Insights into phylogeny, age and evolution of Allium (Amaryllidaceae) based on the whole plastome sequences

Abstract

Background and Aims

The genus Allium L., one of the largest monocotyledonous genera and one that includes many economically important crops with nutritional and medicinal value, has been the focus of classification or phylogeny studies for centuries. Recent studies suggested that the genus can be divided into 15 subgenera and 72 sections, which were further classified into three evolutionary lineages. However, the phylogenetic relationships reconstructed by one or two loci showed weaker support, especially for the third evolutionary lineage, which might not show the species relationships very clearly and could hinder further adaptive and evolutionary study.

Methods

In this study, a total of 39 complete chloroplast genomes of Allium (covering 12 Allium subgenera) were collected, and combining these with 125 species of plastomes from 19 other families of monocots, we reconstructed the phylogeny of the genus Allium, estimated the origin and divergence time of the three evolutionary lineages and investigated the adaptive evolution in this genus and related families.

Results

Our phylogenetic analysis confirmed the monophyly and three evolutionary lineages of Allium, while new species relationships were detected within the third evolutionary lineage. The divergence time of the three evolutionary lineages was estimated to be in the early Eocene to the middle Miocene, and numerous positive selected genes (PSGs) and PSGs with high average Ka/Ks values were found in Allium species.

Conclusions

Our results detected a well-supported phylogenetic relationship of Allium. The PSGs and PSGs with high Ka/Ks values, as well as diversified morphologies, complicated chromosome characteristics and unique reproductive modes may play important roles in the adaptation and evolution of Allium species. This is the first study that conducted phylogenetic and evolutionary analyses on the genus Allium combined with the plastome and morphological and cytological data. We hope that this study can contribute to further analysis of Allium for other researchers.

INTRODUCTION

The genus Allium L. (Amaryllidaceae, Allioideae) currently comprises >900 species (Herden et al., 2016), making it one of the largest monocotyledon genera. They are well known (but not always appreciated) for their specific and commonly intense smell and taste, and are characterized by having bulbs enclosed in membranous (sometimes fibrous) tunics, free or almost free tepals, and often a sub-gynobasic style (Friesen et al., 2006). Many economically important species are included, such as garlic, leek, onion and shallot. Two centres for Allium species diversity were suggested in previous studies (Fritsch and Friesen, 2002; Nguyen et al., 2008), one is from the Mediterranean Basin to central Asia and Pakistan, and the other is in North America. It is widely accepted that the Allium species can be classified into 15 subgenera and 72 sections (Friesen et al., 2006). In addition, based on 195 representative species of Allium, Fritsch and Friesen (2002) suggested that the genus Allium was monophyletic and can be differentiated through a number of evolutionary steps into three evolutionary lineages. The taxonomy of Allium was further revised by Li et al. (2010a) based on morphological characters, internal transcribed spacer (ITS) and rps16 sequence data, and the Chinese Allium species were further classified into 13 subgenera and 34 sections. Nonetheless, the phylogenetic relationships reconstructed by ITS sequences showed weaker support, especially for the third evolutionary lineage, and limited rps16 sequences were included in the study by Li et al. (2010a), which might not show the species relationships of Allium very clearly.

In view of the origin and divergence of Allium, unfortunately only one possible Amaryllidaceae fossil that dated to the latest early Eocene was discovered, in Washington (Pigg et al., 2018), and a few fossils of the Asparagales have been reported from the late Eocene (Couper 1960; Herendeen and Crane 1995; Muller 1981); however, they are all too young to calibrate the crown clade of the order (Wikström et al., 2001; Janssen and Bremer 2004). Li et al. (2016b) estimated that the genus Allium originated during the late Eocene [approx. 34.26 million years ago (Mya), highest posterior density (HPD) 95 % 24.29–45.76 Mya] and suggested that this genus originated from eastern Asia and underwent different biogeographical pathways (Li et al., 2010a, 2016b). This is helpful to understand and assess the evolutionary processes and migration history of the Allium. However, the divergence time of some Allium lineages, especially for the important node branches (such as the nodes of the three evolutionary lineages), were not estimated, which might hinder further studies conducted on subgenera or sections of Allium.

Despite the phylogeny, origin and divergence of Allium, species of this genus are naturally distributed only in the Northern Hemisphere and widely spread from the dry sub-tropics to the boreal zone. The habitat of Allium species varies from dry and well-drained soil to moist and organic soil, which can be even in swamps or in water (Block, 2010). Evolution of this genus species was also accompanied by habitat and ecological diversification, and their adaptation to the various habitat environments has resulted in a remarkable morphological polymorphism (e.g. flowers, leaves and bulbs) (Li et al., 2016a). Additionally, the chromosome numbers of Allium are diversified, including basic chromosomes numbers x = 7, 8, 9, 10 or 11, and thus they produce many polyploids (Friesen, 1992; Xu et al., 1998; Zhou et al., 2007; Zhang et al., 2009; Jones, 2012; Li et al., 2017; Peruzzi et al., 2017), which are often considered to have many advantages compared with their diploid progenitors in morphological, physiological, life history characteristics and rates of adaptation (Mayrose et al., 2010). It has been generalized that chromosomal characteristics (e.g. chromosome number, ploidy level, genome size, karyotype asymmetry, etc.) prove to be informative and significant for understanding the relationships, evolution and adaptation of taxa (Sharma and Sharma, 2014). Therefore, Allium is a successful taxon from the point of view of its wide distributions, diversified morphologies and complex chromosome characteristics. Recent transcriptome or genome-wide studies on multiple species that succeed in the evolutionary processes have identified various evolutionary and adaptive processes that may be responsible for adaptation, including for humans (Beall et al., 2010; Simonson et al., 2010; Yi et al., 2010; Peng et al., 2011), animals (Savolainen et al., 2007; Qiu et al., 2012; Ge et al., 2013; Gou et al., 2014) and plants (Leimu and Fischer, 2008; Hancock et al., 2011; Jia et al., 2013; Yates et al., 2014; Zhang et al., 2016b; Zhang et al., 2019). In addition, adaptation to the environment can allow population persistence (Aitken et al., 2008; Franks and Hoffmann, 2012). Local adaptation has become an important component for species responses to changing environments (Davis and Shaw, 2001), and the phenotypic plasticity in a changing environment has received much attention to date (Loarie et al., 2009; Nicotra et al., 2010; Chevin and Hoffmann, 2017; Van Buskirk, 2017; Oostra et al., 2018; Villemereuil et al., 2018). However, several challenges exist in analysing and interpreting the genetic basis of evolution and adaptation of Allium species. For instance, poor genomic information is available in Allium because of the enormous size of the genome (approx. 16.3 Gb) (Duangjit et al., 2013), which is 32 times larger than the genome of rice (Arumuganathan and Earle, 1991). The diploid (2n = 16) genome sizes of garlic (A. sativum) and onion (A. cepa) were estimated to be >30 Gb (Egea et al., 2017; Peska et al., 2019). Although sequencing technologies have undergone rapid development in the past 10 years, analysis of such a complex and huge genome of the Allium species, a non-model plant, remains a Herculean task (Schatz et al., 2012; Nagarajan and Pop, 2013).

Complete chloroplast (cp) genomes are known to be highly conserved in both gene order and gene content (Raubeson and Jansen, 2005), and exhibit a much lower substitution rate than nuclear DNA (Wolfe et al., 1987). Due to their conserved structure, small effective population size, lack of recombination and usually uniparental inheritance (Henry et al., 2016), cp genomes have been extensively used in phylogenetic reconstruction (especially in taxonomically complex groups) (Jansen et al., 2007; Moore et al., 2010; Barrett et al., 2014; Malé et al., 2014; Shaw et al., 2014; Dong et al., 2015; Yu et al., 2017; Ye et al., 2018) and selection pressure analysis (Allen et al., 2011; Carbonell-Caballero et al., 2015; Hu et al., 2015; Xie et al., 2018b). Many cp genomes have been reported in Allium species (Kim and Yoon, 2010; Lee et al., 2017; Filyushin et al., 2018; Jin et al., 2018; Xie et al., 2019a, b; Yang et al., 2019), and it is necessary to performe a comprehensive cp genome analysis on the genus Allium. Here, a total of 39 complete cp genomes of Allium were collected, covering 12 of 15 Allium subgenera. Combining these with 125 species plastomes from another 19 families of monocots, we wish to accomplish the following: (1) to reconstruct the phylogeny of the genus Allium and analyse lineage relationships at the cp genome level; (2) to estimate the origin and divergence time of Allium by using more cp genome regions and more fossils; and (3) to investigate the adaptative evolution of Allium and allied families by using selective pressure analysis and morphological characteristics. Overall, this study will contribute to a comprehensive understanding of plastome evolution in Allium plants.

MATERIALS AND METHODS

Taxon sampling and DNA extraction

A total of 164 species representing 19 families of monocots were included in this study. Thirty-nine species from three evolutionary lineages of Allium (Friesen et al., 2006; Fritsch and Keusgen, 2006) were selected, ranging from the basal to the top lineages of the whole Allium phylogenetic tree, including all types of inflorescences of the genus Allium, and showing the complicated ploidy of chromosome and other morphological characteristics. Thus, we think that these species to some degree can act as representatives of the Allium genus. Among the 164 sampled species, the whole cp genomes of 22 Allium species were reported from our laboratory (Supplementary data Table S1), and the data of the remaining species were downloaded from NCBI (Supplementary data Table S2). Total genomic DNA was extracted from either fresh or silica gel-dried material using the DNeasy Plant Mini Kit following the manufacturer’s protocol (Biomed, Beijing, China). A NanoDrop spectrophotometer (ND-1000; Thermo Fisher Scientific, USA) and agarose gel electrophoresis were used to determine DNA quality and purity.

Sequence assembly and annotation

All examined DNAs of 22 Allium species were sent to Novogene (Beijing, China) for library construction and sequencing. Paired-end libraries were generated on an Illumina Hiseq 2500 platform. The raw reads obtained from Novogene were filtered using Trimmomatic 0.3.2 with default parameters (Bolger et al., 2014). The program MITObim v1.7 (Christoph et al., 2013) was used for plastome assembly, and the whole cp genomes of several Allium species were downloaded from GenBank as references. Reads were assembled de novo using Velvet (Zerbino and Birney, 2008) with k-mer sizes ranging from 27 to 145, and the coverage cut-offs were auto-adjusted. In order to obtain accurate plastomes, each of the species was assembled four times with the reference genomes A. cepa (KM088014), A. sativum (KY085913), A. victorialis (NC_037240) and A. obliquum (LT699701). Gaps that appeared in the assembled cp genomes were further confirmed and corrected by Sanger sequencing and the primers were designed using Lasergene 7.1 (DNASTAR, Madison, WI, USA). The primers and amplifications are shown in Supplementary data Table S3. The modified plastomes were annotated using the program DOGMA (Wyman et al., 2004), and subsequently corrected within GENEIOUS R11 (Biomatters, Ltd, Auckland, New Zealand). Finally, we used OGDRAW (Lohse et al., 2013) to draw circular plastome maps.

Sequence divergence analysis

The cp genomes of all 39 Allium species were aligned, and the alignments were visualized using mVISTA (Frazer et al., 2004) with A. altaicum as reference to detect sequence divergence. Furthermore, in order to evaluate the nucleotide diversity (Pi) of each gene, DnaSP version 5.1 (Librado and Rozas, 2009) was used to calculate the nucleotide diversity of genes in LSC (large single copy) regions, SSC (small single copy) regions and inverted repeat (IR) regions.

Phylogenetic analyses

In order to investigate the phylogeny of Allium species and allied families, the 22 assembled Allium species and another 142 allied plastomes were analysed together. First, all single-copy genes (SCGs) of the 164 taxa were extracted and then aligned using MUSCLE v3.6 (Edgar, 2004), manually examined and adjusted. These alignments were then concatenated as a super locus of single-copy genes, which were further used for phylogenetic analysis. Maximum parsimony (MP) was performed using PAUP* version 4.10 (Swofford, 2003). All characters were equally weighted, gaps were treated as missing and character states were treated as unordered. A heuristic search was performed with TBR branch swapping and the Multrees option, and random stepwise addition with 1000 replications. All analyses used the best-fitting models of nucleotide substitutions selected in jModelTest v2.1.4 (Darriba et al., 2012) under the Akaike information criterion (AIC). Maximum likelihood (ML) analyses were conducted using RAxML v8.0 (Stamatakis, 2014) based on the best-fit GTR + G model and 1000 bootstrap replicates. Bayesian analyses were performed with MrBayes v3.2 (Ronquist and Huelsenbeck, 2003). Three independent Markov chain Monte Carlo (MCMC) runs of different lengths, but under the same estimation conditions, were conducted. Each chain ran 1 × 10⁸ generations with the sample frequency of 50, and the initial 20 % of the samples were discarded as burn-in to confirm the stationarity. Tracer v1.5 (Rambaut and Drummond, 2009) was used to assess the quality of the MCMC simulations and stability of runs. Effective sample size (ESS) values were >200 for all parameters, suggesting that sufficient sampling occurred. In addition, phylogenetic analyses were also performed for the coding sequences (CDSs) that were shared in all 164 species.

Molecular dating and fossil calibration

The combined single-copy gene data set was used to estimate the origin times of Allium and other allied families. Bayesian searches for tree topologies and node ages were conducted in BEAST (Drummond and Rambaut, 2007) using a GTR + G substitution model selected by jModelTest 2.1.4 (Darriba et al., 2012) and an uncorrelated log-normal relaxed clock (Drummond et al., 2002). A Yule process was specified as tree prior, and the MCMC algorithm was run for 5 × 10⁷ generations with sampling every 2000 generations, following a burn-in of 10 % of the initial cycles. MCMC samples were inspected in Tracer to confirm sampling adequacy and convergence of the chains to a stationary distribution. Three fossils used to calibrate time in BEAST were as follows. (1) According to the studies of Friis et al. (1994, 1997, 1999), Magallón et al. (2015), Eklund et al. (2004) and Li et al. (2019), 121 Ma was implemented as a minimum age in the penalized likelihood analysis and as the zero offset of a log-normal distribution with log mean of (120.7 + 10 %), and s.d. of 1 in the uncorrelated log-normal analysis. This time is equal to the crown group of the Chloranthaceae. (2) From the references of pollen fossils (Doyle and Hickey, 1976; Doyle and Robbins, 1977; Hickey and Doyle, 1977), we set the minimum age in the penalized likelihood analysis as 112 Ma (calibrate the monocot crown node), and as the zero offset of a log-normally distributed prior with log mean of (112 + 10 %) and s.d. of 1. (3) Based on studies of Bell et al. (2010), Friis (1988) and Magallón et al. (2015), the node of Zingiberales was set as follows: 77.8 Ma was implemented as a minimum age in the penalized likelihood analysis, which equalled the minimum age of the Zingiberales fruits and seeds fossils, and as the zero offset of a log-normal distribution with log mean of (77 + 10 %) and s.d. of 1 in the Bayesian analysis.

Selective pressure analysis

In order to detect the sites that are under positive selection in the protein-coding genes in the plastid genomes of Allium species, an optimized branch-site model (Yang and Dos, 2011) and Bayesian empirical Bayes (BEB) methods (Yang et al., 2005) were conducted. The CDSs of all 164 taxa were extracted and aligned using the software MUSCLE (Edgar, 2004) and the ‘gaps’ in the alignments were further checked. The alignment sequences were further trimmed by Trimal v1.2 (Capellagutiérrez et al., 2009) with parameters Trimal -in $i -out $i.fasta -fasta -noallgaps, and the bona fide alignments were used to perform the positive selection analyses. The Allium lineage was selected as a specifically designated branch to assess potential positive selection in the CODEML program implemented in the PAML package (Yang, 2007). The non-synonymous (Ka) and synonymous (Ks) nucleotide substitution rates and their ratio (ω = Ks/Ks) were used to measure the selective pressure. The ratios ω > 1, ω = 1 and ω < 1 suggest positive selection, neutral selection and negative selection, respectively (Yang and Nielsen, 2002). The log-likelihood values were calculated and tested with a neutral model and an alternative model according to Yang (2007). The right-tailed χ ² was used to calculate P-values according to the difference in log-likelihood values between the neutral model and alternative model with one degree of freedom to assess the model fit. Afterwards, the BEB method was applied to compute the posterior probabilities of amino acid sites to identify whether these specific sites were under positive selection (codon sites with a high posterior probability) (Yang, 2007; Lan et al., 2017). A gene with a test P-value <0.05 and with positively selected sites was considered as a positively selected gene (PSG). Moreover, in order to compare the differences in selection pressure that were experienced in allied families of Allium, all family lineages included in this study were separately subjectd to positive selection analysis.

Statistics of morphological and chromosomal characteristics

Morphological traits and chromosomal characteristics are important sources for analysing relationships and evolution of taxa (Schneeweiss et al., 2004). Our laboratory members have researched the genus Allium for >20 years, and accumulated a lot of morphological and karyotypic data. Therefore, in order to understand and analyse the species relationships and the evolutionary processes better, we collected and exhibited the morphological and karyotypic traits of the 39 Allium species (the karyotypes are summarized according to previously published references), including the inflorescences, ploidy level, existence or absence of aneuploids chromosome, and so on.

RESULTS

Characteristic of the Allium chloroplast genomes

Chloroplast genome structures are conserved and similar in gene order across 39 Allium species. The genome size ranges from 145 819 to 159 125 bp, and the GC content varied from 36.7 to 37.8 %. The length of the coding regions changes from 64 581 bp to 81 609 bp, and the minimum and maximum lengths were 72 410 bp and 84 711 bp in non-coding regions. Information on LSC, SSC and IR regions, and gene number is given in Table 1 and Supplementary data Table S4.

Table 1.

Summary of major characteristics of Allium plastomes, including genome size, GC content and gene number.

Taxon	Total genome		LSC length (bp)	SSC length (bp)	IR length (bp)	Gene number	Protein coding	tRNAs	rRNAs	Coding region		Non-coding region
	Length (bp)	GC (%)								Length (bp)	GC (%)	length (bp)	GC (%)
Allium altaicum	153 129	36.8	82 197	17 912	26 510	131	85	38	8	78 183	37.3	74 946	36.3
Allium ampeloprasum	152 732	36.7	81 775	17 905	26 526	127	81	38	8	77 289	37.2	75 443	36.2
Allium caeruleum	153 267	36.8	82 389	18 058	26 410	131	85	38	8	79 195	37.2	74 072	36.4
Allium cepa	153 440	36.8	82 543	17 929	26 485	132	86	38	8	79 263	37.3	74 177	36.3
Allium chinense	152 525	36.8	81 324	18 205	26 498	126	87	31	8	80 115	37.3	72 410	36.2
Allium chrysanthum	153 621	36.8	82 744	17 985	26 446	132	86	38	8	79 269	37.2	74 352	38.4
Allium chrysocephalum	153 710	36.8	82 688	17 998	26 512	132	86	38	8	79 280	37.2	74 430	38.4
Allium cyathophorum	154 174	36.8	83 359	17 881	26 467	132	86	86	8	79 382	37.3	74 792	36.3
Allium fetisowii	154 018	36.9	83 657	17 941	26 210	132	86	86	8	79 300	37.3	74 718	36.5
Allium fistulosum	152 859	36.9	81 930	17 921	26 504	132	86	86	8	79 307	37.3	73 552	36.5
Allium forrestii	153 186	36.8	82 339	17 959	26 444	132	86	86	8	79 194	37.2	73 992	36.4
Allium herderianum	153 605	36.8	82 658	17 983	26 482	132	86	38	8	79 276	37.2	74 329	38.4
Allium macleanii	152 633	36.9	82 890	17 213	26 265	131	85	38	8	78 988	37.3	73 645	36.5
Allium macranthum	152 876	37.1	83 600	18 959	25 095	132	86	38	8	78 723	37.6	74 153	36.6
Allium mairei	152 913	36.9	82 493	18 762	25 829	132	86	38	8	78 914	37.3	73 999	36.5
Allium maowenense	153 608	36.8	82 668	18 000	26 470	132	86	38	8	79 256	37.2	74 352	38.4
Allium monanthum	154 804	37.0	83 835	18 007	24 551	132	86	38	8	79 305	37.5	75 499	36.5
Allium nanodes	154 077	37.0	84 274	20 075	24 864	131	85	38	8	70 029	37.3	84 048	36.8
Allium neriniflorum	154 280	37.0	83 131	18 191	26 479	132	86	38	8	79 532	37.5	74 748	36.5
Allium nutans	153 456	36.9	82 533	17 951	26 486	132	86	38	8	79 299	37.2	74 157	36.6
Allium obliquum	152 597	36.8	81 798	18 059	26 370	132	86	38	8	79 325	37.2	73 272	36.4
Allium oschanini	153 580	36.8	82 522	18 030	26 514	132	86	38	8	79 311	37.3	74 269	36.3
Allium paradoxum	145 819	37.1	80 919	13 504	25 698	133	86	39	8	64 581	37.3	81 238	36.9
Allium platyspathum	152 529	36.8	81 544	17 955	26 515	131	85	38	8	78 514	37	74 015	36.6
Allium polyrhizum	153 086	36.9	82 438	19 188	25 730	132	86	38	8	79 127	37.3	73 959	36.5
Allium praemixtum	153 226	36.8	82 163	18 041	26 511	132	86	38	8	79 302	37.3	73 924	36.3
Allium prattii	154 482	37.0	83 428	18 056	26 499	131	85	38	8	69 771	37.3	84 711	36.8
Allium przewalskianum	153 245	36.9	82 428	18 703	26 057	132	86	38	8	79 239	37.3	74 006	36.5
Allium_pskemense	153 788	36.7	82 721	18 033	26 517	132	86	38	8	79 197	37.2	74 591	36.2
Allium rude	153 697	36.7	82 815	17 978	26 452	132	86	38	8	79 284	37.2	74 413	38.2
Allium sativum	153 118	36.7	82 048	17 986	26 542	128	83	37	8	77 808	37.2	75 310	36.2
Allium schoenoprasides	153 583	36.7	82 551	18 082	26 475	132	86	38	8	79 540	37.2	74 043	36.2
Allium schoenoprasum	153 014	36.8	82 059	17 937	26 509	132	86	38	8	79 296	37.2	73 718	36.4
Allium spicatum	153 225	36.9	82 382	17 930	26 456	132	86	38	8	79 184	37.3	74 041	36.5
Allium strictum	152 962	36.8	82 880	20 562	24 760	132	86	38	8	79 169	37.2	73 793	36.4
Allium tuberosum	157 735	36.9	86 472	19 079	26 092	132	86	38	8	81 609	37.4	76 126	36.4
Allium ursinum	159 125	37.3	88 056	18 105	26 482	131	84	39	8	77 859	37.5	81 266	37.1
Allium victorialis	154 074	37.0	83 173	17 853	26 526	131	85	38	8	78 996	37.5	75 078	36.5
Allium xichuanense	153 673	36.7	82 797	17 950	26 463	132	86	38	8	79 269	37.2	74 404	38.4

Sequence divergence

The sequence divergence analysis showed high sequence similarity across the Allium plastid genomes (Supplementary data Fig. S1). In addition, the IR regions and coding regions were more conserved than the LSC, SSC and non-coding regions (Supplementary data Fig. S2). The nucleotide diversity values of the LSC regions ranged from 0 to 0.02525 with a mean value of 0.00963 (the values varied from 0.00354 to 0.02584 with an average value of 0.01493 in SSC regions), while the values were from 0.0000 to 0.00765 with a mean value of 0.00229 in the IR regions (Supplementary data Fig. S3). Ten genes with high nucleotide diversity (>0.02) were detected, namely ndhK, ndhE, ndhA, rps16, matK, psaI, rpl22, ndhF, rpl32 and trnK-UUU.

Morphological and chromosomal characteristics

In order to exhibited the morphological and chromosomal traits reasonably and aesthetically, we combined them with the phylogenetic results and they are presented in Fig. 1. Most of the species possess an umbel inflorescence with various flower densities, while individual species have a spike inflorescence (A. spicatum) and an inconspicuous umbel inflorescence (A. mairei). In view of the ploidy levels of the species, the basic chromosome number of 37 Allium species is x = 8, and only in A. macranthum and A. ursinums is it x = 7. In addition to diploidy, many Allium species have multiple ploidies, such as A. monanthum (2–4x, x = 8), A. tuberosum (2–4x, 8x, x = 8), A. ampeloprasum (2–6x, x = 8) and A. nutans (2–10x, 13x, x = 8). Moreover, some Allium species are aneuploid, such as A. victorialis, A. nutans, A. strictum and A. schoenoprasum (Fig. 1).

Fig. 1.

Phylogenetic relationships, flower morphologies and chromosome characteristics of Allium species collected in this study. The tree was constructed by maximum parsimony (MP), maximum likelihood (ML) and Bayesian inference (BI) based on 48 shared SCGs. Support values marked above the branches follow the order MP (bootstrap values)/BS (bootstrap support)/PP (posterior probability); *Maximum support in all three analyses. Three evolutionary lineages (L1–L3) are marked with different colours: L1, pink; L2, purple; L3, red. The chromosome ploidies are marked using different coloured circles.

Characteristics of the SCG and CDS data sets and phylogenetic analysis

In order to test whether additional characters or taxa were responsible for any changes observed in resolution and support of the integrated phylogenies, the 48 SCGs and the 43 CDSs were used to perform phylogenetic analysis, respectively. Alignments of the SCG data set showed a length of 32 055 bp with 14 336 variable sites (44.72 %), and 12 025 parsimony-informative characters (PICs; 37.51 %). The CDS data set possesses 30 093 characters with 13 276 variable sites (44.12 %) and 11 152 PICs (37.06 %).

Bayesian inference, MP and ML analyses of the SCG and CDS data sets shared between the 164 plastomes (four species from family Chloranthaceae were set as outgroups: Sarcandra glabra, Chloranthus japonicus, C. erectus and C. spicatus) generated almost identical topologies with generally high bootstrap support and posterior probability (Fig. 2; Supplementary data Fig. S4). Monophyly of each family was strongly confirmed, and Narcissus poeticus and Agapanthus coddii showed a close relationship with the genus Allium (Figs 1 and 2). The family Asparagaceae was identified as being closest to the Amaryllidaceae, and the Acoraceae was located at the stem of the phylogeny followed by the Tofieldiaceae and the Araceae. Following the studies of Friesen et al. (2006) and Li et al. (2010a), the Allium species were divided into three lineages (L1–L3; Figs 1 and 2; Supplementary data Fig. S4): L1 was composed of A. monanthum, A. macranthum, A. paradoxum and A. ursinum; L2 consisted of A. nanodes, A. prattii, A. victorialis, A. macleanii, A. fetisowii and A. neriniflorum; and L3 was formed by the remaining 29 Allium species. The flower morphologies of 39 Allium species and the chromosome characteristics of all collected Allium species are shown in Fig. 1 and Supplementary data Table S5.

Fig. 2.

Phylogenetic relationships inferred from 164 species based on 48 shared SCGs. Tree constructed by maximum parsimony (MP), maximum likelihood (ML) and Bayesian inference (BI) based on 48 shared SCGs. Support values marked above the branches follow the order MP (bootstrap values)/BS (bootstrap support)/PP (posterior probability); red circles in nodes represent maximum support in all three analyses. Accessions from different families are written using different colours and labelled at the top of the tree. *Species sequenced by our laboratory research group.

Estimation of divergence time

Divergence time analyses based on fossils indicated that the crown groups of monocots arose 148.989 Mya (95 % HPD: 108.006–166.103 Mya; Fig. 3; Table 2; Supplementary data Fig. S5). In addition, the divergence time of the Amaryllidaceae was estimated as approx. 49.399 Mya (95 % HPD: 48.165–50.588 Mya). The genus Allium originated approx. 41.932 Mya (95 % HPD: 34.549–47.605 Mya) and then diverged into three lineages (L1–L3; Fig. 3, Table 2). The L1 lineage separated around 22.184 Mya (95 % HPD: 15.232–34.812 Mya) and the divergence time of L2 and L3 was approx. 19.655 Mya (95 % HPD: 13.491–31.145 Mya). From the results, we found that most of the Allium species differentiated in the late Tertiary to middle Miocene (approx. 10–15 Mya).

Fig. 3.

Divergence time estimation based on 48 shared SCGs. (A) The maximum credibility tree from the divergence times estimated with BEAST. The 95 % highest posterior density (HPD) estimates for each well-supported clade are represented by bars, and white triangles with a black outline represent compressed clades. Letters (A–C) in a black circle represent fossil calibration points (see the Materials and Methods). The node ages are given for each node. (B) Phylogeny of three evolutionary lineages of Allium; L1–L3: the first to the third evolutionary lineages, the origin time of Allium and three evolutionary lineages are shown with arrows, and red background represents the main periods of time during which Allium species differentiated.

Table 2.

Age estimates for Allium and allied families based on the combined single-copy gene data sets

Number	Median (Ma)	95 % HPD		Node labels
		Lower (Mya)	Upper (Mya)
1	49.399	48.165	50.588	Asparagaceae
2	49.399	48.165	50.588	Amaryllidaceae
3	41.932	34.549	47.605	Allium
4	22.184	15.232	34.812	Allium (Line 1)
5	19.655	13.491	31.145	Allium (Line 2–Line 3)
6	54.225	50.408	60.443	Asphodelaceae
7	62.724	58.043	67.976	Iridaceae
8	74.633	70.766	78.327	Orchidaceae
9	64.151	34.63	69.312	Poaceae
10	64.151	34.63	69.312	Typhaceae
11	78.509	76.553	81.051	Zingiberaceae
12	70.428	44.204	89.435	Liliaceae
13	46.541	34.959	65.071	Colchicaceae
14	46.541	34.959	65.071	Alstroemeriaceae
15	81.281	67.609	98.668	Campynemataceae
16	53.764	39.71	71.134	Stemonaceae
17	53.764	39.71	71.134	Cyclanthaceae
18	101.205	67.051	124.253	Nartheciaceae
19	110.645	82.94	131.2	Araceae
20	110.645	82.94	131.2	Tofieldiaceae
21	148.989	108.006	166.103	Acoraceae
22	162.446	120.369	174.875	Chloranthaceae

Selective pressure analysis

A total of 43 CDSs that were shared by 164 taxa were eventually used for positive selection analysis, and the branch site model in CODEML was independently performed for Allium species and allied families. For the Allium lineage, ten genes with at least one positively selected site were detected from the BEB test, and seven of them showed a significant P-value (P < 0.05). For family Amaryllidaceae, 27 genes were found with positively selected sites and eight of them have significant P-values. Among the 19 families, the Orchidaceae possessed the maximum number of genes (i.e. 36 genes) that had sites under positive selection, and nine of them had significant P-values. The Typhaceae possessed the minimum number of genes (seven genes) that have positively selected sites, but none of them had a significant P-value. The detailed results of positive selective analysis are shown in Fig. 4A and Supplementary data Table S6.

Fig. 4.

Frequency of genes that possessed positively selected sites and the curve of the average Ka, Ks and Ka/Ks values of each gene. (A) The frequency of genes with positively selected sites in each family. Each grey square indicates the positively selected gene which occurs in the corresponding family, and black squares indicate genes whose positively selected sites are significant (P < 0.05. (B) The curve of the average Ka, Ks and Ka/Ks values of each gene; different colours of the bars are consistent with the curve colour of Ka, Ks and Ka/Ks.

In terms of the gene (owning positively selected sites) frequency that was detected in genus Allium and the 19 families, we found that atpA and psbC possessed the highest frequency and appeared in 17 families, following by rpl11 (15 families), rpoC2 (14 families), petA (14 families), psaB (12 families) and atpI (12 families). The detailed gene frequencies are listed in Fig. 4A and Supplementary data Table S6.

The average Ka, Ks and Ka/Ks values of the 43 CDSs were calculated. The average Ka value of psaA (0.38991) was highest, following by rpoC2 (0.38348), rpoC1 (0.3581), atpA (0.3254), atpB (0.172), rpoA (0.15243), rps11 (0.13878) and matK (0.13379). psaA (0.54487) possessed the highest average value of Ks, followed by petN (0.53934), psbI (0.5232), rpoC1 (0.46461) and rpoC2 (0.4294). The average Ka/Ks ratio was highest in psbC (6.31579), followed by rps11 (2.03818), psaI (1.27620), rpoC2 (0.89306) atpA (0.78090) and rpoC1 (0.77075) (Fig. 4B; Supplementary data Table S7).

DISCUSSION

The plastome variation of the genus Allium

Recently, cp genomes have been used to evaluate the genetic variation in genera or families (e.g. Myrtaceae, Corylus, Podophylloideae and Apodanthaceae) (Bayly et al., 2013; Bellot and Renner, 2016; Ye et al., 2018; Yang et al., 2018). The cp genome size, gene order and structure of the 22 Allium species were similar to those reported in previous Allium plastid genomes (Table 1), with sizes ranging from 145 to 160 kb (Kim and Yoon, 2010; Lee et al., 2017; Jin et al., 2018; Yang et al., 2019). Additionally, the GC contents of the Allium species varied from 36.7 to 37.8 %, and the GC contents in non-coding intergenic regions were much lower than those in coding regions, which is similar to the case in most land plants (Bock, 2007). The overall cp genome assemblies obtained from 39 Allium species indicated that there is a high sequence similarity across the Allium cp genomes (Supplementary data Fig. S2), which also suggested that the Allium cp genomes are relatively well conserved. In addition, the sequence variations were more conserved in the IR regions than in the LSC and SSC regions, and similar results have been found in most angiosperms (Khakhlova and Bock, 2006; Zhang et al., 2016a; Wu et al., 2018). Ten genes (rpl32, trnK-UUU, ndhF, rpl22, psaI, matK, rps16, ndhA, ndhE and ndhK) with nucleotide diversity of >0.0200 were detected (Supplementary data Fig. S3). Among them, ndhF, rpl22, matK, rps16 and ndhE have been reported as highly variable regions in many plants (Fu et al., 2017; Fan et al., 2018; Wu et al., 2018; Ye et al., 2018). These genes with high nucleotide diversity may be good sources for interspecies phylogenetic analysis in the future.

Phylogenetics of Allium

Appropriate and multiple gene combination is one of the most important determinants of accurate phylogenetic estimation. The nuclear ribosomal DNA genes [e.g. ITS and external transcribed spacer (ETS)] and many cpDNA fragments (e.g. matK, rps16 and trnL–trnF) have been used to infer the phylogeny of Allium (Friesen et al., 2006; Li et al., 2010a, 2016b; Huang et al., 2014; Herden et al., 2016; Li et al., 2016a; Hauenschild et al., 2017). Li et al. (2010a) reconstructed the phylogeny of Allium based on ITS and rps16; however, the phylogenetic relationships reconstructed by ITS showed weaker support, especially for the third evolutionary lineage, and limited rps16 sequences were included in that study. Here, our plastome phylogenomic analysis of the monocots, based on the shared SCGs and CDSs, provided strong support for the monophyly of Allium and Amaryllidaceae (Figs 1 and 2; Supplementary data Fig. S4), in agreement with previous molecular evidence (Friesen et al., 2006; Li et al., 2010a). This phylogeny also confirmed the three evolutionary lineags of Allium (L1, L2 and L3) that were provided by Friesen et al. (2006), but detected new species relationships within the third evolutionary lineage with high support values. The phylogenetic placements of some species (such as A. caeruleum, A. schoenoprasoides and A. platyspathum), however, differed from the results that were found in the study of Li et al. (2010a), and a similar phenomenon was also revealed in the species A. forrestii, A. strictum and A. obliquum. Notably, in previous phylogenetic analyses, some species of Allium formed weakly supported clades (bootstrap support/posterior probability <50 %), particularly in the third evolutionary lineage (L3) (Friesen et al., 2006; Li et al., 2010a), which was also consistent with the study of Hauenschild et al. (2017), who suggested that some of the subgenera in the third evolutionary clade are not monophyletic. Allium species possess greatly varied morphologies, and, of them, flower changes are most conspicuous (Fig. 1). Previous studies suggested that the diversity of flower form can be attributed to a range of evolutionary novelties that change the appearance of the flower in ways that influence their perceptions by animal pollinators (Darwin, 1859; Moyroud and Glover, 2017). However, changes in flower morphologies are not always driven by pollinators (e.g. Polemonium viscosum) (Galen and Butchart, 2003), which may hint that interspecies relationships are not always consistent with morphologies, such as A. cyathophorum and A. spicatum; the inflorescence of the former is an umbel, and that of the latter is a spike, but they show close a relationship in phylogeny analysis (Figs 1 and 2) (Friesen et al., 2000). All these results may indicate that species relationships of Allium are complex. Although we detected some new species relationships and obtained high support for each branch in this study, relationships among species of Allium are still not well resolved (especially for species in the third evolutionary lineage), and more extensive geographic and genomic sampling for further resolution is required in the future. Here, we conducted the first cp genome analysis on the whole of the Allium genus; we hope this study can contribute to the further analysis on Allium for other researchers.

Divergence time analysis

We estimated that the Amaryllidaceae origin occurred 49.399 Mya (95 % HPD 48.165–50.588 Mya) and the genus Allium originated during the late Eocene [41.932 Mya (95 % HPD: 34.549–47.605 Mya)] (Fig. 3; Table 2; Supplementary data Fig. S5), which was roughly the same time scale as in the study of Li et al. (2016b), who estimated that Allium originated during the late Eocene [34.26 Mya (95 % HPD: 24.29–45.76 Ma)]. This time was also consistent with the possible fossil time of Allium or Amaryllidaceae (49.42 ± 0.54 Mya) that was found in Washington (Pigg et al., 2018). Although the divergence time of many subgenera in Allium has been estimated (Herden et al., 2016; Li et al., 2016b; Xie et al., 2018a), hitherto no study has been conducted on the origin times of the three evolutionary lineages. In this study, we first detected the divergence times of three evolutionary lineages, which are in the early Eocene to the middle Miocene, and most of the Allium species differentiated in the late Tertiary to middle Miocene (approx. 10–15 Ma). Many species divergence or speciation events occurred during this time (Li et al., 2010b; Xu et al., 2010; Zhang and Fritsch, 2010; Gao et al., 2013; Qin et al., 2013; Yu et al., 2014; Zhang et al., 2015) due to factors such as orogeny and climatic oscillations; this may suggest that these factors also play important roles in species divergence of Allium during that time. Moreover, our study suggests that crown group monocots arose in the lower Cretaceous [148.989 Mya (95 % HPD: 108.006–166.103 Mya)]; this estimate is generally congruent with several estimates from other relaxed molecular clock analyses, for example 156 Mya (139–167 Mya) (Smith et al., 2010), 161 Mya (141–176 Mya) (Foster et al., 2017) and 154 Mya (131–184 Mya) (Li et al., 2019), although younger estimates such as 134 Mya (125–145 Mya) (Magallón et al., 2013) and 138 Mya (127–149 Mya) (Zeng et al., 2014) have also been suggested. Despite possible limitations, this analysis provided new insights into the divergence and origin of the genus Allium and other allied families. The detected divergence time of Allium and three evolutionary lineages (L1–L3) may contribute to future studies on subgenera or sections of the genus Allium.

Evolution and positive selection of Allium

Adaptative evolution is a process enabling an organism to fit its habitat better by means of natural selection (Lan et al., 2017). By positive selection analysis, we detected seven significant PSGs (P < 0.05) in the Allium lineage, which is just lower than in the Orchidaceae (nine PSGs) (Fig. 4A; Supplementary data Table S6). As we know, the Orchidaceae have successfully colonized almost every habitat on Earth; the number of its species accounts for approx 10 % of flowering plant species and it has been regarded as a taxon which is highly evolutionary and with great adaptability (Roberts and Dixon, 2008; Givnish et al., 2015). In consideration of the numerous PSGs found in the Allium lineage, its wide species distributions (De Wilde-Duyfjes, 1976; Choi and Oh, 2011; Govaerts et al., 2016), diversified habitat and morphologies (Fig. 1) (Fritsch and Friesen, 2002; Friesen et al., 2006; Block, 2010) and complex chromosome characteristics (Friesen, 1992; Xu et al., 1998; Zhou et al., 2007; Zhang et al., 2009; Jones, 2012; Li et al., 2017; Peruzzi et al., 2017), it may be suggested that Allium is also a successful taxon (the same as the Orchidaceae) in evolution and adaptation, and underwent strong positive natural selection.

By analysing the functions of these PSGs, we found that three significant PSGs (rpoA, rpoB and rpoC2) are associated with RNA polymerase, one gene rpl20 is associated with the large subunit of ribosomal proteins (LSU), and another three genes, petD, rbcL and matK, are related to the subunits of the cytochrome b/f complex, the large subunit of Rubisco and maturase, respectively (Fig. 4A; Supplementary data Table S4). A previous study revealed that RNA polymerase can control the process of gene transcription and affect the pattern of gene expression, thereby allowing species to adapt to a changing environment, and maintain basic metabolic processes necessary for survival (Ishihama, 2000). Furthermore, petD and rbcL are essential in the electron transport chain for generation of ATP and play important roles in plant photosynthesis (Weiss et al., 1991; Allahverdiyeva et al., 2005; Cramer et al., 2011; Xiao et al., 2012). rpl20 is involved in translation, which is an important part of protein synthesis (Krause, 1995), and the matK gene encodes a maturase that is involved in splicing type II introns from RNA transcripts and has been recommended frequently in phylogenetics and barcoding (Hilu, 2000; Hilu et al., 2003; Dunning and Savolainen, 2010). Most of the genes mentioned above have been reported to be under positive selection in previous studies (Dong et al., 2013; Carbonell-Caballero et al., 2015; Ivanova et al., 2017; Xie et al., 2018b; Ye et al., 2018). We also found that most of the genes in Allium with non-significant (P > 0.05) positively selected sites are associated with photosynthesis (e.g. psbC, psbE, petG, petL and atpB) and self-replication (e.g. rps11, rps14 and rps18), which are extremely important processes for plant growth and development (Bryant and Frigaard, 2006; Ewaschuk and Turney, 2006). Therefore, all these genes with positively selected sites may have played key roles in the adaptation of Allium species during the evolutionary process.

Additionally, we found that most of the genes with higher frequencies in these 19 families tend to possess higher average Ka/Ks values (Fig. 4; Supplementary data Table S7). For instance, psbC is present in 17 families and has the highest value (Ka/Ks = 6.31579), followed by rps11, which appeared in 15 families and had the second highest Ka/Ks value (2.03818). Such genes were also detected in the Allium lineage (e.g. psbC, rps11 and rpoC2). A previous study suggested that adaptive evolution may preferentially occur at the molecular level, expressed by an increased value of Ka/Ks (Bakewell et al., 2007). Many studies have confirmed that the higher the Ka/Ks ratio, the stronger the positive selection that species underwent (Hurst, 2002; Fay and Wu, 2003; Ai et al., 2015; Yang et al., 2015). Thus, those genes with high a Ka/Ks rate may play important roles in the adaptation and evolution of Allium.

Moreover, what needs to be pointed out are the chromosome characteristics of Allium species. It has been generalized that chromosome characteristics (chromosome number, ploidy level, karyotype asymmetry, etc.) are crucial to investigate the species relationships and evolution (Sharma and Sharma, 2014). According to previous studies (see the statistical results of Fig. 1 and Supplementary data Table S5), polyploidization is a common karyological feature for Allium species (Ohri et al., 1998; Jones, 2012; Peruzzi et al., 2017). Wu et al. (2010) suggested that the tetraploids of A. przewalskianum arose independently from diploids at least eight times, and that those in A. mairei arose at least three times (Yang, 2010). It has been considered that polyploids possess many advantages compared with their diploid progenitors in morphological, physiological, and life history characteristics, and rates of adaptation (Ramsey and Schemske, 2002; Mayrose et al., 2010). In addition, it has been reported that about 20 out of 97 Allium species with B chromosomes are polyploids (Vujošević et al., 2013), and the B chromosomes were considered to play important roles for species to adapt to harsh environments (e.g. cold, drought or high elevation) (Plowman and Bougourd, 1994, Chen et al., 2005, Wu et al., 2010). Hong (1990) considered that the growth habit and breeding system are the main factors that influence the polyploid frequency of a species. However, a reduction in fertility or even infertility may have occurred when changes in chromosome number generated aneuploids or odd chromosomes (e.g. triploid and pentaploid) (Hong, 1990). Species in Allium are perennial and bulbiferous herbs, and sometimes possess well-developed rhizomes, which can combine asexual and sexual propagation very well (Xu and Kamelin, 2000), and thereby overcome the disadvantages from their chromosomal abnormalities (Zhang et al., 2009). Therefore, the polyploidization, B chromosome and unique breeding system are very important for Allium species to adapt to various environments in the evolutionary processes.

SUPPLEMENTARY DATA

Supplementary data are available online at https://academic.oup.com/aob and consist of the following.

Figure S1: gene maps of the Allium species’ chloroplast genomes.

Figure S2: alignment of 39 Allium chloroplast genome sequences.

Figure S3: the nucleotide diversity of genes in the whole chloroplast genomes of the Allium species.

Figure S4: phylogenetic tree reconstruction based on the 43 shared CDSs of the 164 species.

Figure S5: detailed results of divergence time based on 48 shared SCGs.

Table S1: information for sample collection.

Table S2: the GenBank accessions of all 164 taxa cp genome sequences used this study.

Table S3: primers used for gap closure in this study.

Table S4: gene content in Allium species’ cp genomes.

Table S5: chromosome data statistics of Allium species collected in this study.

Table S6: the potential positive selection test on Allium and allied families.

Table S7: the average values of Ka, Ks and Ka/Ks for each gene.

FUNDING

This work was supported by the National Natural Science Foundation of China (grant nos 31872647, 31570198 and 31700176), and the Chinese Ministry of Science and Technology through the ‘National Science and Technology Infrastructure Platform’ project (grant no. 2005DKA21403-JK) and the fourth national survey of traditional Chinese medicine resources (grant no. 2019PC002).