Biology Reference
In-Depth Information
TABLE 8.2 N
e
Estimates from household-by-year with n
21. The “Best” estimate-N
e
is the harmonic mean of the LD-N
e
and SA-N
e
methods
House ID_year
>
House intensity
LD-
N
e
SA-
N
e
“Best” estimate-
N
e
097_2000
32
42.5
116
62.2
097_2003
43
66.7
133
88.8
119_2002
28
41.3
117
61
121_1999
89
90.1
180
120.1
122_1999
173
314.8
251
279.3
122_2002
48
271.8
265
268.3
123_1999
115
59.2
92
72
135_1999
132
240
208
222.9
135_2002
85
184.5
173
178.6
140_1998
31
40.7
67
50.6
148_2002
24
89.1
185
120.3
Harmonic mean
76.9
137.9
98.8
accurate estimates of the parental breeding population
N
e
that contributed
to the infections in each household. This is especially true given the two
methods utilize very different methods (linkage disequilibrium versus
identification of pedigree structure) to estimate
N
e
. The harmonic means
of
the household
N
e
point estimates (
n
11) were 76.9 (95% CI:
¼
55.6
183.6) for the LD-
N
e
and SA-
N
e
,
respectively (CI based on 1000 bootstraps over the point estimates of the
household-by-year samples). The harmonic mean is used because the
distribution of
N
e
can be highly skewed.
56
Waples and Do
56
also suggest
that if two single-sample estimators are independent and are estimating
the same parameter from a population, then a more precise or “best”
estimate of
N
e
can be obtained by taking the harmonic mean of the two
single-sample estimators. The “best” estimate-
N
e
's for the 11 household-
by-year samples with
n
e
116.8) and 137.9 (95% CI: 108.2
e
21 are given in
Table 8.2
. The harmonic mean
of these “best” estimates is 98.8 (bootstrap 95% CI: 73.5
>
139.1).
One of the questions that can be asked with these data is whether or not
drug treatment impacted
N
e
. Simulations have shown that LD-
N
e
esti-
mates from two time points can be used to detect a population genetic
bottleneck.
58
After omitting samples with small
n
, I only had three houses
(97, 122, 135) with estimates from both time periods (people in households
were treated and worms collected, then three years later this was
repeated). This is a small sample size, but visual (i.e. not statistical)
e