Diversity Analysis of Selected Solanum Species in Sri Lanka using Molecular and Morphological Descriptors

: The family Solanaceae is composed of a broad list of species, which includes both commercial and wild accessions with enormous medicinal importance. The published phylogenies on Sri Lankan wild and naturalized Solanum spp. confers that S. hispidum, and S. torvum are sister species. However, this has not been proved using multiple barcoding markers. Moreover, S. torvum, an underutilized crop is expected to contain multiple varieties. However, varietal descriptions using morphological and DNA markers have not been done so far. Therefore, our study was aimed at accurate identification of four Solanum spp. found in Sri Lanka (S. torvum ‘Bindu’ and landraces S. hispidum and S. pubescens) using chloroplast marker based DNA barcoding combined with a morphological description. We used 29 quantitative and 22 qualitative parameters, comprised of vegetative and reproductive traits along with eight DNA barcoding markers. We sequenced rbcL and trnL-F to check the DNA polymorphisms. According to the combined morphological analysis of reproductive and vegetative parameters, the four species clustered separately. There were no separation observed between S. torvum and S. hispidum. Solanum pubescens diverged out and cladded with old world nightshades in molecular analysis. These results are in line with published literature and validates that trnL-trnF could be used as a potential marker to carry out a combined DNA barcoding analysis with matK. However, trnL-trnF could not be used to discriminate between S. torvum and S. hispidum species.


INTRODUCTION
The genus Solanum belongs to the tribe Solaneae, subfamily Solanoideae, and family Solanaceae which is comprised of species with substantial economic importance worldwide (Mace et al., 1999;Seguí-Simarro, 2016). The members of Solanaceae species found in the tropics are generally considered as annual herbs while those in the temperate regions are annuals. They grow as a bush up to a height of 0.5-2.5 m. The fruits produced may vary from a small number to several (Rubatzky and Yamaguchi, 1997).
The members of the family Solanaceae are well known for the remarkable medicinal properties found in leaves, fruits and roots (Jeyakumar et al., 2016). The fruits of Solanum torvum Swartz are used as a cure for cough, liver and spleen disorders (Siemonsma and Piluek, 1994). It has further been discovered to have tranquilizing and diuretic properties and a broad array of antibacterial and antifungal effects on human and animal pathogens (Gousset et al., 2005;Chah et al., 2000;Abas et al., 2006). The phenolic compound present in the fruits of S. torvum is used as an anti-diabetagenic and antioxidant (Gandhi et al., 2011;Loganayaki et al., 2010). The leaves are utilized as a stabilizing agent of blood flow. Tonics and blood cell regeneration agents are prepared from mature fruits and they are also employed for the curing pains and skin diseases (Kala, 2005). Solanum hispidum has been evaluated for the anti-fungal activity due to the presence of a novel spirostanol and several other saponins (González et al., 2004;Chakravarty et al., 1982).
The species of Solanaceae grown in Sri Lanka include. S. melongina, S. torvum-Bindu ('Thibbatu)', S. torvum-Landrace ('Thibbatu'), S. hispidum ('Gonabatu') and S. pubescens ('Walthibbatu'). Majority of the research conducted in Sri Lanka are focused on 'wambatu' or S. melongena eggplant which thrives well in the rain-fed and minimum irrigation facilities due to great stress resilience of the plant (DOA, 2018). Solanum melongena is cultivated as a commercial crop in Sri Lanka. However, the ability of the varieties of eggplant to resist biotic stress such as bacterial and fungal and other nematode and insect attacks is poor. Thus, the development of resistance by identification of resistant genes from the wild relatives is highly recommended (Collonnier et al., 2001). Certain wild relatives of S. melongena such as S. torvum harbors a consortium of resistant genes which can be readily employed in breeding programmes. S. torvum has been identified as resistant to wilt diseases and certain other pests such as nematodes. The resistance has been transferred to the commercial varieties of Solanum spp. through various breeding programs (Jarl et al., 1999). Interspecific hybridization has been carried out between commercial eggplant cultivars and S. torvum (Collonnier et al., 2003;Clain et al., 2004;Gousset et al., 2005). However, the precise identification of species/varieties play a crucial role in subsequent breeding protocols as most of the wild relatives possess undesirable traits which can be easily transferred to the genomes of improved cultivars when subjected to breeding (Prohens et al., 2017).
DNA barcoding can be identified as a key and robust strategy in successful species identification (Group et al., 2009). Several research studies have been carried worldwide for successful identification of the Solanum spp. especially to detect the adulterations when used as medicinal plants (Techen et al., 2014). Solanum species have also been used in DNA barcoding projects carried out for conservation of local biodiversity in certain other regions of the world (Boessenkool et al., 2014). However, studies have not yet been carried out for the varieties/species of the genus Solanum found in Sri Lanka especially for the wild relatives such as S. torvum. Thus, the present study was conducted to accurately identify the varieties of Solanum spp. found in Sri Lanka for the future use of the species in wide hybridization protocols and conservation of the wild germplasm of Solanum spp.

Collection of the plant material
The study was carried out with four species of wild relatives of brinjal (Solanum melongena) (S. torvum 1: 'Thibbatu' variety Bindu recommended by the Department of Agriculture (DOA), Sri Lanka; S. torvum 2: landrace of 'Thibbatu'; S. hispidum: 'Gonabatu'; S. pubescens: 'Walthibbatu'). Breeder seeds of S. torvum 1 were sourced from the Regional Agriculture Research and Development Center, Aralaganwela, Sri Lanka and S. torvum 2 from a farmer in Girandurukotte. Solanum hispidum was collected from roadsides at Haggala, Nuwara Eliya, while S. pubescens were collected from roadsides at Baragama, Hambanthota.

Establishment of plant material
The nursery bed was prepared by using soil and coir dust (1:1), treated with Topspin and was kept for 24 hours for disinfection. Seeds were soaked in pure water and was allowed to germinate. Pits of 0.09 cm 2 dimensions were prepared on 10 m 2 plots with a space of 1 m between rows and 2 m between plants. Plots were maintained at the open fields of the Regional Agriculture Research and Development Centre, Bandarawela (IU3a-average annual rainfall of > 1900 mm, maximum average temperature of 27 ºC, minimum average temperature of 14 ºC), Welimada, and farmer fields at Masnawatta and Mirahawatta, Welimada (U3d-Average annual rainfall of > 1300 mm, maximum average temperature of 28 ºC, minimum average temperature of 15 ºC) (Reddish yellow Podsolic soil in both locations). The pits were filled with a mixture of decomposed cattle manure and topsoil in 1:1 ratio and the basal fertilizer was added according to the recommendation of the Department of Agriculture, Sri Lanka. Four weeks old seedlings were transplanted according to an incomplete block design (IBD) with four genotypes, and five blocks in Bandarawela and three blocks in Welimada sites.
Furthermore, 22 qualitative parameters on cotyledon, plant, leaf, stem, flower, seed, fruit, were descriptively assessed for three and five replicates from Bandarawela and Welimada sites respectively. The color of the cotyledon and the plant growth habit were recorded to distinguish among the species/varieties according to the RHS Color Chart (Royal Horticultural Society, 2001). The parameters recorded to assess the leaf variability were leaf blade color, blade lobbing and the leaf blade tip angle. The stem characteristics recorded in this study were the color of the stem, and presence of hairs and prickles on the stem. The floral parameters included the type of the inflorescence, the color of corolla, position of the stigma and the presence of prickles in the pedicel and color of the seeds. The shape of the fruit apex, fruit flesh density, relative fruit calyx length, color of the petiole, the position of the fruit, the color of fruits, the nature of fruit curvature, cross section and locules, and the number of prickles in calyx were recorded as fruit characteristics. These parameters were used to assess the variability among the species/varieties.

DNA extraction and PCR
Immature, fresh leaf samples were used in the DNA extraction process. DNA was extracted using modified CTAB (hexadecyl trimethyl ammonium bromide) protocol (Porebski et al., 1997). PCR amplification was performed in TP600: Takara (Otsu Shiga, Japan) thermal cycler using eight DNA barcoding primer pairs (Table 1) in a 15 µl the PCR reaction mixture with Go Taq Green Master Mix (7.5 µl), 10 µM forward and reverse primers (0.5 µl each) and 10 µM Spermidine (3.5 µl). The PCR profiles used were as given in Table 1 (Levin et al., 2003;Shaw et al., 2005). The volume of template DNA for each marker was optimized considering each marker profile. The standard PCR protocol was modified for certain markers for an optimal amplification (Table 1).

DNA sequencing
The viable PCR products of two selected DNA markers, which are commonly used for more profound phylogenetic relationship analysis in plants were purified using Qiagen Qiaquick PCR purification kit (Catalog number: 28104, Qiagen, Hilden, Germany) and sequenced using ABI 3500 automated sequencer (Catalog number 622-0010).

Data analysis
All the quantitative morphometric data were subjected to normality testing in Minitab 17 (Minitab Inc. USA), and General Linear Model (GLM) and LS means/pdiff mean separation procedures using statistical package SAS 9.4 (SAS Institute, NC, Cary, USA). Following the data analysis, a dendrogram was constructed for the vegetative and reproductive data using the algorithms of Complete Linkage and Pearson Distance method in Minitab 17 (Minitab Inc. USA). Principal component analysis (PCA) was conducted using reproductive, vegetative and combined data in Minitab 17 (Minitab Inc. USA). The principal component 1 (PC 1) and principal component 2 (PC 2) were employed in drawing scatter plots. All the PCs were used to draw dendrogram using Euclidean distances.

Phylogenetic analysis
The raw sequencing data generated for two markers rbcL and trnL-trnF were first visualized in MEGA 7 (Kumar et al., 2016) in order to define the initial and end noise. Then, the trimmed datasets were subjected separately to a BLAST search and homology sequences with high E values (threshold E value = 10 -5 ) were retrieved ( Table 2). The selected reference sequences were then aligned with the sequences by ClustalW algorithm (Thompson et al., 1994) in MEGA 7 (Kumar et al., 2016) with manual editing. The evolutionary relationship of S. torvum, S. pubescens and S. hispidum was inferred based on the phylogenetic tree reconstructed based on rbcL and trnL-trnF. A tree search was conducted in Maximum Likelihood (ML) framework in RAxML (Stamatakis, 2006) through a rapid bootstrap algorithm (Stamatakis, 2008) for 1000 iterations and GTRGAMMA (Rodriguez et al., 1990) as the evolutionary model to compensate the dataset. All the bootstrap bipartitions were used to get a single tree topology by constraining the bootstrap values to ML consensus tree. Further modification of the final majority rule consensus tree was carried out using Figtree 1.4.3 (Rambaut, 2014). Furthermore, to support the analysis, a tree was built in the Bayesian framework and interpretations on the best nucleotide substitutions were made by carrying out a model selection in Akaike Information Criteria (AIC) (Akaike, 1974) and corrected Akaike Information Criteria (AICc) (Cavanaugh, 1997) in J model test (Posada, 2008). Then, the criteria for the best model were used to construct the tree in the Bayesian framework using MrBays (Huelsenbeck and Ronquist, 2001). Four Markov Chain Monte Carlo (MCMC) were implemented for 50 million cycles to generate precise posterior probability distributions. Burn-in point of the analysis was set to 5000 generations and 10% of the trees generated initially were omitted as burn-in. The 50% majority rule consensus tree out of the remaining trees was drawn. The posterior probability values from each branch were subsequently inferred. The model selection and all the tree constructions were carried out in CIPRES super computer (Miller et al., 2010). Finally, the resemblance of Bayesian and ML trees was modelled, and the posterior probability values were incoporated to the nodes of the Bayesian tree.

Variation of the quantitative morphological parameters
Mean separation of a total of 29 quantitative parameters including 11 vegetative and 18 reproductive parameters for three Solanum species are given in the  Jeyakumar et al. (2016) and Bello et al. (2013) showing higher genetic determination in the considered traits.

Variation of the reproductive traits
Out of the 18 reproductive parameters, NFPI, FD, PL, PW, AW, FB, LS, TS, yield, SD, and HSW were not significantly different between two sites. However, NFPI was the highest in S. torvum 1 and S. torvum 2 varieties (41.31 and 41.00, respectively) and the lowest in S. pubescens (12). In contrast, NFPI was highly variable regarding the site × species/variety interaction (  Jeyakumar et al., (2016).

Analysis of qualitative parameters
Qualitative characters provided an indispensable source for the differentiation of species. The variability of the qualitative morphometric parameters was descriptively assessed and reported in Table 5 and Table 6. Morphological variability of cotyledon, plant, leaf, stem, flower, seed and fruit of Solanum species/varieties were descriptively evaluated at vegetative, flowering and harvesting stages.
The cotyledon color at the seedling stage was variable among the species. S. torvum were found to have green cotyledons, while those of the S. pubescens and S. hispidum were light violet. The growth habits of Solanum spp. were different, as the plants of S. torvum and S. pubescens species developed an upright growth habit and the stems of S. hispidum plants showed an intermediate growth pattern (Table 5). Three parameters of leaf were assessed to detect the variability among the species. Accordingly, the color of the leaf blade of S. torvum sp. was moderately olive green, whereas S. pubescens and S. hispidum possessed leaves with greyish green. Leaf blade lobbing was slightly variable among the species. The blade lobbing of S. torvum and S. hispidum were strong, while that of the S. pubescens was intermediate. However, it was noticed that there was no any difference in the leaf blade tip among the species, such that all the leaves had an acute tip (Table 5) (Plate 1 A and B).    The stem color at the maturity stage of S. torvum was yellowish green, and that of the S. pubescens and S. hispidum were light olive grey and moderate olive grey respectively. Stems of the trees were moreover assessed for the presence of hairs and prickles (Plate 1 C). Accordingly, S. torvum and S. hispidum were found to have hairs and prickles on the stem while the stems of S. pubescens only possessed the prickles (Table 5). All the species possessed flowers in clusters. S. torvum and S. hispidum possessed inflorescence with compound dichasium whereas, the type of inflorescence of S. hispidum was helicoid cyme. The color of the corolla was also variable among the species. S. torvum had flowers with greyish white whereas those of the S. hispidum and S. pubescens were white and strong purple respectively. The position of the stigma in S. torvum was either included/exerted while those of the S. pubescens and S. hispidum were found to be exerted. However, none of the flowers possessed prickles in their pedicels (Plate 1 D). Moreover, the color of the seeds was found to be of variable yellows among the species (Table 6). It was also noted that most of the fruit characters were not variable among the three species revealing their less importance in discriminating of species. All the Solanum species possessed strong yellow-green, semi-erect fruits with a straight curvature, circular cross-section, no grooved locules and no prickles in fruit calyx. Nevertheless, the fruits of S. torvum and S. pubescens possessed a round apex and average dense fruit flesh while those of the S. hispidum possessed a protruded apex and dense fruit flesh. The relative length of the fruit calyx of S. torvum and S. hispidum was short whereas that of the S. pubescens was comparatively long. Furthermore, the color of the petiole at the maturity stage in S. torvum species was moderately yellowish green while those of the S. pubescens and S. hispidum were found to have a strong yellow-green (Table 6) (Plate 1 E and F). Qualitative traits however, played a key role in differentiating the three species thus providing a better way to develop easily utilizable morphometric markers for species identification within the same genera. Although a variation was detected, it is relatively difficult to distinguish the species based on the quantitative traits, as compared to qualitative morphological traits. Our results are in conformity with the previous findings in Yousaf et al., (2010).

Cluster analysis of quantitative parameters
The dendrogram constructed considering both vegetative and reproductive characteristics showed a clear separation between three different Solanum spp. S. torvum 1 and S. torvum 2 which belong to S. torvum were clustered together at 79.78 % similarity level. S. pubescens was highly diverse from the other varieties based on the morphological characteristics ( Figure 1). The PCA yielded 29 components corresponding to 29 morphometric parameters including 11 vegetative and 18 reproductive traits (Figure 2). A total variance of 98.2% was obtained collectively for the first two components of the PCA performed where PC1 and PC2 accounted for 67.8% and 30.4% respectively. The third component produced a variance of 1.9% which did not improve the outcomes of the study and all the other components were uninformative. Therefore, third and all the other components were not considered. In the PCA scatter plot, the four species/varieties were divided into three distinct groups as S. hispidum (black), S. pubescens (brown) and S. torvum (red and blue) where the S. torvum; Bindu and S. torvum; Landrace were overlapped with no distinct separation. The extensive overlap of the two S. torvum varieties indicates that the morphological data cannot provide discrimination between the two varieties.

Assessing the DNA length polymorphism
The DNA bands resulted a polymorphic banding pattern for four markers namely, trnL, trnS GCU -trnG UUC , matK-trnT and trnL-trnF. Considering the aforementioned loci, S. pubescens indicated a banding pattern which is clearly distinguishable from the other vaieties indicating the genetic variability of S. pubescens. All the species produced monomorphic bands for atp B gene, atpB-rbcL, trnH-psbA and rbcL (Plate 2). A total of 16 alleles were detected for eight loci. Thereby an average of two alleles per loci was identified.

Assessing the DNA sequence polymorphism
Out of the two (rbcL and trnL-trnF) DNA barcoding markers employed in this study, there were only one SNP detected in trnL-trnF marker for all four species/varieties. Among the individuals of nightshades, a less polymorphism was observed in rbcL gene region, thus a shallow separation was obtained in the phylogeny. The 50% majority rule consensus tree drawn in the Bayesian framework had low divergence between each sequences. The ML tree constructed for rbcL maker had many polytomies and low support value (Figure 3) thus the employment of rbcL marker would not be suitable for species/varietal discrimination. However, both the phylogenies (ML consensus tree and the Bayesian 50% majority rule consensus tree) constructed for trnL-trnF marker had congruence in the branching pattern with a higher node support ( Figure 4). The branching pattern in the trnL-trnF phylogeny is mostly similar to previously published phylogenies where the S. torvum and S. pubescens were cladded as sister groups (Ranaweera et al., 2018;Särkinen et al., 2013). Thus the trnL-trnF could be used as candidate marker for phylogenetic studies. It is essential to combine the trnL-trnF sequence data with other potential markers such as trnH-psbA and ITS to obtain a robust phylogeny. The generated sequences were submitted to GenBank under the Accession numbers MK122636-MK122643.

CONCLUSIONS
The analyses of morphological diversity of Solanum spp. revealed that the species identification can be successfully accomplished by considering both the vegetative and reproductive morphological parameters together. There was a less sequence polymorphism observed in rbcL coding region, thus could not be used as a marker for characterization of Solanum spp. The trnL-trnF marker provides the required separation for genus Solanum and it separates S. pubescens from S. hispidum and S. torvum.

ACKNOLEDGEMENT
Regional Agriculture Research and Development Centre, Bandarawela for providing the necessary field space for establishment of plant material.