Figure 1 Avicennia marina distribution range and sampling
locations. Ranges of the three varieties are shown in colors as
indicated in the legend. Sampling locations are indicated by black
circles with numbers. Location name and population abbreviation are as
follows: 1, Meed Creek, maMC; 2, Laemson, MaLS; 3, Penang, maPN; 4,
Thongnian, maTN; 5, Samut Sakon, maSS; 6, Ban Kunsha, maBK; 7, Sanya,
maSY; 8, Wenchang, maWC; 9, Cebu, maCB; 10, Sabah, maSB; 11, Bali, maBL,
12, Bunbury, maBB; 13, Darwin, euDW; 14, Cairns, euCA; 15, Brisbane,
auBS; 16, Auckland, auAK. Detailed information for these sampling
locations is in Table 1. Morphological differences between the three
subspecies in leaf, flower and fruit are presented on the right.
Figure 2 Genetic divergence and differentiation amongAvicennia marina populations. (a) Clustering of the A.marina populations using principal component analysis (PCA). PCA
was performed on the SNP frequency matrix. Colors indicate subspecies.
(b) Multi-dimensional scaling analysis of the FSTand DXY matrices of 16 A. marinapopulations. (e-d) boxplots of DXY andFST values. “au”, “ma” and “eu” indicateA. m. australasica , A. m. marina and A. m.
eucalyptifolia respectively. “maWest” and “maEast” refer to the two
recognized groups of A. m. marina populations west and east of
the Malay Peninsula (see the Results section). “BB” refers to the maBB
population from Bunbury, Australia.
Figure 3 The subspecies evolved independently. a) Boxplots of θ
computed for each gene in each population (upper graph) and barplots of
mean θ and π values computed by pooling all SNPs in a population (lower
graph). (b) Simulations reconstructing demographic history ofAvicennia marina populations. Graphical presentation of the ten
models of the three subspecies. N stands for effective size and T stands
for time of split. Black arrows in models 8-10 indicate gene flow.
Figure 4 Gene flow between subspecies. (a) Graphical
presentation of the four models to investigate the contrast between
morphological and genetic characters of the maBB population in western
Australia. vT0 and vT1 indicate
divergence time points and Neu, Nbb, and
Nma indicated effective population size. The constant
bi-directional migration rates are denoted by ma and
mb. (b) TreeMix to capture gene flow events on a
population splitting graph. On the Maximum likelihood tree, each yellow
line indicates a gene flow event between branches it links, with color
indicating migration weight. Horizontal branch lengths of the tree are
proportional to the amount of genetic drift that has occurred on the
branch. The triangle matrix on the top-right indicates residual fit from
the maximum likelihood tree. Residuals above zero imply candidate
admixture events.
Figure 5 Networks and geographical distribution of haplotypes
inferred in eight Avicennia marina populations. Haplotypes are
indicated by different colours. Lines linking haplotypes reflect
mutations, with mutations exceeding a single step marked. The geographic
distribution of haplotypes is also indicated. The presented a to f cases
are six typical ones to represent six types of haplotype networks. Among
the 231 segments, 134, 66, 14, 11, 5 and 1 segments are classified to
each type of a to f respectively.