Genetic Composition of Korean Ginseng Germplasm by Collection Area and Resource Type

: To improve crops, it is important to secure plant genetic source material and evaluate the genetic diversity. Ginseng ( Panax ginseng C.A. Meyer) has long been used as a medicinal herb in Korea and China. Since ginseng originated from wild ginseng with low genetic diversity, it is also expected to have low genetic diversity. In this study, the genetic diversity of 451 ginseng accessions conserved in the National Agrobiodiversity Center (NAC) at Korea was analyzed using 33 SSR markers. Another objective was to establish a strategy for NAC to manage ginseng germplasm based on these results. The 451 accessions were collected from 22 cities in six provinces in South Korea. Among the 451 ginseng accessions, 390 (86.5%) and 61 (13.5%) were landraces and breeding lines, respectively. In the STRUCTURE results for the accessions, there was no relationship between assigned genotypes and collection areas, but there was a population genetic structure. In addition, genetic di ﬀ erentiation within populations of each analysis was low, indicating that the ginseng accessions conserved at NAC are extensively dispersed throughout the collection areas. The results of this study suggest that NAC should increase the genetic diversity of ginseng accessions for breeding programs, and alternatives are needed for securing ginseng genetic resources.


Introduction
Plant breeding in agriculture has decreased the diversity of many crops, which has caused bottlenecks in crop domestication, dispersal, and modernization. Researchers, in particular, have identified significant negative effects of plant breeding on diversity following the modernization bottleneck [1]. The loss of crop variation caused by the modernization of agriculture has been described as genetic erosion [2]. Since loss of genetic variation could decrease the potential of species to persist in the face of abiotic and biotic environmental changes, genetic erosion could pose a severe threat to long-term global food security [3]. Crop improvement largely depends on immediate conservation of genetic resources for their effective and sustainable utilization. For this, it is necessary to conserve and breed the vast genetic variation found in populations of the wild progenitors and landraces of cultivated plants [4,5].
Plant genetic resources are the most important components of agro-biodiversity, which encompasses primitive forms of cultivated plant species and landraces, modern and obsolete cultivars, breeding lines and genetic stocks, weedy types, and related wild species [6]. Since the International Board for Plant Genetic Resources was established in 1974, over 1750 gene banks have been activated worldwide, about 130 of which hold more than 10,000 accessions each [7,8]. In Korea,

Profiling of SSR Markers in Korean Ginseng Accessions
A total of 226 alleles were detected among the 451 ginseng accessions (Table 1 and Table S4). Polymorphic information content (PIC) of 33 SSR markers ranged from 0.025 (GM104) to 0.770 (GM116), with an average of 0.469. In the 390 landraces and 61 breeding lines, there were 223 and 148 alleles, respectively. The number of observed alleles (Na) ranged from 2 to 24, with an average of 6.85, of which there were 6.76 and 4.49 ginseng landraces and breeding lines, respectively. The number of genotypes (Ng) ranged from 2 to 100, with an average of 17.79, of which there were 16.83 and 6.18 landraces and breeding lines, respectively. The Shannon-Wiener index (H) showed the range from 0.06 to 1.998, with an average of 0.92. Observed heterozygosity (Ho) and Nei's genetic diversity (GD) ranged from 0.02 to 1.00 (mean: 0.88) and 0.02 to 0.80 (mean: 0.52), respectively. H, Ho, and GD were similar in the ginseng landraces and breeding lines.

Profiling of SSR Markers in Korean Ginseng Accessions
A total of 226 alleles were detected among the 451 ginseng accessions (Table 1 and Table S4). Polymorphic information content (PIC) of 33 SSR markers ranged from 0.025 (GM104) to 0.770 (GM116), with an average of 0.469. In the 390 landraces and 61 breeding lines, there were 223 and 148 alleles, respectively. The number of observed alleles (Na) ranged from 2 to 24, with an average of 6.85, of which there were 6.76 and 4.49 ginseng landraces and breeding lines, respectively. The number of genotypes (Ng) ranged from 2 to 100, with an average of 17.79, of which there were 16.83 and 6.18 landraces and breeding lines, respectively. The Shannon-Wiener index (H) showed the range from 0.06 to 1.998, with an average of 0.92. Observed heterozygosity (Ho) and Nei's genetic diversity

Population Structure of Korean Ginseng Accessions
To understand the pattern of the population structure, a Bayesian clustering analysis in STRUCTURE was performed. The STRUCTURE results suggested the best grouping number (K = 2) based on the delta K ( Figure 2A (Table S1). In the ginseng landraces, STRUCTURE determined 2 to be the best grouping number based on the delta K ( Figure 3A). Population 1 (L-Pop 1) and 2 (L-Pop 2) consisted of 201and 189 accessions, respectively. Due to the further increase of LnP(D), simulation was performed, and LnP(D) was maximized at K = 12 ( Figure 3A). All the ginseng accessions could be grouped into 12 populations, L1 to L12 ( Figure 3B), ranging from 2 (L6) to 53 (L9) accessions; the distributions of the 12 groups varied by region ( Figure 3C). Among 12 groups, L6 was in GG and JB only. All groups could be found in GG, there were 11 in GG and CN (excluding L6). JB contained ten groups (not L1 and L10), and there were seven in CB: L2, L3, L4, L7, L9, L10, and L11. GB showed the fewest groups, L1, L2, L3, L4, L8, and L11. structures. Among the 61 breeding lines, 14 accessions (23%) were assigned to POP4, followed by 12 (19.7%) in POP2, and 11 (18%) in POP6, while only one (IT239306) was in POP5. The 390 ginseng landraces could also be assigned into the seven population clusters: 84 accessions (21.5%) in POP1, 76 (19.5%) in POP7, and 63 (16.2%) in POP4. The population structure was also estimated to display the regional differentiation of ginseng accessions ( Figure 2D). Results showed that 26 (21.0%) and 24 (19.4%) of the ginseng accessions collected from GW clustered into POP4 and POP5, respectively. Among the 70 accessions from GG, 22.9% and 24.3% were in POP1 and POP7, respectively. In GB, 50% were in POP7, 30% were POP2 and 20% were in POP4. Among the 24 ginseng accessions from JB, 33.3% were in POP6 and 20.8% were in POP5. In CN and CB, 25.1% of the accessions in POP1 and 35.0% were in POP7.
In the ginseng landraces, STRUCTURE determined 2 to be the best grouping number based on the delta K ( Figure 3A). Population 1 (L-Pop 1) and 2 (L-Pop 2) consisted of 201and 189 accessions, respectively. Due to the further increase of LnP(D), simulation was performed, and LnP(D) was maximized at K = 12 ( Figure 3A). All the ginseng accessions could be grouped into 12 populations, L1 to L12 ( Figure 3B), ranging from 2 (L6) to 53 (L9) accessions; the distributions of the 12 groups varied by region ( Figure 3C). Among 12 groups, L6 was in GG and JB only. All groups could be found in GG, there were 11 in GG and CN (excluding L6). JB contained ten groups (not L1 and L10), and there were seven in CB: L2, L3, L4, L7, L9, L10, and L11. GB showed the fewest groups, L1, L2, L3, L4, L8, and L11. In ginseng breeding lines, STRUCTURE proposed 2 as the best grouping number based on the delta K ( Figure 4A). Breeding lines population (B) 1 and 2 consisted of 33 and 28 accessions, respectively ( Figure 4B). In ginseng breeding lines, STRUCTURE proposed 2 as the best grouping number based on the delta K ( Figure 4A). Breeding lines population (B) 1 and 2 consisted of 33 and 28 accessions, respectively ( Figure 4B). In ginseng breeding lines, STRUCTURE proposed 2 as the best grouping number based on the delta K ( Figure 4A). Breeding lines population (B) 1 and 2 consisted of 33 and 28 accessions, respectively ( Figure 4B).

Genetic Diversity and Population Differentiation of Korean Ginseng Accessions
AMOVA was collected for the genetic differentiation among and within the ginseng accession populations by collection area and STRUCTURE results ( Table 2). In the collection area findings, AMOVA revealed that 97% of the total genetic variation was attributable to differences within populations, which were notably and significantly greater than the 3% total variation found among populations. PhiPT and gene flow (Nm) for 451 ginseng accessions were 0.028 (p < 0.0001) and 17.213, respectively. Pairwise population PhiPT for four clusters ranged from 0.007 (GG-CB) to 0.085 (CB-JB) ( Table S5). Pairwise population estimates of Nm for the four clusters ranged from 5.395 (CB-JB) to 67.845 (GG-CB) migrants per cluster. In the STRUCTURE analysis, 97% of the total genetic variation was again attributable to differences within populations. PhiPT and Nm for the 451 accessions in these results were 0.031 (p < 0.0001) and 15.431, respectively. Pairwise population PhiPT for the four clusters ranged from 0.000 (POP1-POP2) to 0.106 (POP6-POP7) (Table S6). Pairwise population estimates of Nm for the four clusters ranged from 4.223 (POP6-POP7) to 39.744 (POP1-POP3) migrants per cluster. In results of, POP1 and POP2 showed no genetic differentiation in PhiPT and Nm. In the ginseng landrace AMOVA findings, 97% of the total genetic variation was contributed by differences within populations ( Table 2). PhiPT and Nm for the 390 landraces were 0.034 (p < 0.0001) and 14.322, respectively, and the Pairwise population PhiPT for four clusters ranged from −0.097 (L1-L6) to 0.084 (L7-L11) (Table S7). Pairwise population estimates of Nm for the four clusters ranged from 5.455 (L7-L11) to 403.3 (L1-L5) migrants per cluster.
In ginseng breeding lines, AMOVA revealed that 99% of the total genetic variation was attributable to within-rather than across-population differences ( Table 2). PhiPT and Nm for 61 ginseng breeding lines were 0.012 (p < 0.0001) and 40.761, respectively.

Distribution of Korean Ginseng Accessions
The collection areas for the ginseng accessions used in this study were mainly distributed in northern and central Korea. In South Korea, the ginseng cultivation area is from 36 • to 38 • N [14]. According to the 15th-century Korean history book Sejong Silokjiriji, Korean ginseng is native to all of Korea except Jeju Island. In the Joseon Dynasty, agriculturalists began cultivating ginseng in areas where native ginseng was found and gradually expanded cultivation throughout the country. However, based on differences in quality, ginseng is cultivated mainly in CN, CB, and GG [13]. The collection for the 390 ginseng landraces used in this study were distributed across six provinces (Figure 1 and Table S3); among them, the main areas were CN (36.4%), GW (31.8%), and GG (17.9%).
At the NAC, the seeds of plant germplasm are generally collected from farmers in the main cultivation areas of the target crops, except wild relatives. CN and GG are main ginseng production areas in the south, centered around Daejeon (CN), and in the north, the major area, bounded by the demilitarized zone and the two coasts (GG) [14]. GW is a mountainous region and is famous for mountain cultivated ginseng, while ginseng in GG and GN is cultivated in the field using shading systems [25][26][27][28][29][30][31][32][33][34][35][36][37][38][39]. Three provinces accounted for 40.3% of all ginseng production in South Korea in 2018 [40]. In traditional agriculture, farmers create and conserve new varieties, the preponderance of which are landraces. They decide which materials to conserve or discard, and they recycle seed on their farms, which are very often quite small. Such systems are highly dynamic in terms of genetic content and adaptation of the landraces [41].

Genetic Differentiation of Korean Ginseng Accessions
Although all of the ginseng accessions could be clustered into one of seven populations using STRUCTURE software, the accession clustering did not show clear separation according to geographic origin ( Figure 2). In addition, AMOVA showed low genetic differentiation (PhiPT) and gene flow (Nm) between collection area (PhiPT = 0.028, Nm = 17.213) and STRUCTURE group (PhiPT = 0.031, Nm = 15.431) ( Table 2). The ginseng landraces also showed low PhiPT (0.034) and high Nm (14.322), although STRUCTURE divided them into 12 groups.
Ginseng seed and seedlings necessarily originate from wild ginseng [42], and wild ginseng, which is rare, has been increasingly endangered because of high demand [43]. Natural populations of rare and threatened species are typically small and geographically isolated [44], and the growth and persistence of small populations are highly influenced by stochastic events such as genetic drift and inbreeding; such events result in reduced genetic diversity and fitness [45,46]. Karron [47,48] reported that some species with restricted distributions showed less genetic diversity than congeneric species with widespread distributions, but Cole [49] observed that it is not clear that rare plant species will necessarily have less gene flow than common species. Ellstrand [44] summarized arguments for expecting that gene flow might increase in rare plant populations, resulting in less differentiation. In Korea, farmers have been cultivating, self-seed gathering, and seed-selling the ginseng shoots themselves [50,51], and it seems that this method of ginseng cultivation is responsible for the low PhiPT and Nm in the Korean ginseng landraces at NAC. Cultivated ginseng seeds obtained from a few wild ginsengs are amplified and self-seed gathered by local farmers, and then they are sold on the ginseng market. As this cycle repeated over time, ginseng seeds of the same genetic background would have spread to other regions; it is assumed that this is the reason the ginseng landraces collected from different regions showed low PhiPT and high Nm.
The ginseng breeding lines also showed low PhiPT (0.012) and high Nm (40.761), although STRUCTURE divided the breeding lines into only two groups. Interestingly, both groups included two ginseng varieties-gumpoong and chunpoong. The two varieties, which were registered in 2002 (cv. Chunpoong) and 2003 (cv. Gumpoong), were bred by selecting individual plants from Korean landraces [11]. In Korea, 27 ginseng varieties have been registered with the Korea Seed & Variety Service, and all were bred by selecting individual plants from landraces [52]. In general, selecting individual plants is more common for developing new ginseng varieties than is crossbreeding because of the long generation period and small number of seeds with plant selection . However, these varieties were selected for their excellent root form, but their genetic background has been little analyzed [11]. Meanwhile, the low genetic diversity in ginseng breeding lines might be because plants have been selected for morphological traits only; because morphological traits are easily affected by environmental conditions [54], their gene profiles could be similar in spite of their different phenotypes.

Increasing the Genetic Diversity of the NAC's Ginseng Germplasm
As mentioned above, the population of Korean ginseng landraces is expected to be very small, and it has been reported that small populations increase genetic drift and inbreeding [26], which could in turn influence these small populations by changing patterns of genetic diversity. One concerning drift-induced genetic change is this erosion of genetic variation; loss of genetic variation may decrease the potential for a species to persist in the face of abiotic and biotic environmental change [55,56]. The primary solution to the genetic impoverishment of crop germplasm is genetic conservation and utilization in breeding the vast genetic variation found in populations of the wild progenitors and landraces of cultivated plants [3]. However, it is very difficult to collect wild ginseng because it is such a rare plant. To overcome the shortage of genetic diversity in ginseng, researchers in Korea have attempted crossing interspecies, tissue culture, and mutation breeding [11]. Ahn et al. [57] and Kim et al. [58] attempted to cross P. ginseng and P. quinquefolius, but they did not get seeds from F2 plants. Kim et al. [56] studied the tissue culture of F1 plant propagation in a ginseng hybrid (P. ginseng × P. quinquefolius) to solve the seedless problem. In addition, some research groups have attempted Agronomy 2020, 10, 1643 10 of 13 mutation breeding using irradiation or chemicals, but they did not develop new ginseng varieties; they did confirm plants with high ginsenosides [59][60][61][62][63].
Bang et al. [11] highlighted the need to develop new varieties that are strong against abnormal weather conditions and reflect different trends. To develop such new varieties, NAC should attempt to secure the genetic diversity of ginseng. Collecting and securing P. ginseng and related species will be required for increasing the diversity, and the effort will also require collaboration with other ginseng research institutes to establish effective breeding methods.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4395/10/11/1643/s1; Table S1. List of 451 ginseng accessions in this study; Table S2. List of 33 SSR markers in this study; Table S3. Distribution of collected areas in 451 ginseng accessions; Table S4. Genetic diversity parameters of 33 SSR markers; Table S5. Pairwise population PhiPT values (above diagonal) and Nm values based on 999 permutations (below diagonal) from AMOVA using collection area. All PhiPT values were significantly greater than 0 (p < 0.0001); Table S6. Pairwise population PhiPT values (above diagonal) and Nm values based on 999 permutations (below diagonal) from AMOVA using STRUCTURE result of 451 ginseng accessions. All PhiPT values were significantly greater than 0 (p < 0.0001); Table S7. Pairwise population PhiPT values (above diagonal) and Nm values based on 999 permutations (below diagonal) from AMOVA using STRUCTURE result of 39 ginseng landraces. All PhiPT values were significantly greater than 0 (p < 0.0001).