Kusaka et al.: Genetic population structure of Dipturus chinensis around Japan 
103 
© Danjo Isls. 
@ Kochi 
Goto Isls. | @ Koshiki-jima Isls. 
@ Aomori @Taiwan @kKorean Penin. 
Figure 3 
Minimum spanning network of haplotypes based on phylogenetic analysis of the mito- 
chondrial cytochrome 6 gene sequences (931 base pairs) identified in 214 polkadot skate 
(Dipturus chinensis). Sequences were obtained from specimens caught at 7 locations 
around Japan, in the East China Sea, Sea of Japan, and Pacific Ocean, between 2010 
and 2017 or were obtained through the International Nucleotide Sequence Database 
Collaboration for 2 other populations, in Taiwan or off the Korean Peninsula. Each circle 
represents a unique haplotype. Haplotypes are combined into groups A and B, and group 
B is further divided into subgroups B1 and B2 (all of which correspond to the clades and 
subclades in the maximum likelihood tree in Figure 2). Slices of the circles represent 
fractions of the number of individuals of a haplotype that are associated with 1 or more 
of the 9 locations. The size of circles is proportional to the number of individuals. Each 
branch equals 1 nucleotide substitution. Isls.=Islands; Penin.=Peninsula. 
were excluded from analysis because the sample size for 
each location was only 1 individual. Results of hierarchi- 
cal AMOVA (Suppl. Table 1) (online only) indicate that the 
gr value was 0.8976 (P<0.001) and that 89.76% of genetic 
variability was distributed among the 6 geographic popu- 
lations that were included in analysis. When the 6 popu- 
lations were divided into 2 groups, clade A (populations in 
the Danjo and Goto Islands in the East China Sea and pop- 
ulations in the Kyoto and Niigata Prefectures in the Sea of 
Japan) and clade B (populations in the Kochi and Aomori 
Prefectures along the southern and northern Pacific coasts 
of Japan, respectively), the ®op value was 0.8529 (P=0.068), 
and 85.29% of genetic variability was distributed between 
the groups. When the 6 populations were divided into 3 
groups, namely, clade A, clade B1 (population in Kochi Pre- 
fecture), and clade B2 (population in Aomori Prefecture), 
the ®op value was maximized (®,7=0.8964), and 89.64% 
of genetic variability was distributed between the groups. 
However, the P-value was barely not significant (P=0.064). 
The pairwise ®g7 values among the 6 geographic popula- 
tions estimated from the mt cyt b gene sequences are shown 
@ Kyoto 
in Table 3. The values between 
each of the 2 populations along 
the Pacific coasts (in Kochi and 
Aomori Prefectures) and each 
of the 4 populations in the East 
China Sea (in the Danjo and Goto 
Islands) and the Sea of Japan 
(Kyoto and Niigata Prefectures) 
were high (pairwise gy: 0.8847— 
0.9342) and significant after a 
Bonferroni correction (P<0.001). 
The ®gp value between the pop- 
ulations in the Kochi and Aomori 
Prefectures was also high (pair- 
wise Mgr: 0.8507) and significant 
after a Bonferroni correction 
(P<0.001). 
1 
I 
I 
I 
I 
I 
I 
I 
I 
I 
I 
I 
I 
i] 
Demographic history The mis- 
match distributions for the sam- 
ples in clade A (Fig. 5A) and clade 
B2 (Fig. 5C) follow a unimodal 
curve, and the distribution for 
the samples in clade B1 (Fig. 5B) 
has roughly an L shape. Historic 
demographic parameters  esti- 
mated for each clade are shown 
in Table 4. The 6, in clade B1 
(0.299) was an order of magni- 
tude smaller than that in clade A 
(3.916) and clade B2 (6.597). The 
sum of squared deviations and 
Harpending’s raggedness index 
values were low and not signif- 
icant in any of the distributions. 
The negative and significant val- 
ues of Tajima’s D were detected 
for the samples for clade A (Taji- 
ma’s D: -1.518, P<0.05) and clade B1 (Tajima’s D: -1.679, 
P<0.05). The negative and significant values of Fu’s F's were 
detected for only the samples for clade A (Fu’s F's: -10.959, 
P<0.005). 
The BSPs for clade A (Fig. 5D) and clade B2 (Fig. 5F) 
indicate that N, increased after the Last Glacial Max- 
imum (LGM, which occurred approximately 0.01—0.02 
MYA), but such an increase of NV, with regard to LGM was 
not observed for clade B1 (Fig. 5E). However, the possi- 
bility of a constant population, a population that has not 
experienced rapid population growth or decline, could not 
be rejected for any clades because the 95% highest poste- 
rior density interval was not indicated to exclude zero by 
the parameter sum (indicators.alltrees). 
@ Niigata 
Microsatellite analyses 
Genetic variation within populations In the 6 geographic 
populations included in analysis, the mean number of 
alleles per locus ranged from 1.00 at locus LERI33 to 
14.00 at locus LERI44 (Suppl. Table 2) (online only). No 
