ecological & Evolutionary Perspective

signed for geographically distinct subpopulations that oc-
cur within a regional landscape, they can be applied to more
local levels of population structure. In fact, F-statistics
are hierarchical, so that several levels of population struc-
ture may be considered at once (reviewed by Chesser et al.
1996). Thus dividing a geographic subpopulation into so-
cial groups creates an additional level of population sub-
structure, and additional F-statistics can be defined. The
rate of inbreeding compared to what would be expected
if mating were random among the members of the subpop-
ulation is still FIS, as noted previously. But the rate of in-
breeding compared to that expected if the members of the
social group were mating randomly can be defined as FIL,
where Lindicates the lineage of the social group. By the
same token, the degree of genetic differentiation among the
social groups can be called FLS. Since effective population
size was based on the change in the average inbreeding co-
efficient from one generation to the next, it should be the
same value at any level of hierarchical organization for pop-
ulations, subpopulations, and social groups that are in ge-
netic equilibrium.
Measures of gene dynamics reflect not only patterns of
inbreeding in populations, but degree of kinship among in-
dividuals as well. Cockerham (1967, 1969, 1973) pointed
out that F-statistics are closely allied to genetic correlations
among individuals and can be calculated from them. The
inbreeding coefficient, F,is the genetic correlation of an
individual’s parents. In a closed population, inbreeding ac-
cumulates over time, and thus Fwill increase steadily over
generations. Fcan be averaged over all the individuals in
a population to yield the average rate of inbreeding. If the
value of Fis compared between parents and offspring, the
rate of change in Fcan be measured and used to estimate
inbreeding effective population size (e.g., Blackwell et al.
1995). When applied to empirical data, however, this esti-
mate assumes that the population is in genetic equilibrium,
a condition that may seldom be met in nature (Dobson et al.
2004).
Genetic correlations, or gene correlations, can be mea-
sured between pairs of individuals (viz., dyads). Individu-
als to compare may be chosen from geographically distinct
subpopulations, different social groups, or within social
groups (Chesser 1991a, 1991b). The dyadic gene correla-
tion is also called the co-ancestry of the individuals, because
genetic correlations are measured by descent, preferably
from a pedigree. Co-ancestry is closely linked to kinship. In
the absence of inbreeding, co-ancestry is half the degree of
kinship between two individuals. When there is inbreeding,
however, co-ancestry takes it into account, thus measuring
all sources of genetic identity by descent. So co-ancestry,
like the inbreeding coefficient, increases over generations.
With pedigree data, co-ancestry of any dyad (individuals i
and j) can be measured as:

(1)

where Srepresents sires and Dis for dams. If this value is

averaged among all the dyads within a social group or sub-

population, the average gene correlation of the social group

or subpopulation can be calculated.

It is probably best to use uas the average co-ancestry

within a social group, and use another symbol for the aver-

age co-ancestry within a geographic subpopulation. Ches-

ser (1991a, 1991b) defined aas the average co-ancestry

of a geographic subpopulation or colony of smaller social

groups. It (a) is averaged over all dyads of individuals that

are in different social groups, but within the geographic

subpopulation. At the next higher level of hierarchical pop-

ulation structure, co-ancestry of dyads of individuals from

different subpopulations could be defined and given a sym-

bol, such asd(fig. 14.1). With the previously cited gene cor-

relations, it is then possible to define F-statistics at several

levels of population structure, for societies that are based

on family social groups of related females:

(2)

These F-statistics define the gene dynamics of a population

with three levels of substructure: individuals within social

groups, social groups within geographical subpopulations,

and subpopulations within a region. The values of FIL, FIS,

and FITcompare the inbreeding coefficient to what would

be expected if mating were random at three levels of pop-

ulation structure: the social group, the subpopulation, and

the regional population. FLSindicates the degree of genetic

differentiation of social groups within a subpopulation and

FSTis the degree of genetic differentiation of subpopulations

within the regional population.

FIL, FIS, and FITvary between 1 and 1. A significant

negative value indicates less inbreeding than expected if

mating were random within the social group, subpopula-

tion, and regional population, respectively. The sorts of

things that often lead to lower inbreeding than expected are

natal dispersal out of the group, subpopulation, or popula-

tion, and behavioral avoidance of mating with close rela-

tives. Dispersal can occur at any level of population struc-

ture, though rates of movement should become lower as

larger geographic areas are considered. Avoiding close rel-

atives as mates, while they are still in spatial proximity,

should occur at the level of social groups or small local sub-

populations. Positive values of these F-statistics occur when

the inbreeding coefficient is greater than expected under ran-

FLS

ua

1 a

FST

ad

1 d

FIL

Fu

1 u

FIS

Fa

1 a

FIT

Fd

1 d

ui,j

1

4 1 uSiSjuSiDjuSjDiuDiDj^2

164 Chapter Fourteen