Genotyping by sequencing (GBS) is a simplified genome sequencing technology based on second-generation sequencing technology with the advantages of high throughput and low cost, which could detect a large number of variation sites of single nucleotide polymorphism (SNP) for the study of genetic relationship, population structure, genetics and reproduction of animals and plants (Poland and Rife, 2012). SNP refers to DNA sequence polymorphism caused by single nucleotide variation in the genome, which has the advantages of high density, wide distribution, high stability, simple typing, rapid and high throughput (Kumar et al., 2012). At present, SNP, as a third-generation molecular marker, is more and more widely used in the research of animal and plant genetics and reproduction. For example, Baral et al. (2018) used the SNP obtained by GBS to study the genetic diversity of crested wheatgrass (Agropyron cristatum (L.) Gaertn.), revealing the potential of GBS application in the characterization of non-model polyploid plants with complex genomes. Luo et al. (2019) used SNPs generated by GBS to analyze the genetic diversity and genetic structure of Camelina sativa Spring Panel.

Dendrobium is the third largest genus in the family of Orchid. There are about 74 species and 2 varieties in China, of which more than 40 species are traditional Chinese medicine, especially Dendrobium huoshanense C.Z. Tang and S.J. Cheng and Dendrobium officinale Kimura et Migo, which are valuable Chinese medicinal herbs. There is a recorded in “Ben-Cao Gang-mu (Compendium of Materia Medica)”, that “Dendrobium has the effects of ‘Chubi Xiaqi (Terms of traditional Chinese Medicine, means removing bi (A TCM disease refers to the abnormal symptoms of joints and muscles caused by external adverse stimulation) and descending qi (A TCM disease refers to the most fundamental and subtle substances that constitute the human body and maintain life activities))’, tonify the ‘Wuzang (means heart, liver, spleen, kidney and lung)’, strong body, nourish Yin (Terms of TCM) and benefit qi. Taking it for a long time can enhance the function of the spleen and stomach, so as to make people flexible, enhance immunity and prolong life.” Modern pharmacological research shows that Dendrobium has the functions of anti-oxidation, anti-cancer, antithrombotic, hypoglycemic, enhancing human immunity, anti-aging and so on (Song et al., 2014). Dendrobium is also a kind of plant “Yaoshi Tongyuan” (means that the homology of medicine and food). It is a valuable health product, and the market demand is very large. Due to over excavation, habitat destruction, large market demand, and the slow growth of Dendrobium itself, low seed germination rate, strict requirements on habitat and other reasons, all kinds of Dendrobium species have been listed as endangered species. Therefore, to meet the market demand, Dendrobium species, especially Dendrobium huoshanense and Dendrobium officinale, are cultivated on a large scale. And the research on various aspects of Dendrobium has also been widely valued, such as the identification of Dendrobium species (Duan et al., 2019), genetic relationship (Ren et al., 2013), genetic diversity (Song et al., 2016), resistance gene research (Li et al., 2013), gene cloning and identification (Li et al., 2018), medicinal components and medicinal value (Si et al., 2016). At present, the whole genome of Dendrobium officinale (Yan et al., 2015) and Dendrobium catenatum (Zhang et al., 2016) has been sequenced, so researchers can use it as a reference genome to make it easier to use GBS technology to study Dendrobium plant genetic diversity, reproduction, gene cloning and identification, as well as medicinal components. For example, Ryu et al. (2019) used the whole genome of Dendrobium officinale as the reference genome to analyze the genetic characteristics of Dendrobium mutants and cultivated species using SNPs produced by GBS. It is rarely reported to analyze the differences and genetic relationships of different species of Dendrobium by using SNPs produced by GBS. Therefore, this study analyzed the differences and genetic relationships of cultivated Dendrobium huoshanense, Dendrobium officinale, Dendrobium moniliforme and Dendrobium fanjingshanense by using genome-wide SNP markers obtained by GBS technology, to provide references for the study of Dendrobium.

1 Results and Analysis

1.1 Data filtering and alignment to reference genomes

The number of clean reads per each sample in Dendrobium huoshanense (abbreviation: HS), Dendrobium officinale (abbreviation: TP), Dendrobium moniliforme (abbreviation: XJ) and Dendrobium fanjingshanense (abbreviation: FJS) were 1 390 262, 1 286 162, 1 380 009 and 1 170 337, respectively, and the clean bases were 39 557 949 bp, 366 277 414 bp, 391 592 914 bp and 333 221 985 bp, respectively. The TP reads mapped to reference genome of D. catenatum accounted for 99.08% of the total reads, which was the highest. And there was no significant difference between HS, XJ and FJS, which were 96.45%, 96.98% and 96.91%, respectively. The proportion of clean reads pairs mapped to the reference genome to the total clean reads and the sequencing length meets a certain threshold to the total number of reads is also the highest in TP, which was 91.01%. There was no significant difference between HS, XJ and FJS, which were 80.15%, 80.08 and 79.21%, respectively (Table 1). The highest coverage of sequencing was TP, which was 6.09%, and the lowest was FJS, which was 4.74% (Table 2).

Table 1 Summary of GBS clean reads per sample and percentage of alignment to reference genome sequence

Table 2 The result of average sequencing depth, coverage (%) of 4 species of Dendrobium

1.2 Variation detection of SNP

GATK (McKenna et al., 2010) software was used to detect that the average number of SNPs in each sample of HS, TP, XJ and FJS were 1 507 746, 893 333, 1 364 605 and 1 227 006, respectively. The number of transition SNP and transvertion SNP were 897 857 and 609 668, 546 045 and 344 283, 810 614 and 551 968, 727 566 and 495 436, respectively. And only the ratio of transition SNPs to transvertion SNPs of TP was 1.58, and the rest were 1.47. The average number of heterozygous SNPs produced by each sample of HS, TP, XJ, and FJS were 196 880, 229 222, 168 830 and 136 823, respectively. The proportion of heterozygous SNPs in the total number of SNPs was 0.257 for TP, which was the highest, and 0.13 for HS, 0.12 for XJ and 0.11 for FJS. The average number of homozygous SNPs produced by each sample of HS, TP, XJ and FJS were 1 310 866 (highest), 664 110 (lowest), 1 195 774, and 1 090 183, respectively. The lowest proportion of homozygous SNP number in the total SNP number was TP, which was 0.74, the highest was FJS (0.89). HS and XJ were 0.87 and 0.88 (Table 3).

Table 3 Summary of average SNPs per sample detected by GBS

1.3 Analysis of population structure and genetic distance

The total number of SNPs for population structure analysis obtained from Clean reads after filtration was 420 445, including 359 521 in D. huoshanense, 118 484 in D. officinale, 153 325 in D. moniliforme and 65 231 in D. fanjingshanense. The genetic distance matrix was constructed with Tree BeST software (Vilella et al., 2009), and the Neighbor-joining tree of the system was constructed with the adjacency method (Figure 1). It showed that the samples of D. officinale were clustered into one branch, as well as D. moniliforme. D. fanjingshanense was clustered between D. moniliforme and D. officinale alone, and the samples of D. huoshanense were divided into two branches except one sample, with a support rate of 100%. It also revealed that the samples from different cohabitation groups of D. huoshanense were obviously mixed, that was, the samples from cohabitation groups were not completely clustered together, or even divided into different branches, indicating that there was a phenomenon of gene hybridization or gene transfer among D. huoshanense samples.

Figure 1 Neighbor-joining tree based on pairwise distance matrix representing the grouping of the 91 Dendrobium plants obtained from 420 455 SNPs-GBS

The genetic structure based on Admixture software (Alexander et al., 2009) revealed that the best K value was 2, samples of D. officinale were clustered into one branch, and the remaining Dendrobium samples were clustered into the other branch, that was, D. huoshanense, D. moniliforme and D. fanjingshanense were clustered together, suggesting that their genetic relationship was closer. When K = 3, the samples of D. officinale were grouped into one cluster, and the remaining Dendrobium samples were clustered into two branches, some samples of D. huoshanense, all samples of D. moniliforme and D. fanjingshanense were grouped into the second cluster, and some samples of D. huoshanense were grouped into the third cluster; When K = 4, the samples of D. officinale were grouped one branch, the samples of D. moniliforme and D. fanjingshanense were grouped one branch, and the samples of D. huoshanense were divided into two branches, which was similar to the result of an unrooted neighbor-joining phylogenetic tree (NJ tree) (Figure 2; Figure 3).

Figure 2 Genetic structure of 91 Dendrobium samples for K=2-8 based on the Admixture software

Figure 3 Cross-validation error and K value Principal coordinates analysis (PCoA) (Yang et al., 2011) was not exactly the same as Structure analysis, which showed that D. officinale were clustered together, D. moniliforme and D. fanjingshanense were clustered together, while D. huoshanense were clustered together, with the total genetic variation of 22.14%, while PC1 and PC2 were 18.79% and 3.35%, respectively (Figure 4).

Figure 4 Principal coordinates analysis (PCoA) of pairwise simple matching dissimilarities between individuals Note: Axis1 and Axis 2 explain 22.14% of total variation

According to the genetic distance between populations calculated by GenAlEx software (Table 4), it was found that the maximum genetic distance was 0.306 2 between SGL and TP3, and the minimum was 0.129 3 between TP2 and TP3. Among them, the genetic distance between D. huoshanense and D. officinale populations was the largest, with a range of 0.290 1~0.306 2, followed by D. huoshanense and D. fanjingense populations with a range of 0.214 0~0.2233, followed by D. huoshanense and D. moniliforme with a range of 0.206 9~0.214 2, while the genetic distance between D. huoshanense populations was small, with a range of 0.182 7~0.199 1. D. moniliforme and D. officinale were 0.259 1～0.279 2, D. moniliforme and D. fanjingense were 0.201 2, D. officinale and D. fanjingense were 0.257 2～0.259 9. D. officinale populations are the smallest, with a range of 0.129 3~0.133 1. According to the genetic distance, it could be concluded that if the genetic distance between D. officinale populations was smaller than that between D. huoshanense populations, D. officinale has a far genetic relationship with the other three Dendrobium species, and in order of distance, D. huoshanense> D. fanjingense>D. moniliforme.

Table 4 The Genetic distance between populations

2 Discussion

Since the whole genome determination of D. officinale and D. catenatum in 2015 and 2016, it has provided convenience for the future research on the evolutionary history, genetic relationship, genetic reproduction, gene cloning and identification, gene function and so on of Orchid or Dendrobium Plants. Botanists could use abundant genomic data and genomic tools to better understand the physiological, molecular and genetic mechanisms of Dendrobium plants, and conduct research on comparative genomics, gene editing and so on. For example, Zhang et al. (2016) believed that the extensive ecological niche of Dendrobium is related to the expansion of resistance related genes, and the synthesis of medicinal polysaccharides seems to be related to the extensive replication of the gene encoding glucomannan synthase. In this study, SNP analysis produced by GBS was used to compare the differences and genetic relationships of four Dendrobium species. The results showed that the percentage of D. officinale clean reads mapped to D. catenatum reference genes was the highest, and the number of SNPs of D. officinale was the lowest, suggesting that D. officinale and D. catenatum had the closest genetic relationship, which was consistent with the research results of Li et al. (2013). Principal coordinates analysis (PCoA) found that D. moniliforme and D. fanjingense clustered together, suggesting that they were closely related, while D. officinale, D. huoshanense and D. moniliforme were far separated, suggesting that their genetic relationship was far away. Structure analysis showed that when K=2, samples of D. officinale were clustered into one branch, D. moniliforme, D. fanjingense and D. huoshanense gathered into another branch, indicating that D. officinale was far away from these three kinds of Dendrobium. When K=3, it was divided into three branches, one branch of D. officinale, and the rest were divided into two branches. Some D. huoshanense, D. moniliforme and D. fanjingense gather into one branch, and some D. huoshanense gather into another branch, which was consistent with the results of Li et al. (2013) that D. huoshanense and D. moniliforme have a relatively close genetic distance. When K=4, it was divided into four branches, D. moniliforme and D. huoshanense were separated, D. moniliforme and D. fanjingense were grouped into one branch, D. officinale gather into one branch. D. huoshanense was divided into two branches, and there was confusion among the populations of D. huoshanense, indicating that D. moniliforme and D. fanjingense were closely related, and there were certain differences between D. huoshanense and D. moniliforme. NJ tree analysis showed that D. officinale, D. moniliforme and D. fanjingense were gather into one branch, respectively, and D. fanjingense was between D. officinale and D. moniliforme. Except for one sample, D. huoshanense samples were divided into two branches, of which some population samples were divided into two branches, and the same population samples were not completely gathered, that was, the samples between populations were mixed. It was found that D. huoshanense was divided into D. huoshanense 1 and D. huoshanense 2, that was, D. huoshanense has two species, which was consistent with the results of this study that that D. huoshanense was divided into two branches. However, in this study, it was found that some populations have these two branches, indicating that these two branches were closely related and have a mixed situation. According to the genetic distance analysis, it was confirmed that D. officinale was far away from the other three kinds of Dendrobium, and that D. moniliforme was close to D. fanjingense, D. moniliforme and D. huoshanense. In addition, there was little difference in genetic distance between D. huoshanense populations.

In this study, the differences and genetic relationships of 4 Dendrobium species were analyzed based on SNPs obtained by GBS with the Dendrobium catenatum as the reference genome. The results of NJ, Structure and PCoA analysis were similar. D. officinale was far away from the other three kinds of Dendrobium. D. moniliforme was close to D. fanjingense and a branch of D. huoshanense. D. huoshanense could be divided into two branches in general, but there was confusion between the two branches, and the genetic distance analysis verified the genetic relationship of four species of Dendrobium. This study confirmed that the high throughput SNP markers obtained by GBS could be used to study the genetic relationship of different Dendrobium species or other different plants.