La. Zhivotovsky, Estimating divergence time with the use of microsatellite genetic distances: Impacts of population growth and gene flow, MOL BIOL EV, 18(5), 2001, pp. 700-709
Genetic distances play an important role in estimating divergence time of b
ifurcated populations. However, they can be greatly affected by demographic
processes, such as migration and population dynamics, which complicate the
ir interpretation. For example, the widely used distance for microsatellite
loci, (delta mu)(2), assumes constant population size, no gene flow, and m
utation-drift equilibrium. It is shown here that (delta mu)(2) strongly und
erestimates divergence time if populations are growing and/or connected by
gene flow. In recent publications, the average estimate of divergence time
between African and non-African populations obtained by using (delta mu)(2)
is about 34,000 years, although archaeological data show a much earlier pr
esence of modern humans out of Africa. I introduce a different estimator of
population separation time based on microsatellite statistics, T-D, that d
oes not assume mutation-drift equilibrium, is independent of population dyn
amics in the absence of gene flow, and is robust to weak migration flow for
growing populations. However, it requires a knowledge of the variance in t
he number of repeats at the beginning of population separation, V-0. One wa
y to overcome this problem is to find minimal and maximal bounds for the va
riance and thus obtain the earliest and latest bounds for divergence time (
this is not a confidence interval, and it simply reflects an uncertainty ab
out the value of V-0 in an ancestral population). Another way to avoid the
uncertainty is to choose from among present populations a reference whose v
ariation is presumably close to what it might have been in an ancestral pop
ulation. A different approach for using T-D is to estimate the time differe
nce between adjacent nodes on a phylogenetic population tree. Using data on
variation at autosomal short tandem repeat loci with di-, tri-, and tetran
ucleotide repeats in worldwide populations, T-D gives an estimate of 57,000
years for the separation of the out-of-Africa branch of modern humans from
Africans based on the value of V-0 in the Southern American Indian populat
ions; the earliest bound for this event has been estimated to be about 135,
000 years. The data also suggest that the Asian and European populations di
verged from each other about 20,000 years, after the occurrence of the out-
of-Africa branch.