Polymorphism of newly developed SSR markers
We detected a total of 118 alleles with 23 SSRs segregating in the 272 Pueraria accessions in Guangxi, with an average of 5.13 alleles per locus. This value is lower than the number of alleles per SSR locus reported in Zhou et al. . The number of effective alleles per locus (4.87) obtained in the Guangxi Pueraria accessions appears to be higher than the number of effective alleles per SSR locus found in other crops, such as the value of 2.26 reported in rice , 3.17 in olive , but lower than the values of 7.2 in maize . Moreover, our results showed that SSR allelic diversity of Pueraria germplasm was moderate (Na =1.9492, Ne = 1.2841, h= 0.1778). Zhou et al.  reported an average of Ne = 1.4503 and h = 0.2865 in a collection of 184 Pueraria accessions from Jiangxi. The number of markers and individuals, the sexual propagules and type of plant material, the population size may be responsible for the level of polymorphism and discrimination power.
Genetic diversity of Pueraria species
Our results revealed that Pueraria species display moderate genetic variation throughout Guangxi. In our study, the STRUCTURE results revealed that the 272 accessions could be divided into two main clusters and admixed individuals, while the UPGMA dendrogram showed that 272 accessions were divided into two main clusters with 37.8% genetic similarity, four main clusters with 68.4% genetic similarity. With evidence for several admixtures within cluster I (code_collection number: 30_JCJ-30, 32_JCJ-32, 196_GL-32, 197_GL-33) or cluster II (code_collection number: 12_YZ-12, 26_LC-26, 27_LC-27, 28_HJ-28, 113_GP-21, 149_BS-13, 160_BS-24, 195_GL-31, 270_Y10), the genetic differentiation was not in congruence with their eco-geographical distribution in this study. Furthermore, the overall clustering patterns generated by the STRUCTURE and PCA did not clearly distinguish the sampling areas, which is consistent with the previous results [6,9,11,21]. The low genetic differentiation indicated that geographical isolation may not restrict gene exchange among Pueraria species populations in Guangxi. It is susceptible to external factors even though there was a certain correlation between genetic variation and geographical distribution based on RAPD in several studies [8,22,23]. As a result, it is thought that Pueraria species has been cultivated and utilized for a long period in Guangxi since native cultivars of Pueraria still exist in the major regions, which is similar to Perilla in Korea . The selection by humans could be responsible for this clustering pattern and moderate genetic diversity.
However, previous studies revealed that Pueraria species possessed from moderate to the high level of genetic diversity with high clonal reproduction and perennial [10,11,20,21,25-29]. The inconsistencies observed, except for various taxon sampling and markers, could have originated by 1) the populations were found by sexual propagules could contribute to the maintenance of high genetic variation in clonal populations regardless of recruitment of sexual offspring ; 2) introductions from across its multiple native populations into novel habitats from seed stock ; 3) clonal populations with fewer genotypes still maintain higher genetic diversity at each locus .
Moreover, Pueraria species, as strictly self-pollinating and clonally persisting clumps plants, have considered heterozygosity (Table 2), like many clonal plants, e.g. Castanea dentata  and Musa balbisiana . Our results showed that relatively low Ht (0.1841) and Hs (0.1435), which suggest that accessions were inbred due to little outcrossing during maintenance . Moreover, we could not be ruled out a case that the existence of ancient clonality and the somatic mutation, which accumulates genetic variation within clonally persisting clumps may account for some of the heterozygosity, especially given rapid mutation of SSR fingerprints.
Implications for utilization of core germplasm resources in Guangxi
Core germplasm plays a key role in the conservation, management, and utilization of germplasm resources, which is critical for the development of plant breeding . Individuals reflecting genetic information can be selected to build the core germplasm resources. China is the center of distribution of Pueraria, with a long history of growing Pueraria species. However, fewer excellent Pueraria germplasm has been established due to artificial over-mining, lack of conservation, and management of resources. Pueraria species are abundant in Guangxi, especially in Tengxian and Wuzhou . Pueraria resources have a low level of genetic differentiation (Nm = 1.7690). Previous researches have shown that sampling proportion between 5 and 30% is enough to include at least 80% of the alleles representing the genetic diversity of the entire collection [36,37]. According to dynamic extracted results, our results revealed that when the samples collected reached 7.35% (20/272) of Pueraria accessions accounted for 105 alleles, accounting for approximately 93.75% of all alleles loci. Interestingly, the retention value of Pueraria core collection genetic diversity was lower than the allele retention values of 100%, 100%, and 97.5% in rosewood, licorice, eggplant, with sampling ratios reaching 12.4% , 16.84%  and 12.03% , respectively. The most likely reason was that the breeding of a majority of Pueraria accessions in Guangxi was still from layering breeding and self-crossing, and lacked extensive gene exchanges from cross-breeding, which led to a decrease in the ratio of the core collection. Our findings will be useful in breeding programs for the introgression of noble alleles into modern cultivars by exploiting natural genetic variation existing in Pueraria genetic resources. Combined with the analysis of phenotypic diversity (e.g. puerarin, starches) of Pueraria species, we may detect the important polymorphic loci associated with the traits based on correlation analysis, which could provide a foundation for developing the molecular marker-assisted breeding or detection of target genes soon .
Complex genetic relationships among Pueraria species
Our results include new clues in genetic relationships among Pueraria species based on SSR markers. 13 Pueraria individuals of 272 accession were admixture. Meanwhile, the genetic clusters were not consistent with species delimitation and geographic distribution, which imply the complex evolutionary history with the human process blur the phylogenetic relationship among these species. Pueraria DC. (1825; Fabaceae, Phaseoleae) comprises ca. 20 species, occurring in tropical and E. Asia. Eight species and two varieties have been recorded in China , with four groups or three sections as infrageneric classification based on morphological traits [42,43]. However, molecular studies have revealed that Pueraria is not a monophyletic group [44,45]. At the species level, the problem is more severe, where morphological intermediacy, artificial selection, biochemical characteristics, and insufficient informative characters contribute to the lack of agreement between morphological-based classification schemes and recent molecular phylogenies. For example, taxonomically kudzu is placed under the genus Pueraria. Pueraria lobata var. thomsonii and P. lobata var. montana were treated as varieties for P. lobata in Flora of China. However, the phylogenetic relationship and classification among these three species are still confused based on various molecular markers and sampling taxon. There were two main taxonomic treatments: firstly, Pueraria lobata var. thomsonii and P. lobata var. montana should be treated as a single species respectively, other than varieties ; secondly, Pueraria lobata var. thomsonii should be treated as the variety to Pueraria lobata, while P. lobata var. montana may be a single species . A wider taxon sampling with higher resolution genetic markers would increase confidence for the phylogenetic relationship among Pueraria species, efforts that are currently underway.