Figure 2: Multidimensional scaling plot (MDS) showing the
genetic structure of scrubtit samples by population based on Prevosti’s
genetic distance. (a) shows all sampled individuals across mainland
Tasmania and King Island; (b) shows only King Island samples. Numbers in
parentheses on axis titles denote the proportion of total variance
explained by the first two coordinates.
Pairwise FST estimates between the King Island
subpopulations (0.12 - 0.18) were larger than the majority of the
pairwise estimates between the mainland subpopulations (0.01 – 0.18)
except the Tasman Peninsula (> 0.22, Table S5). Genetic
isolation was positively correlated with geographic distance (glm β =
494.7, se = 101.7, p < .001, McFadden’s R2 =
0.36, Figure S5).
Increasing values of k resulted in decreased values of the
cross-entropy criterion (Figure S6), with changes in the criterion value
suggesting that the best number of ancestral populations for
interpretation within the current sampling design ranged from 2 to 4.
Admixture plots showed strong differentiation of the King Island
subspecies from the Tasmanian scrubtit regardless of the estimated
number of ancestral populations (Figures 3 & S7). When k = 3,
the east coast subpopulations in the Tasman Peninsula and north east
separated with a high degree of confidence from the remainder of the
Tasmanian scrubtit population, and when k = 4 the Tasman
Peninsula and north east subpopulations separated from each other.
Despite being isolated from the mainland by a 4 km wide sea strait, the
Bruny Island subpopulation was less isolated from the rest of the
Tasmanian scrubtit population than were the north-eastern and Tasman
Peninsula subpopulations (Figures 3 & S7, Table S5). When analysed
independently, King Island scrubtits showed a high degree of
differentiation by subpopulation when k = 3 (Figures 4, S8 –
S9).