Genetic diversity is an important factor in environmental change and plant adaptation. Narrowly distributed and endemic plants show lower genetic diversity than plants in wider distribution areas (Hamrick and Godt 1990). Low diversity has been attributed to inbreeding and drifting in small populations (DeJoode and Wendel 1992). In this study, we genotyped 84 P. amphibia individuals collected from 7 populations (Fig. 1; Table 1). The observed and expected heterozygosity of P. amphibia populations were both less than 0.5, particularly the expected heterozygosity of the GS population (0.27266) was significantly lower than that of other populations. Inbreeding leads to an increase in homozygosity, reduced performance of production traits, and reduced fitness (Ouborg et al. 2010). The signs of within-population inbreeding were identified at three sites (HR, JT, and GS populations), which is likely because the three habitats (HR, JT, and GS populations) of P. amphibia were smaller than those of the other populations.

P. amphibia have lower genetic diversity than other endangered Polygonaceae species, such as Rheum tanguticum (He = 0.515) (Chen et al. 2009). In the present study, the mean He value (0.34082) of P. amphibia was similar to those of endangered plants in Korea, such as Pelatantheria scolopendrifolia (H e = 0.356) (Yun et al. 2020), Saussurea polylepis (He = 0.43) (Yun and Kim 2021) and Abeliophyllum distichum (He = 0.319) (Lee et al. 2022). Moreover, genetic variation among P. amphibia populations (60.08%) was higher than within populations (39.92%). Hamrick and Godt (Hamrick and Godt 1990) have revealed that selfing species have over 50% of the variation among populations. P. amphibia is self-incompatible species (Partridge 2001), which explains our results of the high genetic variation (over 60%) among populations of P. amphibia.

The fixation index FST has been widely used as a measure of population structure; Wright (1984) has suggested that an FST in the range of 0 to 0.05 indicates no genetic differentiation among populations whereas that in the range of 0.05–0.15 indicates moderate differentiation (Slatkin 1985; Shu et al. 2021). In the present study, the FST values between four populations (HR, GS, SE, and JC or HR-GS, HR-SE, and JC-SE) were smaller than 0.05 or close to 1 (JC-GS, JC-HR, and JT-GS) whereas those between KD and DP populations were considerably high, indicating that these two populations are highly differentiated. This phenomenon is considered to be related to the pollinator (Gamba and Muchhala 2023); for P. amphibia, insects are known pollinators (Harms and Grodowitz 2009). In addition, the distance between the KD population and other populations was over 300 km, and that between the DP population and other populations was over 50 km, which may explain the considerably high FST values of the two regions.

In the present study, the spatial clustering pattern suggests three ancestral populations. At K = 3, five populations (JC, SE, HR, JT, and GS) were assigned to a single cluster and two populations (KD and DP) were clearly separated. For the KD population, we suggest that it represents a different cluster because it is located on an Island 500 km away from other populations. Moreover, although the DP population was not geographically far (in terms of distance), it was not connected to other populations in the same river. Notably, both phylogenetic and principal component analyses showed results similar to those of the genetic analysis. The neighbor-joining tree also revealed that the KD population of P. amphibia formed one clade whereas the other populations (DP, JC, SE, HR, JT, and GS) formed one clade. Five populations (JC, SE, HR, JT, and GS) were closely clustered into one group, and the DP population sister group clustered with five populations (JC, SE, HR, JT, and GS). This result may be because these five populations (JC, SE, HR, JT, and GS) were closely located, indicating that their genetic relationships were also close.

GBS is an NGS protocol used for the discovery and genotyping of SNPs in genomes and populations. GBS methods have been applied in a wide range of studies, including genetic diversity, phylogenic classification, and hybridization analysis. This study detected 2,469 SNPs to estimate the genetic diversity and population structure of 84 P. amphibia individuals from 7 populations. We found that the seven populations were clustered into three groups. In particular, the FST values and structural analyses of the KD population were highly different among the seven populations. These phenomena can be explained by their geographic location. This is the first study of the genetic diversity of P. amphibia, providing important insights into the population history, genetic structure, and population conservation of P. amphibia.