Genetic diversity among fig genotypes revealed by SSR markers
In the present study, 60 fig accessions were examined using eight microsatellites (SSRs), designated MFC1 to MFC8 (Khadari et al. 2001). All figs revealed PCR banding pattern for seven out of the eight SSRs (Suppl. Fig. 1). For MFC6, no banding pattern (no PCR products) was observed for 12 out of 60 fig accessions, namely seven from Greece (Acc. No 118, 124, 131, 139, 140, 142, 146), three from Italy (Acc. No 251, 252, 253), one from Cyprus (Acc. No 236) (Suppl. Fig. 1), and one from Turkey (Acc. No 230). PCR reactions for the 12 above-mentioned fig accessions were repeated confirming the initial observation.
For each fig accession showing a banding pattern, one allele (homozygous individual) or two alleles (heterozygous individual) were identified and their size was estimated. For fig accessions from Greece, a total of 28 alleles were revealed for the eight SSRs studied, with an average of 3.5 alleles per SSR locus. In fig accessions from other countries, two more alleles were revealed, both in MFC1, namely the 198 bp in Columbra nera (Acc. No 253), originating from Italy, and the 170 bp in Blank Klirou (Acc. No 210), White Klirou (Acc. No 236), Winter Blank Cyprus (Acc. No 220) and White Prodromou (Acc. No 233), all originating from Cyprus. The number of alleles per locus (Na) varied between 2 (for MFC4, MFC5, and MFC8) to 6 (for MFC3) (Table 1).
Alleles with a frequency < 5% were designated as rare alleles (Table 1). In particular, allele 192 bp (Acc. No 109) of MFC1, 158 bp (Acc. No 145) of MFC2, 248 bp (Acc. No 127) and 290 bp (Acc. No 161) of MFC6 was observed only once, in fig accessions originating from Greece. In addition, allele 194 bp was shared between two fig accessions from Greece (Acc. No 117 and 118) and one from Italy (Acc. No 244). Furthermore, allele 126 bp (Acc. No 118 and 131) and 154 bp (Acc. No 142 and 155) were identified as two rare alleles for MFC3. Lastly, alleles 170 bp and 198 bp of MFC1 were observed only in fig accessions originating from Cyprus and Italy, respectively.
For each SSR locus, several genetic parameters were computed, with the results shown in Table 1. The observed heterozygosity for the fig accessions originating from Greece using seven SSRs (MFC6 was not included) ranged from 0.220 in MFC8 to 0.902 in MFC4 (with mean value 0.537) and the expected heterozygosity ranged from 0.195 in MFC8 to 0.604 in MFC2 (with mean value 0.449). For each SSR, except MFC1 and MFC3, the observed heterozygosity was higher than the expected. Consequently, most of the F values for the SSR loci are negative (average value -0.184) indicating a tendency for heterozygosity selection. The above data indicated that allele frequencies did not significantly deviate from the Hardy-Weinberg equilibrium (P>0.05) (Table 1).
MFC6 was not included in the determination of the mean value of the genetic parameters because in 7 out of 41 the fig accessions from Greece no banding pattern was observed, and genetic variability parameters were estimated only for the remaining 34 fig accessions. Nevertheless, MFC6 is a highly informative locus, possessing five alleles, He and Ho of 0.676 and 0.622, respectively (Table 1) and a PIC value of 0.584 (Table 2).
The MicroCheker results indicated that the presence of null alleles was found to be statistically nonsignificant, with their frequencies per SSR ranging from 0 to 3.53%. No new alleles appeared for any of the SSRs and the genotyping procedure was repeated for approximately 20% of fig accessions. The genotyping error is non-significant since it was found that allele size shift of two base pairs was observed only in 3% of the cases, indicating that the methodology used is reproducible.
Genetic relationship among fig genotypes - Population structure of fig germplasm collection
The genetic relationship among the fig genotypes is presented in a dendrogram produced using UPGMA (Suppl. Fig. 2). Fig accessions formed two groups, designated as group I and II. Group I included all fig accessions, except two, both originating from Greece, and which constituted group II. Group I could be separated into four subgroups, namely A, B, C, and D. Moreover, subgroup A was further divided into two major clusters, AI and AII, including 21 and 19 fig accessions, respectively. Based on the above clustering, AMOVA analysis indicated that 69% of the total population variation could be accounted from within-subgroup individuals and 31% from among subgroup difference. The robustness of the branches was evaluated by bootstrap analysis. The first node was significantly supported (100%) by bootstrap analysis, indicating that the two groups (I and II) could clearly cluster. In contrast, the rest of the early nodes were not significantly supported by bootstrap analysis indicating that subgrouping, clusters (AI and AII) and lower clustering present a weak separation.
From the dendrogram, it became obvious that fig accessions from different countries or from different regions of Greece were mixed into the same subgroup, group or cluster (Suppl. Fig. 2). PCoA analysis supported this observation since the percentage of variability that could be explained by grouping figs by geographical origin (country) was only 45.28% of the total variation in the established fig population (Fig. 1).
In cluster AII, Maurosykia (Acc. No 108) and Zakynthos (Acc. No 160) were pooled together; the branch was supported by significant bootstrap (86%), suggesting that they are two representatives of the same genotype and could be synonymous. However, it should be noted that these two fig accessions are morphologically different (Ntanos et al. 2015) and were collected from remote regions of Greece, namely an island in Ionian Sea (southwestern Greece) and Macedonia prefecture (northern Greece), respectively, and thus their close genetic relationship should be investigated further. The second instance of a very close genetic relationship, which was supported by a 65% bootstrap value, is that of the San Pietro (Acc. No 213) and Dottato (Acc. No 234) from Italy. Both are worldwide well-known varieties, which produce fruit twice per year, and differ in some morphological characteristics, such as fruit color (Ntanos et al. 2015). Their close genetic relationship was also shown in a previous study using RAPD (Papadopoulou et al. 2001). On the other hand, Maurosykia (Acc. No 108), originating from Macedonia prefecture, and Maurosykia 19 (Acc. No 155), originating from Eleia prefecture, are two fig accessions possessing the same name but originate from two remote regions of Greece (northern and southern Greece) and were found to be genetically different. The only caprifig accession included in the population, namely Ag. Anargyri (Acc. No 124), was grouped in a separate group (C) together with a fig accession from Greece (Acc. No 132, Tzavelas Black Large), which is gynodioecious producing edible figs. To further confirm the results described in Fig. 2, a WNJ analysis was produced that is reported in Fig. 3 and Suppl. Fig. 2.
Genotypes of the figs originating from Greece
Allele patterns, per SSR locus, in fig accessions, revealed different genotypes, ranging between two for MFC4, MFC5 and MFC8 to nine for MFC3, as shown in Table 2. Combining the data for all eight SSR markers indicated the presence of 40 different genotypes for the 41 fig accessions originating from Greece (Suppl. Table 1). Only two fig accessions, namely Zakynthos (Acc. No160) and Maurosykia (Acc. No 108), could not be distinguished genotypically; but could be phenotypically distinguished. Similarly, 18 out of 19 fig accessions from other countries revealed different genotypes. Only Dottato (Acc. No 234) and San Pietro (Acc. No 213), both from Italy, exhibit the same genotypic pattern in the present analysis.
The genetic structure of the fig tree cultivars was investigated by a Bayesian-based population assignment analysis using STRUCTURE software (Pritchard et al. 2000). The aim is to group different individuals into K different clusters based on genetic similarity, so that the individuals within a cluster are more similar to each other than to individuals outside the cluster. In our case, we do not have a clear prior hypothesis for the number of clusters to expect, and thus the clustering algorithm was run several times with different values of K. Each run gave a different probability reflecting the fit of the clusters to the data, assuming Hardy–Weinberg equilibrium. The K value that gives the best fit to the data is the best estimate of the true number of clusters, or populations. Our results show a clear maximum for ΔK at K = 3 (Fig. 2A). As shown in Fig. 2B, most cultivars were divided into three sub-populations.
Identification key for figs
In this study, an identification (Id) key was generated based on the discriminating power (Dj) of each SSR. The SSR loci were hierarchically ordered according to their Dj values. MFC3 was placed first, as it is the most discriminative among the seven SSRs (MFC6 was not included at this stage). As a result, the Id key produced was: MFC3-MFC1-MFC2-MFC7-MFC8-MFC5. MFC4, having the smallest Dj, was left out from the Id key since it was not able to differentiate further figs (Table 2). The sixty fig accessions in this study produced 1770 pairs of comparisons. Using the above Id key, only 12 from the 1770 pairs were not differentiated; 54 different fig genotypes were differentiated in the 60 fig accessions.
MFC6 was not included initially in the Id key due to the lack of banding pattern in 12 out of 60 fig accessions, regardless of its high Dj (0.85). Based on this Dj value, MFC6 would have been ranked first in the Id key, but this may introduce ambiguity due to the lack of the locus in 12 accessions. Therefore, we propose to introduce MFC6 at the end of the Id key (step 6, Table 2), where most figs are already differentiated. Using the MFC6 data, it was possible to further discriminate six fig accessions of the 12 non-differentiated pairs. If the absence of banding pattern for MFC6 is considered a genetic characteristic, then four more pairs could also be differentiated. As a result, from the 1770 pairs of fig accessions only two pairs could not be differentiated (Table 3); therefore, 58 different fig genotypes could be distinguished among the 60 fig accessions. In summary, the proposed identification key for figs is as follows: MFC3-MFC1-MFC2-MFC7-ΜFC8-MFC5-MFC6. MFC4 did not reduce the number of non-differentiated pairs (Table 2) and thus was not included in the Id key.