The specific DNA barcodes based on chloroplast genes for species identification of Theaceae plants

Abstract

More than 600 species in over 40 genera have been identified in family Theaceae worldwide. The accurate identification of Theaceae plants can ensure the market economic order, and it plays a vital role in achieving the sustainable utilization of germplasm resources. DNA barcoding, one of the most potential species identification technologies at present, has advanced in the rapid, accurate and repetitive discrimination of species. In this study, matK + ndhF + ycf1 was observed as the optimal combined candidate gene sequence of DNA barcodes by analyzing genetic information of four single chloroplast DNA sequences, including matK, rbcL, ndhF and ycf1, as well as six combined gene sequences. Subsequently, the experiments were performed on phylogenetic analysis based on genetic distance to study the phylogenetic relationship of Theaceae plants and evaluate the species identification accuracy of matK + ndhF + ycf1. Lastly, the species-specific DNA barcodes were designed by searching the variable sites (one type of single nucleotide polymorphism sites) for the accurate identification of Camellia amplexicaulis, Franklinia alatamaha, Gordonia brandegeei and Stewartia micrantha. The previous methods of screening and testing candidate gene sequences were optimized, and innovation was made in the above methods. The process of making visual DNA barcodes was standardized. Besides, DNA barcoding technology increased the accuracy of species identification and DNA barcoding was analyzed in accordance with the theories of population genetics (e.g., neutral theory of molecular evolution). The results of the study will lay a basis for the identification and protection of Theaceae species and germplasm resources.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12298-022-01175-7.

Introduction

There are nearly 40 genera and 600 species of Theaceae extensively grown in tropical and subtropical Asia and America (Zhang et al. 2020). Plant materials from Theaceae are employed for producing cosmetics (Camellia sinensis) (Koch et al. 2019), dietary supplements (Camellia) (Wang et al. 2016) and decorations (Camellia japonica) (Zhang et al. 2020). In addition, Theaceae plants are important biological resource reservoirs. Some species preserve quality germplasm resources which arise from interspecific breeding compatibility and ability to reproduce asexually (Wang et al. 2016). Over the past few decades, increasing counterfeit or adulterated Theaceae-related products and a wide variety of destructive activities by humans have threatened the diversity of Theaceae plant species and human health (Wang and Zhang 2018; Chen and Sun 2018). Accordingly, an efficient and stable method of identifying Theaceae plants is urgently required to standardize the market action and facilitate the protection of species diversity (Meng et al. 2019).

Most Theaceae plants are identified using the combination of conventional morphology and molecular technology (PCR) (Lu et al. 2012) or stoichiometric methods (Liu et al. 2020). The above methods have high performance in the identification of most species, while the limitation of expensive materials and time cost significantly decreases the efficiency of identification. DNA barcoding technology was initially reported by Paul Hebert in 2003 (Hebert et al. 2003), which employs short sequence segments derived from ribosomal, mitochondrial or chloroplast DNA (cpDNA) sequences for the rapid, accurate and automated identification of species (Hebert et al. 2003; Stoeckle et al. 2011). Compared with technologies of biological identification based on conventional taxonomy, DNA barcoding is capable of deeply excavating genetic information and obtaining more species-specific molecular markers (Sheth and Thaker 2017). The DNA barcodes are finally available online, and the above barcodes help researchers investigate the different populations of phylogenetic and evolutionary relationships (Kroymann et al. 2011), operation of food webs (García-Robledo et al. 2013) and measures for global biodiversity investigation and protection (Fontaine et al. 2012). Mitochondrial DNA sequences had been generally employed for the identification of animal species (Gonzalez et al. 2009). Some ribosomal DNA segments (e.g., ITS region in nuclear ribosomal DNA) are present in paralogous gene copies within the respective cell, and divergent copies may result in messy sequences in some groups (Hollingsworth 2011). Thus, researchers are inclined to screen appropriate DNA barcode regions from cpDNA sequences by evaluating high repeatability and sequence quality (Kroymann et al. 2011). matK gene sequence is characterized by high operational stability and repeatability (Newmaster et al. 2009). rbcL gene sequence was initially employed as a single gene sequence for species identification with an accuracy of 90% (Hebert et al. 2003). However, as revealed by the results of recent studies, the species identification ability of rbcL is low (Hebert et al. 2003; Hollingsworth 2011). The experimental results (Zhang et al. 2015; Amar 2020) suggested that ndhF and ycf1 gene sequences have high mutation rates and low mutation saturation, thus becoming suitable for phylogenetic analysis and construction of DNA barcodes.

Existing cross-sectional studies of Theaceae species identification have primarily investigated their traits of morphology and ecological distribution (Zhang et al. 2020). Recently, researchers gradually carried out the work of constructing barcodes within Camellia (Li et al. 2019) and Schima (Yu et al. 2017) genera. However, to the best of our knowledge, there has been rare research determining the DNA barcodes and phylogenetic relationship of this important plant family. Furthermore, several attempts to construct barcodes have often ignored the tests for identification accuracy, and results had been hardly applied to practical tests (Chaveerach et al. 2016).

This study aimed to construct DNA barcodes of Theaceae species and test the identification accuracy of barcodes. Some sequences on online databases are incomplete and have indels due to human impact. Genetic features and diversity of Theaceae species provide exact reference data to screen suitable gene sequences and confirm the long-term identification potential in the above sequences (Gong et al. 2021). Besides, the results of genetic distance and phylogenetic analysis confirmed that the screened barcodes accurately identify most Theaceae species (Chaveerach et al. 2016). Based on the above analysis, the above barcodes obtained will be more reliable, thus laying a foundation for the identification and protection of Theaceae species and germplasm resources.

In this study, a new methodology was proposed to construct specific cpDNA barcodes and increase the accuracy of species identification in barcodes. Four single gene sequences (matK, rbcL, ndhF and ycf1) and six combined sequences (matK + rbcL, matK + ndhF, matK + ycf1, rbcL + ndhF, rbcL + ycf1, ndhF + ycf1) were specially processed to obtain gene sequence databases based on previous investigations (Bhargava and Sharma 2013; Yu et al. 2016). Subsequently, the combined gene sequences were screened based on the results of major genetic features and the genetic diversity analysis. Lastly, the combined gene sequences with more accurate identification levels were investigated to test accuracy and reliability for obtaining specific DNA marker segments through the genetic distance and phylogenetic analysis.

Materials and methods

Multiple sequence alignments

All complete chloroplast single gene sequences (e.g., matK, rbcL, ndhF and ycf1 of Theaceae plants) were collected from the NCBI databases (Yu et al. 2016) (https://www.ncbi.nlm.nih.gov/) to establish species-specific gene sequence databases using MEGA11.0 software (https://www.megasoftware.net/) (Tamura et al. 2021). Synonymies in Theaceae plants were combined into a single species name when using barcodes for species identification (Zhang et al. 2021). Accordingly, The Plant List (TPL) (www.theplantlist.org) and Flora of China (FOC) (http://www.iplant.cn/foc) in English version were referenced to modify species names of Theaceae plants.

In accordance with four single gene sequences, six combined gene sequences (matK + rbcL, matK + ndhF, matK + ycf1, rbcL + ndhF, rbcL + ycf1, ndhF + ycf1) were obtained through directed assembly. Multiple sequence alignment was performed using Muscle in MEGA11.0 (Tamura et al. 2021). For large sequence databases and gene sequences with interior vacancies (e.g., matK), Muscle operated faster and more accurately (Edgar 2004).

Data analysis

Major genetic features were summarized with MEGA11.0 and TBtools v1.09852 software (Chen et al. 2020a) for reference, include the proportions of bases, the number of C (Conserved), V (Variable), Pi (Parsimony informative) and S (Singleton) sites, as well as base substitution saturation effect of all gene sequences after alignment. Genetic diversity analysis including nucleotide diversity and neutrality tests was performed with DnaSP v6 (http://www.ub.edu/dnasp/index_v5.html) (Fu 1997; Chen et al. 2020b). Genetic parameters in nucleotide diversity tests consisted of nucleotide substitution rate (θ) and haplotype diversity (Hd). Neutrality tests involved Tajima’s D tests (D) and Fu and Li’s F tests (Fu's Fs), constructed with the correlation between segregating sites (S) and nucleotide diversity (π) (Chen et al. 2020b).

Genetic distance has been found as a vital standard to screen candidate gene sequences for making barcodes (Hall 2013). Thus, the difference between the interspecific, intraspecific and overall average distances of all gene sequences was compared using MEGA11.0. Phylogenetic trees constructed using the NJ, ML and Bayes methods according to the K2P model were evaluated based on MEGA 11.0 and MrBayes v3.2.6 (http://nbisweden.github.io/MrBayes/download.html) (Hebert et al. 2003; Ronquist et al. 2012). In addition, the bootstrap value was set above 75% to ensure the reliability of branches (Tamura et al. 2021). To reduce the complexity of trees, they were finally modified in iTOL (https://itol.embl.de/) (Letunic and Bork 2021) to clearly compare genetic characteristics.

Construction of barcodes based on SNP sites

A multi-segment specific barcode was combined from 150-bp candidate gene sequence segments starting with V sites. Besides, the above appropriate segments were blasted into the NCBI databases (https://blast.ncbi.nlm.nih.gov/Blast.cgi) in accordance with the standards of the NCBI Blast output.

Results

Basic information of gene sequences

In this study, 160 matK sequences (12 genera, 101 species), 133 rbcL sequences (11 genera, 96 species), 156 ndhF sequences (11 genera, 102 species), and 125 ycf1 sequences (10 genera, 92 species) of all Theaceae species were retrieved on NCBI databases (Table 1). After sequence alignment and editing, the average length of matK, rbcL, ndhF and ycf1 included 1533 bp, 1428 bp, 2314 bp and 5817 bp, respectively, and that of the combined gene sequences, including matK + rbcL sequences (11 genera, 96 species), matK + ndhF sequences (11 genera, 88 species), matK + ycf1 sequences (10 genera, 87 species), rbcL + ndhF sequences (11 genera, 89 species), rbcL + ycf1 sequences (10 genera, 88 species) and ndhF + ycf1 sequences (10 genera, 88 species), was 2691 bp, 3846 bp, 7354 bp, 3741 bp, 7239 bp and 8110 bp, respectively. Tables S1–S10 list the supplementary information, including accession numbers, version numbers, accepted names, synonyms and definition of above species-specific gene sequences on the NCBI online database.

Table 1.

Basic information of gene sequences

Sequences	Genera	Species	Number of sequences	Average length in NCBI	Average length after alignment
matK	12	101	160	1527	1533
rbcL	11	96	133	1428	1428
ndhF	11	102	156	2247	2314
ycf1	10	92	125	5628	5817
matK + rbcL	11	96	130	2928	2691
matK + ndhF	11	88	118	3780	3846
matK + ycf1	10	87	118	7215	7354
rbcL + ndhF	11	89	118	3708	3741
rbcL + ycf1	10	88	118	7044	7239
ndhF + ycf1	10	88	115	7941	8110

Major genetic features of gene sequences

The mutation rate at C and G sites was significantly higher compared to that at A and T sites and the majority of base mutations belonged to GC → AT transitions in Theaceae plants (Fig. 1). rbcL and ycf1 had the highest and lowest GC content (44.0% and 29.8%, respectively) individually among single gene sequences, so rbcL was highly conserved. The bases in ycf1 were more prone to mutation. Some combined gene sequences also showed similar features, including matK + rbcL (GC content 38.3%) and ndhF + ycf1 (GC content 30.3%) (Fig. 1). Average AT and GC content at different coding positions of codons are listed in Table S11 for reference.

Fig. 1.

Average AT and GC content of candidate gene nucleotide sequences in Theaceae plants

ycf1 had the most V sites and the maximum proportion of V sites (1047, 18.1%) among single gene sequences, followed by ndhF (368, 16.2%), matK (219, 14.3%), rbcL (98, 6.9%) (Table 2). Besides, the proportion of Pi sites was the highest in matK (79.0%), followed by those in ndhF (75.5%), rbcL (73.5%) and ycf1(71.1%) (Table 2). The genetic stability of ycf1 with more mutations was relatively low. In contrast, matK had fewer mutations, while its genetic stability was 8% higher compared to that of ycf1. Accordingly, ycf1 and matK were key single gene sequences to affect the genetic features of combined gene sequences (e.g., ndhF + ycf1 with 1401 V sites and matK + rbcL with the highest proportion of Pi sites (76.9%)).

Table 2.

The number and proportion of Conserved (C) sites, Variable sites (V), Parsimony informative sites (Pi) and Singleton sites (S) in Theaceae plants

Sequences	Conserved sites		Variable sites		Parsimony informative sites		Singleton sites
	Number	Total proportion (%)	Number	Total proportion (%)	Number	Variable sites proportion (%)	Number	Variable sites proportion (%)
matK	1314	85.7	219	14.3	173	79.0	46	21.0
rbcL	1330	93.1	98	6.9	72	73.5	26	26.5
ndhF	1909	83.8	368	16.2	278	75.5	84	22.8
ycf1	4722	81.9	1047	18.1	744	71.1	303	28.9
matK + rbcL	2656	89.9	299	10.1	230	76.9	69	23.1
matK + ndhF	3249	85.5	549	14.5	418	76.1	125	22.8
matK + ycf1	6057	83.0	1242	17.0	874	70.4	368	29.6
rbcL + ndhF	3247	87.8	452	12.2	329	72.8	117	25.9
rbcL + ycf1	6063	84.2	1137	15.8	802	70.5	335	29.5
ndhF + ycf1	6652	82.6	1401	17.4	958	68.4	433	30.9

As depicted in Fig. 2, ycf1, matK + ycf1, rbcL + ycf1, ndhF + ycf1 had high-frequency base substitution mutations, and the main form of mutations in Theaceae species was base transition. All R values were higher than 1.0 (Fig. 3), which revealed that the main form of gene mutation in Theaceae plants was base transition. matK and ndhF had a lower saturation effect and less evolutionary murmur, and the above features were conducive to constructing phylogenetic trees and collecting correct genetic information (Fig. 3). Moreover, matK and ndhF mitigated the above effect on ycf1 in combined gene sequences. Table S12 lists the number of base substitutions and R values at different coding positions of codons.

Fig. 2.

Significant correlation between GC content and overall mean genetic distance. Stacked Bar of the number of base substitution including base transversion and base transition in candidate nucleotide sequences of Theaceae plants

Fig. 3.

Analysis of ratios of transitionsal pairs to transversional pairs (R value). Y axis takes y = 1.0 as the baseline, which suggests that all of candidate nucleotide sequences have the effect of base substitution saturation

Genetic diversity analysis

ycf1 had high levels of nucleotide diversity with the maximum θ and Hd values (0.03826, 0.997) among single gene sequences (Table 3). Thus, the high mutation rate of ycf1 contributed to abundant genetic diversity and genetic resources within Theaceae plants. Besides, θ and Hd values of matK + ycf1 (0.03583, 0.997), rbcL + ycf1 (0.03314, 0.997) and ndhF + ycf1 (0.03644, 0.998) were significantly higher compared to that of other combined gene sequences (Table 3). As revealed by the above results, ycf1 gene sequences could be suitable for the species identification of Theaceae plants.

Table 3.

Genetic diversity of gene sequences

Sequences	Nucleotide diversity		Neutrality tests
	θ	Hd	S	π	Fu’s Fs	P value	D	P value
matK	0.02837	0.976	212	0.02063	− 0.98564	P > 0.10	− 0.88178	P > 0.10
rbcL	0.01371	0.931	98	0.00761	− 1.56362	P > 0.10	− 1.43028	P > 0.10
ndhF	0.03094	0.982	317	0.01779	− 1.71519	P > 0.10	− 1.38427	P > 0.10
ycf1	0.03826	0.997	969	0.02571	− 1.85941	P > 0.10	− 1.09645	P > 0.10
matK + rbcL	0.02054	0.984	291	0.01324	− 1.32595	P > 0.10	− 1.17354	P > 0.10
matK + ndhF	0.02918	0.993	492	0.0203	− 1.19153	P > 0.10	− 1.01837	P > 0.10
matK + ycf1	0.03583	0.997	1165	0.02467	− 1.84762	P > 0.10	− 1.04623	P > 0.10
rbcL + ndhF	0.02424	0.991	403	0.01595	− 1.53279	P > 0.10	− 1.14323	P > 0.10
rbcL + ycf1	0.03314	0.997	1069	0.02232	− 1.85036	P > 0.10	− 1.09645	P > 0.10
ndhF + ycf1	0.03644	0.998	1278	0.02466	− 1.97306	0.10 > P > 0.05	− 1.08892	P > 0.10

In accordance with P values of Fu’s Fs and D (Table 3), the mutational pattern of chloroplast genes in Theaceae plants was observed as having random mutations. Thus, Theaceae species were observed as less susceptible to environmental, human, or other external factors. Among single gene sequences, rbcL, ndhF and ycf1, all facilitated the population expansion of Theaceae plants with higher absolute values of Fu’s Fs (− 1.56362, − 1.71519, − 1.85941) and D (− 1.43028, − 1.38427, − 1.09645).

Genetic distance analysis

There was no overlap between the maximum intraspecific genetic distance and the minimum interspecific genetic distance among single and combined gene sequences (Fig. 4), which revealed that chloroplast genes of Theaceae plants had significant genetic differences and could help obtain barcodes. Moreover, as indicated by the significant differences between intraspecific and interspecific distances of ycf1, matK + ycf1, rbcL + ycf1 and ndhF + ycf1, the above gene sequences were more suitable for species recognition and identification. However, rbcL had difficulty in making barcodes due to the smaller corresponding difference of rbcL + ndhF gene sequence.

Fig. 4.

Bar chart of interspecific distances and intraspecific distances of Theaceae plants based on K2P genetic distance of gene sequences

Phylogenetic analysis

In accordance with the results of major genetic features, genetic diversity and genetic distance analysis, matK, ndhF and ycf1 gene sequences were combined to construct NJ, ML and Bayes phylogenetic trees. NJ tree contained 9 blocks, 10 genera and 90 different species (Fig. 5). Camellia was observed as the first largest genera in Theaceae plants, while Apterosperma, Eurya, Ternstroemia, Gordonia, and Franklinia were small genera. Eurya, Ternstroemia and Apterosperma underwent a long evolutionary process, as revealed by the estimated genetic distances of species at terminal branches. Camellia, Schima, Stewartia and Pyrenaria followed the same evolutionary transition pattern with similar genetic distances. Since Polyspora and Apterosperma were small genera, they were probably new genera evolved from Camellia and Pyrenaria. Moreover, Eurya and Ternstroemia had a related ancestry with similar genetic distances.

Fig. 5.

The NJ tree of Theaceae plants coming from matK + rbcL + ycf1 gene sequences based on the K2P model. The bootstraps of tree branches were all greater than 75%. Triangle marks (color in red) on branches represent bootstrap values; the bigger the triangles, the higher the bootstrap values would be. At the same time, displaying branch lengths on the main branches, namely the length of evolutionary distance, which retained four significant digits

matK + ndhF + ycf1 gene sequences had excellent identification ability at genera level. There were obvious genetic nodes on branches of all genera in Theaceae plants for segmentation (Fig. 5). For species-level identification, species in monotypic genera (Apterosperma, Ternstroemia, Franklinia), small genera (Eurya, Gordonia) and species in large genera (species ≤ 10, Polyspora) had obvious internal branches. However, species in large genera (species > 10, Stewartia, Pyrenaria, Camellia, Schima) were reported with limitations in terms of species-level identification ability, especially for Camellia; their genetic distances at terminal branches were below 0.0001.

The same results were acquired in ML and Bayes phylogenetic trees (Figs. S1 and S2). Impacted by the algorithm, some branches with low bootstrap values in Bayes tree would be difficult to exclude. Supplementary information regarding matK + ndhF + ycf1 gene sequence database is listed in Table S13.

Specific barcodes based on matK + ndhF + ycf1 gene sequences

It was observed that when “Query Cover” and “Per.Ident” were both 100% for the candidate gene sequence segments, the accuracy of species-level identification increased markedly by repeated experiments (Table 4). Subsequently, species-specific barcodes based on the above segments effectively identified Camellia amplexicaulis (Ca), Franklinia alatamaha (Fa), Gordonia brandegeei (Gb) and Stewartia micrantha (Sm) from other species of Theaceae plants (Fig. 6) with the highest “Max score” and “Total score” values. Furthermore, alternative segments with the same identification ability were provided to increase efficiency (Fig. S3 and Table S14).

Table 4.

NCBI Blast output of candidate gene sequence segments

Barcodes	Query cover (%)	Per.Ident (%)	Max score	Total score	E value	Accession on NCBI	Identified species
Ca-matK-1	100	100	278	278	2E−70	NC_051559.1; MT317095.1	Camellia amplexicaulis
Ca-ndhF-1	100	100	278	278	2E−70	NC_051559.1; MT317095.1	Camellia amplexicaulis
Ca-ycf1-1	100	100	278	556	2E−70	NC_051559.1; MT317095.2	Camellia amplexicaulis
Fa-matK-1	100	100	278	278	2E−70	KY406774.1; AF380082.1	Franklinia alatamaha
Fa-ndhF-1	100	100	278	278	2E−70	KY406774.1; HM100319.1; HM164089.1	Franklinia alatamaha
Fa-ycf1-1	100	100	278	278	2E−70	KY406774.1	Franklinia alatamaha
Gb-matK-1	100	100	278	278	2E−70	KY406761.1; AF380084.1	Gordonia brandegeei
Gb-ndhF-1	100	100	278	278	2E−70	KY406761.1	Gordonia brandegeei
Gb-ycf1-1	100	100	278	278	2E−70	KY406761.1	Gordonia brandegeei
Sm-matK-1	100	100	278	278	2E−70	NC_041471.1; MH782186.1	Stewartia micrantha
Sm-ndhF-1	100	100	278	278	2E−70	NC_041471.1; MH782186.1	Stewartia micrantha
Sm-ycf1-1	100	100	278	556	2E−70	NC_041471.1; MH782186.1	Stewartia micrantha

Fig. 6.

Combinatorial DNA barcodes including one-dimensional and two-dimensional DNA barcodes of Theaceae plants based on matK, ndhF and ycf1. In one-dimensional DNA barcodes, base A, T, C, G in green, red, blue and black, respectively

Discussion

Theaceae-containing products which further include precious herbal medicines (Wang et al. 2016) and healthy food (Koch et al. 2019) spread across the world. Moreover, germplasm resources of Theaceae plants with low genetic diversity should be identified and protected (Niu et al. 2019). Existing studies have confirmed that barcodes distinguish plants at the species and population levels using chloroplast gene segments (Cui et al. 2020; Besse et al. 2021). Methods of constructing barcodes cannot be limited to a single standard in accordance with genetic algorithms and ignore tests for identification accuracy of barcodes. Accordingly, as revealed by the results of barcode stability and accuracy analysis in this study, chloroplast barcodes based on matK + ndhF + ycf1 combined gene sequences which efficiently identified Camellia amplexicaulis, Franklinia alatamaha, Gordonia brandegeei and Stewartia micrantha from Theaceae species. Constructing species-specific barcode databases will be a more accepted Theaceae species identification technology by rapidly sequencing technologies (Li et al. 2015) exhibiting high stability and accuracy.

Design of specific barcodes

In accordance with previous barcodes regarding Rehmannia (Duan et al. 2019) and Orchidaceae species (Li et al. 2021), whether combined gene sequences increase the accuracy of identifying Theaceae plants should be further examined. Moreover, novel synonyms among species of Theaceae plants would lead to difficulty in making accurate species-level identifications (Skvarla et al. 2020). In this study, sequences with confused synonyms were deleted, or some synonyms were replaced with more common species names.

GC content in complete chloroplast genomes of Theaceae plants is 36–38% (Kim et al. 2017; Xu et al. 2021; Yang et al. 2021), which is similar to 37.2% GC content in Orchidaceae plants (Trávníček et al. 2019) and 37.7% GC content in Asteraceae plants (Lv et al. 2020; Xie et al. 2021). In fact, the population expansion of Orchidaceae and Asteraceae species have been affected by GC content of own sequences, while there is no correlation between Theaceae species and GC content of own sequences (Wang et al. 2016). The reason is probably that the severe growth environment with relatively high temperatures of some Theaceae species instead of the population expansion enriched genetic diversity in chloroplast genomes (Wang et al. 2022). Thus, as indicated by the results of 34.6% GC content of 4 single gene sequences (matK, rbcL, ndhF and ycf1) (Fig. 1), some specific sequences with lower GC content instead of complete chloroplast genomes artificially improved the success of species identification. Studies on C, V, Pi and S sites revealed single-nucleotide polymorphisms (SNP) in Theaceae plants to identify de novo mutations of sequences (Neininger et al. 2019). Recent studies have demonstrated V sites as the most common SNP sites that played a role of segregating sites to construct barcode segments (Gong et al. 2021; Li et al. 2021), and barcodes based on Pi and S sites were found to appear sometimes due to their rare genetic traits (Delabye et al. 2019; Smidt et al. 2020). In our studies, S and Pi sites (the sum of Pi and S sites is V sites in Table 2) were considered the standards of screening gene sequences, and V sites were adopted to achieve barcode segments. ycf1 was confirmed as a new focus on making barcodes (Fig. 2) (Li et al. 2021), and matK and ndhF gene sequences could increase the identification accuracy by the combination with ycf1.

Genetic distance refers to a critical factor of judging whether a combined gene sequence reduces identification accuracy because the four single gene sequence fragments share high sequence homology (Li et al. 2021). Besides, since Theaceae databases on NCBI have more limited datasets, this method was employed instead of “Best Close Match” (Collins and Cruickshank 2013; Li et al. 2021). In fact, tests for genetic distances have confirmed that matK (Chaveerach et al. 2016), ndhF (Korotkova et al. 2014) and ycf1 (Gogoi et al. 2020) cannot reduce identification accuracy of combined gene sequences.

Camellia amplexicaulis, the main raw materials of new neolignan component, has significant potential for officinal development (e.g., osteoblast differentiation treatment) (Tung et al. 2009). Franklinia alatamaha used as an ornamental plant is primarily distributed in the United States (Luna and Ochoterena 2004). The feasibility of using genetic traits of Gordonia brandegeei and Stewartia micrantha and establishing them in creating germplasm resources of Theaceae species is high since Gordonia and Stewartia genera species have great abilities to asexual propagation and interspecific crossing (Wang et al. 2016). Constructing barcodes of above species could ensure the utilization and protection of biological resources of Theaceae plants.

Identification accuracy of barcodes

Results of major genetic features (Table 2) and genetic diversity (Table 3) have showed that ndhF and ycf1 have a positive significance for population expansion and facilitate the generation of abundant SNP sites, while matK + ndhF + ycf1 cannot effectively identify some species in large genera (such as Camellia) (Fig. 5), which is similar to other researches (Li et al. 2021). According to studies regarding phylogenetic analysis in Piper species (Chaveerach et al. 2016) and Orchidaceae species (Li et al. 2021), some species with low distance values in large genera were difficult to distinguish. The results were achieved since nucleotide diversity among species in large genera was low (Feng et al. 2017). Accordingly, multiple SNP sites were screened in matK, ndhF and ycf1 gene sequences to expand the scope of species identification. Besides, Eurya, Ternstroemia and Apterosperma species were observed as rare species (Wang et al. 2016; Zhang et al. 2020), and Camellia, Schima, Stewartia and Pyrenaria species had a similar evolutionary degree of internal species individually (Luna and Ochoterena 2004).

Several limitations in this study should be solved. One limitation was that a lower number of gene sequences on NCBI and ambiguous codes, indels and missing symbols in sequences affected genetic features and diversity of Theaceae species. As a result, some significant SNP sites were omitted, thus causing a narrower spectrum of identifiable species. Furthermore, in Blast tests for screening candidate gene sequence segments, the recognition rate of numerous segments with lower max or total scores could not approach 100%, so we could derive candidate gene sequence segments from only a few Theaceae species.

Notwithstanding the above unavoidable limitations, this is the first research that construct barcodes based on genetic information of Theaceae species. Since the emergence of the next-generation sequencing technology facilitates the widespread use of online sequence databases and pursuits of highly efficient species identification technology (Besse et al. 2021), we consider that DNA barcoding still applies to Theaceae species identification, and matK + ndhF + ycf1 will serve as a vital candidate combined gene sequence.

Lastly, DNA barcoding identifies species in accordance with the principle of differential alignment between sequences, which pertains to the branch of genetics in essence. Thus, this technology should be continuously optimized to ensure its innovation and universality, and it should be gradually combined with other theories and technologies of genetics to form a complementary data analysis system about genetics (e.g., the analysis of codon usage bias (Ganie et al. 2019), gene flow and genetic differentiation (Presgraves 2018)). Besides, we suggest that researchers upload complete chloroplast sequences to databases, and NCBI presents up-to-date sequence information and corresponding supplementary references for the protection of some important species.

Conclusions

In this dissertation, the aim was to construct species-specific cpDNA barcodes of Theaceae plants and assess identification accuracy of barcodes. matK + ndhF + ycf1 with high accuracy and stability identifies Camellia amplexicaulis, Franklinia alatamaha, Gordonia brandegeei and Stewartia micrantha from other Theaceae species. To the best of our knowledge, this study has been considered the first comprehensive investigation of barcodes of Theaceae species. Methods of constructing barcodes involved in this study are more theoretically supportive through the genetic information analysis. Moreover, matK + ndhF + ycf1 was tested for its accuracy as compared with combined gene sequences through the phylogenetic analysis. Lastly, barcoding provided in this study was used as references for identifying Theaceae species and optimizing existing barcoding technologies to protect and make rational use of biological resources.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

cpDNA

Chloroplast DNA

C sites

Conserved sites

V sites

Variable sites

Pi sites

Parsimony informative sites

S sites

Singleton sites

SNP sites

Single nucleotide polymorphism sites

ii

Identical Pairs

si

Transitionsal pairs

sv

Transversional pairs

R value

Transitionsal pairs/Transversional pairs 

θ

Nucleotide substitution rate

Hd

Haplotype diversity

S

Segregating sites

π

Nucleotide diversity

Fu’s Fs

Representative value of neutrality tests

Tajima’s D

Representative value of neutrality tests

P value

Hypothetical probability

ML tree

Maximum likelihood tree

NJ tree

Neighbor joining tree

Ca

Camellia amplexicaulis

Fa

Franklinia alatamaha

Gb

Gordonia brandegeei

Sm

Stewartia micrantha

Author’s contributions

XHG and YLL designed the research. SJ, FLC, PQ, HX, GP and YLL collected and analyzed data. SJ, FLC, YLL and XHG wrote the main manuscript text. SJ prepared all figures and tables. All authors read and approved the manuscript.

Funding

Authors would like to acknowledge the funding received from Key Research & Development Project of Hunan Provincial Department of Science and Technology (2019NK2081), National Natural Science Foundation of China (31872866) and National Key Research and Development Program of China (2017YFF0210301) to carry out this assignment.

Data availability

The data supporting the finding of this study is provided in the manuscript and its supplementary material.

Declarations

Conflict of interest

The authors declare that they have no competing interests.