Rodent Societies: An Ecological & Evolutionary Perspective

such that FLScannot be greater than u. Since uis an average
genetic correlation, we can consider what its maximum
value should typically be. If a social group was made up
of full siblings, the average degree of relatedness would be
0.50, and the average genetic correlation in the absence
of inbreeding would be 0.25. However, there are two im-
portant caveats to this generalization. One is that unlike
the simple measure of kinship, co-ancestry accounts for in-
breeding. Thusucan be greater than 0.25 if there is consid-
erable inbreeding, as may occur in the plateau pika (where
FLS0.29; Dobson and Smith et al. 2000). The second
consideration is that it is unlikely that females will all be full
sisters in a polygynous social group, even with only one
breeding male. Rather, some mixture of full and half sib-
lings is likely, and uamong offspring will converge on 0.17
(1/6; Dobson and Chesser et al. 2000). Thus in the ab-
sence of inbreeding, co-ancestries among philopatric fe-
males in matrilineal social groups should fall to near 1/6,
or lower if there is more than one breeding male.
Chesser (1991a, 1991b) envisioned a buildup of genetic
relatedness among philopatric females from two sources
of genetic correlation. First, polygyny may often result in a
single male mating with all the females in a breeding group.
In this case, offspring from unrelated mothers are half sib-
lings. This source of genetic correlation can be stronger
than the second source, which is that breeding females may
be related, such as full or half sisters, through their mother,
depending on whether they have the same father. In general,
increasing the number of successfully breeding adult males
in a group will lower the gene correlations among daugh-
ters. Increasing the number of adult females in a group, how-
ever, has little effect on average gene correlations of daugh-
ters (Dobson and Chesser et al. 2000). Of course, variation
among breeding groups in their gene correlations may well
be increased by sampling events in small groups, or due to
demographic events as individuals age (e.g., Dobson and
Chesser et al. 1998).

Estimating Gene Dynamics

There are several ways that gene dynamics can be estimated
at the level of social breeding groups and at higher levels of
hierarchical population structure. F-statistics may be esti-
mated from pedigrees using equations (1 and 2). Pedigrees
provide the soundest measure of gene dynamics, because the
gene correlations F,u, a, and dcan be accurately measured.
A good example of this technique comes from John Hoog-
land’s long-term study of black-tailed prairie dogs (Dob-
son and Chesser 1997, 1998, 2004). Building a pedigree
requires that both paternity and maternity be known. Pedi-

gree data might be most available from captive breeding

programs in zoological parks and for domestic animals.

Cases of multiple paternity, however, indicate that obser-

vations of matings alone cannot identify sires. Biochemical

techniques of paternity assessment provide a possible so-

lution to this problem (e.g., Westneat and Sherman 1997;

Goossens et al. 1998).

If biochemical data are available, F-statistics can also be

estimated from heterozygosity of selectively neutral alleles

at variable genetic loci, under the assumption of genetic

equilibrium (Slatkin 1987). In a hierarchically structured

population, Wright’s (1969) F-statistics would be:

(3)

His heterozygosity of variable genetic loci at the indicated

level of population structure, and average levels of hetero-

zygosity are calculated among individuals, lineage groups,

and subpopulations. Several statistical software packages

are available for calculating F-statistics from allelic scores

at polymorphic loci (e.g., Goudet 1995). Of course, gene

correlations and levels of heterozygosity are related:

(4)

A final way of estimating F-statistics for lineage groups

is to apply Chesser’s (1991a, 1991b; Chesser et al. 1993)

breeding-group model of gene dynamics. This model draws

on the sorts of data collected in studies of “behavioral ecol-

ogy,” and was specifically designed for polygynous species

that exhibit natal philopatry by females (Sugg et al. 1996).

The breeding-group model is thus particularly appropriate

for studying the gene dynamics of the more social rodent

species. An extension of the model takes multiple paternity

into account (Sugg and Chesser 1994). The basic idea be-

hind the model is that gene dynamics can be estimated from

patterns of mating, dispersal, and demography. Chesser de-

rived mating parameters that quantify the degree of polyg-

yny, the probability that offspring share the same mother,

and the degree of multiple paternity (both within a litter

and among offspring born in different years). He combined

these mating parameters with demographic data such as the

number of males and females in social groups, the number

of social groups, mean female breeding success, and natal

dispersal rates of males and females. These and a few other

variables are then entered into transition matrices that are

used to predict changes in gene correlations (viz., F,u, and

HI 1 F HL 1 u HS 1 a HT 1 d

FLS

HSHL

HS

FST

HTHS

HT

FIL

HLHI

HL

FIS

HSHI

HS

FIT

HTHI

HT

166 Chapter Fourteen