Analysis of Self-Incompatibility and Genetic Diversity in Diploid and Hexaploid Plum Genotypes

During the last decade, S-genotyping has been extensively investigated in fruit tree crops such as those belonging to the Prunus genus, including plums. In plums, S-allele typing has been largely studied in diploid species but works are scarcer in polyploid species due to the complexity of the polyploid genome. This study was conducted in order to analyze the S-genotypes of 30 diploid P. salicina, 17 of them reported here for the first time, and 29 hexaploid plums (24 of P. domestica and 5 of P. insititia). PCR analysis allowed identifying nine S-alleles in the P. salicina samples allocating the 30 accessions in 16 incompatibility groups, two of them identified here for the first time. In addition, pollen tube growth was studied in self-pollinated flowers of 17 Tunisian P. salicina under the microscope. In 16 samples, including one carrying the Se allele, which has been correlated with self-compatibility, the pollen tubes were arrested in the style. Only in one cultivar (“Bedri”), the pollen tubes reached the base of the style. Twelve S-alleles were identified in the 24 P. domestica and 5 P. insititia accessions, assigning accessions in 16 S-genotypes. S-genotyping results were combined with nine SSR loci to analyze genetic diversity. Results showed a close genetic relationship between P. domestica and P. salicina and between P. domestica and P. insititia corroborating that S-locus genotyping is suitable for molecular fingerprinting in diploid and polyploid Prunus species.


INTRODUCTION
Plums, belonging to the Prunus genus (family Rosaceae, subfamily Prunoideae, sub-genus Prunophora), have been cultivated for at least 2000-4000 years, being among the first fruit species to attract human interest and appear to have been domesticated very early (Gautier, 1977). In fact, one of the reasons for the frequent domestication of Prunus species might have been the coincidence between the location of the center of diversity of Prunus and the first ancient high civilizations of human history (OCDE, 2002). Within Prunophora, the section Euprunus contains the Asian (such as P. salicina Lindl. and P. simonii Carr.) and European (such as P. domestica L., P. cerasifera Ehrh., P. insititia L. and P. spinosa L.) plums, whereas the section Prunocerasus contains the North American species (such as P. americana Marsh., P. angustifolia Marsh., P. hortulana Bailey, P. munsoniana Wight & Hedr., P. maritima Marsh. and P. subcordata Benth).
The basic somatic chromosome number for Prunus is x = 8. However, natural interspecific hybridization is responsible for several cases of polyploidy in this genus (Das et al., 2011). Thus, chromosome numbers of different plum species varies from diploid to hexaploid: P. salicina, P. cerasifera, P. americana and P. simonii are diploid with 2n = 16, P. spinosa is tetraploid with 2n = 32 whereas P. domestica and P. insititia are hexaploid with 2n = 48. Polyploidy is of widespread occurrence in plants and a major mechanism of adaptation and speciation. It is thus recognized as a major force in evolution ( Van de Peer et al., 2017). Polyploidy is important, too, from a practical point of view, since polyploid plants are often more vigorous and may be more resistant to frost and the attacks of parasitic fungi than their diploid counterparts (OCDE, 2002). In contrast, the complexity of the polyploid genome makes molecular studies in these species a complex and elusive task.
Plums are mainly self-incompatible species and show, like other Prunus species, a Gametophytic Self-Incompatibility system (GSI). Within Rosaceae, the self-incompatibility mechanism (SI) is a widespread genetic system that promotes outcrossing by enabling hermaphrodite plants to avoid self-and cross-fertilization with close relatives. This system comprises recognition of self-related pollen, by cells of the pistil, followed by rejection of the incompatible pollen. In GSI, pollen tube growth is aborted during pollen tube growth in the style (De Nettancourt, 1977). GSI is the most common self-incompatibility system in angiosperms, and has been reported in more than 60 families of flowering plants (Kao and McCubbin, 1996). The recognition in GSI is genetically determined by a polymorphic locus (S), encoding at least two linked genes that determine the pistil and pollen phenotypes (Kao and Tsukamoto, 2004). In Prunus species, the S-locus is less than 70 kb (Ushijima et al., 2001) and contains two linked genes; one, S-RNase, is expressed in the pistil while the second, SFB (S-haplotypespecific F-box), is expressed in the pollen (Ushijima et al., 2001;Entani et al., 2003). Genotypes sharing the same Sgenotypes are genetically inter-incompatible and included in the same Incompatibility Group (IG), while cultivars having different S-genotypes are inter-compatible and are allocated to different IG. As consequence, assignment of each cultivar to its corresponding IG is essential for orchard planning and appropriate crosses in breeding programs (Guerra and Rodrigo, 2015;Halász et al., 2017).
In the last two decades, cloning and characterization of the S-locus genes in species of the Rosaceae have allowed the development of fast and accurate PCR methods for Sallele typing based on the polymorphism of amplified fragment length. This strategy does not require controlled pollinations to evaluate the percentage of fruit set in orchard conditions or microscopic observations of pollen tube growth in flowers of controlled crosses (Guerra et al., 2009;Guerra and Rodrigo, 2015;Matsumoto and Tao, 2016). Several PCR S-allele typing methods (Beppu et al., 2002(Beppu et al., , 2003 were widely and successfully developed in many Prunus species such as Japanese plum , sweet cherry , almond (Tamura et al., 2000), European apricot (Romero et al., 2004) and Japanese apricot (Yaegaki et al., 2001).
Publications regarding SI in polyploid plums are scarcer. Most of these studies were conducted to test whether polyploidy influences self-compatibility (SC). The limited works of Sgenotyping in polyploid plums might be explained by the fact that SI occurs less frequently in polyploid than in diploid species (Halász et al., 2014). The first study regarding the molecular genetics of SI in P. domestica (2n = 48) was published by Sutherland et al. (2004). Later, six S-locusspecific markers previously used for other Prunus species were successfully tested on 33 domestic plum cultivars (P. domestica) by Kota-Dombrovska and Lācis (2013). Preliminary information regarding the S-genotypes of 16 P. domestica cultivars grown in Hungary was conducted by Halász et al. (2014). More recently, Halász et al. (2017) provided an S-allele profiling of 17 SI polyploid Hungarian plums (10 P. spinosa L., 4 P. insititia L. and 3 P. spinosa X P. domestica) and identified the sizes of 23 S-RNase alleles using PaConsII consensus primers spanning the second intron of the S-RNase gene.
Due to the economic importance, risk of vulnerability and the high diversity of Tunisian plum germplasm, the first objective of this study was to establish the S-genotypes, the incompatibility groups and the self-(in)compatibility phenotypes of 17 Tunisian P. salicina plums with unknown S-genotypes by combining PCR analysis and observation of pollen tube growth in selfpollinated flowers. The second objective was to fingerprint the S-alleles in hexaploid plums. Thus, 29 polyploid plums (5 P. insititia from Tunisia and 24 P. domestica accessions from Spain) with unknown S-genotypes were S-genotyped for the first time. The third objective was to elucidate the pattern of genetic diversity based on nine SSR loci. Hence, the results of S-locus and those of SSR markers were combined to investigate the genetic relationships between the studied Prunus species.

Plant Material
A total of 59 plum accessions were used in this study. Twenty four accessions belonged to hexaploid P. domestica (2n = 2x = 48), eighteen of them conserved at the CITA-Aragon germplasm collection in Zaragoza (Spain) and six collected in La Palma Island (Spain). Additionally, five accessions belonging to hexaploid P. insititia (2n = 2x = 48) were collected from non-cultivated populations in Tunisia. Thirty accessions belonging to diploid P. salicina (2n = 2x = 16) were collected from Tunisia, thirteen of them from two collections (Sodon and Sadira) and represent commercial varieties, and seventeen from growers orchards. Information related to plant material is summarized in Table 1.

S-Locus Analysis
Young leaves from each of these accessions were used as the starting material to extract DNA using the cetyltrimethyl ammonium bromide (CTAB) following the protocol described by Doyle and Doyle (1987). DNA was quantified using a Nanodop 1000 and diluted to 10 ng/µL.
S-genotyping of plum cultivars was performed using three pairs of consensus primers. The second intron of the S-RNase gene was amplified using the primer pair PruC2/PCER Yamane et al., 2001). The primer pairs PaCons1F/PaCons1R2 (Sonneveld et al., 2006) and Fbox5'F/Fbox-IntronR (Vaughan et al., 2006) were used to amplify the first intron of the S-RNase gene and the SFB gene intron, respectively.
PCRs were performed in a volume of 15 µl, with 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 2 mM MgCl2, 0.1 mM of each dNTP, 0.2 µM of each primer, 40 ng of genomic DNA and 0.45 U of BioTaq TM DNA polymerase (Bioline, London, United Kingdom). All PCR reactions were carried out in an I-cycler (Bio-Rad Laboratories, Hercules, CA, United States) thermocycler using the following temperature profile: an initial denaturation step at 94 • C for 1 min, 35 cycles of 94 • C for 3 min, 56 • C for 1 min, and 72 • C for 3 min. A final extension was programmed at 72 • C for 7 min.
Fragments amplified with the PruC2/PCER primer combination, with sizes were greater than 500 bp, were separated using a 2% agarose gel electrophoresis, stained with SYBRGreen and visualized with UV light. Fragment size estimation was done using a size standard (1 kb DNA Ladder; Invitrogen, Carlsbad, CA, United States). Three P. salicina cultivars ("Fortune, " "Santa Rosa, " and "Beauty" with known S-genotypes (SbSc, ScSe, and SbSh, respectively) were used as references.
Fragments amplified with PaCons1F/PaCons1R2 andFbox5'F/Fbox-IntronR primers with sizes smaller than 500 bp, were analyzed by capillary electrophoresis. Forward primers were labeled with a fluorescent dye on the 5-end and PCR products were detected and sized with a Beckman Coulter Genome Lab GeXP TM capillary DNA analysis system. Samples were denaturalized at 90 • C for 120 s, injected at 2.0 kV for 30 s, and separated at 6.0 kV for 35 min.

Pollination Tests
To establish self-(in)compatibility in the Tunisian diploid cultivars, 17 genotypes were self-pollinated in the laboratory and pollen tube growth was observed under the microscope. Flowers of each genotype were collected at the balloon stage and pollen was obtained by manually removing and drying the anthers at room temperature during 24 h. The pollen was then sieved through a fine mesh and stored at −20 • C until required (Guerra et al., 2009;Guerra and Rodrigo, 2015).
To obtain the pistils, flowers were collected at balloon stage 24 h before anthesis. These flowers were emasculated by removing petals, sepals and stamens with fingernails, then the pistils were maintained on wet florist foam at laboratory temperature (Rodrigo and Herrero, 1996). On the following day, a group of 25-45 flowers was hand-pollinated using a fine paintbrush. Three days later, the pollinated pistils were fixed in FAA [70% ethanol: acetic acid: formaldehyde (18 : 1 : 1, v/v/v)] (Guerra et al., 2009).
For microscope preparations, the fixed pistils were washed three times for 1 h with distilled water and left in 5% sodium sulphite at 4 • C. To soften the tissues, the pistils were autoclaved at 1 kg/cm2 during 10 min in sodium sulphite (Jefferies and Belcher, 1974), and stained with 0.1% (v/v) aniline blue in 0.1 N K 3 PO 4 to stain callose (Linskens and Esser, 1957). Pollen tube growth in the style was observed under an Olympus BH2 microscope with UV.

SSR Analysis
The 59 plum genotypes were also analyzed using nine SSR primer pairs previously developed in Prunus (Supplementary Table S1). PCR reactions were carried out in a volume of 20 µl, with 20 mM Tris-HCl, pH 8.4, 50 mM KCl, 4 mM MgCl2, 0.1 mM of each dNTP, 0.2 µM of each primer, 40 ng of genomic DNA and 0.45 U of BioTaq TM DNA polymerase (Bioline, London, United Kingdom). PCR reactions were run in an I-cycler (Bio-Rad Laboratories, Hercules, CA, United States) thermocycler using the following temperature cycles: an initial step of 2 min at 94 • C, 35 cycles of 45 s at 94 • C, 45 s at 57 • C, 1 min at 72 • C, and a final step of 5 min at 72 • C. The obtained fragments were analyzed by capillary electrophoresis. Forward primers were labeled with a fluorescent dye on the 5-end and PCR products were detected and sized with a Beckman Coulter GenomeLabGeXP TM capillary DNA analysis system. Samples were denaturalized at 90 • C for 120 s, injected at 2.0 kV for 30 s, and separated at 6.0 kV for 35 min.

Diversity Analyses
To explore the genetic diversity of the studied cultivars, different diversity parameters of the S-locus and SSR markers were calculated with Arlequin ver. 3.5 (Excoffier et al., 2005) for each plum species. Detected alleles for both SSR and S-locus markers were scored as present (1) or absent (0). Genetic distances were calculated with Maximum Composite Likelihood (Tamura et al., 2011) in Popgene ver. 1.31 software, and then imported to Mega5 to construct a UPGMA tree. Numbers on major branches represent bootstrap support from 2000 replicates. For MDS analysis, genetic distances were calculated with the Euclidean index (Taguchi and Oono, 2005) with threedimensional coordinate system using the PAST software (Ver. 3.21) (Hammer et al., 2001).

S-Locus Analysis
S-Allele Typing in Diploid P. salicina PruC2/PCER markers, spanning the second intron of the S-RNase gene, showed a high degree of length polymorphism. Hence, nine fragments were amplified with sizes ranging from 400 to 1600 bp ( Figure 1A). The sizes of these bands were consistent with the S-alleles previously identified in P. salicina. As band sizes were larger than 500 bp, results were analyzed by agarose gel electrophoresis which make difficult to distinguish differences between some alleles with similar sizes such Sg-Sq (1250 and 1270 bp, respectively) and Sa-Sh (470 and 500 bp, respectively).
To confirm these results, a second round of amplification was conducted with the PaCons1F/PaCons1R2 primer pair, spanning the first intron of the S-RNase gene. Eight fragments were amplified with sizes ranged from 208 to 389 bp. Although the differentiation between Sg and Sq alleles was successful (Supplementary Table S2), it was not possible to distinguish between the Sa and Sh alleles since they showed similar sizes (388 and 389 bp).
Fbox5'F/Fbox-IntronR primer pair was additionally used to amplify the SFB gene intron. Capillary electrophoresis analysis showed nine fragments whose sizes ranged from 170 to 190 bp. Despite the reduced level of length polymorphism between bands (ranging between 1 and 2 bp), it was possible to differentiate the Sa (174 bp) and Sh (187 bp) alleles (Supplementary Table S2).
Hence, after combining the results of the three primer pairs used, nine S-alleles were identified in 17 Tunisian P. salicina genotypes of unknown S-allele composition ( Supplementary  Table S2). Additionally, the S-alleles of 13 P. salicina cultivars with previously reported S-genotypes were amplified in order to confirm the size of the S-alleles.
In all, 30 S-genotypes were reported, 17 of them for the first time. These cultivars have been assigned to 16 incompatibility groups (IG), three of them allocated to group 0 identified as rare genotypes and two new IG reported here for the first time (IG-F, IG-H) ( Table 2).

Pollination Experiments in Diploid P. salicina
To explore self-(in)compatibility of the 17 P. salicina accessions analyzed, self-pollinations were carried out in the laboratory. For each cultivar, 25 to 45 flowers were collected at balloon stage (Figures 2A.a,b), emasculated (Figures 2A.c,d) and hand pollinated with the help of a paintbrush. Then, pollen tube growth in the style was observed in self-pollinated flowers under the microscope. As control, self-pollinations were tested in two cultivars: "Santa Rosa" (ScSe), known to be self-compatible, and "Fortune" (SbSc), considered as self-incompatible. Pollen germination on the stigma was successful in the 17 cultivars tested (Figure 2B.a). In fifteen cultivars (IG-A-B-C-E-F-H-I), pollen tube growth was arrested at halfway the length of the style (Figures 2B.b,c). These self-pollination experiments confirmed the SI phenotypes of these plum accessions revealed by PCR ( Table 3).
As expected, in the cultivar "Bedri" (IG-G), which carries the Se allele, previously known as related to the SC phenotype (Beppu et al., 2005), pollen tubes reached the base of the style in 75% of self-pollinated flowers confirming the SC phenotype in this cultivar (Figure 2B.d). However, an unexpected result was observed in the three accessions of "Cidre" (IG-D). Pollen tube growth was arrested at halfway the length of the styles despite the identification of the Se allele by PCR, which suggested that "Cidre" is rather a self-incompatible cultivar ( Table 3).

S-Genotyping in Hexaploid P. insititia and P. domestica
The three tested primer pairs exhibited successful PCR amplifications for the 29 hexaploid plum accessions ( Figure 1B). Furthermore, the same fragments were amplified in both P. insititia and P. domestica.
PCR amplifications performed with PruC2/PCER showed a multiallelic profile with genotypes carrying up to six fragments ( Figure 1B). The S-typing revealed 12 alleles with sizes ranging between 300 and 1600 bp. These alleles were named with numbers to allow distinguishing them from the alleles of diploid plums, which are named with letters ( Table 4). Three alleles showed the same sizes than those reported in diploid plums: S 12 -Sb (1580 bp), S 9 -Sf (1090 bp) and S 1 -Sa (470 bp).
To confirm these results, PCR amplifications were conducted using PaCons1F/PaCons1R2 primers. Capillary electrophoresis revealed 11 fragments with sizes ranging from 228 to 413 bp ( Table 4). The use of the PaCons1F/PaCons1R2 primer pair allowed amplification of all S-alleles previously detected with PruC2/PCER, except the S4 allele for which no amplification with PaCons1F/PaCons1R2 was obtained. Moreover, the amplified alleles showed similar sizes between S 12 -Sb (368 bp) and S 9 -Sf (326 bp) alleles, but a difference of 15 bp was revealed between S 1 and Sa (374 and 389 bp, respectively).
A third round of amplification was conducted using the Fbox5'F/Fbox-IntronR primer pair. Amplified bands showed small differences in length (in order of 1 or 2 bp) which required several repetitions to ensure the exact size of each band. All the Salleles previously defined were detected with the Fbox5'F/Fbox-IntronR primer pair with sizes ranging between 172 and 208 bp ( Table 4), except the three S-alleles S 2 , S 8 and S 10 . In addition, the alleles S 12 and S 9 showed the same lengths than Sb (185 bp) and Sf (176 bp), respectively.

Genetic Diversity at the S Locus
Using the three S-markers, nine S-alleles have been identified in 30 diploid P. salicina cultivars; among them, the Sf allele was the most frequent (0.22). In hexaploid plums, among 12 S-alleles, S 1 (0.19) and S 10 (0.21) were the most frequent in P. domestica and P. insititia, respectively ( Table 5). In P. salicina, allelic richness (Ar) ranged between 0.11 and 0.48 with an average of 0.3. In P. domestica, Ar ranged between 0.35 and 0.49 with an average of 0.42, while in P. insititia it ranged between 0.37 and 0.51 with an average of 0.44 (Table 6).
A total of 32 S-genotypes were identified, 16 of them belonging to P. salicina, 11 to P. domestica and 5 to P. insititia. Genotypic diversity (Gd) showed high values and ranged between 0.54 and 1 in P. salicina (an average of 0.77), 0.72 and 1 in P. domestica (an average of 0.86) and 1in P. insititia (Table 6). This Gd depends on allele frequency in each genotype (whether the allele is frequent or rare) and on the number of accessions sharing the same genotype. These high GD reflected an important frequency of the heterozygous fraction. The observed heterozygosity (Ho) ranged from 0.8 to 0.94 in diploid plums with an average of 0.87. In hexaploid plums, Ho ranged between 0.83 and 0.91 with an average of 0.86 for both P. domestica and P. insititia. These values were higher than the expected heterozygosities (He), which ranged between 0.71 and 0.78 in diploid plums with an average of 0.74 and between 0.74 and 0.77 in hexaploid plums with an average of 0.75. Consequently, negative values of Wright's fixation index were observed in P. salicina (f = −0.18) as well as in hexaploid plums (f = −0.15) ( Table 6).
The differentiation index (F ST ) between the three groups was 0.19. F ST values were statistically significant between P. domestica and P. salicina and between P. insititia and P. domestica (0.18, 0.16, respectively with P < 0.05); however, F ST values were not significant between P. salicina and P. insititia (0.23, 0.05 < P < 0.1) ( Table 6). In fact, a value of F ST ranging between 0.15 and 0.25 indicated that the analyzed populations are genetically different, independently from their geographic origins (Weir and Cockerham, 1984).

SSR Analysis
The nine SSR markers proved to be highly polymorphic in the three species tested. The BPPCT012 primer pairs amplified the highest number of alleles while the UDP96-008 amplified the lowest number of alleles in the different species.
In diploid P. salicina, 94 alleles (Ao) were amplified in the 30 accessions resulting in a value of 10.5 alleles per locus (Al) ( Table 7). A total of 33.1 of them represent effective alleles (Ae) indicating that an important number of alleles appeared at low frequency (less than 5%) which explains the high allelic richness observed (Ar = 0.78). The loci analyzed allowed identifying 24 genotypes ( Table 7).
In hexaploid P. domestica, the number of amplified alleles per genotype ranged from 1 to 6. The used SSR markers amplified 195 alleles (Ao) giving a value of 21.6 alleles per locus (Al) with 50.4 effective alleles (Ae) leading to an important allelic richness (Ar = 0.82). With the whole set of nine SSRs used it was possible to identify 20 different genotypes among the 24 cultivars studied ( Table 7).
In hexaploid P. insititia, 73 alleles were amplified (Ao) in the 5 tested cultivars with an average of 8.1 alleles per locus (Al). A total of 29.3 were defined as effective alleles (Ae) reflecting the high allelic richness (Ar = 0.73). The five accessions had different genotypes using the whole set of SSR markers (Gn = 5) ( Table 7).
In the three tested species, observed heterozygosity values were higher than those of expected heterozygosity leading to negative values of f (−0.06). In P. salicina, the observed heterozygosity ranged between 0.65 and 0.78 with an average of 0.71, whilst the expected heterozygosity ranged between 0.61 and 0.73 with an average of 0.67. In P. domestica, the observed heterozygosity varied from 0.71 to 0.76 with an average of 0.73, while the frequency of expected heterozygosity ranged between 0.66 and 0.73 with an average of 0.69. In P. insititia, the average of observed heterozygosity was 0.74 and the expected heterozygosity was 0.7 (Table 7).

Genetic Relationships Among the Accessions
Based on the defined S-alleles, the constructed UPGMA tree divided the cultivars into two heterogeneous groups (Figure 3A.a). The first group (G1) clustered together Japanese plums (P. salicina) with some European plums (P. domestica). This result suggests a close genetic relationship between P. salicina and P. domestica. The second group (G2) grouped the remaining P. domestica accessions together with P. insititia plums.
Multidimensional scatter plots (MDS) yielded similar results than the UPGMA tree (Figure 3A.b). Using dimensions 2 and 3, the first MDS plot (1) clearly separated P. insititia accessions from P. domestica and P. salicina, and confirmed a  close genetic relationship between P. domestica and P. salicina. In contrast, using dimensions 1 and 2, the second MDS plot (2) showed a close relationship between European plums (P. domestica and P. insititia), while Japanese plums (P. salicina) were relatively separated. Based on SSR loci, the UPGMA dendrogram (Figure 3B.a) showed the same topology than that constructed with S-markers and clearly divided the tested accessions into two main groups: G1 and G2. The first group (G1) gathered P. salicina and P. domestica accessions. The first sub-group in that main group clustered the diploid P. salicina accessions and the second sub-group gathered 10 P. domestica accessions. The remaining P. domestica accessions and the five P. insititia cultivars fell into the second group (G2). In addition, MDS plots were drawn based on these SSR markers. Using dimensions 1 and 2, the first MDS plot (1) confirmed the close genetic relationship between P. salicina and P. domestica and between P. domestica and P. insititia. Using dimensions 2 and 3, the MDS plot (2) showed an overlap between the three species (Figure 3B.b).

DISCUSSION
This study was conducted in order to report the S-genotypes of 59 plum accessions covering two levels of ploidy: diploid P. salicina (2n = 16) and hexaploid P. domestica and P. insititia (2n = 46). Among them, thirteen P. salicina accessions were previously Sgenotyped and 46 accessions were analyzed for the first time. In addition, S-locus and SSR data were combined to investigate the genetic relationships between the species studied.

Applicability of Intron Length Polymorphism Technique (ILP) in Diploid and Hexaploid Plums
The ILP technique, the most common method for S-allele profiling in Prunus species (Sonneveld et al., 2003), was used with three S-locus specific consensus primer pairs. Two primer pairs amplified the first and second introns of the S-RNase coding region and one primer pair amplified the SFB intron. These markers were developed and used earlier to study SI in several Prunus species (Ushijima et al., 1998;Tao et al., 1999;Sutherland et al., 2004;Sonneveld et al., 2006;Vaughan et al., 2006;Guerra et al., 2009;Herrera et al., 2018) and showed good applicability for plum germplasm characterization and cultivar discrimination which confirmed the high conservation of S-genes among Prunus species and primer pair transferability among species (Kota-Dombrovska and Lācis, 2013).
PruC2/PCE-R, spanning the second intron of the S-RNase gene, amplified up to six fragments in P. salicina, P. insititia and P. domestica. Although it has been suggested that primers

Variety
Detected S-genotype

Prunus insititia
Meski Kbira Kahla S 6 /S 7 /S 8 /S 10 Meski Sghira Hamra S 6 /S 7 /S 10 /S 11 Chaaraouia S 7 /S 8 /S 10 /S 11 Meski Kbira Hamra S 6 /S 5 /S 8 /S 10 Zenou S 1 /S 2 /S 10 /S 11 specific to the second intron of S-RNase gene were not suitable for S-genotyping (Sutherland et al., 2007), this primer pair was sufficient for successful genotyping in the species tested. Primer pairs PaCons1F/PaCons1R2, specific to the first intron of the S-RNase gene, and Fbox5'F/Fbox-IntronR, spanning the SFB gene intron, showed limited polymorphism and amplified alleles with similar sizes, which make the amplified fragments hardly distinguishable on agarose gel electrophoresis. However, these markers should be employed to ensure a more accurate characterization. When amplification results are analyzed by capillary electrophoresis, these primers can be used as supplementary markers in the characterization of plum genetic resources.

Self-Compatibility in Japanese Plums
Pollination experiments provided complementary information to the PCR amplification in order to establish the self-(in)compatibility phenotypes of Tunisian diploid plums. Although Japanese plum is known to be SI, the SC reversion is not excluded in this species as in other Prunus species. The SC in species of this genus has been associated to particular S-alleles; examples include apricot (Vilanova et al., 2006;Halász et al., 2014;Herrera et al., 2018), almond (Fernandez i Marti et al., 2014;Kodad et al., 2015), peach Hanada et al., 2014), sour cherry  and sweet cherry (Wünsch and Hormaza, 2004;Cachi and Wünsch, 2014). In P. salicina, the SC was firstly related to the Se haplotype expression (Beppu et al., 2005(Beppu et al., , 2010Guerra et al., 2009;Guerra and Rodrigo, 2015), for which an accumulation of insufficient levels of Se-RNase due to very low transcription levels in the pistil has been described Guerra et al., 2009). After that, other S-haplotypes have been identified as SC related, such as Sb (Guerra et al., 2009;Beppu et al., 2010), St (Beppu et al., 2012a) and Sg (Beppu et al., 2012b). Pollen tube growth and PCR analysis showed that "Bedri, " carrying the Se allele, is a SC cultivar, which may confirm the implication of this allele in SC. In contrast, three accessions of "Cidre, " also carrying the Se allele, were revealed as SI. Similar results were observed in previous studies where some Japanese plum cultivars, described as SI, which carried one of the alleles related to the SC in their S-genotypes (Beppu et al., 2002(Beppu et al., , 2003Guerra et al., 2009;Guerra and Rodrigo, 2015). Thus, SC in Japanese plum could be also related to a double phenotypic expression of the same allele, as it has been reported in almond (Fernandez i Marti et al., 2010, than associated uniquely with specific S-alleles (Guerra and Rodrigo, 2015).

S-Genotyping in Hexaploid Plums
Sixteen S-genotypes were assigned for polyploid individuals. Only one genotype carrying six alleles (IG-3) was defined, while the remaining S-genotypes carried four or five alleles. Similar results were observed using different molecular approaches (Sgenotyping and SSR markers) in polyploid Prunus species such P. domestica, P. spinosa, and P. insititia (Kota-Dombrovska and Lācis, 2013;Halász et al., 2014Halász et al., , 2017Urrestarazu et al., 2018;Baraket et al., 2019). These incomplete S-genotypes in some accessions can be the consequence of the multiple appearance of a given allele (Walsh et al., 1992;Kota-Dombrovska and Lācis, 2013). This homozygosity leads to the question of whether polyploidy can cause SC in these species. Hence, direct association between polyploidy and SC has not been observed in some polyploid Prunus species. In the allotetraploid P. cerasus, the heteroallelic pollen retained its SI phenotype and SC was related to the co-expression of two non-functional S-haplotypes in the same genotype (Hauck et al., 2006). Additionally, polyploidy seemed not to cause SC in P. spinosa and P. insititia (Nunes et al., 2006). In contrast, polyploidy influences SC in Solanaceae and Scrophulariaceae (Mable, 2004;Nunes et al., 2006;Vieira et al., 2008). In this work, an average of 0.41 alleles per polyploid accession was reported using three S-markers. This number is relatively low for species whose allopolyploid genome structure tends to increase the number of alleles and promote the emergence of new alleles (Ainouche and Wendel, 2014;Halász et al., 2017). Additionally, 23 S-alleles were assigned in 17 polyploid plums in Halász et al. (2017). This result may be explained by the low heterogeneity of our tested samples. Halász et al. (2017) used the PaConsII primer pair, spanning the second intron of the S-RNase gene. Compared to the result of PruC2/PCE-R specific to the same region, four S-alleles (S 2 , S 4 , S 6 , and S 11 ) showed sizes similar with S R , S Z , S S , and S J , respectively (Halász et al., 2017). In addition, four alleles (S 1 , S 3 , S 5 , and S 9 ) showed differences less than 20 bp with S O , S B , S D , and S M , respectively, and two alleles (S 7 and S 12 ) showed differences less than 80 bp with S I and S V , respectively (Halász et al., 2017).

SSR Polymorphism
The level of SSR polymorphism observed in 30 P. salicina (10.5) is comparable to that revealed in 29 P. salicina accessions using eight SSR loci (12) by Carrasco et al. (2012). This variability is higher than that reported for P. salicina in other works, i.e., Mnejja et al. (2004) and Klabunde et al. (2014). It is also higher than that observed in other Prunus species such as almond (Fathi et al., 2008), cherry (Wünsch and Hormaza, 2002;Baris et al., 2017) and apricot (Gürcan et al., 2015). Due to their polyploid genomes, few studies have focused on evaluating the genetic diversity of hexaploid plum species (Horvath et al., 2011;Sehic et al., 2015;Halász et al., 2017;Urrestarazu et al., 2018). A high degree of polymorphism has been revealed in hexaploid plums in previous works (Urrestarazu et al., 2018).
In general, among the Prunus genus, plums are considered as one of the most polymorphic species (with almond) (Salazar et al., 2014) due to their SI system. In SI populations, the fact that self-fertilization is avoided decreases the probability of obtaining homozygote offspring (Balloux, 2004;Stoeckel et al., 2006) and tends to increase the genetic diversity as well as heterozygosity levels elsewhere in the genome (Hartl and Clark, 1997;Stoeckel et al., 2006).

Genetic Richness and Allelic Singularity of Tunisian Plums
The characterization of 17 Tunisian P. salicina plums has allowed assigning them into nine IG (IG-A→IG-I). Three Sgenotypes, previously defined as rare genotypes and included under IG-0 (Beppu et al., 2003;Guerra et al., 2009;Guerra and Rodrigo, 2015) were amplified in ten accessions and assigned to IG C, D, and L. Additionally, two new genotypes (IG F and H) were reported here for the first time. SSR results showed a significant genetic diversity in the P. salicina tested genotypes. These results highlight the allelic singularity and the rich variability characterizing the Tunisian plum germplasm as an unexplored source of genetic variation. In fact, several Tunisian Prunus varieties have disappeared during the last years due to the lack of programs addressed to preserve these local germplasm (Nabli, 2011;Abdallah et al., 2016;Baraket et al., 2019), which require the establishment of appropriate germplasm collections. The results obtained in this study could complement previous work of Baraket et al. (2019) in which 23 Tunisian plums accessions were phenotypically and genetically characterized.

Genetic Relationships in Prunus Species
Phylogenetic tree and Multidimensional scatter plots (MDS) for both S-locus and SSR loci underlined genetic relationships between the three plum species studied, especially between P. domestica and P. salicina and between P. domestica and P. insititia. Similar results were reported in previous works such in Hegedus and Halasz (2006) where the same alleles were reported in P. domestica, P. domestica × P. cerasifera hybrids, P. spinosa and P. salicina.
The close genetic relationship found between P. domestica and P. insititia is expected. In fact, P. insititia has been reported in wild form in three continents (southern Europe, Western Asia and northern Africa) (Malcolm, 2006). Although the ancestral form of this species remains still unknown, P. insititia has often been proposed to be a subspecies of P. domestica and chloroplast DNA analysis seems to corroborate this hypothesis (Reales et al., 2010).
The origin of P. domestica has also been debated for over a century (Crane and Lawrence, 1929;Bajashvili, 1991;Eryomine, 1991;Zohary, 1992;Hegedus and Halasz, 2006;Liu et al., 2007;Zhebentyayeva et al., 2019). The most accepted hypothesis is that of Crane and Lawrence (1929) in which P. domestica would be an allopolyploid hybrid species between tetraploid P. spinosa (2n = 2x = 32) and diploid P. cerasifera (2n = 2x = 16). Thus, P. domestica is believed to have arisen in the Caucasian region where these species are originated (Crane and Lawrence, 1929) via either chromosome doubling of the hybrid triploid or as a product of unreduced gametes from both parents (Liu et al., 2007). Bajashvili (1991) suggested that European plum could have also been originated directly from hexaploid P. divaricata (=P. cerasifera) or from P. spinosa by alloploidization. Moreover, Zohary (1992) reported that comparative morphology and available cytogenetic evidence do not support the assumption that P. spinosa contributed its two genomes (or even a single genome) to hexaploid P. domestica and suggested that P. domestica would be an autopolyploid P. cerasifera (Zohary, 1992). Eryomine (1991) proposed that P. domestica originated by artificial selection by early Eurasian societies and that several species participate in the origin of P. domestica probably explaining why P. domestica has never been found in the wild. He suggested that hybrids between (P. cerasifera × P. salicina) × P. spinosa and (P. cerasifera subsp. macrocarpa) presented the highest morphological similarities with P. domestica. In addition, P. cerasifera (CC) × P. salicina (SaSa) hybrids developed unreduced gametes (Eryomine, 1991). Hence, the genotypic formula suggested by Eryomine (1991) for P. domestica is SSCCCSa, where part S represents the genome of P. spinosa, part C represents the genome of diploid P. cerasifera and part Sa represents the genome of diploid P. salicina inherited into domestic plum. Recent results seem to support the hypothesis that P. domestica could be a hybrid of interspecific hybrids including P. cerasifera, P. spinosa, and other Eurasian plum species (Zhebentyayeva et al., 2019). Additionally, the close genetic relationship between P. domestica and P. salicina obtained by S-locus and SSR markers may represent an argument in favor of Eryomine proposal and may confirm that diploid P. salicina contributed to P. domestica formation.

CONCLUSION
The main objective of this study was to analyze the S-genotypes of 30 diploid P. salicina and 29 hexaploid plums (24 of P. domestica and 5 of P. insititia). The results obtained are relevant at three levels: (i) The identification of new S-alleles and new S-genotypes in the Prunus species analyzed should facilitate future work focused on exploring and understanding the SI system in diploid and, especially, in polyploid Prunus species. (ii) The identification of S-genotypes and the genetic characterization of Tunisian local plums showed that the Tunisian germoplasm could represent a new source of variability useful in future breeding programs. In addition, such results could be used for defining a Tunisian core collection aiming to preserve the extant diversity in local Tunisian plums. (iii) Combined with SSR markers, the information obtained from the S-locus may supply additional information regarding the genetic relationships among Prunus species.

AUTHOR CONTRIBUTIONS
DA conducted the experiments and statistical analyses, developed the genetic analyses, and wrote the manuscript. GB provided some plant material, and discussed and corrected the content. VP provided some plant material and discussed the results. SBM conducted the pollination experiments and the microscope observations. AS-H and JH provided experimental instructions, supervised the work, and assisted in writing the manuscript.

FUNDING
This research was supported by the Tunisian "Ministère de l'Enseignement Supérieur et de la Recherche Scientifique" and "Ministerio de Economía y Competitividad -European Regional Development Fund, European Union (AGL2016-77267-R)."