Introduction

Molecular markers based on DNA polymorphisms are of great utility for different applications in biological and agricultural sciences (Avise 2012; Grover and Sharma 2014). There is a wide array of available molecular markers (e.g., isozymes, RFLPs, RAPDs, ISSRs, AFLPs, SSRs, SNPs, etc.), which have different characteristics depending on their nature and the techniques employed to identify them. Among them, microsatellites or simple sequence repeats (SSRs) have been amongst the most used ones for germplasm management, selection and breeding (Kalia et al. 2011). SSRs are codominant, abundant, robust and highly polymorphic (Varshney et al. 2005). Although a few years ago their identification was relatively expensive and time-consuming through the development of genomic libraries, with the debut of new generation sequencing (NGS) platforms their isolation has become straightforward and cost-effective (Fernandez-Silva et al. 2013; De Barba et al. 2016; Zhan et al. 2016). In fact, thanks to the tremendous advances in sequencing of transcriptomes and genomes, hundreds or thousands of SSRs can be identified (Xiao et al. 2013; Goodwin et al. 2016). However, these large-scale sequencing projects also allow the identification of thousands to millions of molecular markers of single nucleotide polymorphism (SNPs) markers with a reasonable amount of resources (Van Tassell et al. 2008; Davey et al. 2011; Scheben et al. 2017). As occurs with the SSRs, SNPs are codominant, very reliable, ubiquitous and universal molecular markers, and although are generally less informative than SSRs (Yang et al. 2011; Filippi et al. 2015; Gonzaga, 2015), they are much more abundant and easy to automate (Thomson et al. 2014; Kim et al. 2016).

SNPs are more uniformly spread across the genomes, while the SSRs tend to distribute more frequently in heterochromatic regions, although this strongly depends on the species (Li et al. 2002; Hong et al. 2007). In addition, the validation of SSRs is generally more time-consuming and expensive due to the need of detection through an agarose or polyacrylamide gels or capillary sequencing (Jones et al. 2007). In this respect, Yan et al. (2010) estimated that the resources and time needed to build a genetic map using a high-throughput SNP genotyping may be 75% cheaper and 100-fold faster than SSR gel-based methods.

Because of their different characteristics, SSRs and SNPs sample different levels of genetic diversity. In this respect, SSRs have a much higher mutation rate than SNPs (Hamblin et al. 2007; Coates et al. 2009; Fischer et al. 2017), because the mutations can be derived from a variation of the number of repeats, motif length or motif sequence (Ellegren 2004; Kashi and King 2006). In fact, the estimation of mutational rates of SSRs (1 × 10−5) is several orders of magnitude higher than that of SNPs (1 × 10−9) (Li et al. 1981; Kruglyak et al. 1998; Martínez-Arias et al. 2001). Although at the transcriptomic level the mutation rate of both SSRs and SNPs are lower than at the genomic level, the mutation rate of SSRs is also much higher than that of SNPs in the expressed sequences (Li et al. 2002).

Cultivated eggplants and their wild relatives are very variable (Vorontsova et al. 2013; Kaushik et al. 2016; Acquadro et al. 2017), with three different eggplant complexes (Daunay and Hazra 2012; Plazas et al. 2014; Syfert et al. 2016), and may be an appropriate material for comparing SSR and SNP markers for fingerprinting, evaluation of genetic identities and distances and for studying their relationships. Although the brinjal (or common) eggplant (Solanum melongena L.) is the most economically important, two other cultivated eggplant species exist, namely the gboma (S. macrocarpon L.) and the scarlet eggplants (S. aethiopicum L.), which are mostly grown in sub-Saharan Africa (Lester et al. 1990; Sunseri et al. 2010). Solanum melongena together with its ancestor (S. insanum L.) and other close relatives, like S. incanum L., are part of the so-called brinjal eggplant complex (Knapp et al. 2013). Similarly, the gboma eggplant together with its ancestor S. dasyphyllum Schumach. & Thonn. on one side, and the scarlet eggplant together with its ancestor S. anguivi Lam. on the other constitute, respectively, the gboma and scarlet eggplant complexes (Lester and Niakan 1986; Bukenya and Carasco 1994). Among the latter, the scarlet eggplant complex is more variable than the gboma eggplant and four cultivar groups (Aculeatum, Gilo, Kumba, and Shum) are considered within the cultivated S. aethiopicum (Lester and Daunay 2003). In addition, intermediate forms between the wild S. anguivi and cultivated S. aethiopicum are also common (Plazas et al. 2014).

Although independently of the markers used the three eggplant complexes are often genetically differentiated clearly (Sakata and Lester 1997; Furini and Wunder 2004; Isshiki et al. 2008; Tumbilen et al. 2011; Acquadro et al. 2017), the genetic proximity and phylogenetic relationships among them frequently depend on the markers and plant materials used. In this respect, there is wide discrepancy in the studies over which of the three eggplant complexes is phylogenetically basal or genetically more distant to the two others (Sakata et al. 1991; Sakata and Lester 1997; Furini and Wunder 2004; Levin et al. 2006; Isshiki et al. 2008; Weese and Bohs 2010; Tumbilen et al. 2011; Meyer et al. 2012; Vorontsova et al. 2013; Särkinen et al. 2013; Acquadro et al. 2017). Regarding the relationships within the hypervariable scarlet eggplant complex there have been few molecular studies evaluating them (Sunseri et al. 2010; Adeniji et al. 2013; Acquadro et al. 2017), but in general, they reveal that they present a low genetic differentiation.

Given the lack of studies comparing SSR and SNP markers in Solanum crops for genetic resources, breeding and genetic relationships, in this work we have genotyped a collection of accessions from the brinjal, gboma and scarlet eggplants with SSR and SNP markers obtained from the transcriptomes of S. aethiopicum and S. incanum (Gramazio et al. 2016). Although genomic markers tend to be more informative and accurate because are less prone to selection (Gadaleta et al. 2011), several studies with markers derived from transcriptome have shown that they are useful and reliable for phylogenetics and establishment of relationship among and within populations (Vogel et al. 2006; Castillo et al. 2008; Choudhary et al. 2009). Our objective is to confirm the potential utility of these markers as well as to compare them for fingerprinting, evaluation of genetic identities and distances, and for the establishment of relationships among these three groups.

Materials and methods

Plant material

The collection of accessions used in this study encompasses accessions from different origins (African, Asian and European), as indicated in Table 1. All materials are maintained at COMAV germplasm bank (Universitat Politècnica de València, Spain). Thirty-nine accessions belonged to the scarlet eggplant complex and according to Lester and Niakan (1986) classification key and Plazas et al. (2014), morphological characterization, belong to S. anguivi (2), to S. aethiopicum groups Gilo (16), Kumba (7), Aculeatum (5), Shum (3), or to an intermediate group between S. anguivi and S. aethiopicum (8). Seven other accessions used in this study corresponded to the gboma complex, of which six are of S. macrocarpon and one of S. dasyphyllum. Finally, one accession of S. incanum and one S. melongena were used to represent the brinjal complex. The seeds were germinated following a protocol which is especially recommended for Solanum species that may present dormancy (Ranil et al. 2015).

Table 1 Plant materials used for molecular characterization with SNP and SSR markers of a collection of accessions of brinjal, gboma and scarlet eggplant complex, including the species and cultivar group and the country of origin
Full size table

DNA extraction

Total genomic DNA was isolated from leaves of 3–4 true leaves stage plantlets, according to the CTAB protocol (Doyle and Doyle 1987) with slight modifications. The extracted DNA was dissolved in Milli-Q water and general quality was confirmed in agarose gel at 0.8%. After a concentration measurement using a Qubit® 2.0 Fluorometer (Thermo Fisher Scientific, Waltham, USA), the DNA was diluted at 30 ng/uL for PCR and High Resolution Melting (HRM) amplification.

SSR genotyping

The identification of SSRs from the transcriptomes of one S. aethiopicum and one S. incanum accessions (Table 2) was reported in Gramazio et al. (2016). The selection of 11 highly reliable polymorphic SSRs was performed through filtering them for quality parameters and checking their coverage and length in the IGV visor (Thorvaldsdóttir et al. 2013). Primers pairs were designed using Primer3 tool (v. 0.4.0, http://bioinfo.ut.ee/primer3-0.4.0/primer3/).

Table 2 Characteristics of the SRR and SNP markers used, including the chromosome in which they are situated, the S. incanum and S. aethiopicum unigenes to which they correspond (Gramazio et al. 2016), the corresponding scaffold in the eggplant genome (Hirakawa et al. 2014), and the forward and reverse primers
Full size table

The SSRs amplification was performed by PCR in a volume of 12 μL including 7.21 μL water, 1.2 μL 1 × PCR buffer, 0.6 μL MgCl2 50 mM, 0.24 μL dNTPs 10 mM, 0.3 μL 10 μM, 0.06 μL forward primer with M13 tail 10 μM, 0.24 μL fluorochrome (FAM, VIC, NED and PET) 10 μM, 0.15 μL Taq DNA Polymerase (5 U/μL), 2 μL DNA template 20 ng/μL. The PCR program used was the following: 95 °C for 3 min for a denaturation, 30 cycles of 30 s at 95 °C followed by 30 s at 65 °C and of 30 s at 72 °C and finally 72 °C for 5 min for the last step of extension. The PCR products were subsequently diluted in formamide and sequenced by capillary electrophoresis through an ABI PRISM 3100-Avant sequencer (Thermo Fisher Scientific, Waltham, USA) using a 600 LIZ GeneScan size standard (Thermo Fisher Scientific, Waltham, USA). The fragments were analyzed using the GeneScan software (Thermo Fisher Scientific, Waltham, USA) to obtain the electropherograms and polymorphisms were analyzed with Genotyper DNA Fragment Analysis software (Thermo Fisher Scientific, Waltham, USA).

SNP genotyping

The identification of 35 SNPs and primer pairs design (Table 2) was similar to the one for SSRs as indicated above by using S. aethiopicum and S. incanum transcriptomes (Gramazio et al. 2016). Validation of SNPs was performed through Real-Time PCR in a LightCycler 480 (Roche, Basel, Switzerland). The reactions were performed in a 10 μL volume comprising 5 μL Master Mix 2X, 0.8 μL MgCl2 25 mM, 0.25 μL each primer, 1.7 μL water and 2 μL DNA 30 ng/μL with the following touchdown PCR program: denaturation at 95 °C for 10 min, followed by 55 cycles of 10 s at 95 °C, 15 s at 65 °C (decreasing 1 °C each cycle until 55 °C) and of 15 s at 72 °C, finally the melting step at 1 min at 95 °C, 1 min at 40 °C, 1 s at 60 °C and rising the temperature at 0.02 °C/s until 95 °C.

Data analyses

The molecular marker analysis for SNPs and SSRs was performed using the software packages PowerMarker (Liu and Muse 2005) and GenAlEx 6.5 (Peakall and Smouse 2012). The following parameters were calculated using the PowerMarker package: number of alleles per locus, major allele frequency, number of genotypes, polymorphic information content (PIC) values calculated as

$${\text{PIC }} = 1 - \mathop \sum \limits_{i = 1}^{n} p{}_{i}^{2} - \mathop \sum \limits_{i = 1}^{n - 1} \mathop \sum \limits_{j = i + 1}^{n} 2p_{i}^{2} p_{j}^{2} ,$$

where n is the total number of alleles detected, \(p_{i}\) the frequency of the i th allele, and \(p_{j}\) the frequency of the jth allele) (Botstein et al. 1980), expected heterozygosity (H e ), calculated as H e  = \(1 - \mathop \sum \nolimits_{i = 1}^{n} p_{i}^{2}\) (where \(p_{i}\) is the frequency of the ith allele) (Nei 1972), observed heterozygosity (H o ), calculated as the number of heterozygous alleles/number of alleles and fixation index (f), calculated as f = 1 − (Ho/He) (Wright 1965).

The consensus tree was calculated from the genetic similarity to illustrate the level of relatedness between the accessions using the UPGMA method (unweighted pair-group method using arithmetic averages) (Sneath and Sokal 1973) and it was reconstructed using the software TreeView (Page 2001). Branch support on the phenogram was tested by bootstrap analysis with 1000 replications using the PHYLIP version 3.67 software (Felsenstein 2007). Bootstrap values of 50% or higher were used to indicate support for the phenogram topology at a node (Highton 1993). The genetic distance matrix (Nei 1972) among the different accessions was calculated with the GenAlEx 6.5 software package for both for SSRs and SNPs. The correlation between pair-wise genetic distances calculated for both markers was investigated by the Mantel test (Mantel 1967) of matrix correspondence.

Results and discussion

Validation of the SSR and SNP markers for fingerprinting

All SSR and SNP markers tested amplified and were polymorphic in the collection evaluated, although in a few cases there were missing data. In other studies in which eggplant SSR markers were developed de novo, the levels of polymorphism were not as high. For example, Vilanova et al. (2012) found an 85.5% of polymorphism in SSRs derived from a genomic library of S. melongena. A lower level of polymorphism (56.7%) was detected in the genetic SSRs derived from a library developed by Nunome et al. (2009), where 598 out of 1054 markers were polymorphic. When comparing the SSR and SNP profile of the two accessions used for obtaining the transcriptomes of S. aethiopicum (accession BBS135) and S. incanum (accession MM577) (Gramazio et al. 2016), only one SNP marker (SNP_14499) was found to be monomorphic among the two accessions. This confirms the high quality of the transcriptome sequences and in silico analysis performed by Gramazio et al. (2016) and reveals that both SSR and SNP markers obtained from in silico analyses of transcriptomes in materials of cultivated and wild eggplants are reliable and transferable to related species for being used in genotyping and fingerprinting. Other studies confirmed the reliability of the molecular markers discovered in silico from transcriptomes. For example, a set of SNPs identified in silico from a pepper (Capsicum annuum L.) transcriptome were validated in 43 pepper lines and accessions resulting in a rate of 89.9% polymorphic markers (Ashrafi et al. 2012). Also 86.7% of a subset of SSRs identified in silico in a zucchini (Cucurbita pepo L.) transcriptome resulted polymorphic in a set of ten accessions of genus Cucurbita, nine of which were representative of the diversity within C. pepo and one accession C. moschata accession (Blanca et al. 2011). Up to now, few transcriptomes have been sequenced in genus Solanum and in just a few of them molecular markers have been identified and primers pairs designed in silico for a subsequent validation in a wide range of related materials. An example is the S. dulcamara L. transcriptome where the SSR identified in silico were validated in seven plants and all of them resulted polymorphic (D’Agostino et al. 2013).

When considering the 11 SSRs, a total of 34 different genetic profiles were found among the 48 accessions evaluated, while for the 35 SNPs, the number of different genetic profiles was 36 (Table 3). This confirms the highest discrimination potential of SSRs compared to SNPs for genetic fingerprinting (Hamblin et al. 2007; Varshney et al. 2007; Yang et al. 2011), as the number of different profiles obtained is almost the same with 11 SSRs or 35 SNPs. In this respect, several authors considered that in order to obtain a similar genetic power to discriminate individuals from different populations the number of SNPs required might be 8–15 times the number of SSRs (Hess and Matala 2011; Yu et al. 2009).

Table 3 Accessions having a shared genetic profile with 11 SSR or 35 SNP markers, or using all of them (11 SNP plus 35 SNP markers)
Full size table

When considering a combination of both SNP and SSR markers 47 unique genetic fingerprints were obtained for the 48 accessions, and only two of them (both from S. aethiopicum Gilo group) shared the same genetic profile for the markers that amplified in both of them. Although in general (Hu et al. 2011; Nandha and Singh 2014; Thiel et al. 2003) and in the particular case of eggplant (Muñoz-Falcón et al. 2011) genomic SSRs and SNPs are frequently more polymorphic than transcriptome-derived SSRs and SNP, in our case we have found a considerable level of polymorphism in the markers we tested. Particularly, the combination of both SSR and SNP markers has been highly efficient for genetic fingerprinting. Probably the fact that both types of markers sample different levels of genomic diversity (van Inghelandt et al. 2010) increases the efficiency of fingerprinting when combining both types of markers.

In all cases, accessions sharing a single SSR or SNP profile corresponded to the same eggplant complex (brinjal, gboma, or scarlet eggplant complexes) (Table 3), indicating that both types of markers provide consistent results. The number of accessions having a single genetic profile ranged between two and six for SSRs, between two and seven for SNPs, and only two when combining both SSR and SNP markers. When considering only SSR or SNP markers, accessions from different groups of the scarlet eggplant complex (i.e., S. anguivi, the Intermediate S. anguivi-S. aethiopicum group, and the four groups of S. aethiopicum) often shared a same profile. Our data are in agreement with those of other authors who used AFLPs and SSRs (Sunseri et al. 2010), RAPDs (Aguoru et al. 2015), and SNPs (Acquadro et al. 2017) and also found that accessions of S. aethiopicum did not cluster according to the cultivar group. This is an additional indication that these groups, which are distinguished on the basis of morphology (Lester 1986; Lester and Daunay 2003; Plazas et al. 2014), are not genetically differentiated.

SSR and SNP diversity statistics

In general, the diversity statistics for the SSRs had higher average values than those of the SNPs (Tables 4, 5). This is common because due to its nature, the potential variability of SSRs is larger than that of SNPs (Ellegren 2004; Kashi and King 2006; Fischer et al. 2017). In our study, the number of alleles obtained with SSRs ranged between three and seven, with an average value of 5.00 (Table 4), while for SNPs it ranged between two and four with an average value of 2.34 (Table 5). Also, the major allele frequency was generally lower with SSRs, ranging from 0.531 to 0.857, with an average value of 0.689, while for SNPs the values were considerably higher, with a range from 0.510 to 0.989, and an average value of 0.851. These values reveal that, despite a having a great diversity, there is always a major allele with a frequency over 50%, both for SSRs and SNPs. The number of genotypes, as occurs for the alleles, was also higher for SSRs, with a range from four to eight and an average of 5.91 (Table 4), than for SNPs, which presented a range from two to five, with an average of 3.14 (Table 5). As a consequence of the larger number of alleles and lower frequency of the major allele, the PIC values were generally higher for SSRs, with a range from 0.249 to 0.636 and an average of 0.419, than for SNPs, which ranged from 0.021 to 0.386 and an average value of 0.191. Our results are consistent with previous works in eggplants, which find similar values to ours. For example, Vilanova et al. (2012), using genomic SSRs in a collection of 20 accessions of S. melongena plus one accession of S. aethiopicum and S. macrocarpon found similar levels of number of alleles (4.72) and PIC (0.47) than us. Otherwise, other authors found lower values of average PIC rate and number of alleles per locus. For example, Stàgel et al. (2008) found a mean PIC rate of 0.38 and number of alleles per locus of 3.1 when assessed 11 EST-SSRs in 38 S. melongena accessions. Also, Nunome et al. (2009) found lower values for the diversity statistics when they genotyped eight lines of S. melongena using 1054 genomic SSRs (mean PIC value = 0.27, mean number of alleles = 2.2) and 66 EST-SSRs (mean PIC value = 0.13, mean number of alleles = 1.4).

Table 4 Genetic diversity statistics, including major allele frequency, number of genotypes, number of alleles, expected heterozygosity (H e ), observed heterozygosity (H o ), polymorphic information content (PIC), and coefficient of inbreeding (f) for the 11 SSR markers evaluated in a collection of brinjal, gboma and scarlet eggplants accessions
Full size table
Table 5 Genetic diversity statistics, including number of alleles, major allele frequency, number of genotypes, polymorphic information content (PIC), expected heterozygosity (H e ), observed heterozygosity (H o ), and coefficient of inbreeding (f) for the 35 SNP markers evaluated in a collection of brinjal, gboma and scarlet eggplants accessions
Full size table

The expected heterozygosity (H e ) was, on average, higher for SSRs (0.456) than for SNPs (0.224), and much higher than the observed heterozygosity (H o ), which had average values of 0.039 for SSRs and 0.025 for SNPs. The much higher values for H e compared to H o result in high levels for the fixation index (f), with average values of 0.907 for SSRs and 0.837 for SNPs. In fact, for some SSR and SNP markers, the f value was 1, and so all materials were homozygous for these loci. Our values are similar to those obtained by others with SSRs and SNPs in common eggplant collections (Muñoz-Falcón et al. 2009; Vilanova et al. 2012, 2014, Ge et al. 2013; Augustinos et al. 2016). Although it is known that cultivated brinjal eggplant is fundamentally autogamous (Arumuganathan and Earle 1991; Pessarakli and Dris 2004; Daunay and Hazra 2012) our results also provide evidence that gboma and scarlet eggplants present a reproductive system similar to that of brinjal eggplant.

SSR and SNP-based genetic relationships

Genetic identity values among groups within each of the eggplant complexes had high values, both with SSR and SNP markers (Table 6). Not surprisingly, low values of genetic identities have been obtained between S. incanum and S. aethiopicum groups, as markers were selected for polymorphism between the transcriptomes of two accessions of these taxa (Gramazio et al. 2016). When considering each of the eggplant complexes, the genetic identities within eggplant complexes have been larger than the between eggplant complexes identities. For example for the brinjal eggplant complex, the within complex genetic identity (i.e., between S. incanum and S. melongena) has been 0.810 for SSRs and 0.588 for SNPs, while the range for genetic identities with other eggplant complex groups has been between 0.045 (S. melongena vs. S. macrocarpon) and 0.439 (S. melongena vs. S. anguivi) for SSRs and between 0.107 (S. incanum vs. S. aethiopicum group Gilo) and 0.565 (S. melongena vs. S. macrocarpon) for SNPs (Table 6). Amazingly, in this case, the genetic identity between S. melongena and S. macrocarpon has been largest with SNPs and lowest with SSRs, reflecting that different levels of genetic diversity are sampled by both markers (Ellegren 2004; Kashi and King 2006; Fischer et al. 2017). In the case of the gboma eggplant complex, the within complex identity has been very high, with a value of 0.935 for SSRs and 0.911 for SNPs, and it has ranged between 0.045 (S. melongena vs. S. macrocarpon) and 0.229 (S. macrocarpon vs. Intermediate S. anguivi-S. aethiopicum) for SSRs and between 0.343 (S. incanum vs. S. dasyphyllum) and 0.753 (S. macrocarpon vs. S. aethiopicum group Shum). For this complex group the genetic identities with other complex groups have been much larger with SNPs than with SSRs (Table 6). Finally, for the scarlet eggplant complex the within complex identity has ranged between 0.704 (S. anguivi vs. S. aethiopicum group Aculeatum) and 0.970 (S. aethiopicum group Gilo vs. S. aethiopicum group Kumba) for SSRs, and between 0.859 (S. anguivi vs. S. aethiopicum group Kumba) and 0.976 (S. aethiopicum group Gilo vs. S. aethiopicum group Kumba) for SNPs; when considering genetic identities with other groups it has ranged between 0.098 (S. incanum vs. S. aethiopicum group Shum) and 0.439 (S. melongena vs. S. anguivi) for SSRs and between 0.107 (S. incanum vs. S. aethiopicum group Gilo) and 0.753 (S. macrocarpon vs. S. aethiopicum group Shum) (Table 6). In this case, the genetic identities of scarlet eggplant with the brinjal eggplant have been larger than with the gboma eggplant with SSRs, while the contrary occurred with SNPs. Overall, these results provide evidence that important differences exist among SSR and SNP markers for genetic identities among groups. This may have important consequences for the establishment of relationships among eggplant complexes based on genetic profiles, as depending on the markers used the results may be very different.

Table 6 Nei (1972) genetic identities among the different groups of the brinjal, gboma and scarlet eggplant complexes based on SSR (above the diagonal) and SNP (below the diagonal) markers
Full size table

SSR and SNP pair-wise genetic distances among all individual accessions displayed a moderate correlation (r2 = 0.529; b = 0.341; P < 0.0001) (Fig. 1). However, it is evident from Fig. 1 that there is a difference in the relationship between both markers depending on the value of the SNP genetic distance. In this way, for values of SNP genetic distance below 70, the correlation is much higher (r2 = 0.592; b = 0.534; P < 0.0001) than for values for SNP genetic distance above 70 (r2 = 0.080; b = 0.045; P = 0.0047). This shows that, for the materials used, when genetic distances based on SNPs are low to intermediate (i.e., within complexes), SNPs and SSRs provide similar levels of information, while when the genetic distances are higher (i.e., among complexes), SNPs provide better resolution, as SSR-based genetic distances seem to reach a saturation. This different performance of both types of markers is very likely due to the much higher rate of mutation of SSRs compared to SNPs (Hamblin et al. 2007; Coates et al. 2009). In this way, when certain levels of phylogenetic distance are reached, the SSR alleles seem to have diverged so much that they are not good to establish relationships based on phylogenetic distance. This has important implications for phylogenetic and germplasm conservation studies.

Fig. 1
figure 1

Relationship between SNP (X-axis) and SSR (Y-axis) pair-wise genetic distances among 48 individual accessions of the brinjal, gboma and scarlet eggplant complexes

Full size image

Cluster analyses

The topology of the SSR-based and SNP-based phenograms is different (Fig. 2). In other studies, the results obtained on the relationships among the three eggplant complexes depended on the markers used (Sakata et al. 1991; Furini and Wunder 2004; Levin et al. 2006; Isshiki et al. 2008; Weese and Bohs 2010; Meyer et al. 2012; Särkinen et al. 2013; Vorontsova et al. 2013; Acquadro et al. 2017). Although in both cases three major clusters are identified, corresponding to each of the three eggplant complexes, in the SSR-based phenogram the gboma eggplant complex cluster is basal to the brinjal and scarlet eggplant complexes, while in the SNP-based phenogram the brinjal eggplant cluster is basal to the gboma and scarlet eggplant complexes. Amazingly, Acquadro et al. (2017), using genomic SNPs found that the scarlet eggplant complex cluster is basal to the brinjal and gboma eggplant complexes. In our case, because markers used here were selected for polymorphism between expressed sequences of one accession of each of S. incanum and S. aethiopicum (Gramazio et al. 2016), it was expected that the largest distance in the tree should have been between the clusters of brinjal and scarlet eggplants, but this was only true for SNP markers (Fig. 2). This provides evidence that SNP markers may provide better resolution than SSRs when evaluating phylogenetic relationships among Solanum taxa not belonging to the close primary genepool. When considering the scarlet eggplant complex, in both phenograms the wild S. anguivi is basal to the other groups, although the other groups present a different clustering pattern depending on the markers used (Fig. 2).

Fig. 2
figure 2

UPGMA hierarchical clustering consensus phenograms based on Nei (1972) genetic distances for the brinjal (S. melongena and S. incanum), gboma (S. macrocarpon and S. dasyphyllum) and scarlet eggplant (S. aethiopicum and S. anguivi) complex groups according to SSR (left) and SNP markers (right). Bootstrap values (based on 1000 replications; expressed in percentage) greater than 50% are indicated at the corresponding nodes

Full size image

Conclusions

The transcriptome-derived SSR and SNP markers have been highly polymorphic in the eggplants collection evaluated and proved useful for genetic fingerprinting. As expected, SSRs were on average more informative than SNPs, but a similar number of fingerprints were obtained with 11 SSRs and 35 SNPs. Both SSR and SNP markers confirmed the high fixation index of the eggplant materials and clearly distinguished the three eggplant complexes. However, different results depending on the type of marker were obtained for the relationships among eggplant complexes, indicating that they sample different levels of genetic variation. In this respect, SSRs and SNPs presented a moderate correlation for low to intermediate values of SNP pair-wise genetic distance, but a low correlation for high SNP genetic distances. This suggests that both markers are complementary in the information provided, although SNPs seem more appropriate to evaluate materials genetically distant in the eggplant complexes. This information will be useful for eggplants germplasm management, phylogenetic and relationships studies, as well as for genetic fingerprinting and breeding.