Introduction

Citrus is one of the most cultivated fruit-crops all around the word (Singh 2002) because of its economical and agronomical values, nutritional properties and health benefits. The exact number of natural species and the taxonomic relationships within the Citrus genus is unclear due to a very long history of domestication that has included frequent hybridizations and mutations among the different species and cultivars. Several classifications of Citrus have been reported due to the enormous difficulty in characterizing the high diversity of this genus. For instance, Swingle (1943) identified 10 species of Citrus whereas Tanaka (1954) includes a minimum of 147. Other alternatives have also been presented (Scora 1988; Barrett and Rhodes 1976).

DNA molecular markers are widely used as powerful tools for the discrimination of crops and plants with different purposes (Weising et al. 2005; Garcia-Lor et al. 2013). Thus, single nucleotide polymorphism (SNP) and to a lesser extend, smalls insertions and deletions (InDels) (Primmer et al. 2002), are regularly used to detect polymorphisms (Bielsa et al. 2014) and genotype/phenotype associations (Xing et al. 2005) to identify cultivars (Mammadov et al. 2012; Ramadugu et al. 2015) to construct linkage maps (Ling et al. 2000) and to establish phylogenetic relationships (Whitkus et al. 1994).

In Citrus, a high number of molecular markers have been developed for the identification of cultivars (Coletta Filho et al. 1998; Garcia-Lor et al. 2013) and for the elucidation of their phylogenetic relationships (Chen and Gmitter 2013; Wu et al. 2014; Carbonell-Caballero et al. 2015; Oueslati et al. 2016; Shimizu et al. 2016; Fujii et al. 2016; Yu et al. 2017; Wu et al. 2018). These previous studies are a show of the high genetic diversity that involves the Citrus genus.

Markers derived from the chloroplast genome have been used to study the phylogeny of Aurantioideae subfamily, including Citrus (Penjor et al. 2013; Oueslati et al. 2016). The identification of Citrus species (Mahadani and Ghosh Sankar 2014) and hybrids (Zhen-Hua et al. 2011) has also been approached through the analysis of polymorphic fragments of the Citrus chloroplast. The results of these studies have been applied to detect mandarin in commercial orange juice (Aldeguer et al. 2014; Pardo 2015). However, the most comprehensive and detailed study on the evolution and variability of the genus Citrus has been recently presented (Carbonell-Caballero et al. 2015; Wu et al. 2018). This report is based on the phylogenetic analysis of the whole DNA sequence of the chloroplast genomes of 34 Citrus genotypes and utilized the Sanger sequence of the Citrus sinensis (L.) Osbeck genome chloroplast (Bausher et al. 2006) as the reference genome for the comparisons. The phylogenetic tree derived from this study displayed three major clusters that further diversified into different subclades. The first cluster split into citrons, while mandarins and pummelos dominated the other two clades. The mandarin cluster separated in two sub-clusters the more “recent mandarins” from the “traditional mandarins”. The pummelo cluster grouped several species that separated in three sub-clusters: pummelos and grapefruits, sour orange and lemons and sweet oranges (Carbonell-Caballero et al. 2015). In this study, specific molecular markers for the two main sub-cluster of mandarin and the three sub-clusters of pummelo were selected. The amplification of fragments that include each chloroplast molecular markers displayed new genetic variants, evidence of the high diversity that are still unknown on germplasm bank.

The availability of the nuclear genome sequence of these Citrus varieties and the study of their comparisons has also demonstrated that most horticultural Citrus groups are actually formed by genetic admixtures of a few “true Citrus species” (Wu et al. 2014; Wu et al. 2018), an idea that has recurrently appeared in the literature in many reports (Scora 1975; Barrett and Rhodes 1976; Luro et al. 1995; Nicolosi et al. 2000; Barkley et al. 2006; Uzun et al. 2009; Jena et al. 2009).

The conclusions derived from these studies provide a plausible explanation to the high diversity contained within the genus Citrus, as observed in many Citrus germplasm banks. Accessions in these banks have been classified mostly by botanical or agronomical traits, as usual during the pre-genome sequencing era. Currently, it is well accepted that the use of molecular markers can help to understand complex phylogenetic relationships and, in this way, to assist in the proper classification and re-formulation of plant genera. Recently a few Citrus ancestors have been reported and the admixture of them drafted the actuals lemon, oranges and mandarin varieties (Wu et al. 2018). Our results suggested a bigger diversity on pummelo group than has been reported before.

In the present study, we investigate the cpDNA (chloroplast DNA) haplotypes of 24 Algerian accessions from the ITAFV (Institut Technique de l′Arboriculture Fruitière et de la Vigne, Algeria) germplasm bank and elucidated their phylogenetic relationships in base on previous reports (Carbonell-Caballero et al. 2015) and new genetic variants detected on this study. The accessions were positioned on the phylogenetic tree of the genus Citrus reported in Carbonell-Caballero et al. (2015). The allocation of the Algerian accessions was additionally supported with the inclusion into the examinations of a set of five Spanish accessions from the IVIA (Instituto Valenciano de Investigaciones Agrarias, Moncada, Valencia) germplasm bank.

Materials and methods

Plant material

For PCR analysis, 29 Citrus cultivars were used. Analyses were performed from fresh, young leaves of twenty-four Algerian accessions obtained from the ITAFV (Institut Technique de l′Arboriculture Fruitière et de la Vigne, Algeria) and of five Spanish Citrus cultivars collected from the IVIA (Instituto Valenciano de Investigaciones Agrarias, Moncada, Spain) (Table 1).

Table 1 Citrus varieties and accessions used in this study
Full size table

For the “in silico” analyses 27 genotypes from ENA database were used (Table 1).

Mapping, variant calling, and selection of variants

Chloroplast sequences of 27 Citrus genotypes were download from ENA database (identifier ERP005411). Mapping and variant calling were performed as in Carbonell-Caballero et al. (2015). Base on Carbonell-caballero et al. (2015), 23 Algerian and 5 Spanish accessions were a priori classified on mandarin and pummelo chloroplast haplotype. The Algerian accession of Citrus lumia Risso et Poit. hasn’t been included on previous studies. From the total genomic variants reported previously, 23 diagnostic molecular markers for mandarin and pummelo type were selected (Supplemental Table 1) and 13 of them were checked by PCR and used for further analysis (Table 2-A, Supplemental Table 1).

Table 2 Position, variant type: single nucleotide polymorphism (SNP) or small deletion or insertion (INDEL), Gene effect (I = Intergenic, N = Non-synonymous, S = Synonymous, T = intron), gene description and gene identifier (Gene ID) of 37 molecular markers used to infer the phylogenetic relationship of the Algerian and Spanish Citrus accessions
Full size table

Design of primers

Thirteen pairs of primers (Table 3) were designed to study chloroplast genetic diversity among the Citrus accessions. Chloroplast genome sequences of Citrus sinensis (L.) Osbeck were used as reference (Bausher et al. 2006). Primer design of was performed with the Primer 3 online program (Untergasser et al. 2012; Koressaar and Remm 2007). Primers conditions were design with the minimum, optimum and maximum for a primer oligo length of 19, 20, 21; melting temperature (Celsius) of 59, 60, 61 and GC% of 45, 50, 55. Product size range was 150–1000 base of pairs.

Table 3 Molecular markers used to infer the phylogenetic relationship of Algerian Citrus accessions
Full size table

Molecular markers

Based on the variants detected “in silico” on previous study (Carbonell-Caballero et al. 2015), 13 genomic variants, 10 SNPs and 3 INDELs, discriminating among different haplotypes of pummelo and mandarin chloroplasts were selected (Table 2-A, Supplemental Table 1). Thus, molecular markers mmp-1 and mmp-2, SNPs at positions 15439 and 83821, respectively, identified Citrus maxima (Burm.) Merr. and Citrus paradisi Macfad. Molecular markers mmo-1 and mmo-2, SNPs at position 38215 and 44357, respectively, identified Citrus sinensis (L.) Osbeck. Molecular markers mml-1 and mml-2, a 12 pb deletion starting at position 17705 and a SNP at position 55629, respectively, differentiated Citrus aurantium L. and Citrus limon (L.) Burm. f. Molecular markers mmm1-1, mmm1-2, mmm1-3, mmm1-4 and mmm1-5, indels at positions 13080 and 70549 and SNPs at positions 9471, 71899 and 117149, respectively, identified mandarins of the Citrus clementina and Citrus reticulata types (Citrus clementina Hort. ex Tan., Citrus deliciosa Ten., Citrus reticulata Blanco, Citrus unshiu (Mak.) Marc., Citrus tangerina Hort. ex Tan., Citrus nobilis (Lour.). Molecular markers mmm2-1 and mmm2-2, SNPs at 77108 and 116207, respectively, identified mandarins of the Citrus reshni Hort. ex Tan., Citrus sunki (Hayata) Hort. ex Tan. and Citrus limonia Osbeck. Further study of the sequences surrounding these 13 fragments rendered 24 additional variants to complete a set of diagnostic variants composed of 37 molecular markers (Table 2). For “in silico” genotypes, the new 24 additional variants were genotyped through the visualization of their accessions sequences using the integrative genome viewer (IGV) (Robinson et al. 2011; Thorvaldsdóttir et al. 2013) software, using the Citrus sinensis (L.) Osbeck chloroplast genome (Bausher et al. 2006) as the genome of reference.

DNA extraction, rt-PCR and sequencing

DNA from fresh leaves of Citrus accessions was extracted with the CTAB method (Doyle and Doyle 1990) with minor modifications. About 100 mg of plant material were ground using liquid nitrogen. Eight hundred μL of preheated extraction buffer containing CTAB 2% and 2µL β-mercaptoethanol were incubated for 15 min at 65 °C and shaken three times during incubation. Thereafter, one volume of chloroform:isoamylalcohol (24:1) was added. The resulting mixture was then placed in a centrifuge for 15 min at 13,000 rpm (4 °C). One volume of isopropanol was added to the aqueous phase and the solution was stored at − 20 °C for four hours. The mixture was centrifuged for 15 min at 13,000 rpm and the isopropanol was discarded. DNA was cleaned adding 400 μL of ethanol (70%) and centrifuged for 5 min. The DNA was dried at 37 °C and then 65 μL of 1 × TE buffer and 1 μL of RNAase were added before incubation at 37 °C for 30 min. DNA was stored at − 20 °C. Finally, the concentration and quality of extracted DNA were estimated on a Nanodrop 1000 spectrophotometer.

PCR analyses were determined through Real-time quantitative PCR, on a LightCycler2.0 instrument (Roche) using the LightCyclerFastStart DNA MasterPLUSSYBRGreen I kit (Roche) essentially as described in Terol et al. (2015). Each individual PCR reaction contained 2 ng of genomic DNA extract. Cycling protocol consisted of 10 min at 95 °C for pre-incubation followed for 45 cycles of 10 s for denaturation, 10 s at 60 °C for annealing and 20 s at 72 °C for extension. Specificity of the PCR reaction was assessed by the presence of a single peak in the dissociation curve after amplification and through size estimation of the amplified product.

Amplification PCR’s product was purified using E.Z.N.A® Cycle-Pure Kit. The purified product was sequenced by Sanger method.

Aligment of sequences

The alignment of the product PCR’s Sanger sequences was carried out with the BioEdit (https://www.mbio.ncsu.edu/BioEdit/bioedit.html) software and manually curated.

Phylogenetic analysis

For the phylogenetic analysis, 56 Citrus varieties, the 24 Algerian and 5 Spanish accessions described above and the 27 genotypes also listed above from Carbonell-Caballero et al. (2015) were used. The tree was rooted with Severinia buxifolia (Poir.) Ten. The phylogenetic analysis was performed with the MEGA7 software (Kumar et al. 2016). The Jakes-cantor model (Jukes and Cantor 1969) was found to optimal by the program MEGA7 to calculate dissimilarity between samples (Supplemental Table 2). The phylogenetic tree was performed using Maximum-likelihood iterative model and the bootstrap of 1000 repetitions was used to assess the reliability of the phylogeny reconstructed. A total of 37 variants including 34 SNPs and 3 Indels were used for the analysis. The length of the indels were considered. A total of 48 variants in a total high-quality Sanger fragment of 2873 obtained by PCR amplification were used.

Results

Chloroplast molecular markers

In a previous study, Carbonell-Caballero et al. (2015) reported the chloroplast genome sequences of several horticultural Citrus groups. Base on their results, three different haplotypes of pummelo: one for orange group (C. sinensis (L.) Osbeck), another for lemon group (C. limon (L.) Burm.f. and C. aurantium L.) and a third type for pummelo and grapefruit group (C. maxima (Burm.f.) Merr. and C. paradisi Macfad.) and two different haplotypes of mandarin: one for ancestral mandarin (C. limonia Osbeck, C. reshni Hort. ex Tan. and C. sunki (Hayata) Hort. ex Tan.) and one for modern mandarin (C. clementina Hort. ex Tan., C. deliciosa Ten., C. reticulata Blanco, C. unshiu Marc., C. tangerina Hort. ex Tan. and C. nobilis Lour.) were stablished according the number of variants. The 24 Algerian and 5 Spanish accessions (Table 1) belong “a priori” to the mandarin or pummelo type. Based on the variants detected “in silico” in the previous work (Carbonell-Caballero et al. 2015), 23 diagnostic variants for the three pummelo and two mandarin haplotypes were selected as molecular markers (Supplemental Table 1) and 13 of them were checked by PCR and used for the current work (Table 3). PCR amplification, alignment and detailed study of sequences containing these 13 variants in the 29 accessions under study (Fig. 1), revealed a set of 24 additional variants that were probably filtered or discarded on the variant calling process. Thus, the complete set of diagnostic variants used in the analyses of the Algerian and Spanish Citrus accessions was composed of total of 37 molecular markers (Fig. 1).

Fig. 1
figure 1

Alignment of 37 PCR products sequenced by Sanger method of 29 varieties: 24 Algerian and 5 Spanish accessions. Variant positions according to the Citrus sinensis (L.) Osbeck chloroplast genome (Bausher et al. 2006) are pointed

Full size image

The PCR products of the 13 pair of primers (Table 3) provided a total of 2873 pb sequence with high quality. Twenty-six molecular markers were located in intergenic regions, 3 in introns, 5 involved synonymous substitutions while only 3 were non-synonymous (Table 2).

The PCR marker analyses performed on the 24 Algerian and 5 Spanish varieties identified two kinds of mandarin chloroplasts, MA1, characterized by 22 diagnostic variants, distributed between 20 SNPs and 2 Indels and MA2, characterized by 17 diagnostic SNPs (Table 1, Fig. 2). These analyses also discriminated 4 types of pummelo chloroplasts, PU1, the reference chloroplast (Bausher et al. 2006); PU2, characterized by 6 diagnostic SNPs and 1 indel; PU3, characterized by 7 diagnostic SNPs; PU4, characterized by 4 diagnostic SNPs and a new chloroplast haplotype for the accessions Pomme d’Adamm LUM with 15 diagnostic SNPs and 1 indel (Table 1, Fig. 2).

Fig. 2
figure 2

Distribution of variants along the 56 accessions studied: chinese box orange (SVB: Severinia buxifolia (Poir.) Ten.), mandarins (MA1: Citrus sunki (Hayata) Hort. ex Tan., Citrus limonia Osbeck and Citrus reshni Hort. ex Tan.; MA2: Citrus nobilis Lour., Citrus reticulata Blanco, Citrus unshiu Marc., Citrus tangerina Hort. ex Tan., Citrus deliciosa Ten. and Citrus clementina Hort. ex Tan.), mangshan (MSM: Citrus mangshanensis S. W. He et G. F. Liu), ichang papeda (IPA: Citrus ichangensis Swingle), fortunella (FOR: Fortunella margarita (Lour.) Swingle), citrons (CIT: Citrus medica (L.) Swingle), micrantha (CMC: Citrus micrantha Wester), pummelo (PU1: Citrus sinensis (L.) Osbeck; PU2: Citrus aurantium L., Citrus limon (L.) Burm. f. ‘Le doux d’afrique du nord’, Citrus limon (L.) Burm. f. ‘Dellys’ and Citrus limon (L.) Burm. f. ‘Eureka’; PU3: Citrus limon (L.) Burm. f. ‘Verna’, PU4: Citrus limon (L.) Burm. f. ‘Bèni Abbés’; Citrus maxima (Burm. f.) Merr. and Citrus paradisi Macfad.) and lumia (LUM: Citrus lumia Risso et Poit.). Positions of variants are depicted according the Citrus sinensis (L.) Osbeck reference genome of the chloroplast (Bausher et al. 2006). Dark grey bars indicate the occurrence of variants different from the reference while grey bars are as in the reference. The 3 indels, in bold letters, spread over one position, except 13080 indel that is a deletion in some accessions and an insertion in others and therefore is represented by two positions. Accession colors are as in Fig. 3

Full size image

The occurrence of these 37 molecular markers also was verified “in silico” using the set of variants reported in Carbonell-Caballero et al. (2015) on the varieties listed in Table 1. Chloroplast haplotypes of both sets were compared and the differences found are reported below.

MA1 and MA2 chloroplast types identified by PCR consistently contained one more SNP that those determined “in silico” (Table 1, Figs. 1, 2), strongly suggesting that the missing SNP in the “in silico” analyses was very probably filtered out or discarded and therefore it was not considering during the analyses. The PU4 chloroplast haplotype showed 4 diagnostics SNPs in Chandler pummelo, Foster and Tompson Pink grapefuits and Beni Abbass lemon when analyzed by PCR but 7 SNPs in Low acid, Guanxi and Shatian pumelos and in Marsh grapefruit, in the “in silico” analyses (Table 1, Figs. 1, 2). Since Foster and Tompson Pink are somatic mutations of Marsh grapefruit and Low acid pummelo is the parental mother of Chandler pumelo, these observations suggest that these additional 3 SNPs are probably located in a variable region and therefore are not powerful diagnostic SNPs. On the other hand, all other differences in the pummelo chloroplast haplotypes found in that comparison were apparently due to real differences in the sequence of these chloroplasts. These affect to the variability of lemons since in addition to Beni Abbass (PU4), the chloroplast of Verna lemon (PU3) shows 7 SNPs, and those of Le Doux d´Afrique du nord and Dellys (PU2) contained 6 SNPs and 1 indel (Table 1, Figs. 1, 2).

The chloroplast of the lumia, Pomme d′Adam, LUM (Table 1), was quite unusual since it contained 15 diagnostic SNPs and 1 indel, showing a new haplotype.

Phylogenetic relationship of Citrus accessions

A phylogenetic tree of 56 Citrus accessions based on 37 molecular markers was produced using the Maximum-likelihood from genetic distances Jukes-Cantor method. Variants used in the construction of the tree were obtained from PCR determinations on the 29 Algerian and Spanish accessions.

The phylogenetic tree obtained basically reproduced the tree of the genus Citrus as reported with the addition of the new Citrus accessions (Fig. 3). All Algerian accessions, except the Pomme d′Adam clustered into two main clades, one characterized by the occurrence of a mandarin chloroplast and the other of a pummelo chloroplast. Between these two major groups, the Pomme d′Adam nested together with the pummelo clade but clearly separated in a distinct monophyletic clade. The reliability and consistence of the clustering distribution of the Algerian accessions was further supported with the inclusion into the examinations of a set of 5 Spanish accessions from the IVIA germplasm bank representative of mandarins, pummelos, lemons and oranges.

Fig. 3
figure 3

Phylogenetic tree of 56 Citrus accessions based on 34 SNPs and 3 Indels. The tree was produced using the Maximum Likelihood method from genetic distances by the Jukes-Cantor method. Different chloroplast haplotypes determined by PCR (Table 2) in the Algerian and Spanish samples are represented by different colors (red, Citrus sinensis (L.) Osbeck; green Citrus limon (L.) Burm.f.; blue, Citrus maxima (Burm.f.) and Citrus paradisi Macfad.; fuchsia Citrus clementina Hort. ex Tan.; orange Citrus reticulata Blanco and pink Citrus sunki (Hayata) Hort. ex Tan.). Chloroplast haplotypes of accessions with normal black lettering was deduced from “in silico” information of available chloroplast genomes (ENA database). The phylogenetic tree was rooted with Chinese box orange (Severinia buxifolia (Poir.) Ten.) as outgroup. Bootstrap (10000) support values for nodes are displayed in the corresponding branching points of the tree

Full size image

The set of Algerian mandarins analyzed that included 2 varieties of C. reticulata and 5 varieties of C. clementina were grouped in the same clade. Although this cluster appears to be further divided in two subclades, as explained in the previous section, this appears to be a consequence of an artifact due to the filter process of the analyses presented in Carbonell-Caballero et al. (2015). This clade grouped the Algerian clementines Pourcy messerghine, Messerghine 48, Trabut Taous, Cadoux and Clone 63 (2749) and the reliability of this cluster was supported by the allocation of the Clementina Fina from the Spanish germplasm bank (Fig. 3). The clade also includes the two Algerian mandarins, Mandarine de Blida and Satsuma Saigon.

The pummelo cluster including 2 grapefruits (Citrus paradisi Macfad.), 1 pummelo (Citrus maxima (Burm.f.) Merr.), 4 lemons (Citrus limon (L.) Burm.f.) and 11 sweet oranges (Citrus sinensis (L.) Osbeck) was separated in 3 different groups. The first one clustered three lemon accessions separating the Spanish Verna lemon in one side from two Algerian lemons, Le Doux d´Afrique du nord and Dellys, that clustered together, in the other side. The widest pummelo cluster was divided in two homogenous clusters. Pummelos and grapefruits were included in one of them whereas all sweet oranges grouped in the other. Thus, the Algerian grapefruit accessions Thompson Pink and Foster were clustered together with the Chandler pummelo Spanish accession in a single clade. As explained in the previous section, the sub-cluster observed in this grapefruit-pummelo cluster may be a consequence of 3 non-diagnostic SNPs and therefore an artifact. It is well known for instance that Low acid-Chandler pummelo have a mother–child relationship (Cameron and Soost 1961, Wu et al. 2014) while the two Algerian grapefruit accessions are spontaneous mutations of the Marsh grapefruit. On the other hand, it is worth to notice that the Algerian lemon Beni Abbass shares the same chloroplast haplotype as grapefruits and pummelos (Fig. 3). The other cluster grouped the Spanish sweet orange and the Algerian ones in an independent and homogenous clade. This cluster included all kind of sweet oranges such as early (Navelina), mid-season (Navel) or late (Valencia) ones.

Discussion

In this work, 37 chloroplast markers (Table 2) have been used to establish the phylogenetic relationships of 24 Algerian and 5 Spanish Citrus accessions from ITAFV (Institut Technique de l′Arboriculture Fruitière et de la Vigne, Algeria) and IVIA (Instituto Valenciano de Investigaciones Agrarias, Spain), respectively (Table 1). The variants were obtained (Figs. 1, 2) from a previous description reporting the chloroplast genomes of 34 Citrus genotypes (Carbonell-Caballero et al. 2015). Analyses were performed by PCR. Most of the variants involved in the 37 molecular markers used in this study were preferentially located in intergenic regions while only a few genes (ATP synthase, Photosystem II, Photosystem I P700 chorophyll a, Ribosomal protein L33, NADH dehydrogenase subunit 5, RNA polymerase beta subunit, Ribosomal protein S12, Ribosomal protein S12) showed different versions (Table 2). Furthermore, molecular markers only introduced non-synonymous substitutions in three of these genes, Photosystem II (ID: 427120); Ribosomal protein L33 (ID: 4271121) and Photosystem II (ID: 4271201). This high proportion of SNPs located in intergenic region is coherent with the previous studies (Carbonell-Caballero et al. 2015; Pessoa-Filho et al. 2017) because it is accepted that most of the chloroplast genes are rather essential for plant.

The purpose of the current study was to include these 29 accessions in the phylogenetic tree of the genus Citrus (Fig. 3) determined by the “in silico” analyses of the Citrus chloroplast genome in Carbonell-Caballero et al. (2015). This Citrus phylogenetic tree is basically divided into three major clusters dominated essentially by mandarins (first cluster), pummelos (second cluster) and citrons (third cluster).

The mandarin cluster was classified into two sub-groups: one cluster was formed by “traditional” non-edible mandarins such as Sunki and Cleopatra while the other cluster included more “modern” and edible mandarins, for instance, Clementine, Satsuma, King, Willowleaf, Ponkan, Dancy, W. Murcott … etc (Fig. 3). Our data confirmed the occurrence of two mandarin chloroplast haplotypes (MA1 and MA2; Table 1) and the assignation of the Spanish accession of Sunki to group MA1. Furthermore, all Algerians mandarin accessions (Mandarine de Blida, Pourcy messerghine, Messerghine 48, Trabut Taous, Cadoux, Satsuma Saigon and Clone 63) and the Spanish “Clementina Fina” contained the MA2 chloroplast and clustered together in a second mandarin clade (Fig. 3). It is worth to mention that “Clementina Fina” that is the origin of all current Clementine clones, was imported to Spain from Algeria at the beginning of XX century (Zaragozá et al. 2011).

Our results, also confirmed that the pummelo cluster also grouped grapefruits, sour and sweet oranges and lemons in addition to pummelos (Fig. 3), indicating that C. maxima is the maternal ancestor of these important horticultural groups (Wu et al. 2014; Carbonell-Caballero et al. 2015). In addition, the lumia chloroplast appears to be a new Citrus chloroplast phylogenetically related to the chloroplasts of the pummelo group. The data obtained in the current work, therefore, suggest that there is at least four pummelo or pummelo-related chloroplast types (PU1–PU4; Table 1). In principle, this is not surprising since it is well known that C. maxima, probably one of the oldest species of the genus Citrus (Wu et al. 2014), shows a huge degree of diversity in China. For instance, Liu et al. (2017) identified a high genetic variability among 110 pummelo cultivars while more recently, Yu et al. (2017) divided a collection of 274 pummelo accessions in three subpopulations.

The chloroplast haplotype PU4 (Table 1) that characterized one of the sub-clusters of the pummelo clade, was detected in addition to the Spanish accession of pummelo Chandler, in the two Algerian grapefruits, Foster and Thompson Pink and in the lemon Bani Abbass. In this cluster nested also other pummelo varieties and grapefruits (Carbonell-Caballero et al. 2015). This observation suggests that the same maternal pummelo might give rise to two different horticultural groups, i.e. grapefruits and lemons. On the other hand, the group of sweet oranges formed a homogenous clade that contained haplotype PU1 and included all 11 Algerian varieties (Valencia late station, Mitidja navel, Washington navel case of isolation, Orange de Blida, Orange Djerid, Shamouti Station, El-Miliah, W. navel 251, Sid Alí, Hamline clone 1073, and Alger navel) and the Spanish accession of Navelina orange.

Interestingly, the 4 lemons investigated in this work were clustered in 3 different sub-clusters of the pummelo clade (Fig. 3). Aside from the lemon Bani Abbass (PU4) that nested with pummelos and grapefruits, Verna lemons exhibited a different chloroplast haplotype (PU3) than Le doux d’Afrique du nord and Dellys lemons both grouped together (PU2) in a different clade.

One of the most interesting results obtained in this study is probably linked to the phylogenetic relationships of the lumia (C. lumia) Pomme d´Adam in the Citrus tree (Fig. 3). The data indicated that this genotype cannot be classified with strong certainty in any of the Citrus groups defined so far. Actually, Pomme d´Adam contains a different chloroplast type (LUM) to those defined in Carbonell-Caballero et al. (2015) and generally accepted in the literature. Interestingly, this chloroplast appears to be more related to the pumelo or micrantha chloroplast than to any other kind of chloroplast.

The lumias are a group of varieties of unknown genetic origin that in some way resemble citrons. The lumias of the Mediterranean basin are thought to be natural hybrids showing a mixture of lemon and pummelo characters and traits (Hodgson 1967). They differ from lemons in several characteristics such as fruits of large size (Fig. 4) although as with lemons, lumias are acidic and exhibit large purple-tinged flowers and purple-tinted shoots. Our data clearly precludes citrons as parental mother of lumias.

Fig. 4
figure 4

Leaves and fruits of Dellys (Citrus limon (L.) Burm.f.) and Pomme d’Adam (Citrus lumia Risso et Poit.)

Full size image

Conclusions

Our results display 24 new variants in a fragment of the chloroplast of the Citrus accessions studied, a proof of the high diversity that are conserved on germplasm accessions. The results suggest that Algerian mandarins share the same mandarin chloroplast haplotype and the sweet oranges share the same pummelo chloroplast sequence. However, the diversity of current lemon accessions is at least partially correlated with the identity of different pummelo progenitors. It is worth to mention also that a particular pummelo progenitor identified here as PU4 (represented by Chandler pummelo) was the parental mother of the lemon Bani Abbass and of the Marsh grapefruit and therefore of Tompson Pink and Foster grapefruits. Thus, the data appear to suggest that different pummelo ancestors were the origin of current lemon varieties while a certain pummelo could have given rise to different horticultural groups such as lemons and grapefruits. These observations reinforce the idea that Citrus domestication and evolution was based on a few ancestral genotypes with restricted phenotypic diversity and specific degrees or levels of genome admixture (Wu et al. 2018).

This work has also identified a new kind of Citrus chloroplast present in the lumia Pomme d′Adam that might be associated with a new ancestral Citrus. Although not strongly related to any of the known chloroplast genomes, this haplotype was phylogenetically closer to the pummelo group of chloroplasts.