Introduction

Citrus (L.) is one of the most economically important fruit crops of the world, belonging to the subfamily Aurantioideae of the family Rutaceae. It is widely distributed throughout the tropical and subtropical regions of the world and believed to have originated in Southeast Asia, particularly northeast India, the Malayan archipelago, China, Japan, and Australia [17, 23]. India has an enormous diversity of Citrus genetic resources, both cultivated and wild. Citrus occupies the second position in terms of area (987 ha) and third position in terms of production (9.64 mt) of fruit crops in India [18].

Among the cultivated species, Citrus sinensis (L.) Osbeck (sweet orange) is the most important commercial fruit crop of Citrus and occupies the second position after mandarins in India. C. sinensis is believed to be a hybrid between pummelo (Citrus maxima) and mandarin (Citrus reticulata) [7, 20]. It is a highly polyembryonic species. Its fruit pulp is used for preparing fresh juice which is rich in vitamin-C and protein content. The peel of the fruit is used for making perfume and soaps. Cooking oil is extracted from its seeds. Juice extracted from its leaves is used to control several diseases like ulcers, sores, etc. [11].

The genetic diversity of C. sinensis is diminishing rapidly because of a number of factors, such as displacement of the natural gene pool due to selection and introduction of genotypes suitable for intensive horticulture forming a limited gene pool [2]. There is urgent need to retain the essential characters of varieties/cultivars and to characterize and evaluate the existing genotypes to achieve significant improvement in C. sinensis cultivars.

The use of molecular markers has been a valuable and precise strategy to identify Citrus species, cultivars and biotypes and to investigate the genetic diversity of Citrus species. Molecular marker techniques like RAPD, ISSR, RFLP, SSR, AFLP and other markers have been used for germplasm characterization, studies of genetic diversity, systematics and phylogenetic analysis [24]. Among them, random amplified polymorphic DNA (RAPD) markers have been employed most widely for characterization of plant species [25]. RAPD have gained more attention due to the simplicity of the procedure, the low cost and the very small amount of DNA required for analysis. In Citrus, RAPD markers have been used for cultivar identification, genetic mapping, genetic diversity assessment and other breeding programs [1, 36, 9, 14, 16, 19, 21]. In the present study, RAPD and morphological markers have been applied to characterize indigenous as well as exotic cultivars of C. sinensis and to establish genetic relationships among these cultivars.

Materials and Methods

Plant Material and Sample Collection

A total of 22 cultivars of C. sinensis were collected from field genebank collection of Regional Research Station, Punjab Agricultural University, Abohar, Punjab and used for morphological and molecular studies (Table 1). A selective sampling strategy was employed, where samples collected from a single plant of a cultivar was given an indigenous collection number (IC number) and treated as an individual accession. Leaf and fruit samples of each accession were collected for confirmation of taxonomic identity, characterization and DNA extraction. Detailed passport information of each accession was recorded in the NBPGR database.

Table 1 Citrus sinensis cultivars used for the morphological and molecular analyses

Morphological Characterization

Morphological characterization of 22 cultivars of C. sinensis was done using descriptors developed for Citrus by International Plant Genetic Resources Institute (IPGRI), Rome, Italy (now Bioversity International). Characterization data of 43 characters (29 qualitative and 14 quantitative) of leaf, fruit and seed were recorded for the collected germplasm. All the 43 morphological characters were converted into bi- and multi-state code. A pair-wise similarity matrix was generated based on simple matching coefficient method using software NTSYS ver. 2.10e [22]. A cluster analysis was performed using the unweighted pair group method with arithmetic average (UPGMA) based on simple matching coefficient in NTSYS software. Principal component analysis (PCA) was also carried out to study correlations among the variables and establish relationships among cultivars using the same software. The two-way Mantel test [15] for goodness of fit for the UPGMA cluster was also performed using the same software.

DNA Extraction

Total genomic DNA was extracted from all the 22 cultivars of C. sinensis through the cetyl tri-methyl ammonium bromide (CTAB) method [10]. Quantitation of isolated DNA was done spectrophotometrically and its quality checked by electrophoresis on 0.8 % agarose gel.

RAPD-PCR Amplification

The RAPD primers of Operon Technologies Alameda, CA, USA were used for molecular analysis. A total of 60 primers were screened in C. sinensis, of which 20 primers were selected for final profiling based on banding patterns and reproducibility. The basic protocol reported by Williams et al. [25] for RAPD-PCR amplification was followed, in which a final reaction volume of 20 μl contained 1× Assay buffer, 2.5 mM MgCl2, 0.2 mM of each dNTP, 1U Taq DNA polymerase (Life tech, India), 10 pmol RAPD primer and 20 ng of template DNA. The PCR amplification conditions were as follows: Initial denaturation step at 94 °C for 4 min followed by 40 cycles of denaturation at 94 °C for 1 min, annealing at 35 °C for 1 min, and extension at 72 °C for 2 min followed by final extension at 72 °C for 7 min. Amplification products were separated by electrophoresis (80 V for 3 h) in 1.5 % agarose gel containing ethidium bromide (10 mg/ml). A photographic record was taken under a UV gel documentation system (Mega Biosystematica, UK).

Data Analysis

Amplified fragments were scored for each accession as presence (1) or absence (0) of homologous bands on the basis of size comparison with standard DNA ladder. Molecular weight of the amplified bands was estimated by using a 1 Kb DNA ladder (Gibco BRL Life Technologies, New York, USA) as standard. A pairwise genetic similarity matrix between cultivars was estimated using Jaccard’s coefficient and a dendrogram was constructed based on UPGMA using Software NTSYS ver. 2.01e [22]. The two-way Mantel test [15] for goodness of fit for the UPGMA cluster to the binary data and PCA were also performed using the same software.

Results

Morphological Characterization

The tree is medium sized, profusely branched with elliptic to ovate leaf lamina and narrowly winged petiole. Fruit shape varies from spheroid to ellipsoid with rounded or truncated apex. Most of the fruits have pitted surface texture with either conspicuous or strongly conspicuous oil glands. C. sinensis shows mostly seeded cultivars and few cultivars like Jaffa and Delta Valencia are seedless. Seeded cultivars have 2–18 seeds with an average of five seeds per fruit. Seeds were clavate in shape with wrinkled seed surface. Mature seeds were cream to brown in colour with cream or light yellow-cream cotyledon and brown chalazal spots. Most of the C. sinensis cultivars are polyembryonic with two to four embryos. Among the studied cultivars, Vaniglia sanguigno reported the highest percentage of polyembryony (70 %) with four embryos per seed. The only cultivar with 100 % monoembryonic seeds was Temple.

Pair-wise similarity values among the cultivars of C. sinensis ranged from 0.18 to 0.64 with an average of 0.39 based on morpho-metric data. A dendrogram generated based on morpho-metric data grouped all the 22 cultivars into five major clusters (Fig. 1). The first cluster comprised of the cultivar Washington navel which was the most distinct from all other clusters and separated with similarity value of 0.30. A second cluster was comprised of two cultivars, namely Jaffa and Delta Valencia, which were closely related with similarity value of 0.52. The third cluster comprised of only one cultivar, Malta, which was separated from the fourth cluster with similarity value of 0.32. The fourth cluster was the biggest one comprising 16 cultivars, viz. Mosambi, Declarbe sweet orange, Valencia late, Parent navel, Mediterranean sweet orange, Teneriffe, Satgudi, Blood red, Temple, Sweet orange, Rhode red Valencia, Morro, Vaniglia sanguigno, Tardiff, Vanale, and Campbell Valencia. Within this cluster the cultivars Campbell Valencia and Vanale were most similar morphologically showing a similarity value of 0.64. A fifth cluster comprised of two cultivars, viz. Olinda Valencia late and Seleta, which were closely related by a similarity value of 0.44.

Fig. 1
figure 1

Dendrogram generated based on morphological traits of 22 C. sinensis cultivars using the unweighted pair group method with arithmetic average (UPGMA)

Based on Mantel Z statistics [15] the correlation coefficient (r) was estimated as 0.84. A value of 0.84 is considered a good fit of the UPGMA cluster pattern to the data. A two-dimensional plot generated from PCA showed five groups which was found to be less similar to the clustering pattern of the UPGMA dendrogram. In a 2D plot, cultivar Delta Valencia alone constituted one group whereas in UPGMA clustering, Delta Valencia and Jaffa were grouped together in one cluster. Jaffa also formed a separate group in 2D plot but in the dendrogram it was grouped along with Delta Valencia. Malta and Washington navel together constitute one group in the 2D plot, whereas in the dendrogram they were present in two different clusters. The analysis gave first ten principal components, which contributed 91.36 % of the total variability of the collected accessions. The first five principal components accounted for 70 % of the total variability and the first three accounted for 53.77 % of the variance, in which maximum variability was contributed by the first component (26.85 %) followed by the second component (14.84 %), and the third component (12.07 %). The first PC was most highly influenced by characteristics of the fruit morphology, viz. fruit weight, fruit length, fruit diameter, fruit rind thickness and TSS (Table 2). In the second PC, the traits contributing to the total variability were fruit adherence of albedo to pulp, leaf apex, leaf lamina shape and petiole wing width. The third PC was mostly influenced by fruit shape, leaf length, leaf width, and oil gland size.

Table 2 Eigenvectors of morphological variables explained by the first three principal components

Genetic Polymorphism among Cultivars

Twenty primers were selected for the RAPD analysis based on the reproducibility and banding patterns. PCR amplification of the genomic DNA isolated from 22 cultivars of C. sinensis yielded a total of 99 bands, of which 51 were polymorphic and 48 were monomorphic (Table 3). Representative gel profiles generated using primers OPC-08 and OPG-17 are shown in Fig. 2. The total number of amplified DNA bands ranged from three to eight, with an average of five bands per primer. The maximum number of polymorphic bands (5) was obtained with two primers, i.e. OPG-08 and OPA-01 and the minimum number (1) was generated with primer OPT-01. The polymorphism percentage ranged from 12.5 (primer OPT-01) to 83.33 % (OPA-01). Average polymorphism across all the 22 cultivars was 51.83 %. Overall size of the PCR amplified fragments ranged from 300 to 3,000 bp.

Table 3 Details of amplified bands generated in 22 cultivars of C. sinensis based on 20 RAPD primers
Fig. 2
figure 2

Representative gel profiles of 22 cultivars of C. sinensis based on random amplified polymorphic DNA (RAPD) primers. M represents 1 kb DNA ladder. a OPC-08. b OPG-17

Five RAPD primers gave seven unique bands in specific sweet orange cultivars. These primers produced a specific DNA fragment which distinguished one cultivar from the rest. Primer OPB-18 generated two unique bands, one in Satgudi and another in Tardiff, while primer OPC-08 also generated two unique bands each for Seleta and Temple. Each OPO-04, OPG-17 and OPA-12 primers also generated single fragments in Satgudi (500 bp), Valencia late (900 bp) and sweet orange (600 bp), respectively.

Genetic Diversity and Relationships

The pair wise similarity values obtained between 22 cultivars of C. sinensis ranged from 0.48 to 1.0. A maximum similarity value of 1.00 was observed between cultivars Declarbe sweet orange and Rhode Red Valencia, indicating that they are most genetically similar, whereas Washington naval and sweet orange, Delta Valencia and Mosambi, and Satgudi and Delta Valencia showed the lowest similarity coefficient value of 0.48. Average similarity across all the cultivars was 0.77. A dendrogram generated based on the UPGMA method grouped all the 22 cultivars into two major clusters (Fig. 3). The first cluster comprised of the cultivars Delta Valencia and sweet orange. The biggest formed cluster was the second cluster consisting of the remaining 20 cultivars. Within this cluster, the cultivars Declarbe sweet orange and Rhode red Valencia were genetically most similar, showing 100 % genetic similarity; while Seleta and Vanale cultivars were individually separated into distinct clades from the rest of the cultivars with similarity value of 0.79.

Fig. 3
figure 3

Dendrogram generated based on random amplified polymorphic DNA (RAPD) data of 22 cultivars of C. sinensis

Based on Mantel Z statistics [15], the correlation coefficient (r) was estimated as 0.93. A value of r > 0.90 is considered a very good fit of the UPGMA cluster pattern to the data. A 2D plot generated from principle component analysis (PCA) of RAPD data also supported the clustering pattern of the UPGMA dendrogram, except Seleta, which was distinctly separated in the 2D plot, while in the dendrogram, it was grouped in cluster II. First and second principal components accounted for 23.18 and 14.35 %, respectively, of the total variation.

Discussion

Experiments with C. sinensis cultivars have demonstrated the potential of RAPD markers as a rapid, reproducible and useful method for distinguishing different cultivars of C. sinensis and clustering genotypes into different groups. The moderate level of genetic polymorphism (51.83 %) was observed among the 22 cultivars of C. sinensis based on 20 primers. This could be explained by the fact that somatic mutations are the main source of variability in cultivars of this species. However, a high level of polymorphism (70.13 %) was reported within the naval sweet orange cultivar based on RAPD markers [8]. In contrast to these results, Natividade Targon et al. [19] and Novelli et al. [21] did not observe polymorphisms among the cultivars of C. sinensis based on RAPD and microsatellites markers.

Pair-wise similarity analysis of 43 morphological characters in the 22 cultivars of C. sinensis revealed that maximum similarity (0.64) occurred between the cultivars Campbell Valencia and Vanale, both of which are exotic in origin and are most similar in terms of qualitative fruit, leaf and seed characters. Minimum similarity (0.18) was observed between the cultivars Washington navel and Mosambi, which may be attributed to their different centres of origin where they have developed their distinct characters. Both cultivars are different in fruit characters, as Mosambi has an ellipsoid shape with rounded apex while the Washington naval has spheroid shape with truncated apex. The average similarity value of 0.39 indicated that cultivars show moderate to significant variability among these cultivars with respect to morphological traits. Based on RAPD markers, a high similarity value of 1.00 was found between two cultivars, Declarbe sweet orange and Rhode Red Valencia, showing very close genetic relationships between them which may be due to their common origin by mutation. Low genetic similarity (0.48) between cultivars Washington naval and sweet orange; Delta Valencia and Mosambi; and Satgudi and Delta Valencia may be due to different sources of origin of these cultivars. High genetic similarity (avg. 0.77) was recorded within this group, which showed a narrow level of genetic diversity existed within C. sinensis. This was also congruent with a moderate level of polymorphism that occurred within this group. Similarly, Fang and Roose [12] also reported low genetic variation among the cultivars of C. sinensis based on ISSR markers. This further supports the view that a majority of C. sinensis cultivars derived from a single ancestor through somatic mutation [13].

A search for unique bands was made for all the cultivars tested, in which a total of seven unique bands were generated in six cultivars by five RAPD primers. In Satgudi, a maximum of two unique bands were given by primer OPB-18 (600 bp) and OPO-04 (500 bp). Similarly, unique bands were generated in Sweet orange, Teneriffe, Seleta, Valencia late and Temple by RAPD primers. These unique fragments can be used as a marker for identification of these cultivars, which will be useful for future conservation, maintenance and breeding programmes. These accessions can also be used for developing the core collection of C. sinensis germplasm.

The UPGMA dendrogram divided all the cultivars into five main clusters based on morpho-metric data. The cultivar Washington naval was the most distinct from rest of the clusters mainly with respect to its fruit and leaf morphology. Jaffa and Delta Valencia were grouped together due to their similarity in fruit morphology and seedlessness. Indigenous cultivars of C. sinensis were clustered together in one group because of their similarity in most of the observed traits. A 2D plot showed five groups which was found less similar to the clustering pattern of UPGMA dendrogram. In the 2D plot, cultivars such as Delta Valencia, Jaffa, Malta, Washington navel and Seleta were found very distinct from rest of the cultivars as all are exotic and might have originated from different sources. The UPGMA clustering pattern based on RAPD data also indicated the genetic relatedness of C. sinensis cultivars by grouping 20 cultivars out of 22 into a single cluster. This shows that most of the cultivars of C. sinensis originated through somatic mutation (bud sports). However, two cultivars, Delta Valencia and sweet orange, were clearly separated from the rest of the cultivars and grouped in the same cluster. This indicated that both cultivars may have originated from the same genotype.

Comparison of morphological and molecular characterization data is of immense importance to conclude the extent of genetic diversity present in the set of cultivars. Although the correlation between the morphological and RAPD data was low in the analyzed cultivars of Citrus sinensis, as correlation values were found to be much less than 0.5, both methods allowed fare groupings of cultivars based on the analyzed traits. This is clear from the clustering pattern of the cultivars, where UPGMA dendrogram based on morphological data divided 22 cultivars into five major clusters whereas the dendrogram based on the RAPD marker divided them only into two major clusters. This shows that in spite of the wide phenotypic variations present within the cultivars they had a very narrow genetic base. The cultivar Washington Navel, which was morphologically most distinct from other cultivars, showed maximum genetic similarity with the rest of the cultivars. In the same way the Delta Valencia which was genetically most distinct showed some extent of morphological similarity with the cultivar Jaffa.

Based on the PCA of morphological characters, the first three principal components accounted for 53.77 % of the variance, in which maximum variation was given by the first component. Morphological characters, viz. fruit shape, fruit weight, fruit length, fruit diameter, fruit rind thickness, TSS, adherence of albedo to pulp, leaf apex, leaf lamina shape, leaf length, leaf width, petiole wing width, and oil gland size, represent maximum variability revealed by the first three components which were identified for developing a minimal descriptor for describing the species.

The moderate level of polymorphisms, in spite of the high morphological variability, could be explained by the fact that somatic mutations may be one of the sources of variability in C. sinensis. These results can be further used to manipulate genetic determinants of horticulturally important traits and to characterize the basis of productivity of C. sinensis cultivars in India. In the present study, RAPD markers proved to be useful for germplasm characterization and diversity analysis in C. sinensis cultivars.