Ag cola Development of an SSR-based identification key for Tunisian local almonds

Ten simple sequence repeat (SSR) loci were used to study polymorphism in 54 almond genotypes. All genotypes used in this study originated from almond-growing areas in Tunisia with different climatic conditions ranging from the sub-humid to the arid and are preserved in the national collection at Sidi Bouzid. Using ten SSR, 130 alleles and 250 genotypes were revealed. In order to develop an identification key for each accession, the data were analysed separately for each microsatellite marker. The most polymorphic microsatellite (CPDCT042) was used as a first marker. Two microsatellite loci (CPDCT042 and CPDCT025) were sufficient to discriminate among all accessions studied. Neighbour-joining clustering and principal coordinate analysis were performed to arrange the genotypes according to their genetic relationships and origin. The results are discussed in the context of almond collection management, conformity checks, identification of homonyms, and screening of the local almond germplasm. Furthermore, this microsatellite-based key is a first step toward a marker-assisted identification almond database.


Introduction
Almond [Prunus dulcis (Miller) D.A. Webb, syn.Prunus.amygdalus Batsch] belongs to the Prunoideae subfamily of Rosaceae, which includes several other species producing fruits of economic importance.Al� mond is a diploid species with 2n = 16 and a small genome size of ∼0.3 pg/1C (Dickson et al., 1992).Traditional almond plantations in Tunisia are char� char� acterised by old selected cultivars such as 'Achaak', 'Abiodh', 'Blanco', 'Fekhfekh', 'Khoukhi', 'Ksontini' and 'Zahaaf'.Local germplasm consists of numerous ecotypes (most unknown to consumers) selected by farmers for high production and adaptation to specif� ic agro�ecosystems (Gouta et al., 2010a).Thus, it was necessary to identify these genotypes as a first step in their conservation and protection against potential genetic erosion.
SSR markers have been used for a genetic di� versity assessment of Tunisian almond germplasm and to determine its allocation in comparison to some Eu� ropean and American cultivars (Gouta et al., 2010b).However, the relatedness among local cultivars was very briefly described.In this study, we examined the relatedness among this germplasm more closely and develop an identification key based on microsatellite polymorphism.

Materials and Methods
Fifty�four almond accessions of different origins were analyzed (Table 1).Most were identified in vari� ous Tunisian growing regions, while others were already preserved in the national collection at Ettaous.Four were of unknown origin and were included in the study set to clarify their origin.All 50 Tunisian local genotypes originated from the regions of Bizerte, Sidi Bouzid, Sfax, Nefta, and Tozeur (Figure 1).Leaf samples for DNA extractions were collected from farmers' fields, but all genotypes were preserved in a new national germplasm collection at Sidi Bouzid.
Young leaves were collected from all accessions for DNA extraction.Total genomic DNA was isolated as described (Doyle and Doyle, 1987).DNA quality was examined by electrophoresis in 0.8 % agarose gel and DNA concentration was quantified by spectrophotom� eter.Extracted DNA was diluted to 5 ng µL -1 with Tris� EDTA buffer (1 mM Tris�HCl, 0.1 mM EDTA, pH 8.0) and stored at �20°C for polymerase chain reaction (PCR) amplification.
DNA was amplified by PCR using ten microsatel� lite primer pairs (Table 2).Pairs one to nine were derived Sci.Agric.v.69, n.2, p.108-113, March/April 2012 from a library enriched for AG/TC motifs constructed from the almond cultivar 'Texas' (Mnejja et al., 2005).The final pair (number 10) was described previously (Joobeur et al., 2000).Amplifi cation reactions were car� Amplification reactions were car� ried out in a final volume of 15 µL containing 10 ng of template DNA, 1 × reaction buffer (20 mM (NH 4 ) 2 SO 4 ,  75 mM Tris�HCl, pH 8.8), 2 mM MgCl 2 , 50 µM each of dATP, dGTP, dTTP, dCTP (Amersham Pharmacia Bio� tech, Spain), 0.15 mM forward and reverse primers, and 0.5 U Tth DNA Polymerase (Biotools Band M Labs, S.A., Spain).PCR amplifications were carried out in a Gene Amp 2700 thermocycler (Applied Biosystems, CA, USA) using the following temperature cycles: one cycle of 3 min at 95 °C; 35 cycles of 1 min at 94 °C, 45 s at the cor� responding annealing temperature, and 1 min at 72 °C.133,141,151,153,155,157,161,163,171,175 27 CPDCT025 15 162,164,172,174,176,182,184,186,188,190,192,194,196,198,200 31 CPDCT027 11 156,158,160,162,164,166,172,174,176,198,180 19 CPDCT033 11 116,120,126,128,130,132,134,136,138,142,150 19 CPDCT038 12 147,149,155,161,163,169,171,177,179,181,185,197 21 CPDCT040 11 138,146,156,160,162,164,166,168,170,172,174 19 CPDCT042 18 160,162,164,166,170,172,174,176,178,182,184,186,188,190,192,194,198,200 36 CPDCT044 16 161,165,169,171,173,175,177,179,181,189,191,193,195,197,199,205 21 CPDCT047 15 170,174,176,182,184,186,190,192,194,198,204,206,212,214 The last cycle was followed by a final incubation for 7 min at 72 °C and the PCR products were stored at 4°C until analysis.Two independent SSR reactions were per� formed for each DNA sample.The DNA amplification products were loaded on 5 % polyacrylamide sequenc� ing gels.Gels were run for 2 h at 65 W and then silver� stained as described (Bassam et al., 1983).Fragment sizes were estimated using 30�330 bp AFLP ladder DNA sizing markers (Invitrogen, Carlsbad, CA, USA) and ana� lyzed by the Quantity One program (Bio Rad, Hercules, CA, USA).The genetic relatedness among Tunisian almond cultivars was described using a phylogenetic analysis.To represent the differences among individuals and con� struct a phylogenetic tree, the simple matching distances were calculated with d ij : dissimilarity be� tween units i and j; L: number of loci; p: ploidy; and ml: number of matching alleles for locus l.The individ� ual distance tree was constructed using Darwin 5.0.148software (Perrier and Jacquemoud�Collet, 2006) and the neighbor�joining method of Saitou and Nei (1987).The robustness of each node was evaluated by bootstrapping data over loci for 10,000 replications.A principal coordi� 10,000 replications.A principal coordi� nate analysis based on the dissimilarity matrix was also performed with the same software.

Results and Discussion
The 10 microsatellite primer pairs revealed 130 al� leles and 250 potential genotypes among the 54 almond accessions studied (Table 2).Overall, primer pairs showed alleles in size ranges larger than those reported (Mnejja et al., 2005).Unique genotypes for all 54 cultivars identified a subset of the best 10 microsatellite primers for almond cultivar differentiation.To develop the identification key, the data were analysed separately for each microsatellite marker.The most polymorphic marker was chosen as the principal marker.The remaining markers were used to separate the genotypes in groups created by the previous marker until all accessions were clearly identified.We be� gan by selecting the most polymorphic loci that revealed the most different genotypes, CPDCT042 (36) and CPD� CT025 (31).We based the identification key first on the primer CPDCT042 and then on CPDCT025.These two SSRs discriminated among all 50 Tunisian genotypes as well as the four of unknown origin.Theoretically, these two loci could encompass a total of 36 × 31 = 1116 possi� ble genotypes, suggesting that there is room to expand our key to discriminate more genotypes.The most polymor� phic SSR primer, CPDCT042, allowed unambiguous dif� ferentiation of 28 of the 54 studied cultivars (52 %).The use of the additional CPDCT025 primer pair was required to identify the remaining cultivars.An identification key was thus established for these local almond accessions (Figure 2).
A similar identification key was obtained for 49 Tunisian date palm cultivars (Phoenix dactylifera L.) based on three microsatellite primers, revealing 25 al� leles and 57 genotypes (Zehdi et al., 2004).For the Tuni� sian apricot landraces, 26 Prunus microsatellite primers formed an identification key for 54 genotypes (Krichen et al., 2006).With only five primers, it was possible to discriminate among all landraces studied, identifying 103 alleles and 155 different genotypes.In fig (Ficus carica L.), it was not possible to discriminate among all 72 Tunisian local ecotypes with six SSR primers, but the identification key revealed 58 alleles and 124 genotypes, for a resolving power of 97.2 % (Saddoud et al., 2007).An identification key for 26 Tunisian olives (Olea europea L.) was successful in discriminating among all local cultivars using three of ten SSR markers (Taamalli et al., 2008).Our finding that in almond, the alleles of only two loci were sufficient to discriminate among all the acces� sions is probably because they were generated from an AG/TC�enriched library constructed from a cultivar of the same species.Thus, while transfer of SSRs among different species in the genus Prunus is well documented (Mnejja et al., 2005;Mnejja et al., 2010), our work stress� es the greater efficacy of using SSRs generated from the same species for genotyping studies.The identification key was used to detect differ� ences among cultivars with the same name.For instance, three accessions of 'Guernghzel' ('Guernghzel', 'Guerng� hzel C.H.', and 'Guernghzel B.N.') that originated from two regions (Sfax and Sidi Bouzid) were distinguished on the basis of one microsatellite primer.In the same way, the accessions 'Khoukhi' and 'Khoukhi Bizerte' were identified as homonyms.Further investigation showed that the accession 'Khoukhi' originated from the region of Bizerte.Additional homonyms were detected for the cultivar 'Bouchouka' (K.F. and B.S. from Sidi Bouzid), which can be distinguished by one primer.Similar cases of homonymy were detected in Tunisian almond germ� plasm on the basis of RAPD analysis (Gouta et al., 2008), in apricot and grapevines using microsatellites (Krichen et al., 2006;Ulanovsky et al., 2002), and in a fig col� lection using morphological traits (Mars et al., 1998).Thus, the identification key for almond established in this study is a continuation of an effort that was started several years ago for the molecular characterisation of the main Tunisian fruit species.
A Neighbor�joining phylogenetic tree based on the simple matching distance clustered the 54 almond geno� types into five major groups: A, B, C, D, and E (Figure 3).
Cluster A grouped 14 genotypes, all from northern Tuni� sia, with 'Mahsouna' from Sfax; 'Tlili 2', 'Tlili 5', 'Tlili 6', and 'Tlili 9' from Sidi Bouzid; and two of the unknowns, 'B200' and 'B202'.The close relatedness revealed between 'Blanco' and 'Dillou' from one side and 'Khoukhi' and 'Abiodh Ras Djebel' from the other side was supported by high bootstrap values of 93 and 98 %, respectively.Moreover, the genotypes 'Blanco' and 'Dillou' shared at least one allele for each SSR used in this work (14 of 20 total alleles) (data not shown).This supposes a common parentage for these two local cultivars of unknown ped� igree.The presence of 'B200' and 'B202' in the group implies genetic closeness between these two genotypes and the local cultivars of cluster A. This also suggests a different origin for these genotypes from the two other unknowns, 'B203' and 'B204', which were included in cluster C.This cluster was the largest, with 18 cultivars.The relatedness between 'B203' and the local cultivar 'Ksontini B.', supported by both a bootstrap value of 99 % and 16 of 20 common alleles, supposes a common parentage and consequently a Tunisian origin for this unknown genotype.
Four of the five most commonly planted local cul� tivars, 'Achaak', 'Fekhfekh', 'Guerneghzel', and 'Kson� tini', were included in cluster C; the exception was 'Zahaaf', present in group D. The low similarity (0.75 dissimilarity) between 'Achaak' and 'Zahaaf' observed in this study agrees with previous results (Fernández Marti et al., 2009).The repartition of the different eco� types 'Tlili' 1 to 9, which originated from the same area (Sidi Ali Ben Aoun in Sidi Bouzid) into all five clusters highlights the importance of an underestimated local diversity and stresses the importance of a continuous collecting effort.The most distant cluster, E, includes three cultivars from Sfax: 'Abiodh de Sfax', 'Elloumi', and 'Sahnoun' and three from Sidi Bouzid: 'Forme en Poire', 'Merghad H.1' and 'Tlili 8'.There is no clear separation between cultivars originating from central and southern Tunisia.This can be explained by their close proximity, while Sfax and Sidi Bouzid have com� mon borders (Figure 1) and their cultivars might have a common origin.Moreover, commercial exchanges between these regions are well documented since an� cient times.
Principal coordinate analysis (PCA) generated two clearly important components, PC1 and PC2, which ex� plained 11 and 7.3 %, respectively, of the total varia� tion in SSR data (Figure 4).This analysis showed some well�defined distribution patterns and relationships among the accessions.The divergence of all northern cultivars (Bizerte) was clearly demonstrated by PC1.In addition, a group composed of 'Abiodh Ras Djebel', 'Blanco', 'Dillo', 'Khoukhi', and 'Khoukhi Bizerte' could be clearly separated from the other accessions.The PC2 principal coordinate separated two distinct groups.The first contained genotypes from Sfax and Sidi Bouzid ('Grosse Tendre de Sfax', 'Bouchouka B.S.', and 'KF.3').The second had cultivars originating from Sfax and Bizerte ('Mahsouna' and 'Porto Farina').The neighbour joining analysis placed these two groups in cluster D and cluster A, respectively (Figure 3).A clear distinction between almond genotypes from northern and southern Tunisia is clearly demon� strated in this paper and supported by previous results (Gouta et al., 2010b).This in turn suggests the presence of two almond gene pools in Tunisia, associated with the geographic position of the cultivars, and implying a dif� ferent initial ancestry.Only two of 10 microsatellite loci were sufficient to distinguish among all Tunisian almond cultivars analysed in this study.The remaining microsat� ellite primers allow assessment of genetic relationships between the studied cultivars.
The homonyms elucidated through this paper illustrate the confusion in nomenclature that can be observed in a given region or even between regions in a species such as almond, which adapts well to dif� ferent climates.Our identification key will aide the description, registration, and certification of plant ma� terial and facilitates the rational management and con� servation of Tunisian almond germplasm.Moreover, since many ecotypes are cultivated world wide and exchanges of materials among breeders are common, the availability of an efficient SSR�based identification system is of interest.

Figure 1 -
Figure 1 -Location of the collection areas for the cultivars used in this study.

Figure 2 -
Figure 2 -Identification key and genotypes of 50 Tunisian and four unknown almond cultivars based on two microsatellite markers: CPDCT025 and CPDCT042.

Figure 3 -
Figure 3 -Unweighted neighbor-joining tree based on the simple matching dissimilarity matrix of 10 SSR markers for 54 almond genotypes.The numbers on the tips indicate bootstrap values expressed in percentages and are shown for all clusters when ≥ 50 %.

Figure 4 -
Figure 4 -Plot of the first two components (PC1 and PC2) of the principal coordinate analysis on the dissimilarity matrix obtained for 54 almond accessions using 10 SSR markers.

Table 1 -
Zone of origin and names of 50 Tunisian almond cultivars used to create the identification key.

Table 2 -
Locus name, number of alleles, allele sizes, and number of genotypes identified by using 10 SSR markers of Prunus species on 54 almond cultivars.