Phylogeny and adaptive evolution of subgenus Rhizirideum (Amaryllidaceae, Allium) based on plastid genomes

Abstract

The subgenus Rhizirideum in the genus Allium consists of 38 species worldwide and forms five sections (A. sect. Rhizomatosa, A. sect. Tenuissima, A. sect. Rhizirideum, A. sect. Eduardia, and A. sect. Caespitosoprason), A. sect. Caespitosoprason being merged into A. sect. Rhizomatosa recently. Previous studies on this subgenus mainly focused on separate sections. To investigate the inter-section and inter-subgenera phylogenetic relationships and adaptive evolution of A. subg. Rhizirideum, we selected thirteen representative species, which cover five sections of this subgenus and can represent four typical phenotypes of it. We conducted the comparative plastome analysis with our thirteen plastomes. And phylogenetic inferences with CDSs and complete sequences of plastomes of our thirteen species and another fifty-four related species were also performed. As a result, the A. subg. Rhizirideum plastomes were relatively conservative in structure, IR/SC borders, codon usage, and repeat sequence. In phylogenetic results, the inter-subgenera relationships among A. subg. Rhizirideum and other genus Allium subgenera were generally similar to the previous reports. In contrast, the inter-section relationships within our subgenus A. subg. Rhizirideum were newly resolved in this study. A. sect. Rhizomatosa and A. sect. Tenuissima were sister branches, which were then clustered with A. sect. Rhizirideum and A. sect. Eduardia successively. However, Allium Polyrhizum Turcz. ex Regel, type species of A. sect. Caespitosoprason, was resolved as the basal taxon of A. subg. Rhizirideum. Allium siphonanthum J. M. Xu was also found in clade A. subg. Cyathophora instead of clade A. subg. Rhizirideum. The selective pressure analysis was also conducted, and most protein-coding genes were under purifying selection. At the same time, just one gene, ycf2, was found under positive selection, and another three genes (rbcL, ycf1a, ycf1b) presented relaxed selection, which were all involved in the photosynthesis. The low temperature, dry climate, and high altitude of the extreme habitats where A. subg. Rhizirideum species grow might impose intense natural selection forces on their plastome genes for photosynthesis. In summary, our research provides new insights into the phylogeny and adaptive evolution of A. subg. Rhizirideum. Moreover, we suggest that the positions of the A. subg. Rhizirideum species A. polyrhizum and A. siphonanthum should be reconsidered.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-022-03993-z.

Introduction

Allium (Allioideae, Amaryllidaceae), one of the largest genera of monocots, has more than 900 verified species on the Earth [1]. Many species in this genus have been used for edible (e.g., A. sativum, A. tuberosum, A. porrum), medicinal (e.g., A. sativum, A. victorialis, A. cepa), and ornamental (e.g., A. giganteum, A. wallichii, A. moly). Genus Allium was originally established by Linnaeus in Species plantarum [2], which initially contained only 30 Allium species sorted into three alliances. Subsequently, many scientists published a large quantity of new Allium taxa, and analyses on the taxonomy and phylogeny of Allium also emerged because of the complicated relationship within this genus. Regel’s monograph included 263 species and distributed them into six sections [3, 4]. Afterwards, Traub [5] sorted 600 Allium species into three subgenera, including 36 sections and subsections. Wendelbo [6] first proposed the subgenus Rhizirideum. After that, Kamelin [7] revised the phylogeny of Allium and classified it into six subgenera (44 sections & subsections). In Kamelin’s taxonomy, the subgenus Rhizirideum contained 150 species, such as A. cepa, A. senescens, and A. ramosum and further sorted into 12 sections and subsections as Sect. cepa, Sect. Butomissa, Sect. Rhizirideum. Later, Friesen et al. [8] reconstructed the phylogeny of Allium based on ITS data and divided it into three main evolutionary lineages. Friesen et al. [8] put the new subgenus Rhizirideum forward (A. subg. Rhizirideum in the following) and distributed approximately 780 Allium species into 15 subgenera (72 sections). At the same time, the previous subgenus Rhizirideum was disproved and found to be nonmonophyletic.

Sixteen species (e.g., Allium senescens L.) in the previous subgenus Rhizirideum were still assorted into the new one, while the others were assorted into other subgenera such as Anguinum (e.g., Allium Victorialis L.), Cepa (e.g., Allium Cepa L.), and Butomissa (e.g., Allium Ramosum L.). At that time, subgenus Rhizirideum included five sections: Rhizirideum, Rhizomatosa, Tenuissima, Eduardii, and Caespitosoprason. Recently, [9] provided adequate evidence for the monophyly of subgenus Rhizirideum based on chloroplast DNA fragments data. Friesen et al. [10] merged section Caespitosoprason into section Rhizomatosa under subgenus Rhizirideum.

Currently, subgenus Rhizirideum consists of four sections (Rhizirideum, Rhizomatosa, Tenuissima, Eduardii) and thirty-eight species in total [8, 11] (Additional file 2: Table S1). It was located in the third lineage of the Allium phylogeny. Species in this subgenus were characterised by obvious rhizome, leaves subcylindrical to flat, perianth white to purple, ovary with two ovules per locule, and inner filaments broadened at the base (Fig. 1, Additional file 1: Fig. S1).

Fig. 1.

Inflorescences of eight species in A. subg. Rhizirideum. (A), A. bidentatum; (B), A. mongolicum; (C), A. anisopodium; (D), A. tenuissimum; (E), A. senescens; (F), A. eduardii; (G), A. przewalskianum; (H) & (I), A. polyrhizum

Studies of species in subgenus Rhizirideum have been conducted frequently in the past century [8, 10, 12–23]. It was found that the chromosome base number of this subgenus was eight, and the ploidy was mainly 2x or 4x. And Species in this subgenus spread over the Eurasian steppe. Sinitsyna et al. [24] divided section Rhizirideum into two geographical groups, the Asiatic and European groups. The diversification and speciation of this section coincide with the history of the modern Eurasian steppe. Meanwhile, the latest study of section Rhizomatosa on biogeography indicated that species in this section were distributed in the Central Asian steppe, and the distribution was in accordance with the history of the landscape and climate [10]. However, some phylogenetic studies were focused on the previous subgenus Rhizirideum, and the others focused on section Rhizirideum and section Rhizomatosa. Phylogenetic analysis on section Tenuissima and section Eduardii were lacking, so more fieldwork and further investigation should be undertaken.

In recent years, the complete chloroplast genome has been popular for its conservative structure, low recombination rate, and enormous genetic information. It has been widely used in the phylogenetic reconstruction and adaptive evolution [25–27]. Several Allium taxa have also been studied on their plastomes, for instance, section Cepa, Daghestanica, and subgenus Cyathophora [28–30]. Xie et al. [31] reconstructed the phylogenetic relationship of the genus Allium with thirty-nine complete chloroplast genomes and revealed the evolutionary features of Allium. However, similar studies on the subgenus Rhizirideum have not yet been reported. Previous phylogenetic studies of subgenus Rhizirideum were primarily based on ITS or plastid DNA fragments, which provided limited information for infrageneric relationships. Furthermore, the analysis of adaptive evolution was also inadequate. Thus it is necessary to investigate further the composition, structure, and evolution of subgenus Rhizirideum plastomes. We collected thirteen species in Subgenus Rhizirideum and combined fifty-four related species to conduct comparative chloroplast genome analyses. Our aims are as follows: (1) to compare the structures and genetic compositions of plastomes of thirteen Rhizirideum species and (2) to reconstruct the phylogeny of subgenus Rhizirideum as well as some related Allium subgenera; (3) to analyse the adaptive evolution of subgenus Rhizirideum species.

Results

Plastome structure of subgenus Rhizirideum species

The subgenus Rhizirideum plastomes shared a quadripartite circular structure with two inverted repeats (IRa & IRb), one large single copy (LSC), and one small single copy (SSC) (Fig. 2, Table 1). The sizes of thirteen Rhizirideum plastomes ranged from 153,723 bp to 153,257 bp, and their overall GC content ranged from 36.8 to 36.9%. Each plastome contained 141 genes, among which 87 or 89 protein-coding sequences, 38 tRNA-coding genes, and 8rRNA-coding genes. Moreover, 26 genes were interpreted by introns (Table 2). The genes clpP, rps12, and ycf3 had two introns inserted into their sequences. Moreover, the trnK-UUU gene had the longest intron, where the matK gene was located. The rps12 is a trans-spliced gene with the 5′-end in the LSC region and the duplicated 3′-ends in the IR regions.

Fig. 2.

Chloroplast map of Allium senescens (the outermost circle and rings p-q) and GView comparison of thireteen A. subg. Rhizirideum plastomes (rings c-o). Genes are shown in different colors based on their functional groups. Genes on the inside of the outermost circle are transcribed clockwise, and those on the outside counter-clockwise. LSC, large single-copy region; SSC, small single-copy region; IR, inverted repeat. Ring a: GC content deviation from the average GC content of A. senescens, waves inside lower than the average, waves outside higher than the average. Ring b: GC skew of A. senescens, GC skew = (G-C)/(G + C), waves inside > 0, waves outside < 0. Ring c: reference of multiple alignments (A. senescens plastome). Rings d-n denote the result of multiple alignments of plastomes outwards in turn: A. polyrhizum, A. bidentatum, A. dentigerum, A. caespitosum, A. mongolicum, A. anisopodium, A. tenuissimum, A. spirale, A. nutans, A. eduardii, A. przewalskianum, A. siphonanthum. Ring p: GC content of A. senescens. Ring q: AT content of A. senescens

Table 1.

Summary of the subgenus Rhizirideum plastomes

Taxon	Total genome length (bp)	GC (%)	IR length (bp)	LSC length (bp)	SSC length (bp)	Gene	CDS	tRNA	rRNA
A. polyrhizum	153,614	36.8	26,450	82,624	18,090	141	89 (9)	38 (8)	8 (4)
A. bidentatum	153,443	36.8	26,459	82,504	18,021	141	89 (9)	38 (8)	8 (4)
A. dentigerum	153,538	36.8	26,625	82,522	17,766	141	89 (9)	38 (8)	8 (4)
A. caespitosum	153,667	36.8	26,490	82,643	18,044	141	87 (9)	38 (8)	8 (4)
A. mongolicum	153,667	36.8	26,490	82,645	18,042	141	87 (9)	38 (8)	8 (4)
A. anisopodium	153,407	36.8	26,491	82,426	17,999	141	87 (9)	38 (8)	8 (4)
A. tenuissimum	153,459	36.8	26,491	82,484	17,993	141	87 (9)	38 (8)	8 (4)
A. senescens	153,516	36.8	26,491	82,548	17,986	141	89 (9)	38 (8)	8 (4)
A. spirale	153,549	36.8	26,493	82,576	17,987	141	89 (9)	38 (8)	8 (4)
A. nutans	153,456	36.9	26,487	82,531	17,951	141	87 (9)	38 (8)	8 (4)
A. eduardii	153,497	36.9	26,732	82,296	17,737	141	89 (9)	38 (8)	8 (4)
A. przewalskianum	153,257	36.9	26,437	82,410	17,973	141	89 (9)	38 (8)	8 (4)
A. siphonanthum	153,723	37.9	26,495	82,752	17,981	141	89 (9)	38 (8)	8 (4)

Numbers in brackets note the number of double-copy genes

Table 2.

Summary of genes interrupted by introns in Rhizirideum plastomes

No.	Gene	Region	Exon I (bp)	Intron I (bp)	Exon II (bp)	Intron II (bp)	Exon III (bp)
1	atpF	LSC	144+	789	411+
2	clpP	LSC	69+	1094	294+	879	252+
3	ndhA	SSC	558+	1128	540+
4	ndhB	IRb	777+	701	756+
5	ndhB	IRa	777−	701	756−
6	petB	LSC	6−	920	642−
7	petD	LSC	8−	746	514−
8	rpl16	LSC	9+	1042	399+
9	rpl2	IRb	387+	659	432+
10	rpl2	IRa	387−	659	432−
11	rpoC1	LSC	432+	759	1623+
12	rps12a	LSC, IRa	114+	69,984	232−	542	26−
13	rps12b	LSC,IRb	114+	28,994	232+	542	26+
14	rps16	LSC	40+	846	197+
15	trnA-UGC	IRb	38−	815	35+
16	trnA-UGC	IRa	38+	815	35+
17	trnG-TCC	LSC	23−	692	49−
18	trnI-GAU	IRb	37−	934	35−
19	trnI-GAU	IRa	37+	934	35+
20	trnL-UAA	LSC	35−	305	50−
21	trnV-UAC	LSC	37+	598	37+
22	ycf3	LSC	129+	722	228+	738
23	ycf68	IRb	42−	31	411−		153+
24	ycf68	IRa	42+	31	411+
25	trnK-UUU	LSC	37+	2564	35+

The data in this table is from Allium senescens plastome

Multiple alignments of Rhizirideum plastomes showed similar structural features (Fig. 2). Matching distribution patterns of GC islands were displayed among thirteen Rhizirideum plastomes (Fig. 2, rings a-b). IR regions showed a GC skew < 0 (G < C) while most areas of LSC and SSC regions showed a GC skew > 0 (G > C) (Fig. 2, ring b). LSC and SSC regions, especially LSC, showed lower sequence identities than IR regions (Fig. 2, rings c-o). There was only one common gap in each IR region, ycf2-trnI CAU and rrn16-trnI GAU, respectively (Fig. 2). However, in the single copy (SC) regions, Rhizirideum species shared several divergent sequence sites (Fig. 2): (1) psbA-rps19, (2) matK-trnK UUU (10 species except for Allium bidentatum Fisch. ex Prokh. et Ikonnikov-Galitzky, Allium dentigerum Prokh. and Allium spirale Willdenow), (3) psbD-trnT GGU (10 species except for Allium mongolicum Regel, Allium anisopodium Ledeb. and A. spirale), (4) trnG UCC – trnfM CAU (12 species except for A.spirale), (5) ndhC-ndhK, (6) petA-psbJ (11 species except for A. spirale and Allium nutans L.). Besides, the diagram drawn by mVISTA (Fig. 3) showed sequence identities of different regions in Rhizirideum plastomes straightforward regarding A. senescens as a reference. As it indicated, exon regions had higher identity values than UTR and CNS regions. IR regions also had higher sequence identities than SC regions.

Fig. 3.

mVISTA comparison of thirteen A. subg. Rhizirideum plastomes (A. senescens as reference).

We selected 111 genes (Fig. 4 A) and 103 intergenetic regions (Fig. 4 B) to compute their nucleotide diversity (Pi) values by using DnaSP software. As the results indicated, the average Pi value of the genes (0.0043) was smaller than that of the intergenetic regions (0.0118). In terms of Pi values, the top three genes were trnW-CCA (0.0266), trnS-UGA (0.0174) and rps16 (0.0160), while the top three intergenetic regions were rpl32-trnL-UAG (0.0353), ndhF-rpl32 (0.0352), and psbC-trnS-UGA (0.0346).

Fig. 4.

Nucleotide diversity (Pi) values of 111 genes and 103 intergenic regions. (A) Nucleotide diversity values of 111 genes. (B) Nucleotide diversity values of 103 intergenic regions

IR/SC borders

Lengths of the IR and SC regions of thirteen A. subg. Rhizirideum plastomes were compared (Fig. 5). In the results, the longest three IRs belonged to A.eduardii (26,732 bp), A.dentigerum (26,625 bp) and A. siphonanthum (26,495 bp), while the shortest three IRs belonged to A. przewalskianum (26,437 bp), A. polyrhizum (26,450 bp), and A. bidentatum (26,459 bp). For SSC regions, A. polyrhizum (18,090 bp), Allium caespitosum Siev. ex Bong. et Mey. (18,044 bp) and A.mongolicum (18,042 bp) had the top three SSCs, while A. nutans (17,951 bp), A. dentigerum (17,766 bp) and Allium eduardii Stearn (17,737 bp) got the last three ones. The longest three LSCs belonged to A.siphonanthum (82,752 bp), A.mongolicum (82,645 bp) and A. caespitosum (82,643 bp), and the shortest three belonged to A. eduardii (82,296 bp), A. anisopodium (82,426 bp), and A. przewalskianum (82,410 bp).

Fig. 5.

IR/SC boundaries of thirteen A. subg. Rhizirideum plastomes

The positions of IR/SC borders were examined in thirteen Rhizirideum plastomes, too (Fig. 5). Gene contents on both sides of the IR/SC borders of thirteen Rhizirideum plastomes were conserved. The LSC/IRb border was rps19/rpl22, and the IRa/LSC border was rps19/psbA. Mostly, rpl22 was interrupted by LSC/IRb border, and rps19 was no less than 63 bp away from LSC/IRb border. There were exceptions anyway that the rpl22 gene of A. bidentatum plastome was located entirely in its LSC region (18 bp away from its LSC/IRb border), and the rps19 gene of A. polyrhizum was just 47 bp away from its IRb/LSC border. In A. bidentatum plastome, the position of rps19/psbA, 30 bp/146 bp away from its IRa/LSC border, was also distinguished from others, which were 63 ~ 82 bp and 67 ~ 85 bp, respectively.

For SSC boundaries, two SSC/IR borders crossed two ycf1 genes in most of the Rhizirideum plastomes. Regarding the IRb/SSC border, a large part of the ycf1b sequence mainly was located in the IRb region, while gene ndhF was completely located in the SSC region. Nevertheless, there were still several exceptions that the whole ycf1b gene of the A. eduardii plastome was in its IRb region (away from the IRb/SSC border by 208 bp). In plastomes of A. anisopodium and A. tenuissimum, IRb/SSC borders overlapped ndhF genes by 7 bp and 8 bp, respectively. Gene ycf1a of most Rhizirideum plastomes was 5295 bp in length, except for A. siphonanthum (5274 bp), A. anisopodium (5313 bp) and A. tenuissimum (5313 bp). Gene ycf1a was divided into two fragments by border SSC/IRa, and its IRa side ranged from 679 bp (A. przewalskianum) to 1309 bp (A. eduardii).

Codon usage bias analysis

Seventy-seven protein-coding sequences (CDS) were extracted from thirteen Rhizirideum plastomes and were concatenated end-to-end to form a tandem CDS dataset. The codon usage bias of the tandem CDS dataset was analyzed by using program codonW (Fig. 6, Fig. 7, Additional file 6: Table S5 & S6). The total numbers of codons in the CDS tandem sequences ranged from 22,838 (Allium caespitosum) to 22,986 (Allium eduardii). The average values of relative synonymous codon usage (RSCU) of each sort of codon in thirteen tandem CDS sequences ranged from 2.08 (UUA) to 0.31 (CUG, AGC). In terms of codon proportions, six synonymous codons coding leucine (Leu) accumulated to a largest proportion of 10.35%, and two codons coding cysteine (Cys) accumulated to a smallest proportion of 1.13% except for three stop codons (0.40%). Methionine (Met, AUG) and tryptophan (Trp, UGG) showed no codon bias and were encoded by only one codon. Thirty codons with RSCU > 1 encoded nearly all kinds of amino acids except for Trp and Met. Among the thirty codons, only UUG for Leu was ended by G/C, and the last twenty-nine were ended by A/U.

Fig. 6.

Codon usage in percentages (the left column) and RSCU values (the right column) of twenty-one amino acids. Each codon for an amino acid is shown with different colours

Fig. 7.

Comparative analysis of codon usage bias in species of five sections in A. subg. Rhizirideum. CAI, codon adaptation index; CBI, codon bias index; FOP, frequency of optimal codons index; ENC, effective number of codons; GC3: GC percentage of 3rd position in synonymous codons

Repeat sequence analysis

We detected 879 simple sequence repeats (SSRs) in thirteen Rhizirideum plastomes (Fig. 8 A). A. mongolicum and A. caespitosum contained the most SSRs (88) whereas A.siphonanthum contained the least (67). SSRs with 1 bp ~ 5 bp could mostly be witnessed across thirteen plastomes, but those with 6 bp (i.e., the hexanucleotides) were rare, only existing in four of the plastomes (A.senescens 2, A.spirale 2, A.eduardii 1, A.siphonanthum 1). Among all kinds of SSRs of all thirteen plastomes, mononucleotides (55.74%) were the most abundant, followed by compound microsatellites (15.95%), dinucleotides (12.16%), tetranucleotides (11.09%), trinucleotides (2.63%), pentanucleotides (1.99%) and hexanucleotides (0.58%). Most of the SSRs were composed of A/T while G/C rarely occurred. Furthermore, SSRs were distributed more in LSC regions than in IR or SSC regions.

Fig. 8.

Numbers of SSRs and repeats of thirteen A. subg. Rhizirideum species. (A), stacking histogram of SSR numbers. Seven sorts of SSRs are shown with different colors: c, compound microsatellites; p1-p6, microsatellites with one to six bases as a repeat unit. (B), stacking histogram of repeats (30–50 bp) numbers

In addition to SSRs, repeats of 30 bp ~ 60 bp were also detected (Fig. 8 B). Four types of repeats were summed up to 528, including forward, reverse, palindromic and complementary. The proportion of palindromic repeats (51.14%) was the highest, while that of the complementary repeats (0.38) was the lowest. A. przewalskianum contained the most repeats (49), and A. anisopodium contained the least (31).

Phylogenetic analysis in subgenus Rhizirideum

Seventy-seven protein-coding sequences of sixty-seven plastomes were extracted and concatenated to establish a tandem CDS dataset. The complete chloroplast genomes (cp) of sixty-seven species were multiple-aligned and trimmed to establish a cp dataset. The CDS dataset and the complete chloroplast genome (cp) were used to reconstruct phylogenetic trees. The CDS tree (Fig. 9 A) and the cp tree (Fig. 9 B) showed a similar topology. Thus, we will take the CDS tree as an example to explain the intra-subgenus (A. subg. Rhizirideum) and inter-subgenera relationships, and the differences between the CDS tree and the cp tree will be explained at the end of this section.

Fig. 9.

Phylogenetic tree reconstruction of 67 species inferred from Maximum likelihood (ML) and Bayesian inference (BI) analyses based on CDS sequences and complete plastomes. (A), CDS tree. (B), complete plastome tree. Tip colors, colors of subgenus names of genus Allium on the clade. Branch colors, colors of three evolutionary clades of genus Allium. The bootstrap support values are listed on the left side of slash (/) and posterior probability values are listed on the right side of slash (/). Null means 100% or 1. The minus sign (−) means parallel branch in BI tree

Within A. subg. Rhizirideum clade, there are twelve species clustered into five small clades. A. anisopodium and A. tenuissimum formed a clade and belonged to A. sect. Tenuissima. Species A. caespitosum, A. mongolicum, A. bidentatum and Allium dentigerum Prokh. formed a clade, and belonged to A. sect. Rhizomatosa. Species A. senescens, A. spirale and A. nutans clustered to form A. sect. Rhizirideum, and A. eduardii and A. przewalskianum formed A. sect. Eduardia. Interestingly, A. polyrhizum, previously belonging to A. sect. Caespitosoprason, was resolved as a sister to clade A. sect. Tenuissima + A. sect. Rhizomatosa + A. sect. Rhizirideum + A. sect. Eduardia. Besides, A. siphonanthum, previously belonging to A. sect. Eduardii, was now resolved as a sister to A. spicatum + A. farreri in clade A. subg. Cyathophora.

Within genus Allium (Fig. 9 A), there was a distinct division of three evolutional clades. For the eleven subgenera involved in this study, A. subg. Microsordum and A. subg. Amerallium formed clade 1, A. subg. Caloscordum, A. subg. Melanocrommyum and A. subg. Anguinum formed clade 2, and the left seven subgenera formed clade 3. In clade 3, only A. subg. Rhizirideum is monophyletic while the other six subgenera are polyphyletic (A. subg. Polyprason, A. subg. Cepa, A. subg. Allium, A. subg. Reticulatobulbosa, A. subg. Cyathophora, and A. subg. Butomissa). Six species in A. subg. Polyprason (A.rude, A.chrysanthum, A.xichuanense, A.chrysocephalum, A.maowenense, A.herderianum) were clustered with six A. subg. Cepa species (A. cepa, A. galanthum, A. altaicum, A. fistulosum, A. cepiforme, A. semenovii) by a support ratio of 100/1.0, then clustered with a branch of five A. subg. Allium species (A.caeruleum, A.schoenoprasoides, A.macrostemon, A.delicatulum, A. tanguticum) + one Reticulatobulbosa species (A. teretifolium) by 100/1.0. Two species from A. subg. Polyprason (A.caricoides, A.obliquum) and one A. subg. Allium species (A. pallasii) formed a small branch (100/1.0) and then became a sister to Clade A. subg. Polyprason + A. subg. Cepa + A. subg. Allium (100/1.0). And A. subg. Rhizirideum was resolved as a sister to the branch just mentioned (50/0.506). Five A.subg. Reticulatobulbosa species (A. paepalanthoides, A. plurifoliatum, A.cyaneum, A.sikkimense, A.forrestii) formed a branch (81/0.9999) and then clustered with a small clade of one A. subg. Reticulatobulbosa species (A. songpanicum) + A. changduense from A. subg. Allium (100/1.0). This branch was clustered with a clade of three A. subg. Cepa species (A. praemixtum, A. oschaninii, A. pskemense) (89/0.6666), then clustered with another A. subg. Cepa species, A. chinense (100/1.0). Clade A. subg. Reticulatobulbosa + A. subg. Cepa was resolved as a sister to the Clade A. subg. Polyprason + A. subg. Cepa + A. subg. Allium + A. subg. Rhizirideum (48/−). Clade A. subg. Reticulatobulbosa + A. subg. Cepa was a parallel clade to the Clade A. subg. Polyprason + A. subg. Cepa + A. subg. Allium + A. subg. Rhizirideum in the BI CDS tree. Then two another A. subg. Allium species (A. sativum, A. ampeloprason) formed a sister to the big clade just mentioned. And one another A. subg. Cepa species (A. condensatum) was resolved in the A. subg. Butomissa clade.

In cp tree, there is several differences in topology. The Clade A. subg. Reticulatobulbosa + three species from A. subg. Cepa (A. praemixtum, A. oschaninii, A. pskemense) possessed a support ratio of 80/−, which means a parallel clade in BI tree, while the ratio of CDS tree came to 89/0.6666. The clade of two A. subg. Allium species (A. sativum, A. ampeloprason) was resolved as a sister to clade A. subg. Rhizirideum by 44/0.7897, and this big clade was then clustered to the mentioned clade A. subg. Reticulatobulbosa + A. subg. Cepa by 38/0.6252. And the Clade A. subg. Rhizirideum + A. subg. Allium + A. subg. Reticulatobulbosa + A. subg. Cepa was resolved as a sister to clade A. subg. Polyprason + A. subg. Cepa + A. subg. Allium by 100/1.0.

Gene selective pressure

We calculated the Ka/Ks ratio (ω) of seventy-seven common protein-coding sequences (CDSs) in thirteen Rhizirideum plastomes (Fig. 9) and then estimated the selective pressure (Additional file 9: Table S9). Most ω values were less than 1, while three CDSs were found 0.5 < ω < 1.0 (rbcL, ycf1a, ycf1b) and one ω > 1 (ycf2) (Fig. 9). Unexpectedly, after selective pressure analysis in EasyCodeMl, none of the seventy-seven CDSs was found significant (P < 0.05) after the likelihood ratio test (LRT). We examined the functions and relative biochemical pathways of the four protein-coding genes mentioned above (Table 3).

Table 3.

Genes under positive and relaxed selection in A. subg. Rhizirideum plastomes

Gene	Ka/Ks ratio	Selection pressure	Gene description	Pathway
ycf2	1.107	Positive selection	Ycf2, part of a 2-MD heteromeric AAA-ATPase complex	Photosynthesis
rbcL	0.563	Relaxed selection	ribulose-1,5-bisphosphate carboxylase/oxygenase large subunit	Calvin-Benson Cycle
ycf1a	0.787	Relaxed selection	Tic214, part of a translocon at the inner envelope membranes of chloroplast called TIC	Photosynthesis
ycf1b	0.544	Relaxed selection	Tic214, part of a translocon at the inner envelope membranes of chloroplast called TIC	Photosynthesis

Discussion

Comparative plastome structure analysis of A. subg. Rhizirideum

Although events of evolution such as genome rearrangement, gene loss, IR expansion, and contraction, have been detected for many times, plastomes are generally highly conserved in genome size, structure, and gene content [32–40]. In this study, the A. subg. Rhizirideum plastomes are of high conservation by large. The quantity of genes, CDSs, rRNA-coding genes, and tRNA-coding genes is 141, 87 (or 89), 8, and 38, respectively, which follows most angiosperms [28, 36, 38, 40–42].

There were 5 of 141 genes pseudogenized (orf56, ycf15, rps2, infA, ycf68) (Table 4). Plastome genes ycf15, ycf68, and infA are also pseudogenized in many other species such as Malus pumila, Morus alba, Cynodon dactylon [38, 41, 43, 44]. The rps2 gene, encoding ribosomal protein S2, is lost in A. sect. Daghestanica plastomes but is pseudogenized in Chlorophytum rhizopendulum [30, 44]. In addition, rps2 production is of great significance to the defense signal transduction process [45]. Thus, in terms of genes coding confirmed products (infA & rps2), their pseudogenization might be used to adjust the transcription and signal transduction of Rhizirideum plants in response to the changing environment.

Table 4.

Summary of pseudogenes and their productions in A. subg. Rhizirideum plastomes

Pseudogene	Position	Production
ycf68	IR	Putative protein RF68
orf56	IR	Putative protein RF56
ycf15	IR	Putative protein RF15
rps2	LSC	Ribosomal protein S2
infA	LSC	Translation initiation factor 1

The SC/IR borders of angiosperm plastomes are generally conservative, lying mostly beside rps19 and ycf1 [46]. Genes trnH-GUG and trnN-ACG are believed to be located at the IR/LSC and IR/SSC borders of the ancestor of monocots, respectively [37]. According to the relative positions of rps19/trnH-GUG and ycf1/trnN-ACG in A. subg. Rhizirideum plastomes (Fig. 2), an expansion of IR regions might occurred. Generally speaking, expansion of IRs can lead to the movement of SC/IR borders. Most terrestrial plants, as A. subg. Rhizirideum species, present movements to a tiny extent, which can make a few genes into or out of IRs [47–49]. Nonetheless, some plants do have their IRs expanding in a large scale. The large expansion can contribute to a large increase or loss of IR genes, such as species in Pelargonium, Psilotum, Leguminosae, and Erodium [32, 33, 50–55]. In A. subg. Rhizirideum plastomes, the duplicated rps19 moved into the IRs from the LSC, while the incompletely duplicated ycf1 moved to cover the IR/SSC borders from the SSC (Fig. 5). In addition, the LSC/IRb boundaries also present a slight shift to the rpl22 gene. The movements of IR/SC borders of A. subg. Rhizirideum plastomes are tiny compared to the species mentioned above. Despite this, the IR expansion of our taxa is somewhat significant for the evolution. It is known that IR regions possess the nature of self-duplication, which has been proven to reduce the synonymous mutation rate (Ks) of genes, resulting in the Ks of IR genes being generally lower than that of SC genes [48]. It can be inferred that in the A. subg. Rhizirideum plastomes, the Ks value of the rps19 gene decreased after moving from the LSC to the IRs. That is, the rps19 gene has been more conserved, as well as its product, ribosomal protein S19, which is a component of the 40S ribosomal subunit. Therefore, it may contribute to the increase of stability of the ribosomal structure when rps19 moved to IRs. This is also true of the gene rpl22 coding ribosomal protein L22, a component of the 60S ribosomal subunit. The moving trend of rpl22 may also influence the ribosomal structure. As is known, the structure of ribosomes can influence the expression pattern of genes, which are often relative with the environment [56]. Consequently, the shift of IR/SC boundaries may be regarded as the adaptive evolution of plastomes. There are twenty-six genes with introns in the plastome of A. senescens, three more than Anena sativa in the family Gramineae (single-copy gene clp, rpoC1 and double-copy gene ycf68). The transpliced gene rps12 has three exons, one in the LSC and two in the IRs. In eukaryotes, intron-splicing enhances gene expression by reducing transcription-associated mutagenesis [57]. Meanwhile, this process imposes selection pressure on genes [58]. Therefore, the intron-existing genes in A. subg. Rhizirideum plastomes indicate that they are also under this kind of pressure.

The GC contents of A. subg. Rhizirideum plastomes range from 36.8 to 36.9%, which is in accordance with those of many other monocots, approximately 37% [59–61]. Additionally, the GC content of Amarillydaceae subfamily Allioideae plastomes is below that of other families, such as Asparagaceae, Iridaceae, Agapanthaceae, Etc. [62]. This decrease can be attributed to the selective pressure caused by either neutral mutation [63–65] or high transcription efficiency [66, 67]. This is the same as the low GC content of the A. subg. Rhizirideum plastomes.

Codon usage bias is a significant feature of plastomes, which influences gene expression and demonstrates natural selection pressure [68, 69]. According to the results, we found that subgenus Rhizirideum have thirty frequently used codons (RSCU > 1), 29 of which ended by A or U. In plastomes, codons often appeared with a higher AU content. The third position of codons have a higher trend of using A/U than G/C [70–72]. Codons encoding leucine were the most of all, and the codon bias showed as UUA > CUU > UUG > CUA > CUC > CUG, consistent with other plants like Ligusticum and Geraniaceae [40, 73]. From Fig. 7, we found that differences in CAI, ENC and GC3 of five A. subg. Rhizirideum sections were small, while differences in CBI and FOP were relatively more significant. CBI and FOP of section Rhizomatosa had the highest values and those of section Rhizirideum had the lowest. The results revealed that the diversity of codon usage patterns of different taxa might also be helpful for the identification and classification of species [74].

SSRs are regarded as potential resources in evolutionary research and are effective in species classification and population genetic analyses that study the biogeography of allied taxa [75–79]. According to the SSR counting results (Fig. 8 A, Additional file 7: Table S7), we found some repeats only in some species, such as hexanucleotides TTTCCC in A. siphonanthum, pentanucleodide TTTAG in A. przewalskianum, and trinucleotides CTT in A. mongolicum and A. caespitosum. These unique SSRs can be used for species identification and classification in subgenus Rhizirideum. There have been SSRs detected for this purpose, like Lycoris, Psidium, and Asparagus [80–82]. Thus, we hope SSRs detected in our study will provide some helpful information for research of Allium in the future. Besides, large repeat sequences can promote plastome rearrangement and play an important part in sequence divergence [83–85]. In our study, 528 repeats of 30–50 bp were detected (Fig. 8 B). Among all kinds of large repeats, forward repeats and palindromic repeats were in the majority, similar to many other species [40, 86, 87]. Moreover, we found that complement repeats were specially owned by A. spirale and A. senescens.

Phylogenetic analysis

Appropriate gene combinations are significant for accurate phylogenetic inference. Nuclear DNA genes (e.g., ETS and ITS), cpDNA fragments (e.g.,matK, trnL-trnF, and psbJ-petA) and plastomes have been used for the phylogenetic estimation of plants. Here, we used two datasets (complete chloroplast genomes and chloroplast CDSs) to conduct ML analysis and BI analysis for the reconstruction of A. subg. Rhizirideum phylogeny. According to previous studies based on ITS sequences, Allium species were divided into three lineages called clade 1, clade 2 and clade 3 [8]. In a study based on ITS [8], the subgenera Cepa, Reticulatobulbosa, Polyprason, and Allium formed parallel branches, which clustered with Rhizirideum. The results of a recent study based on plastomes [31] showed that A. subg. Cepa first clustered with A. subg. Polyprason, then successively clustered with A. subg. Allium, small branch of A. subg. Polyprason species, small branch of A. subg. Allium species, another A. subg. Cepa branch, and A. subg. Rhizirideum. These previous studies indicated that the subgenera A. subg. Cepa, A. subg. Reticulatobulbosa, A. subg. Polyprason, and A. subg. Allium were polyphyletic groups. Concerning the inter-subgenus relationships within the genus Allium, the topology of our phylogenetic trees (Fig. 9) are generally consistent with previous studies based on ITS and plastomes [9, 23, 88]. Phylogenetic analysis results (Fig. 9) demonstrate that A. subg. Rhizirideum is a strongly supported monophyletic group, which corresponds with previous reports [8]. However, other subgenera near A. subg. Rhizirideum, like A. subg. Allium and A. subg. Cepa, are polyphyletic groups. For instance, several species from A. subg. Cepa are clustered with A. subg. Reticulatobulbosa clade (A. praemixtum, A. oschaninii, A. pskemense and A. chinense) and A. subg. Butomissa clade (A. condensatum), and species from A. subg. Allium are clustered with A. subg. Reticulatobulbosa clade (A. changduense). More genomic samples and geographic information are required for further investigation in the future.

In a phylogenetic tree based on ITS-rps16 datasets [9], A. sect. Rhizomatosa clustered with A. sect. Caespitosoprason, and A. sect. Tenuissima clustered with A. sect. Rhizirideum. The mentioned two clades were resolved as sister branches and clustered with A. sect. Eduardia. In our results (Fig. 9), clade A. subg. Rhizirideum has five branches containing 12 of our 13 species (except A. siphonanthum) and each of them represents a section of this subgenus (A. sect. Rhizomatosa, A. sect. Tenuissima, A. sect. Rhizirideum, A. sect. Eduardia, and A. sect. Caespitosoprason). Species in A. sect. Eduardii (A. przwalskianum & A. eduardii) and A. sect. Tenuissima (A. anisopodium & A. tenuissimum) cluster into two individual branches, which is the same as the phylogenetic analysis of Li et al. [9]. Nevertheless, the interspecific relationships of the other two sections, A. sect. Rhizirideum and A. sect. Rhizomatosa, are somewhat different. The section A. sect. Rhizirideum was divided into Asiatic and European geographical groups by Sinitsyna et al. [24] and the species A. senescens, A. spirale and A. nutans were in the former group, but the relationships among A. senescens, A. spirale and A. nutans were not so clearly shown. Our results show that A. senescens and A. spirale form a sister branch and then cluster with A. nutans with a 100/1.0 support rate. Friesen et al. [10] conducted a phylogenetic analysis for species in two A. subg. Rhizirideum sections (A. sect. Rhizomatosa, and A. sect. Caespitosoprason) based on chloroplast DNA fragments, where A. sect. Caespitosoprason, including A. polyrhizum, was merged into A. sect. Rhizomatosa. In contrast, our plastome tree indicates that A. polyrhizum is not clustered with A. sect. Rhizomatosa species but with clade A. sect. Rhizomatosa + A. sect. Tenuissima + A. sect. Rhizirideum + A. sect. Eduardia in A. subg. Rhizirideum. In other words, A. polyrhizum may be separated from A. sect. Rhizimatosa and placed back into A. sect. Caespitosoprason, which is believed to be a basal taxon of this subgenus. Additionally, A. siphonanthum, a member of A. sect. Eduardia, is now clustered with A. subg. Cyathophora. A. siphonanthum characters as umbel densely many flowered, pedicels shorter than perianth, and bulb tunic subreticulate, while A. cyathophorum in A. subg. Cyathophora is charactered as umbel laxly flowered, Pedicels 1–3 times as long as perianth, and bulb fibrous sometimes subreticulate. So there is few similarity between A.siphonanthum and Cyathophora species. This may be an example of disagreements between molecular and morphological analyses, but identifying the phylogenetic position of A. siphonanthum still needs more specimens and molecular evidence.

Molecular relationships are often consistent with morphology characteristics. Except for A. sect. Caespitosoprason, the rest four clades of A. subg. Rhizirideum can represent the four typical phenotypes of this subgenus (Fig. 9, Fig. 1, Additional file 1: Fig. S1). Species in A. sect. Rhizomatosa character as leaf semiterete to terete, bulbs densely clustered, and the outer skin fibrous. A. sect. Tenuissima species character as leaf semiterete, bulb clustered, and outer skin not broken. A. sect. Eduardia species character as leaf semiterete, and bulbs covered with a common reticulate tunic. Species in A. sect. Rhizirideum have a very different morphology from other sections, leaf broadly linear and bulb ovate-cylindric and thicker. Despite this, one species in A. sect. Caespitosoprason (A. polyrhizum) and four species in A. sect. Rhizomatosa (A. caespitosum, A. bidentatum, A. mongolicum, and A. dentigerum) are relatively similar in morphology. Disagreements between molecular and morphological analyses have also been frequently reported in other taxa, for instance, section Daghestanica and subgenus Cyathophora in the genus Allium [30, 89].

As is shown in Fig. 9, intra-section relationships in five sections of A. subg. Rhizirideum are supported by high support (100/1.0). However, the inter-section relationships among some sections show a lower support ratio, such as the node between A. sect. Rhizomatosa and A. sect. Tenuissima (54/0.9907 in CDS tree, 82/1.0 in cp tree). This phenomenon probably occurs because of lacking samples. Both bootstrap support values and posterior probabilities in cp tree (Fig. 9 B) are relatively more prominent than those in CDS tree, especially the node between A. sect. Eduardia and clade A. sect. Rhizomatosa + A. sect. Tenuissima + A. sect. Rhizirideum (35/0.5399 in CDS tree, 59/0.674 in cp tree). This may be due to the fewer genetic sites in CDSs than those in cpDNA sequences. Also, the numbers of support ratio in ML tree are smaller than in BI tree, possibly because of the difference in inference methods.

Adaptive evolution

The Ka/Ks ratio (ω) is used to assess the selective pressure on protein-coding genes. The ω values > 1, = 1, and < 1 indicate that this gene has undergone positive, neutral, and purifying selection, respectively. In addition, there is also a sort of relaxed selection with 0.5 < ω < 1, according to other research [30, 40, 62]. The Ka/Ks calculating results (Fig. 10) showed that most of the genes had a ω < 0.5, while one of the genes had a ω > 1 (ycf2) and three of them had a ω > 0.5 (rbcL, ycf1a, ycf1b). So we consider that ycf2 has been under positive selection and rbcL and ycf1 have been under relaxed selection. In previous studies, those four genes mentioned above have been reported under positive selection [90–94]. Gene rbcL encodes ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCO) large subunit that is vital to CO₂ fixation by plants. RuBisCO catalyzes the production of 3-phosphoglycerate by adding CO₂ onto RuBP, which provides the resource for sugar synthesis [95]. Genes ycf1 and ycf2 have been enigmatic and their functions had not been found for a long time until knockout studies showed that the gene ycf1 is essential for the survival of plants [96–98]. The latest study has proved that ycf1 encodes a protein Tic214 that forms a vital component of a translocon at the inner envelope membranes of chloroplast called TIC, which is indispensable for photosynthetic protein import in green tissues [99]. The gene ycf2 has also been proven to encode a protein, part of a 2-MD heteromeric AAA-ATPase complex, which is closely associated with the TIC complex and functions as a motor for protein import [100]. Thus, these genes with a ω > 0.5 are necessary for photosynthesis, which is essential for plants. Species of A. subg. Rhizirideum are distributed in extreme environments, such as areas with very low temperature, arid climates, and high altitudes [10, 24, 101]. The photosynthesis demands for sufficient light might have exerted relatively intensive pressure on these genes. The other way round, the positively or relaxedly selective genes may help those species fit in the various environments.

Fig. 10.

Ka/Ks ratios of 41 single-copy genes. Three genes > 0.5 and one gene > 1 are noted with red triangles

Methods

DNA isolation, sequencing and plastome annotating

The fresh leaves of eleven species were sampled from public areas and dried with silica gel afterwards (locality see Additional file 5: Table S4). Total genomic DNA was isolated from silica-dried leaf tissues with a modified CTAB method. The voucher specimens (Additional file 5: Table S4) were deposited at the herbarium of Sichuan University (Chengdu, China) (voucher specimens: H11072607 (SZ), De-qing Huang; ZCJ20210821 (SZ), Chun-jing Zhou; FX2020081001 (SZ), Xiao Fu; FX2020080902 (SZ), Xiao Fu; H11072807 (SZ), De-qing Huang; H11070501 (SZ), De-qing Huang; FX2020081401 (SZ), Xiao Fu; FX2020081501 (SZ), Xiao Fu; FX2020081901 (SZ), Xiao Fu; ZCJ2012081910 (SZ), Chun-jing Zhou; FX2021072101 (SZ)). And the DNA sample of A. siphonanthum was from Germplasm Bank of Wild Species and National Wild Plant Germplasm Resource Center, voucher specimen 13CS6776 (KUN) at Kunming Institute of Botany. DNA libraries were prepared and sequenced with the Illumina HiSeq 2500 platform with PE150 bp reads.

Complete chloroplast genomes were reconstructed by NOVOPlasty v2.6.2 [102] using A. cepa (MK335926) and A. sativum (MK335928) as references. Then the plastid genomes were annotated with PGA [103] and manually adjusted with Geneious R11 (Biomatters, Ltd., Auckland, New Zealand). Finally, the plastome circus map was drawn with OGDRAW [104] and Gview [105].

Synonymous codon usage bias

Seventy-seven protein-coding sequences (Additional file 3: table S2) were extracted from thirteen Rhizirideum plastomes with Phylosuite v1.2.2 [106], aligned with MAFFT v7.487 [107] and trimmed with trimAl v1.2 [108]. Then again with Phylosuite, they were concatenated respectively and form thirteen CDS tandem sequences. Afterwards, the thirteen sequences were input into codonW v1.3 to calculate codon contents and RSCU values, which were later sorted and analyzed manually in Microsoft Excel 365.

Sequence divergence

The online program mVISTA [109] was used to generate the whole-genome alignment of the thirteen Rhizirideum plastomes with A. senescens as a reference. All the plastomes were aligned with MAFFT v7.487. The nucleotide diversity (Pi) of genes and intergenic regions was calculated by DnaSP v6 [110].

Repeat structure

REPuter [111] was used to examine plastome repeat sequences. Thirteen plastomes of subgenus Rhizirideum were input into the REPuter website and the list of repeats were exported. Four sorts of repeats were classified: forward, palindromic, reverse, and complimentary matches. The parameters were as follows: repeat size of (1) > 30 bp; (2) > 90% sequence identity between the two repeats; and (3) Hamming distance = 3. Simple sequence repeats (SSRs) of thirteen A. subg. Rhizirideum plastome sequences were mass counted by Perlscript MicroSAtellite (MISA). The setting motif sizes were one to six nucleotides, and the minimum repeat units were defined as 10, 5, 5, 4, 3 and 3 for mono-, di-, tri-, tetra-, penta- and hexa-nucleotides, respectively.

Phylogenetic analysis

In addition to fourteen plastomes newly sequenced (thirteen A. subg. Rhizirideum plastomes and A. condensatum from subgenus Cepa), another fifty-three species were also selected (including thirty-six Allium species from our team, thirteen Allium species and four Amaryllidaceae outgroups downloaded from NCBI) (Additional file 4: Table S3) to infer the phylogenetic relationships. Seventy-seven CDSs were extracted from sixty-seven taxa by using Phylosuite and were multiple-aligned with MAFFT. The alignments were trimmed with trimAl and then concatenated in series for the phylogenetic analysis with Phylosuite to form a CDS dataset. Sixty-seven plastome sequences were multiple-aligned with MAFFT and trimmed with trimAl, leaving LSC, SSC and only one IR region to establish a cp dataset. The CDS and cp datasets were used to perform phylogenetic inferences, respectively. The Maximum Likelihood (ML) analysis was performed by RAxML v8.2.8 [112] with the GTR + G model and 1000 bootstrap replicates. The Bayes Inference (BI) analysis was performed by MrBayes v3.2.7 [113] with the substitution model GTR + I + Γ. The Markov chain Monte Carlo (MCMC) algorithm was run for one million generations, and one tree was sampled every 1000 generations. We then determined the MCMC convergence according to the average standard deviation of split frequencies (ASDSF) below 0.01. The first 20% of the trees were discarded as burn-in, and the remaining trees were used to generate consensus trees. Finally, online software Interactive Tree of Life (iTOL) was used to edit the phylogenetic trees [114].

Selective pressure analysis

Thirteen studied species were used to calculate pairwise Ka/Ks ratios with KaKs Calculator v2.0 [115], and the average values were calculated to represent the Ka/Ks ratio of each gene. Seventy-seven CDSs of thirty-seven taxa were extracted and aligned with the software MUSCLE v5 [116] aligned by codons. The positive selection analyses, measured by the ratio (ω) of the non-synonymous substitution rate (Ka) to the synonymous substitution rate (Ks), were performed using the branch-site model in EasyCodeML v1.4 [117] and our subgenus lineage was designated. Positive, neutral, and purifying selection are demonstrated when the ratio ω > 1, w = 1, and ω < 1, respectively [118]. The log-likelihood values were tested (LRT) in accordance with [119]. The BEB method [120] was applied to compute the posterior probabilities of amino acid sites, and those with a higher posterior probability were determined to be under positive selection.

Conclusions

Our work revealed that (1) the Rhizirideum plastomes have similar structures, (2) the phylogenetic position of the Rhizirideum species A. polyrhizum and A. siphonanthum should be reconsidered, (3) the plastome gene ycf2 is under positive selection, probably contributing to the adaptability to the environment. Much remains to be investigated on the phylogenetic relationships of species in subgenus Rhizirideum, notably improving the sampling of Allium species.

Authors’ contributions

Xiao Fu: Conceptualization, Resources, Data curation, Formal analysis, Writing - original draft, Writing - review & editing. Deng-Feng Xie: Data curation, Writing - review & editing. Yu-Yang Zhou: Resources, Data curation. Rui-Yu Cheng: Resources, Writing - review & editing. Xiang-Yi Zhang: Resources, Writing - review & editing. Xing-Jin He: Resources, Data curation, Writing - review & editing. Song-Dong Zhou: Resources, Writing - review & editing. The authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 32100180, 32070221, 32170209), the Fundamental Research Funds for the Central Universities (20826041E4158), the China Postdoctoral Science Foundation (2020 M683303).

Availability of data and materials

All data generated or analysed during this study are included in this published article and its supplementary information files. The datasets analyzed during the current study are available in the NCBI GenBank repository (See supplementary Additional file 4: Table S3 for accessions).

Declaration

Ethical approval and consent to participate

Plant material complies with local and national regulations. The voucher specimens were deposited at the herbarium of Sichuan University (Chengdu, China) (voucher specimens: H11072607 (SZ), De-qing Huang; ZCJ20210821 (SZ), Chun-jing Zhou; FX2020081001 (SZ), Xiao Fu; FX2020080902 (SZ), Xiao Fu; H11072807 (SZ), De-qing Huang; H11070501 (SZ), De-qing Huang; FX2020081401 (SZ), Xiao Fu; FX2020081501 (SZ), Xiao Fu; FX2020081901 (SZ), Xiao Fu; ZCJ2012081910 (SZ), Chun-jing Zhou; FX2021072101 (SZ), Xiao Fu) and the DNA sample of A. siphonanthum was from Germplasm Bank of Wild Species and National Wild Plant Germplasm Resource Center, voucher specimen 13CS6776 (KUN) at Kunming Institute of Botany (Kunming, China).

Competing of interest

The authors declare that they have no competing interests.

Consent for publication

Not applicable.