Table 4.
Population size, migration and dispersal estimates
| |
Linkage Diseq. |
Temporal (MLNE) |
Temporal (TM3) |
Mean NC |
95% HPD |
|||
|---|---|---|---|---|---|---|---|---|
| |
(LDNe) |
(Likelihood) |
(Bayesian) |
|||||
| Mean NE | (95% CI) | Mean NE | (95% CI) | Mean NE | (95% CI) | |||
| Clusters 3 & 4 |
310 |
(258–380) |
243 |
(181–327) |
240 |
(223–250) |
4121 |
(2422–6109) |
| Cluster 2 |
121 |
(88–180) |
45 |
(35–61) |
33 |
(22–47) |
1199 |
(470–2191) |
| Cluster 1 |
157 |
(98–326) |
212 |
(106–611) |
178 |
(105–285) |
1299 |
(412–2621) |
| |
NE/NC |
σ (km) |
σ (km) |
σ2(km2) |
m |
WN |
|
|
|
(1D) |
(2D) |
(2D) |
(2D) |
(2D) |
|
|
||
|
Based on LDNe estimates |
|
|
||||||
| Clusters 3 & 4 |
0.075 |
15.16 |
14.14 |
200.06 |
0.033 |
64 |
|
|
| Cluster 2 |
0.101 |
4.53 |
3.59 |
12.92 |
0.025 |
19 |
|
|
| Cluster 1 | 0.121 | 3.94 | 2.49 | 6.21 | 0.013 | 13 | ||
Estimates were computed for clusters 1, 2 and the group comprising clusters 3 and 4. Effective population size (NE) was computed using both a linkage disequilibrium method (in LDNe) and two temporal methods (a likelihood approach implemented in MLNE and a Bayesian approach implemented in TM3). Mean NE and parametric 95% confidence intervals are shown. Census size (NC) was computed using a sequential Bayesian method [51]. Mean NC and 95% highest probability density (HPD) intervals are shown. Dispersal distance (σ) was estimated using both a one-dimensional (1D) (FST/(1-FST) ~ a + bGD; GD = geographic distance in kilometers) and a two-dimensional (2D) (FST/(1-FST) ~ a + bln(GD); ln(GD) = log of geographic distance) isolation-by-distance model. Dispersal surface (σ2), migration (m) and Wright’s neighborhood size (WN) estimates from the 2D model are also reported. σ, σ2, m and WN in cluster 1 were only computed for female individuals (the model was not significant when males were considered) and are shown in italic type.