Figure 1 . Map showing the population localities and the sex ratio of each population. A population code (p1–p33) is given to each population. Pie-charts show sex ratio investigated by sex check PCR or crossing experiments: females are represented in black, males are represented in white, and samples from which both female and male markers amplified in the sex check PCR are represented in grey. Pie-chart sizes are proportional to the corresponding sample size. TS, Tsugaru Strait; KC, Kuroshio warm ocean current; TC, Tsushima warm ocean current; OC, Oyashio cold ocean current; ESC, East Sakhalin cold ocean current.
Figure 2 . Cox 1-haplotype network. Population codes [1–33 (without ‘p’)] are given to the haplotypes of Japanese samples. Sampling regions are represented as different colors. For the samples from Japan, samples from sexual and parthenogenetic populations were represented in different colors. Each colored circle represents a haplotype and its size is proportional to its frequency, whereas white circles represent hypothetical unsampled haplotypes. White squares represent median vectors. Haplotypes indicated by arrowheads were found in 6–45 samples of which sex was identified, but none of these samples were identified as male (female or samples having both female and male markers).
Figure 3 . Maximum likelihood (ML) tree based on the nuclearcetn -int2 (604 bp). Numbers on branches indicate bootstrap percentage from ML analysis. Only bootstrap percentages ≥ 60% are shown. The labels of samples Japanese from parthenogenetic populations are indicated in red color, those from Japanese sexual populations are in black, and those from Argentina and Europe are in blue. Asterisks are given to the samples from the Sea of Japan coast (p1–p16), and stars are given to the samples from the Tsugaru Strait and Kyushu (p6, p16, p17, p30, and p31). Arrowheads indicate the two haplotypes found in samples from p24 and p27.
Figure 4 . Histograms showing genetic distances between samples based on the haploid dataset. (a) Genetic distances in parthenogenetic populations. (b) Genetic distances in sexual populations. The threshold distances (10) to recognize clonal lineages is indicated by broken lines. Samples from Europe and Argentina are included in the histogram for parthenogenetic populations. Note that, in the actual calculation in ‘assign clone ’ function of GenoDive, the dataset was not separated into sexual and parthenogenetic populations.
Figure 5 . Result of the STRUCTURE analyses based on the diploid dataset when K = 2 and K =5. The frequency of each cluster in each sample is shown by the bar plot. Sampling region in the Japanese Islands are shown above the bar plot, and population codes [1–33 (without ‘p’), ARG: Argentina, EUR: Europe; parthenogenetic populations from Japan are indicated in red] are shown below the bar plot. (a) The genetic clusters estimated in all runs at K = 2; (b) The genetic clusters estimated in five of 10 runs at K = 5; (c) The Neighbor Joining tree for the five clusters at K = 5 shown in (b). Color for each cluster is consistent in the bar plot of (b) and the NJ tree.
Figure 6 . Result of the PCA based on the diploid dataset showing the locations of each individual on the first three principal components (PC1–PC3), which describe 54% of the genetic variation. (a) Plot for the first and second PC. (b) Plot for the first and third PC. In Sampling localities of each Japanese population are shown in the map of Japan shown above the charts. Colors for each population are consistent in the PCA plots and the map.
Figure 7 . The NeighborNet tree based on the diploid dataset, showing the polyphyly of parthenogenetic populations. Labels of population codes of parthenogenetic populations from Japan are indicated in red. Japanese population are roughly color-coded according to the regions: the Pacific coast (deep blue), around the Tsugaru Strait (orange), around Kyushu (green), and the Sea of Japan coast (rose).
Figure 8 . Results of the GCMS analyses of the volatile compounds released from gametes: extracted ion chromatograms = 148–149m/z . The compounds predicted as sex pheromones were detected with high intensity at around 12.1 and 14.6 min from female gametes of a sexual population (p15; asterisks), not from those of parthenogenetic population p28. For the first compound, the fisrt candidate was ectocarpene and the second candidate was the cis-trans isomor of ectocarpene. For the second compound, the first candidate was ectocarpene and the second candidate was (3E ,5E ,8E )-1,3,5,8-undecatetraene.