Eggplant (Solanum melongena L., Solanaceae) is an important vegetable crop cultivated in the tropical and temperate regions of the world. Domesticated long ago in the Indo-Burma region where the greatest diversity is found, it has been a common crop in the Middle East and around the Mediterranean basin where it was introduced by the Arabs (Daunay et al., 2001; Daunay & Janick, 2007). Eggplant is a fairly good source of phenolics, vitamins and minerals with antioxidant, hepatoprotective, anti-microbial, and cardio-protective properties (Kalloo, 1993; Stommel & Whitaker, 2003; Das et al., 2011).

Intensive selection and breeding have resulted in the development of a large number of varieties with variable morphological, physiological and qualitative characteristics (Daunay et al., 2001; Kashyap et al., 2003; Frary et al., 2007). Among them, fruit color, size and shape are the most distinctive and therefore used to establish the different commercial groups, e.g. black, green, purple, striped and white, long, cylindrical, egg shaped etc. existing today (Kumar et al., 2008; Muñoz-Falcón et al., 2009).

Today global trade is concentrated on a limited number of eggplant elite varieties and F₁ hybrids with increased yield and stability. The progressive dominance of those genotypes results in a considerable reduction in genetic diversity posing a threat of genetic erosion and narrowing the source of useful genes available for breeding (Muñoz-Falcón et al., 2011; Cericola et al., 2013). Local cultivars on the other hand with specific and high quality characteristics represent an elite germplasm which can be used for breeding as well as in scion-rootstock combinations for increased production and tolerance to various pathogens (Bletsos et al., 2004; Arvanitoyannis et al., 2005; Rodríguez-Burruezo et al., 2008).

In Greece, besides the commercial F₁ hybrids and imported varieties, there are several traditional cultivars, e.g. Tsakoniki (short cylindrical fruits with purple stripes), Lagada (long cylindrical purple fruits), Skoutari (medium cylindrical purple fruits), EMI (large spherical purple fruits) and Santorini (large spherical white fruits), which are adapted to the local environment having an excellent texture and cooking quality. These cultivars have not been subjected to intensive breeding so far, and are available through local markets, where the fruits of the heirloom local varieties are highly appreciated (Bletsos et al., 1998a, 1998b).

During the last years, heirloom vegetables produced in their regions of origin, may acquire an added value in the market if protected by geographical indications (PGI) or designations of origin (PDO) (Hurtado et al., 2014). This approach could increase the prospects of benefits for the farmers, therefore certification of the authenticity is essential for the protection and enhancement of these cultivars (Muñoz-Falcón et al., 2011).

Morphological descriptors and molecular markers have been widely utilized for the characterization of eggplant material from different regions of origin, the study of diversity among different varieties and accessions and also the determination of the relationships among the cultivated eggplant with its weedy and wild forms and close relatives (Prohens et al., 2005; Isshiki et al., 2008; Demir et al., 2010; Ge et al., 2011, 2013; Muñoz-Falcón et al., 2011; Tümbilen et al., 2011a, 2011b; Hurtado et al., 2012; Meyer et al., 2012; Vilanova et al., 2012; Cericola et al., 2013; Davidar et al., 2015; Mutegi et al., 2015). Those studies have provided invaluable information regarding the management of germplasm collections, the conservation of eggplant genetic resources and the application of breeding programs. In addition, the combination of morphological descriptors with molecular markers provides a more comprehensive view about the degree of variation among different groups of cultivars since they sample different levels of diversity and thus have a complementary effect (Prohens et al., 2005; Muñoz-Falcón et al., 2009; Tümbilen et al., 2011b; Hurtado et al., 2012; Cericola et al., 2013).

The purpose of this study was to assess the levels of genetic and phenotypic diversity existing within and among six traditional eggplant cultivars and a F₁ hybrid from Greece using both morphological and molecular markers. We also included a commercial variety and three more cultivars from Africa (Togo) and Japan, in order to estimate the degree of variation among cultivars originating from different diversity centers and regions. Genetic variation was studied using nuclear microsatellites, as well as the retrotransposon microsatellite amplified polymorphism (REMAP) markers (Kalendar et al., 1999). REMAP has been successfully applied for the determination of the genetic relationships, the germplasm identification and the assessment of genetic variation of several plant species and at various taxonomic levels (Kalia et al., 2010; Castro et al., 2012). Finally, a comparison between the genetic data and the data obtained from the morphological analysis has also been made.

In this study the morphological and genetic diversity of the most promising Greek eggplant traditional cultivars with high quality and reputation at least at the local markets were examined. Some of those cultivars could achieve an advanced market status as candidates for a Protected Designation of Origin (PDO) or a Protected Geographical Indication (PGI) label (EC, 2006). Usually, products with a PDO or PGI label are perceived as unique having an added value in the market and getting a higher price (Gracia & Albisu, 2001; McLaughlin, 2004). Therefore it is necessary to devise tools for the certification of the uniqueness and authenticity of those products in order to protect and enhance them (Muñoz-Falcón et al., 2011; Hurtado et al., 2014).

Morphological diversity: The eggplant cultivars displayed a relatively high level of diversity for most of the morphological traits studied (total phenotypic diversity (H_T = 0.51). In addition, the mean phenotypic diversity among cultivars was relatively high (G_ST = 0.52), while the mean within population diversity was low (H_s = 0.24), as expected for cultivars.

The morphological differentiation and the proportion of phenotypic variation found among cultivars as indicated by the high G_ST values is more likely due to the artificial selection performed by the farmers, the mating system, the management of the genetic material, and also to the environmental conditions (Prohens et al., 2005; Terzopoulos & Bebeli, 2010; Hurtado et al., 2012). Some of the traits for which significant differences exist e.g. fruit size, shape and color, growth habit, prickliness among others, are controlled by a few genes which are presenting a high degree of penetration and expression. As a consequence the effect on phenotype is rapid and probably linked with response to selection and the different origin of those cultivars (Daunay, 2008; Hurtado et al., 2012).

Understanding which traits contribute the most to the phenotypic variation observed among and within cultivars is very important for breeding purposes as well as for conservation procedures (Terzopoulos & Bebeli, 2008). In the present study traits such as plant height, fruit calyx prickles, fruit length, fruit breadth, fruit length to breadth ratio, fruit predominant, additional and flesh color tend to be highly differentiated among cultivars while most of them exhibited low diversity within the cultivars. Analysis of morphological diversity showed that clustering of cultivars was strongly associated with plant height and fruit appearance such as fruit length, shape and color. Clustering of different eggplant cultivars according to fruit appearance has also been reported by other researchers (Choudhury, 1976; Tümbilen et al., 2011b; Cericola et al., 2013). This was also evident in the PCA analysis where in the first component (30.4% of the total variance) the most important parameters for separating the cultivars were those related to fruit appearance and shape such as fruit breadth, fruit weight fruit calyx prickles and fruit length. Moreover, as shown in the Ward’s dendrogram two major clusters were formed based primarily on those characters. In the first cluster there were the SKO, MELF1, MOMA, WAS, LAG, TSAK cultivars which are predominantly taller with longer fruits, while in the second cluster were those with smaller fruit length and plant height. Within this cluster another sub-cluster was formed by LERWHIT and SANT based exclusively on fruit color.

Genetic diversity: Both REMAP and SSR markers showed that the overall polymorphism revealed is low. This is in agreement with the current literature, where larger collections and more markers have been independently applied (Stàgel et al., 2008; Muñoz-Falcón et al., 2009; Ge et al., 2011, 2013; Muñoz-Falcón et al., 2011; Hurtado et al., 2012; Cericola et al., 2013). Because S. melongena is a largely autogamous species, most heritage and commercial varieties are expected to be highly homozygous (Cericola et al., 2013). In addition extensive phenotypic-based selection and the use of a limited number of individuals as starting material in breeding programs have deprived this species of the bigger part of its original genetic pool, reducing thus the amount of genetic diversity available today (Muñoz-Falcón et al., 2009).

In our study, no significant associations were observed between REMAP and SSR markers, according to the Mantel test employed. This might be due to the type of genetic polymorphism detected and the different type of information provided by each molecular marker system used (Ghislain et al., 2006). Also, poor correlation between the used genetic markers most likely indicates that they refer to different subsets of loci in the genome as for example the SSR and REMAP markers in our case (Biswas et al., 2010; Mandoulakami et al., 2015). It would be surprising to obtain a better insight of the diversity existing in the S. melongena populations with REMAP markers than with microsatellite markers. However, the combination of the REMAP markers used here produced specific patterns of bands that can distinguish among the different cultivars.

Nuclear SSRs have been widely used to study the genetic diversity and the relationships of different eggplant groups and cultivars of so far. SSRs have the ability to detect genetic relationships based on specific traits, probably caused by its sensitivity to neutrality and/or linkage disequilibrium, and therefore are more appropriate in studying specific sets of genetically related materials (Muñoz-Falcón et al., 2009).

The overall variability detected in our study was low since the SSR markers produced an average of 3.67 alleles per locus. Muñoz-Falcón et al. (2011) using also genomic and EST-SSRs revealed low levels of polymorphism. This could be due to the intensive breeding efforts and the narrow genetic background (Ge et al., 2011). On the other hand Behera et al. (2006) found broader diversity in 92 South Asian eggplant accessions but such greater variation was expected since South Asia is a primary center of eggplant diversity. According to Tümbilen et al. (2011a) genetic diversity based on molecular data is highly dependent on the number and type of marker chosen and the accessions tested and this could be the case in our study.

The genetic differentiation as indicated by the genetic distance values estimated among the cultivars studied was high. Also all of them were significantly differentiated from each other, except the two cultivars from Japan. As shown by the UPGMA dendrogram and also by the STRUCTURE analysis our data support the existence of three main clusters, the first with the foreign cultivars, the second with the LERWHIT, SANT and BB, while the third with the remaining Greek cultivars. Interestingly the two Japanese cultivars seem to be more closely related to some of the Greek ones. Hurtado et al. (2012) and Cericola et al. (2013) failed to sufficiently discriminate eggplants from different centers of origin, although this was achieved completely by Prohens et al. (2005) and Muñoz-Falcón et al. (2009) or partly by Sunseri et al. (2010). A possible explanation is that the limited genetic variability of the populations of this species (shown in other studies as well) and also the extensive selection performed on a few elite cultivars for specific traits have probably lead to extensive inbreeding, hitch-hiking effects of microsatellite markers along with traits of choice and severe bottlenecks. All factors have probably acted together, resulting to the ‘masking’ of the signal of the genetic origin of eggplant varieties.

The differentiation of the LERWHIT and SANT cultivars from the remaining Greek ones may be due to the SSR markers tested which could reflect certain phenotypic differences given that these cultivars differ in fruit shape and color from the rest. Also because these two cultivars originate from small islands it is also well supported that microevolutionary forces, including natural selection under local conditions, artificial selection by farmers, genetic drift, and recombination may have favored the isolation of local variants (Portis et al., 2006; Muñoz-Falcón et al., 2011).

In a recent study by Ganopoulos et al. (2015), thirty-six Greek eggplant accessions, were discriminated based on High Resolution Melting (HRM) barcoding profile. In our case, although we studied fewer cultivars but with more SSR markers, we did not succeed in discriminating them using nuclear microsatellites only. This could be due to a) the utilization of different SSR markers, b) the utilization of a different resolution technology, since HRM could theoretically reveal more diversity, as discussed in Ganopoulos et al. (2015) and references therein. However, the interpretation of the results can be more difficult, since melting curves are not exactly ‘descriptive’ of the events generating these differences, c) the methodology used regarding sampling: in the present study, DNA was isolated from 10 individual plants per variety, while in Ganopoulos et al. (2015) DNA per variety were isolated as a bulk.

The Mantel test showed no correlation between the genetic and the morphological data. On the other hand, the stepwise discriminant analysis performed in order to compare the morphometrical descriptors with the molecular genetic data, showed that discrimination between the clusters of the UPGMA dendrogram could be made if we use the morphometrical descriptors related to plant height (PH), number of leaves to first flower (NOLFF), leaf blade length (LBLe), leaf blade lobing (LBLo), leaf surface shape (LSS) and fruit shape (FS). These descriptors and especially those related to FS have also been found to be strongly associated with the clustering of S. melongena cultivars in several other cases (Muñoz-Falcón et al., 2008; Tümbilen et al., 2011b; Hurtado et al., 2012 etc.).

The lack of correlation between morphological and molecular data has been a matter of debate since there are many studies where a similar situation has also been confirmed (Tümbilen et al., 2011b; Hurtado et al., 2012; Cericola et al., 2013), whereas others showed a reasonable level of phenotype/genotype correlation (Prohens et al., 2005; Muñoz-Falcón et al., 2008; Caguiat & Hautea, 2014). The correlations between morphological and molecular data usually vary depending on the plant material, the morphological descriptors and the molecular markers used (Hurtado et al., 2012). Morphological markers in eggplant sample traits for which variation is usually controlled by a few genes or QTLs (Doganlar et al., 2002; Daunay, 2008) therefore a tiny portion of the genome is responsible for the phenotypic changes observed (Cericola et al., 2013). On the other hand molecular markers sample variation from a larger portion of the genome, both coding and/or non-coding (Varshney et al., 2005; Kalia et al., 2010) and thus these two types of markers may follow different evolutionary paths which explains the lack of correlation observed (Hurtado et al., 2012). Morphological descriptors and molecular markers provide complementary information and thus characterization of germplasm with both of them allows a more accurate assessment of variation (Caguiat & Hautea, 2014).

In summary, this study has successfully used morphological descriptors and SSR markers to determine the genetic diversity levels and to characterize the most important Greek traditional cultivars together with others from different centers of origin. Although low variability was detected within all cultivars, at least for the molecular markers used, a considerable amount of variation has been revealed among them with both the morphological descriptors and the molecular markers. Our results show that the plant material studied can be efficiently characterized for breeding and conservation purposes and also that we are able to devise tools for the effective screening of the whole Greek eggplant germplasm collection in the future. Moreover, these local varieties can be distinguished from the commercial ones and thus gain a higher market value, while the high quality standards of the final product are maintained.