*2.1. RAD-Seq Data and SNP Filtering*

Approximately 389.4 billion (389,411,194,200) raw reads were produced from 256 individuals of *H. chrysotricha* sampled across the TIL and ZA regions. After quality control, filtering, and trimming, an average of 1421.5 million and 1638.9 million high-quality reads were retained for the TIL and ZA samples, respectively (Table S1). A catalog containing 9,131,292 loci was constructed, and 9,067,359 loci were genotyped by GSTACKS using the data sets of all *H. chrysotricha* samples. The mean, minimum, and maximum values for effective per-sample coverage were 26.9×, 16.8×, and 53.0×, respectively. After filtering loci of low quality (minor allele frequency < 0.01; missing rate > 0.5), 185 and 188 unlinked SNPs were eventually identified in TIL and ZA populations, respectively, and used for all subsequent analyses (Tables S2 and S3).

#### *2.2. Population Variation*

Based on genome-wide SNP markers, we detected significantly higher levels of genetic diversity in *H. chrysotricha* from the TIL region than from the ZA region (*π* = 0.247 vs. 0.208, *p* < 0.01; *H*<sup>O</sup> = 0.307 vs. 0.256, *p* < 0.01; *H*<sup>E</sup> = 0.228 vs. 0.190, *p* < 0.01; Tables 1 and 2). Within each region, there were no major differences in genetic diversity between island and mainland populations (TIL island vs. mainland populations: *π* = 0.257 vs. 0.237, *H*<sup>O</sup> = 0.313 vs. 0.302, *H*<sup>E</sup> = 0.237 vs. 0.219; ZA island vs. mainland populations: *π* = 0.208 vs. 0.207, *H*<sup>O</sup> = 0.256 vs. 0.255, *H*<sup>E</sup> = 0.189 vs. 0.192). *F*IS values were all negative in all *H. chrysotricha* populations, with a mean value of −0.078 and −0.066 in the TIL and ZA regions, respectively (Tables 1 and 2).


**Table 1.** Genetic characteristics of island and mainland populations of *Hedyotis chrysotricha* in the Thousand-Island Lake (TIL) region.

Abbreviations: *π*, nucleotide diversity; *H*O, observed heterozygosity; *H*E, expected heterozygosity; *F*IS, withinpopulation inbreeding coefficient.

**Table 2.** Genetic characteristics of island and mainland populations of *Hedyotis chrysotricha* in the Zhoushan Archipelago (ZA) region.


Abbreviations: *π*, nucleotide diversity; *H*O, observed heterozygosity; *H*E, expected heterozygosity; *F*IS, withinpopulation inbreeding coefficient.

Pairwise *F*ST values between the TIL populations ranged from 0.001 to 0.393, with about 13% being significant (*p* < 0.05) after sequential Bonferroni correction (Table 3). Among the significant pairwise comparisons, the great majority (11 out of 16) involved population EM04. Pairwise *F*ST values between the ZA populations ranged from 0.001 to 0.436, with about 24% being significant (*p* < 0.05) after sequential Bonferroni correction (Table 4). Here, a large proportion of significantly high *F*ST values (6 out of 13) involved population ZI05. The average *F*ST value across populations was slightly higher in the TIL (*F*ST = 0.103) region than in the ZA (*F*ST = 0.092) region, yet this difference was only marginally significant (*p* = 0.046). No IBD pattern was found in either region (TIL: *r* = 0.181, *p* = 0.174; ZA: *r* = 0.072, *p* = 0.242) (Figure S1).


**Table 3.** Pairwise *F*ST values among the 16 populations of *Hedyotis chrysotricha* in the Thousand-Island Lake (TIL) region.

Values in bold are significantly different from zero after sequential Bonferroni correction.

**Table 4.** Pairwise *F*ST values among the 11 populations of *Hedyotis chrysotricha* in the Zhoushan Archipelago (ZA) region.


Values in bold are significantly different from zero after sequential Bonferroni correction.

#### *2.3. Population Genetic Structure*

Based on a STRUCTURE analysis, the most likely number of genetic groups was *K* = 3 for the TIL populations and *K* = 2 for the ZA populations. In the TIL region, Cluster I ('dark blue') was present at high frequency (80% of all local samples) in seven of eight island populations, whereas the great majority of individuals (89%) from the western mainland populations and the remaining island population (IP08) were assigned to Cluster II ('red') (Figure 1 and Figure S2). In the eastern mainland populations, half of the individuals from EM04 were assigned to Cluster III ('green'); the remaining populations contained variable numbers of individuals belonging to both Cluster I and Cluster II (Figure 1). In the ZA region, nearly all populations (excepting ZI04) consisted of variable numbers of individuals with a probability of ≥ 0.5 of belonging to both Cluster I ('blue') and Cluster II ('orange'), when *K* = 2 (Figure 2). At *K* = 3–5, we also observed that the population ZM02 was assigned to a separate cluster (Figure S3), suggesting potential genetic differentiation between this mainland population and the others.

**Figure 1.** STRUCTURE analysis of 155 individuals (16 populations) of *Hedyotis chrysotricha* from the Thousand-Island Lake (TIL) region, based on RAD-Seq-derived single nucleotide polymorphism (SNP) data. (**A**) Plots of the mean posterior probability [LnP(*D*)] values of each *K* and (**B**) the corresponding Δ*K* statistics. (**C**) Histogram of the STRUCTURE analysis for the model with *K* = 3 (showing the highest Δ*K*). A vertical bar represents a single individual, and each color corresponds to a suggested cluster (Cluster I: dark blue; Cluster II: red; Cluster III: green). The *x*-axis corresponds to population codes. The *y*-axis presents the estimated membership coefficient (*Q*) for each individual in the three clusters.
