Biology Reference
In-Depth Information
loses variation due to drift, each subpopulation will become fixed for
different alleles. Thus, genetic variation is maintained over the entire
metapopulation. However, in metapopulation models where some
subpopulations have greater contributions to the migrant pool than others
or where subpopulations go extinct and are recolonized via founders of
another subpopulation,
N
eT
can be greatly reduced below the sum of
subpopulation effective sizes.
49,61,72,73
If estimates of the three parameters
in Eq.
(8.2)
can be obtained to estimate
N
eT
, then the island model value
can be compared to the single-sample
N
eT
“best” estimate to draw on
conclusions about subpopulation contributions to the migrant pool.
Criscione and colleagues
41
reported that genetic differentiation among
households was 0.023 (the equivalent of
F
ST
). Furthermore, using
a Bayesian clustering method,
23
they identified 13 core clusters, which I
will use as
n
. Obviously
N
e
was not the same across households, but for
the purpose of illustration I will assume they were and use the harmonic
mean (
Table 8.2
), 98.8 (95% CI: 73.5
139.1). Based on the latter values, the
island model
N
eT
is 1314 (possible range from 979 to 1851), which is in
agreement with the single-sample “best” estimate of 1291. Therefore, this
comparison suggests that
Ascaris
subpopulations in Jiri reflect more of
Wright island model rather than a metapopulation where subpopulations
have large unequal contributions to the migrant pool or recolonization
e
e
extinction dynamics. If the latter were true, then it seems like the single-
sample estimators would be producing an estimate well below that
predicted from the island model. Readers are encouraged to delve into the
references above
49,61,72,73
to get an understanding of all model assump-
tions. Here I point out two concerns in this data set. First of which is the
number of subpopulations I used in Eq.
(8.2)
. If the landscape genetics
study
41
did not sample all possible subpopulations, then 13, and thus the
estimate of
N
eT
from the island model, would be an underestimate.
Second, I also assumed that the harmonic mean
N
e
of the households
reflects the central tendency of the
N
e
of the 13 genetic clusters. This seems
reasonable as households were largely composed of individuals
belonging to a single cluster. However, all clusters are not represented by
the houses in
Table 8.2
, and a few houses may represent the same cluster
(i.e. there is pseudoreplication).
I did not have a means to estimate
N
c
for each subpopulation.
However, if I assume stable human population growth and infection
patterns are constant over time, I can estimate a census size for
Ascaris
across the Jiri metapopulation (
N
cT
). This enables me to get a
N
eT
/
N
cT
ratio. Using the average prevalence of 25.7% and intensity of 2.52 worms
per infected host data from Williams-Blangero and colleagues,
62
and the
1991 census count of the Jiri human population of 7138, the
N
cT
of
A. lumbricoides
would be 4623. Accordingly,
N
eT
/
N
cT
¼
0.28 when using
the single-sample “best” estimator for
N
eT
. The single-sample estimators
