Genetic Diversity of Castor Bean ( Ricinus communis L.) Revealed by ISSR and RAPD Markers

: Castor ( Ricinus communis L.), known as castor oil plant or castor bean, is a non-edible oilseed crop. In the present study, the genetic diversity among 54 samples (3 wild and 51 cultivated) collected worldwide was evaluated using inter-simple sequence repeats (ISSRs) and random ampliﬁed polymorphic DNA (RAPD) markers. A total of 9 ISSR primers produced 83 high-resolution bands with 61 (74.53%) as polymorphic. The percentage of polymorphic bands per primer and the genetic similarity coefﬁcient ranged from 54.55% (UBC-836) to 100% (UBC-808) and from 0.74 to 0.96, respectively. A total of 11 out of 20 RAPD primers ampliﬁed unique polymorphic products with an average percentage of polymorphic bands of 60.98% (56 polymorphic bands out of a total of 90 bands obtained). The percentage of polymorphic bands per primer ranged from 25% (OPA-02 and B7) to 90.91% (B21) with the genetic similarity coefﬁcient ranging from 0.73 to 0.98. The unweighted pair group method with arithmetic averages (UPGMA) dendrogram using two molecular markers divided 54 castor genotypes into three groups. Furthermore, based on morphological data, all 54 castor varieties were grouped into three main clusters. The genetic diversity analysis based on two molecular makers showed that most varieties from China were closely related to each other with three varieties (GUANGDONGwild, ZHEJIANGWild, and HANNANWild) belonging to a wild group separated from most of the cultivated castor samples from China, India, France, and Jordan. These results suggested that the cultivated castor contains a narrow genetic base. Accordingly, we recommend that wild castor genetic resources be introduced for breeding novel castor varieties. Furthermore, the Vietnam, Malaysia, Indonesia, and Nigeria accessions were clustered into the same group. The results of principal coordinate analysis (PCoA) and UPGMA cluster analysis were consistent with each other. The ﬁndings of this study are important for future breeding studies of castor. the genetic difference and phylogenetic relationship among castor varieties based on ISSR and RAPD markers. Furthermore, these results help establish the theoretical foundation for selection, genetic preservation, and breeding of the castor varieties. Our results demonstrate


Introduction
Castor (Ricinus communis L.) is an annual or perennial plant in Euphorbiaceae and is indigenous to the southeastern Mediterranean basin, East Africa, and India [1]. The castor is known to be highly tolerant in drought [2,3] and its leaves can be used as the food of silkworm. In addition, castor bean with an oil content of 45-55% in the seeds is considered as one of the top 10 oil crops worldwide [4,5]. Castor oil is the non-drying oil with high viscosity and is not easily clogged even under a low temperature of −56 • C [6]. Therefore, castor oil can be used for the production of biofuel in the chemical industry [7,8]. The demand for castor products such as castor oil and its derivatives has increased worldwide since 1991 [9][10][11]. After 2014, the import rate of castor raw materials in China has increased over 90% [10,12]. To date, many methods or techniques have already been used to improve breeding and productivity of castor [13][14][15][16].
Analyses of genetic diversity are very important for the development of genetic resources and new varieties of castor. In the past, genetic diversity studies based on morphological and biochemical characteristics have been conducted in many animal and plant varieties [17,18]. In particular, morphological traits are important phenotypic indicators in the development of sustainable crop breeding [19]. However, the study of genetic diversity based only on the morphological and geographical characteristics is ineffective, due to the strong influence of environmental complexity and genetic responses of plants. To date, many stable and efficient molecular markers have been developed. For example, studies of genetic diversity using molecular markers, such as simple sequence repeat (SSR), amplified fragment length polymorphism (AFLP), restriction fragment length polymorphism (RFLP), and single nucleotide polymorphisms (SNPs) have been widely applied in castor [20,21]. However, these methods have shown some disadvantages such as the high cost of AFLP and radioactive labeling of RFLP [22,23]. In recent years, developing SSR markers through mining SSR motifs in the whole genome has reduced the cost greatly, but has shown somewhat lower congruence with dominant marker data [24,25]. Both random amplified polymorphic DNA (RAPD) and inter-simple sequence repeats (ISSRs) have been used to study the genetic diversity and evaluate the genetic mutations of many plants [26,27]. ISSR technologies were found to be reproducible and effective to uncover polymorphism and obtain more composite marker patterns [28,29]. In contrast, RAPD has low rate of replicability, but is very efficient in analyzing genetic diversity because it does not need sequence data to design molecular primers [30,31].
Generally, both ISSR and RAPD markers are used in combination to improve their resolving power for genetic analysis of many plants [32]. In the past decade, combined morphological and molecular analyses have been widely used to investigate genetic diversity and phylogenetic relationships. For example, morphological characters and RAPD were demonstrated useful for reconstructing a Citrus phylogeny [33] and morphological traits and both RAPD and ISSR markers were evaluated to discriminate 15 potato cultivars [34] and to establish unique identification profiles in 26 sorghum varieties [35].
To obtain data of high quality, the study of genetic diversity based on both morphological characters and molecular markers (i.e., RAPD and ISSRs) is very important for castor breeding and production. To date, there are only a few studies available on analyzing the genetic variation by using combined RAPD and ISSR markers and morphometric characters. In this study, we aim to resolve the relationship between wild and cultivated varieties of caster from different accessions collected worldwide. These results provide the basic information for cultivating new varieties of caster and contribute to the selection and verification of novel varieties and the preservation of castor germplasms.

Plant Materials
The seeds of 54 castor varieties from different regions of the world were provided by the Inner Mongolia Key Laboratory of Castor Breeding, China. During the experiment, four quantitative traits (seed length (mm), width (mm), thickness (mm), and weight (g)) were measured on ten random samples. Detailed information on these 54 accessions is showed in Table 1 and Figure 1. Seed coat colors were scored by comparison with standards by multiple observers.   Table 1 for abbreviations of the varieties.  Table 1 for abbreviations of the varieties.

DNA Extraction
The castor seeds were first peeled and split in half, then were put into plastic bags and kept frozen in the temperature of -80 • C. DNA was extracted from seed tissue (~0.1 g) ground in liquid nitrogen by using Plant Genomic DNA kit (EP309, Beijing Zoman Biotechnology Co., Ltd., Beijing, China).
The extracted DNA was examined through electrophoresis on 1.0% agarose gel and quantified by a spectrophotometer (BioRad, Germany). Then, the DNA samples were diluted into 50 ng/µL and stored at -20 • C for PCR amplification.
The cycling conditions were as follows: 4 min at 94 • C for an initiation step, followed by 40 cycles of 30 s at 94 • C, 30 s at a primer-appropriate temperature, and 2 min at 72 • C, and finalized at a final cycle of 10 min at 72 • C. PCR products were detected by electrophoresis on a 2.0% agarose gel stained with 0.1 µL/mL of DNAgreen (UV) (Tiandz, China) and run in 1× TAE buffer at 100 V for 40 min. PCR products were photographed by the Bio-image System (BioRad, Germany). Each PCR reaction was repeated three times.

PCR Amplification of RAPD Marker
The RAPD analysis was performed using nine primers as previously described by Powell et al. [37]. The composition of the PCR reaction mixture contained 2.2 µL of DNA template, 2 µL 10×PCR Buffer, 1.5 µL dNTPs, 0.2 µL Taq polymerase, 0.5 µL Primer, and 13.6 µL ddH2O. The cycling conditions were as follows: 4 min at 94 • C for an initiation step, followed by 40 cycles of 30 s at 94 • C, 30 s at a primer-appropriate temperature, and 2 min at 72 • C, and a final cycle of 10 min at 72 • C. Amplified products were separated on 2% agarose gel in 1× TAE buffer by electrophoresis at 100 V for 40 min and photographed by Bio-image System (BioRad, CA, USA).

Data Analysis
SPSS 19.0 was used for one-way analysis of variance (ANOVA) and differences between means were assessed by Duncan's multiple range test (p ≤ 0.05). Each band amplified with ISSR and RAPD primers was manually scored as present (1) or absence (0). Only clear and reproducible bands were counted (Table S1). The genetic relationship between 54 castor varieties were evaluated by UPGMA analyses [25]. The genetic similarity of the castor varieties based on Jaccard's index [38]. The principle coordinate analysis (PcoA) was performed by NTSYS-pc version 2.10 software for three-dimensional distribution analysis [39].

Results
Generally, morphological characters (such as seed size and seed coat color) are an important factor in classification of castor varieties and evaluation of genetic diversity. In Table 1, the 51-Eth variety was superior to other varieties in seed length, width, thickness, and weight. The seed coat color pattern of the 54 varieties is given in Figure 1. Cluster analysis was carried out based on the morphological characteristics of seed size and color. Results showed that the 54 castor samples could be divided into three main groups (i.e., Clusters I, II, and III) with the genetic similarity coefficient ranging from 0.02 to 3.78 with an average of 1.90 ( Figure 2). There were a total of 6, 32, and 15 varieties grouped in Clusers 1, 2, and 3, respectively, with variety 51-Eth in its own group. The genetic similarity coefficient of the 51-Eth variety was evidently different from those of Clusters I, II, and III. Most of the seed color was brown or dark brown, and some varieties were red-brown (9-Jor, 11-Yun, 24-Sha, 25-Sha, 30-Ton, 34-Ton, 35-Ton, 41-Zi, 43-Zi, 46-Vie, 47-Pak, 48-Pak, 49-Par, and 52-Indo), while three varieties (4-Ind, 6-Ind, and 37) were light brown, and two varieties (8-Fra and 26-Ton) were yellow-brown.

ISSR Analysis
A total of 9 out of 100 primers that produced clear and reproducible bands were selected in the present study. Those primers produced 83 bands in total and the lengths of the amplified fragments ranged from 200 bp to 2700 bp (Table 2). Note: TNB, total number of bands; NPB, number of polymorphic bands; PPB (%), percentage of polymorphic bands; R represents either A or G and Y represents either C or T.
The percentage of polymorphic bands per primer ranged from 54.55% (UBC-836) to 100% (UBC-808) with an average of 74.53%. As a representative, the ISSR profile of 54 castor samples using UBC-842 primer was shown in Figure 3.

ISSR Analysis
A total of 9 out of 100 primers that produced clear and reproducible bands were selected in the present study. Those primers produced 83 bands in total and the lengths of the amplified fragments ranged from 200 bp to 2700 bp (Table 2). The percentage of polymorphic bands per primer ranged from 54.55% (UBC-836) to 100% (UBC-808) with an average of 74.53%. As a representative, the ISSR profile of 54 castor samples using UBC-842 primer was shown in Figure 3. The UPGMA dendrogram using ISSR data based on the genetic similarity coefficient showed two main clusters with the genetic similarity coefficient ranging from 0.74 to 0.96 with an average of 0.86 (Figure 4). The UPGMA dendrogram using ISSR data based on the genetic similarity coefficient showed two main clusters with the genetic similarity coefficient ranging from 0.74 to 0.96 with an average of 0.86 (Figure 4).  The UPGMA dendrogram using ISSR data based on the genetic similarity coefficient showed two main clusters with the genetic similarity coefficient ranging from 0.74 to 0.96 with an average of 0.86 (Figure 4).  Two varieties (14-Ton and 16-Ton) showed the highest similarity coefficient (0.97). Clusters I and II included most of the 54 varieties, and Clusters III includes 46-Vie, 52-Indo, 53-Indo and 54-Nig varieties. However, variety 9-Jor was obviously different from those of Cluster I and II, III.
The results of the three-dimensional plot based on the PCoA analysis ( Figure 5) were largely consistent with those revealed on the UPGMA clustering ( Figure 4).  Two varieties (14-Ton and 16-Ton) showed the highest similarity coefficient (0.97). Clusters I and II included most of the 54 varieties, and Clusters III includes 46-Vie, 52-Indo, 53-Indo and 54-Nig varieties. However, variety 9-Jor was obviously different from those of Cluster I and II, III.  (Table 1) based on the principal coordinate analysis using Jaccard similarity coefficient of the ISSR markers.
The results of the three-dimensional plot based on the PCoA analysis ( Figure 5) were largely consistent with those revealed on the UPGMA clustering ( Figure 4).

RAPD Analysis
Genetic diversity analysis was performed on 11 out of 20 RAPD primers that were reproducible and provided a unique polymorphism products. These 11 primers produced a total of 90 bands with the lengths of the amplified fragments ranging from 250 bp to 3000 bp (Table 3).   (Table 1) based on the principal coordinate analysis using Jaccard similarity coefficient of the ISSR markers.

RAPD Analysis
Genetic diversity analysis was performed on 11 out of 20 RAPD primers that were reproducible and provided a unique polymorphism products. These 11 primers produced a total of 90 bands with the lengths of the amplified fragments ranging from 250 bp to 3000 bp (Table 3). The percentage of polymorphic bands per primer ranged from 25% (B7 and OPA-02) to 90.91% (B21) with an average of 60.98%. As a representative, the RAPD profile of 54 castor varieties using B21 primer was shown in Figure 6.
The percentage of polymorphic bands per primer ranged from 25% (B7 and OPA-02) to 90.91% (B21) with an average of 60.98%. As a representative, the RAPD profile of 54 castor varieties using B21 primer was shown in Figure 6.  Table 1 for the abbreviations of these 54 accessions.
A dendrogram based on UPGMA analysis using RAPD data was shown in Figure 7 The genetic similarity coefficient ranged from 0.75 to 0.99 with an average of 0.85.  Table 1 for the abbreviations of these 54 accessions. A dendrogram based on UPGMA analysis using RAPD data was shown in Figure 7. The genetic similarity coefficient ranged from 0.75 to 0.99 with an average of 0.85.

Discussion
Castor bean seeds were used by people dating from about 4000 BC and were thought to have originated in Eastern Africa, particularly Ethiopia [40]. The application of castor in India, Saudi Arabia, Greece, and Rome has been documented as having anti-inflammatory, antiasthmatic, and cathartic effects [41]. Castor seeds have also been used in China for centuries as a traditional medicine [42]. It is reasonable to assume that castor might

Discussion
Castor bean seeds were used by people dating from about 4000 BC and were thought to have originated in Eastern Africa, particularly Ethiopia [40]. The application of castor in India, Saudi Arabia, Greece, and Rome has been documented as having anti-inflammatory, antiasthmatic, and cathartic effects [41]. Castor seeds have also been used in China for centuries as a traditional medicine [42]. It is reasonable to assume that castor might have been introduced to China many centuries ago. However, genetic diversity of varieties cultivated in China has not been systematically studied.
In this study, the analysis of genetic diversity of 54 castor samples collected in India, Jordan, France, Vietnam, Pakistan, Paraguay, Malaysia, Ethiopia, Indonesia, Nigeria, and China was carried out based on morphological characters of the castor seed and two molecular markers (RAPD and ISSRs). For a long time, genetic diversity analysis has been done based on the morphological characteristics. Morphological traits display a continuous phenotypic range, and as quantitative traits, they are dominantly controlled by multi genes [43]. Qualitative traits are reported as equally effective in diversity assessment compared with molecular markers in plants [44]. In particular, the size and weight of seeds contribute to the oil content of seeds and are important morphological indicators. The length-to-width ratio of the castor seeds in different varieties ranged from 1.76 (52-Ind) to 1.21 (51-Eth) with an average of 1.51. This ratio was not significantly different from the ratio reported in previous studies [45]. The 54 castor seeds were divided into five groups (brown, light brown, dark brown, red-brown, and yellow-brown) based on stripe color and background color. The stripe pattern of seed is an important factor in characterizing genetic diversity, but it was difficult to investigate this trait due to its great variability. The 54 castor varieties were clustered into three groups according to six morphological characteristics. This result was different from that of the clustering analysis based on molecular markers. Similar results have been reported in previous studies, indicating that genetic diversity analysis based on combined RAPD and ISSR markers may be a more useful tool for the identification of Brahmi than morphological characters [46]. Morphological and geographical patterns may also reflect genetic characteristics due to the environmental influences during the stages of plant development and reproduction. In contrast, both RAPD and ISSR technology are useful tools for the classification and genetic analysis of plant varieties compared to morphological analysis [47,48]. Both RAPD and ISSR markers show high polymorphism and are simple and cheap to generate. Therefore, they are widely used for the preservation studies of varieties and phylogenetic analysis [49,50].
Results of the genetic diversity of 54 castor varieties demonstrated that the ISSR and RAPD markers showed high polymorphism with an average of 74.59% and 60.98%, respectively. These values are comparable with 68.08% and 80.2% but much higher than 38% and 54% as previously reported by Gajera et al. [51] and Kallamadi et al. [5], respectively. Our results show that the genetic diversity among the 54 varieties of castor collected in different regions in the world is quite large. These results are similar to those reported in the previous studies showing the ample diversity among the castor genotypes [52]. Some different topological arrangements of these 54 varieties were revealed in the UPGMA dendrograms based on ISSR and RAPD markers. For example, the variety 9-Jor was revealed in its own group in the ISSR dendrogram but grouped into cluster I in the RAPD dendrogram. Furthermore, the variety 54-Nig was grouped in Clusters III and II in ISSR and RAPD dendrograms, respectively. These differences shown in UPGMA dendrograms are probably due to the polymorphism in the different regions of the castor genome. By using two molecular markers, the genetic background of 54 castor samples can be identified. The numbers of bands amplified by primers of ISSR and RAPD markers are different, ranging from the minimum of 4 or 5 to the maximum of 10 to 12. The different DNA amplification bands revealed in different varieties reflect the differences in genetic background.
Worldwide, Parzies et al. (2008) have categorized varieties of castor into two groups with Group a containing varieties collected from Asian countries (i.e., India, China, Pakistan, Indonesia, Jordan, and Paraguay), while Group b contains varieties collected from Americas (i.e., USA, Brazil, Cuba, and Argentina) [24]. In our study, out of 54 genotypes, 17 varieties were collected from areas outside China and 37 were distributed in China. Cluster I contains a large number of cultivars collected from China closely related to three wild accessions (1-Gua, 2-Zhe, and 3-Hai), indicating that these wild germplasms could be the putative progenitors of the cultivars of Cluster I (Figures 4 and 7). Furthermore, varieties collected from Vietnam, Indonesia, and Nigeria were identified in Cluster III ( Figure  4), together with varieties collected from Vietnam, Malaysia, Pakistan, Paraguay, Ethiopia, and Indonesia (Figure 7), indicating that these varieties share similar genetic backgrounds. The studies of Kallamadi et al. [5], based on the combined data of the three-marker system, showed similar results with these accessions classified into three major clusters, while the Indian and Nigerian accessions were clustered in two large groups, respectively. These results indicated that gene flow occurred frequently worldwide, which may be attributed to castor being either self-pollinated or cross-pollinated by wind, instead of using outcrossing as a predominant mode of reproduction [53]. However, the alternative hypothesis states that castor oil was not dispersed widely after its initial introduction. Therefore, the dispersal pattern appears to be dependent on human introduction or physical transport, owing to the heavy seeds [54].
The results of PCoA used to determine the spatial representation of genetic distances among the castor varieties were consistent with the results of genetic differentiation based on cluster analysis. The results of PCoA based on ISSR and RAPD markers differed slightly from those of the UPGMA dendrogram. The results of genetic diversity analyses based on both morphological and molecular data showed that three varieties (9-Jor, 46-Vie, and 51-Eth) showed larger genetic differences compared to other varieties. Therefore, the larger genetic distance of these three varieties may provide rich genetic resources to satisfy breeding requirements. Indeed, our results showed that we can effectively discriminate the genetic difference and phylogenetic relationship among castor varieties based on ISSR and RAPD markers. Furthermore, these results help establish the theoretical foundation for selection, genetic preservation, and breeding of the castor varieties. Our results demonstrate the utility of wild germplasm to exploit the unique caster breeding resources. Moreover, the castor resources in China showed a narrow genetic base. Therefore, we recommend that more castor resources be introduced to China to enrich the local breeding materials.

Conclusions
In the present study, we combined ISSR and RAPD markers to assess the genetic relationship of 54 R. communis samples collected worldwide (wild or cultivated varieties). Cluster analysis based on morphological data (seed length, width, thickness, weight, length to width ratio, and seed color) grouped all 54 castor samples into three main clusters. The results of PCoA and UPGMA cluster analysis were consistent with each other. The present study showed genetic divergence between the China elite local lines and foreign sources but the overall genetic variation was not extremely large despite the geographical distance. These results are important references for further resource exploitation and conservation of the genetic resources of castor.
Supplementary Materials: The following are available online at https://www.mdpi.com/2073-439 5/11/3/457/s1, Figure S1: Global distribution of the 54 R. communis L. seed accessions used in this study. The sample numbers are listed in Table 1, Table S1: Scoring of amplified bands of 54 samples of castor seeds based on ISSR and RAPD markers.