Genetic Structure of the Liriope muscari Polyploid Complex and the Possibility of Its Genetic Disturbance in Japan

Anthropogenic activities, such as the movement of plants through greening, can result in genetic disturbance that can interfere with local adaptation in wild populations. Although research is underway to prevent genetic disturbance associated with greening, genetic disturbance of intraspecific polyploidy, which is estimated to be present in 24% of vascular plants, has not been well studied. Liriope muscari is a polyploid complex with known diploid (2n = 36), tetraploid (2n = 72), and hexaploid (2n = 108) forms. The plants of this species tolerate dry and hot conditions and are therefore frequently used for greening and gardening. However, the distribution of this polyploid in Japan, its genetic structure, and genetic disturbance are not known. In this study, we investigated the polyploidy distribution and genetic structure in naturally distributed L. muscari in Japan using chloroplast DNA (cpDNA) haplotypes and nuclear DNA (nDNA). Commercially produced individuals were also studied and compared with natural populations to assess any genetic disturbance of the ploidy complex in this species. Chromosome counts, cpDNA, and nDNA results showed three genetically and cytologically distinct groups in Japan: first, a tetraploid group in mainland Japan; second, a hexaploid group in the Ryukyu Islands; and third, a diploid and tetraploid group in the Ryukyu Islands. Significant isolation by distance was also detected within the three groups (p = 0.001). Genetic disturbance due to greening and gardening should be avoided among the three groups. Genetic disturbance can be reduced by using individuals derived from natural populations that are close to the sites used for greening and gardening. For commercially produced individuals, genetic disturbance is unlikely in the Kanto region, an area of high usage, while genetic disturbance is thought possible in the Ryukyu Islands.


Introduction
The genetic structure of a population is formed by various mechanisms including the interaction of gene flow, genetic drift, and natural selection [1]. The genetic structure of populations may include adaptations to the local environment [2], which may be the starting point for expansion into different environments and speciation. Such genetic structure is important for the maintenance of genetic diversity and biodiversity.
Anthropogenic activity, such as plant transfer through greening, can result in genetic disturbance, which can affect a natural population by disrupting its genetic structure and disturbing any adaptation to local conditions [3]. Several studies have been conducted to prevent genetic disturbance caused by revegetation [4][5][6][7].

Polyploidy Level
The polyploidy level of 116 individuals at 86 sites were identified as follows (Table 1, Figure 1): 18 were diploid (8 sites), 61 were tetraploid (49 sites), and 37 were hexaploid (29 sites) (Table 1, Figure 1). We estimated the chromosome number using multiple individuals at 21 sites. The diploid individuals were found sporadically in the Ryukyu Islands and Taiwan, the tetraploids were found in mainland Japan and the southern Ryukyu Islands, and the hexaploids were found only in the central Ryukyu Islands. All 18 commercial individuals of L. muscari obtained from six nurseries in the Kanto region were tetraploid. Of the three individuals obtained from the nursery on the main island of Okinawa, two were tetraploid and one was hexaploid ( Figure 2).

Chloroplast DNA Haplotypes
By combining the sequence data of four cpDNA regions that were analyzed, we identified nine haplotypes (m01, m02, m03, m04, m05, m06, m07, m08, and m09). The nucleotide substitutions and indels are shown in Table S1. The haplotype diversity (h) and nucleotide diversity (π) of L. muscari are shown in Table 2. TCS haplotype network based on a sequence with four regions combined showed that the central haplotype was m01. Haplotypes m02, m03, and m09 were confirmed as independent haplotypes, each with a single mutation step. Systematic connections were indicated between m04 and m05 and between m06, m07, and m08. In terms of the relationship between the haplotype network and the polyploids, of the nine haplotypes, a different polyploidy level was confirmed to be present within the same haplotype in m01, m03, and m07 ( Figure 3b).

Chloroplast DNA Haplotypes
By combining the sequence data of four cpDNA regions that were analyzed, we identified nine haplotypes (m01, m02, m03, m04, m05, m06, m07, m08, and m09). The nucleotide substitutions and indels are shown in Table S1. The haplotype diversity (h) and nucleotide diversity (π) of L. muscari are shown in Table 2. TCS haplotype network based on a sequence with four regions combined showed that the central haplotype was m01. Haplotypes m02, m03, and m09 were confirmed as independent haplotypes, each with a single mutation step. Systematic connections were indicated between m04 and m05 and between m06, m07, and m08. In terms of the relationship between the haplotype network and the polyploids, of the nine haplotypes, a different polyploidy level was confirmed to be present within the same haplotype in m01, m03, and m07 ( Figure 3b).

Nuclear DNA
The Mantel test for Mash distance to SNP-based genetic distance showed a significant correlation between data sets (p = 0.001). SNP analysis finally called 93 SNPs with a genotyping rate of over 90%, and sequence coverage averages were above 30 for all individuals. Stacks parameters adjusted during SNP call settings were -m = 8, R0.85, -n = 1. The statistics obtained from stacks-2.60 are shown in Table 3.
In the neighbor-joining tree output from Mashtree, three groups related to chloroplast DNA haplotypes and polyploidy were recognized: Group 1 includes mainland Japan and is tetraploid with major chloroplast DNA haplotypes m04 and m09; Group 2 is distributed in the Ryukyu Islands and is hexaploid with the major cpDNA haplotype m02; and Group 3, also distributed in the Ryukyu Islands, is diploid and tetraploid, with the major cpDNA haplotypes m01 and m03 (Figure 3a,b). Group 3 was identified as several closely related clusters composed of diploid and tetraploid (Figure 3a). PCoA plots also distinguished three groups (Figure 3c). Nuclear and chloroplast DNA results were generally consistent, but some discrepancies were observed. Haplotype m01 was found in all three groups, and haplotype m07 was commonly found in groups 2 and 3 ( Figure 3a). Mash distance and geographic distance in the three groups were significantly correlated (p = 0.001). In contrast to the trend of genetic diversity for cpDNA, a trend of increasing genetic diversity with increasing polyploidy level was observed for nDNA (Tables 2 and 3). Table 3. nDNA Genetic diversity of L. muscari. N: number of samples, He: genetic diversity, Ho: observed heterozygosity, F IS : inbreeding coefficient, π: nucleotide diversity.

Genetic Characteristics of Commercially Produced L. muscari
The polyploidy, cpDNA haplotype, and nDNA characteristics of 18 L. muscari individuals obtained from nurseries 1-6 in the Kanto region were consistent with those naturally distributed around the nurseries. All were tetraploid, belonged to Group 1 identified by nDNA, and had cpDNA haplotypes m04, m05, and m09 (Figures 3b and 4). Of the three individuals obtained from the nursery on Okinawa Island, one was hexaploid and had haplotype m02 and was included in Group 2. The remaining two individuals were tetraploid, had cpDNA haplotype m03, and belonged to Group 3. One individual in Group 2 matched the major type obtained from the Okinawa mainland, while two individuals in Group 3 had cpDNA haplotype m03, a type with a more southerly distribution (Figures 3b and 4).   Genetic diversity of commercially produced L. muscari in nurseries in mainland Japan, with the main area of consumption near Tokyo, was comparable to the genetic diversity of Group 1 in mainland Japan in both cpDNA and nDNA (Tables 2 and 3).

Distribution of Polyploidy Complex
All L. muscari individuals sampled from mainland Japan were tetraploid in our study; but the Ryukyu Islands samples comprised a mix of three polyploids. However, the hexaploid L. muscari has been reported in Hiroshima, mainland Japan [17]; the hexaploid form is therefore probably distributed throughout mainland Japan but only at a low frequency. Previous studies have reported the diploid form in Zhejiang Province, China; tetraploid in Korea; and hexaploid in mainland Japan [17,19,20,27]. Combined with the present results, there may be considerable overlap in the distribution of polyploidy, with the diploid distributed from the Ryukyu Islands to Taiwan and Zhejiang Province, China; the tetraploid from mainland Japan to the Ryukyu Islands and Korea; and the hexaploid from mainland Japan to the Ryukyu Islands. In our study we were able to clarify the polyploidy distribution pattern roughly in Japan. However, due to the limited number of survey sites and individuals, the distribution of L. muscari polyploidy throughout its distribution range remains unknown in term of the frequency of polyploidy in each region.
Two closely related diploid-tetraploid pairs were identified in the neighbor-joining tree by Mash distance, suggesting that the tetraploids have multiple origins. The multiple origins of polyploidy has also been reported elsewhere [28,29]. The multiple origins of polyploidy may be one of the factors contributing to the geographic obscurity of the distribution of the L. muscari polyploid complex in the Ryukyu Islands.

Genetic Structure of L. muscari in Japan
The three groups clearly differed in nDNA were also mostly consistent with the results for polyploidy and cpDNA, but partial discrepancies were observed in cpDNA. Incomplete lineage sorting and chloroplast capture are the main causes of mismatching between nuclear and chloroplast DNA [30][31][32]. In our present study, both processes are also possible; but given the wide distribution for haplotype m01, which was one of the two chloroplast DNA haplotypes that did not match the nuclear DNA results, and the fact that it was found in all three groups with different polyploidy, strongly suggests a high probability of incomplete lineage sorting.
The diverse results for polyploidy, chloroplast DNA, and nuclear DNA among the three groups suggest that a genetic barrier to gene flow exists between these groups. In general, geographic isolation and climatic conditions are known to be barriers to gene flow [33][34][35]. The adjacent Groups 1 and 2 are separated by the Strait of Tokara, which is known to have a different flora due to geographic isolation [35]. While Group 1 is tetraploid, Group 2 is hexaploid, and such differences in chromosome number are also a barrier to gene flow [36]. The boundary between Group 1 and Group 2 distributions is unclear, probably due to low sampling density, but multiple barriers may also prevent gene flow.
It is difficult to explain the geographic and climatic barriers to gene flow between Groups 2 and 3, whose distributions overlap in the Ryukyu Islands. Since Group 2 is hexaploid and Group 3 is diploid and tetraploid, a barrier to gene flow due to reproductive isolation at different polyploidy levels can be inferred. Group 3 includes diploid and tetraploid, but within the wild population the ploidy level is fixed to either diploid or tetraploid. Fixation of the polyploidy level in each local population suggests that diploid and tetraploid may be exclusive. It may be possible that gene flow is restricted between polyploids within Group 3.
The relationship between Mash distance and geographic distance showed a significant correlation for all groups. Restricted interbreeding between geographically distant individuals results in IBD [37]. L. muscari is a common species at low elevations within the range of our collection of samples. The possibility of restricted gene flow between diploids and tetraploids in Group 3 should be noted, but, in any case, it suggests that geographic proximity is important for interbreeding within groups.
We found lower cpDNA diversity of hexaploid individuals in the Ryukyu Islands compared to other polyploidy levels (h = 0.399, π = 0.00030). If both haplotype diversity and nucleotide diversity are low, the hexaploid may have experienced a more recent origin or a more restrictive bottleneck compared to other polyploidy levels. In either case, the hexaploid in the Ryukyu Islands is likely dependent on limited genetic sources. On the other hand, the higher genetic diversity of hexaploids compared to diploids and tetraploids in nDNA may reflect the increased diversity associated with genome duplication.

Taxonomic Confusion
The genus Liriope exhibits a certain amount of taxonomic confusion. L. tawadae is characterized by large plant size, broad and long leaf blades, and large flowers and long flower stalks, and has been reported in the Ryukyu Islands [38]. Due to the lack of morphological information, the relationship between L. tawadae and the result of this study is unclear. Some cpDNA haplotypes of L. muscari are shared with those of L. spicata (Watanabe, unpublished data), suggesting past hybridization. The situation is further complicated because L. spicata is known to be diploid, tetraploid, and hexaploid [19,21,39]. Further taxonomic reexamination, including related species, is therefore needed.

Potential of Anthropogenic Disturbance and Countermeasures
The commercially produced L. muscari was abundant near Tokyo; however, there was no obvious risk of genetic disturbance evident from our study. All L. muscari produced in nurseries near Tokyo were confirmed to be tetraploid and genetically close to naturally distributed individuals in the neighborhood. On the other hand, there is a possibility of genetic disturbance of L. muscari in the Ryukyu Islands due to greening. Despite our limited number of samples, we observed that genetically distinct L. muscari were being sold together. In addition, individuals from genetically distinct groups that were sold together also differed in their polyploidy levels, with one of the three individuals studied being Group 2 hexaploid and two being Group 3 tetraploids. The mixing of different polyploidy levels can cause additional problems. It has been noted that orthotopic growth of different polyploidy levels causes the eradication of minor polyploidy levels [40]. In fact, the polyploidy levels were fixed in populations where chromosome counts of multiple individuals were examined in this study. It should also be noted that pentaploids can easily be obtained by artificially crossing tetraploids and hexaploids from the Ryukyu Islands (Watanabe, unpublished data).
An effective means of preventing anthropogenic genetic disturbances of L. muscari is to avoid contact between genetically distinct groups. By selecting L. muscari that belong to the same genetic group as the natural population surrounding the proposed greening site, and by using seed collected from a single group during seedling production, contact between genetically distinct groups can be reduced. Considering within groups, genetic distance between geographically close individuals is close, so using individuals derived from natural populations that are geographically close to the proposed greening site is expected to further reduce genetic disturbance. Group 3 suggests multiple origins of the tetraploid, and although the situation may be complex, it is expected that the supply of individuals with the same polyploidy level from a natural population near the proposed site at the time of greening will reduce genetic disturbance.
Determining a level of genetic disturbance based on genetic information remains difficult for greening officials. With L. muscari in Japan, it is possible to recognize groups that should avoid contact with each other based on their geographic distribution and polyploidy levels. Three groups in Japan should avoid being genetically disturbed. The first group is the tetraploid distributed in mainland Japan. There are also two groups in the Ryukyu Islands that have overlapping geographic boundaries but can be distinguished by polyploidy level. The second group is the hexaploid of the Ryukyu Islands. The third group is the diploid and tetraploid of the Ryukyu Islands. At present it is difficult to estimate the polyploidy level of L. muscari by morphological features, but accumulating such morphological information will be an effective way to test for correspondence with the polyploidy level in order to aid their recognition in the field.

Collection of Materials
We collected between one and five individuals from 86 sites (totaling 116 individuals) in the natural distribution area, ranging from Niigata Prefecture in Japan to Taiwan. A further 21 commercially produced individuals were also collected: three each from two nurseries in Tokyo, four nurseries in Saitama Prefecture, and one nursery in Okinawa Prefecture (Table 1).

Determination of Polyploid Level
For each of the 137 L. muscari individuals collected, we counted their chromosome number using the aceto-orcein squash method. The root tip meristems were placed in a 0.002 M 8-hydroxyquinoline solution and pretreated at room temperature for 4-5 h. Subsequently, they were left in acetic acid alcohol (3:1, anhydrous ethanol-glacial acetic acid) to harden for at least five hours at 4 • C. After hardening, they were disaggregated for approximately 40 s in a 60 • C disaggregation solution (2:1, 1N hydrochloric acid-45% acetic acid) and stained for 1 to 15 min in a 2% aceto-orcein solution, and then squashed on a glass slide. The number of chromosomes was counted in somatic cells at metaphase.

Chloroplast DNA Analysis
The total DNA was extracted following the CTAB method of Doyle and Doyle (1987) after removing polysaccharides using the method of Setoguchi and Ohba (1995) [41,42].
Using four pairs of primers developed by Taberlet Liston and Kadereit (1995) [43][44][45][46], polymerase chain reaction (PCR) amplification was conducted for the following intron and intergenic spacers and regions: trnK 5 intron: (5 -CTCAACGGTAGAGTACTCG-3 , 5 -CCAAAAACTTCCACAGGTTCG-3 ), trnT-trnL: (5 -GCGATGCTCTAACCTCTGAG-3 , 5 -TAGCGTCTACCGATTTCGCC-3 ), trnL-trnF: (5 -ATTTGAACTGGTGACACGAG-3 , 5 -ATTTGAACTGGTGACACGAG-3 ), and atpB-rbcL: (5 -ACTTAGAGGAGCTCCCGTGTCAATC-3 , 5 -GAGTTACTCGGAATGCTGCC-3 ) intergenic regions. PCR amplification was performed using a PCR Thermal cycler SP (Takara), and base sequence determination was performed using a CEQ 8800 capillary DNA sequencer (Beckman Coulter). The base sequence obtained in this manner was aligned using the default parameters of the ClustalW program implemented in the MEGA X software [47]. The chloroplast DNA haplotypes were detected based on the arrangement of 3021 bases for the four combined domains. In addition, using DnaSP version 6.12.03 [48], the haplotypic diversity (h) and nucleotide diversity (π) were calculated for each polyploid and each region in which differences in polyploid distribution had been determined [49]. A parsimony haplotype network diagram of chloroplast DNA was created using the PopART 1.7 TCS network based on the data set in which the base sequences for the trnK intron, the trnT-L, trnL-F and atpB-rbcL intergenic regions [50], and the matK gene region were combined. Insertion-deletion (INDEL) mutations were excluded from analysis on the TCS network, so a dataset was created with INDELs replaced with base substitutions. We excluded from the analysis all repeated insertion-deletion of sequences for which the homology of the mutations was unclear.

Nuclear DNA Analysis
Nuclear DNA was investigated by sequencing with MIG-seq analysis [51,52], which creates a reduced library of genomes for samples with known chromosome numbers and chloroplast DNA haplotypes. The region flanked by SSRs was PCR amplified using 8 primers in the first PCR, and the resulting amplicons were indexed in a second PCR.
The indexed DNA library was sequenced using the MiSeq Reagent Kit v3 (150-cycle) (Illumina, San Diego, CA, USA). The obtained reads were filtered using Trimmomatic 0.39 [53]. Adapter sequences were removed using default settings and short reads were removed (MINLEN:79). Low quality reads were then removed and trimmed (SLIDING-WINDOW:4:15, CROP:79). To investigate genetic distances between individuals, we used Mashtree [54] to create a neighbor-joining tree based on Mash distance between individuals. Mashtree parameters were k-mer length 21 (-kmerlength 21), sketch k-mer count was set to 30,000 (-sketchsize 30,000), and k-mers with a count of less than 2 were excluded from the analysis (-mindepth 2).
The genetic analysis of polyploid data involves certain challenges. The main one is the difficulty in estimating allele frequencies of the polyploid. Many existing analysis methods require allele doses, for example, hundreds of coverages to recover tetraploid alleles with 90% confidence [55]. It is difficult to obtain this amount of information by normal greening. k-mer analysis using the MinHash method does not require allele dosage, thus alleviating the difficulties of polyploid analysis [56]. However, Mashtree-derived Mash distances provide information equivalent to genetic distances. They are obtained by evaluating the similarity of reads resolved into k-mers. To assess the similarity between individuals, Principal Coordinate Analysis (PCoA) using Mash distance was performed in GenAlex 6.502. Mantel tests were performed on GenAlex 6.502 for Mash distance and geographic distance to estimate isolation by distance (IBD) between the groups recognized by PCoA (Tables S2-S4). However, such data should be treated with caution when evaluating between different polyploidy levels using Mash distances, as they can be subject to bias [56]. To assess the plausibility of comparisons between different polyploidy levels using the Mash distance, a Mantel test was performed on the genetic distance obtained from single nucleotide polymorphisms (SNP) and the Mash distance (Table S5). The Mantel test was performed using GenAlex 6.502, and the SNP call was made using the denovo_map.pl pipeline from stacks-2.60 [57]. The parameters for estimating the genetic diversity of L. muscari were also obtained with stacks-2.60. Stacks is an analytical protocol for a diploid model, which is usually difficult to apply to a polyploid. Although several informative alleles are lost, stacks can be used to analyze polyploidy by treating them as diploid by linking copies derived from polyploidy [58].

Conclusions
In Japan, there are three groups of natural populations of L. muscari recognized by polyploidy, cpDNA, and nDNA: the first is the mainland tetraploid group; the second is the Ryukyu Islands hexaploid group; and the third is the Ryukyu Islands diploid and tetraploid group. For the reduction of potential risks regarding the destruction of the local adaptation of natural individuals around the greening area and for the establishment of the planted individuals, genetic disturbance associated with greening between these three groups must be avoided. In the Kanto region near large cities, the possibility of genetic disturbance due to greening is low because the cultivated products and the surrounding natural populations belong to the same group. On the other hand, in the Ryukyu Islands, individuals belonging to different groups were being sold in the same nursery, suggesting the possibility of genetic disturbance between groups due to greening. Within the three groups, distinct IBD could be identified in nDNA. Using individuals derived from natural populations that are geographically close to the proposed greening site is expected to further reduce genetic disturbance.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/plants11223015/s1, Table S1: Nucleotide substitutions and indels observed in the Liriope muscari; Table S2: Geographic distance (below) and Mash distance (above) for Group 1; Table S3: Geographic distance (below) and Mash distance (above) for Group 2; Table S4: Geographic distance (below) and Mash distance (above) for Group 3; Table S5: Marsh distance (below) and SNP-based genetic distance (above) for all individuals.