Genetic and ampelographic characteriza­ tion of grapevine accessions maintained in the Lebanese national collection

: Safeguarding grapevine biodiversity is one of the main concerns in viticulture today. Management of ex situ collections requires a comprehensive characterization of the conserved germplasm to provide basic material for selection and breeding programs. In this study, the local grapevine germplasm conserved at the national collection of the Lebanese Agricultural Research Institute and composed of 43 accessions,was submitted to a genetic and ampel­ ographic study. Nine ISSR primers, previously developed in grapevine, were used in this study. These primers generated a total of 51 bands, of which 77.7% were polymorphic allowing the diﬀerentiation of 41 genetic proﬁles vs. one case of synonymy that was recorded among three accessions. Ampelographic characterization was conducted using a set of 33 descriptors established by the International Oﬃce of Vine and Wine and related to leaf, bunch and berry. Principal component analysis identiﬁed 12 descriptors i.e. veraison date, matu­ rity date, berry length, upper and lower vein pigmentation, bunch density, bunch weight, sugar


Introduction
One of the main concerns for public research in viticulture today is the (*) Corresponding author: mmassaad1@gmail.com need to safeguard grapevine biodiversity by estab lishing well managed national repositories to ensure germplasm availability to breeders, researchers and farmers (Nass et al., 2012). There are presently around 130 grapevine germplasm collections across the word (Dettweiler et al., 2000). One of the first grapevine germplasm collections is the one estab lished in Spain in 1950, which contains more than 1066 accessions (Ortiz et al., 2004). In France, more than 2,200 varieties originally collected from 35 countries are grouped together at the National Institute of Agronomic Research (Tessier et al., 1999). Similar collections of grapevine were also established USA and in Latin America containing grapevine acces sions collected all over the world (Martinez et al., 2003). Grapevine is one of the oldest fruit crops growing in the Eastern Mediterranean region includ ing Lebanon (Zohary, 2003). The country was among the first worldwide to have implemented vineyards which gradually became a traditional culture in Lebanon for the production of both table grapes and wine (Chalak et al., 2016). Today viticulture occupies the eighth rank in the agricultural sector in the coun try, with a total production area of 9,240 hectares and an annual production of about 89,000 tons of table grapes (FA0, 2010;FAOSTAT, 2019) vs. 3,000 hectares dedicated to wine grape with an annual pro duction of approximately 10,000 tones (Rahal, 2015). In addition, about 800 hectares of vineyards are intended for the production of Lebanese Arak (Roby, 2003). Commercial plantations of table grapes have long been constituted of four local varieties, com monly named 'Tfayfihi', 'Baitamouni', 'Maghdouche' and 'Obeidy'. More recently, the new plantations in the country are mostly constituted of improved vari eties imported from Europe and the United States. Around 77 varieties are imported to Lebanon. Nevertheless, the major share went mostly to 'Cabernet Sauvignon Blanc', 'Syrah', 'Viognier', 'Chardonnay' and 'Merlot' which represent 65% of the total saplings quantities imported during 2012 2014 (Tabaja, 2015). The long history of viticulture in Lebanon suggests the existence of a large indigenous germplasm associated to a wide range of traditional varieties that are well adapted to the various agrocli matic conditions of the country (Chalak et al., 2016). In Lebanon and despite the increasing interest on the conservation and characterization of plant genetic resources in general, only few studies have addressed the local germplasm of grapevine, although it is threatened by various anthropogenic pressures including in particular the progressive replacement of local varieties by more advantageous improved for eign varieties that are regularly imported into the country (Riachi, 1998;Madi, 2007;Chalak et al., 2016;Rahme, 2016). To face the threat of genetic erosion and avoid the loss of traditional germplasm, the Lebanese Agricultural Research Institute (LARI, Tal Amara station) established in 1998 a national grapevine collection containing numerous local acces sions collected from different production areas across the country (Madi, 2007). Within the environmental change and the challenging predictions for the Eastern Mediterranean region including Lebanon (Santillán et al., 2019), there is a growing need to address the local genetic resources such as traditional varieties which are recognized to be more adapted to harsh conditions and having better tolerance to vari ous biotic and abiotic stresses (FAO, 2015). Added to this, the wine industry is witnessing a new trend towards the utilization of the local traditional varieties in the production of typical and well prized wines. Therefore clonal selection, with the respect to specific traits, has become the most important way to improve the quality of grape cultivars.
As a consequence, there is a need for reliable and precise methods of clonal characterization for further use by breeders, nurseries and industries (Galet, 1998;Moreno et al., 1998). Ampelography is a scien tific methodology that has long been the single method used for the characterization of grapevine phenotype, based on the description of different morphological, phenological and pomological charac ters (Galet, 1998;Sabir et al., 2009;Laucou et al., 2011). In addition to the ampelograhic description, in order to discriminate the varieties, synonyms, homonyms and the variation among the accessions, molecular methods has become more frequently used especially markers based on Polymerase chain reaction techniques (Riaz et al., 2012;Madhumati, 2014). Among the molecular approaches, the ISSR (Inter Sequence Simple Repeat) has being evaluated for its usefulness in grapevine cultivar identification and in assessing genetic diversity of grapevine germplasms (Moreno et al., 1998;Dhanorkar et al., 2005;Santiago et al., 2005;This et al., 2006;Sabir et al., 2009;Seyedimoradi et al., 2012;Choudhary et al., 2014;Castro et al., 2016).
This study aimed to characterize the 43 local grape accessions preserved at the Lebanese National Collection of Grapevine at LARI using both ISSR mark ers and ampelographic descriptors. This work was conducted in an attempt to evaluate the diversity of the local germplasm for further actions of conserva tion and sustainable utilization of grapevine genetic resources in diversification strategies.

Plant material
A total of 43 accessions of traditional grapevine varieties growing in the Lebanese national collection located in Tal Amara, Bekaa (N 33° 49' 59", E 36° 0' 0", 908 m a.s.l.) established in 1998, were considered in this study (Table 1). Accessions initially consisted of farmer's local varieties surveyed and collected across the Bekaa region (Lebanon). The vine cultivars were grafted with four duplications on B41 root stocks which were planted directly in the collection land a year before the grafting process under the same growing conditions using horizontal trellis train ing system with the spaces 2.75 x 2.75 m. The site of plantation is characterized by a clay fertile soil and an average of precipitation ranging between 600 to 700 mm/year. For each accession, young leaves were sampled from a single selected plant and stored at 20°C for the DNA extraction and molecular analysis. On the other hand, samples of ten leaves, five clus ters and 20 fully mature berries were taken from the same selected plant of each accession between August and October 2020 to undergo the ampelo graphic characterization.

Molecular characterization
Total genomic DNA was extracted from 300 mg of young leaf tissue after been crushed to a fine powder with liquid nitrogen and stored at 20°C until use. DNA extraction of grapevine landraces was per formed according to (Doyle and Doyle, 1990) proto col with minor modifications (Kafkas et al., 2006). DNA concentration was measured by Biospecnano (Shimadzu corp) and stored at 20 C until further analysis. Nine ISSR primers: UBC807, UBC812, UBC815, UBC816, UBC818, UBC828, UBC864, UBC868 and UBC880 (  used in this study based on their good results for amplification and high power of discrimination on grapevine (Seyedimoradi et al., 2012). The ISSR (Inter Simple Sequence Repeat) amplifi cation was carried out as per (Sabir et al., 2009), using 20 µl reaction mixture containing 2 µl of 10 X PCR buffer (750 mM TrisHCl pH 8.8; 0.1% Tween20), 0.2 mM dNTP, 2 mM MgCl 2 ; 200 mM Primer; 50 ng genomic DNA and 1 U Taq DNA polymerase (MBI Fermentas Inc, Hanover, MD21076, USA). The ampli fication program consisted of 94°C for 4 minutes fol lowed by 35 cycles of 94°C for 1minute, 52°C for 1.05 minutes, 72°C for 2 minutes and a final extension of 8 minutes at 72°C. The amplified DNA was visualized on the 3% agarose gel.

Data analysis
To assess the information given by ISSR markers, the following parameters were calculated as follow: number of alleles per locus, power of discrimination (PD= 1 Σ gi2, where gi is the frequency of the i th genotype), polymorphism information content (PIC= 1∑ (Pi) 2 , where Pi is the proportion of samples carry ing the i th allele of a particular locus) (Botstein et al., 1980). Genetic distances were calculated according to Jaccard. Trees clustering the data with the unweighted pairgroup method (UPGMA) with SAHN clustering and tree programs of PAST software (Kriege et al., 2014).
Qualitative characteristics have been described and scored. For quantitative descriptors, the mean was calculated. To assess the degree of similarity between the units tested and understand the rela tionships between them, the data were subjected to a principle component analysis (PCA) in order to con dense the quantitative and qualitative descriptors in a small number of synthetic components (Saporta, 1990). Thus, the degree of contribution of each of the characters to the total variation was calculated in order to indicate the most relevant characters.
Hierarchical Cluster Analyses was executed using Euclidean distance to classify cultivars into different groups based on morphological evaluation. LSD test (SAS Institute Inc, 1995) was done in purpose to com pare means of quantitative characters between dif ferent accessions. The correlation between molecular and ampelographic clustering was studied by per forming a Mantel test using past program.

Genetic clustering of the accessions
The allelic diversity data was used to produce a dendrogram by using distance matrixUPGMA, thus revealing the genetic relationship among grapevine accessions (Fig. 1). The dendogram constructed on the base of the amplification product of ISSR primers of the different accessions showed 41 different mole cular patterns. Tfaifihi variety indicated a unique marker with the primer UBC812 (305 bp). One case of synonymy was observed between Gbaii, Arasani and Gnoubi cultivars and they are very close for the most discriminating ampelographic traits. The ISSR analysis showed four distinct groups: G1, G2, G3 and G4 at the distance of 0.67 of similarity including 6, 26, 2, and 9 cultivars respectively.

Ampelographic description
A total of 43 cultivars were studied. Locale names were commonly given by farmers or nurseries based on berry colour or on the country of origine (Asouad, Souri Aswad, Amarcani Abiad) and maturity date (Houzairani). For each cultivar, a descriptive list was established with 33 morphological traits related to the leaf, bunch and berry (Tables 3, 4, 5).
Anthocyanin coloration of main veins on both the upper and the lower side of blade varied from absent (e.g. 'Al Mir'), limited to the petiolar point (e.g. 'Arasani'), extended to the first bifurcation (e.g. 'Khodre'), and beyond the second burifaction (e.g. 'Zaghzaghani').

Principal component analysis
The first three components presented 31.9% of the total variation of the different descriptors ( Table  6). The first component consisted of 12.2% of the total variation and included berry skin color, firmness of flesh, bunch weight, harvesting and veraison times, in addition to the brix level at harvesting. Thus, the first component was dominated by the fruit characteristics more than the leaf ones. The second component represented 10.9% of the total variation and included four variables namely berry weight and length, leaf lower vein pigmentation and lower face pilosity. The third component was characterized by a percentage of variation of 8.8% and was dominated by the flesh juiciness and the bunch density. Based on this validation through PCA, this shortlist of dis criminating descriptors may be further considered to study the relationships between the grapevine acces sions.

Ampelographic accession clustering
The dendogram illustrating the relationship among the 43 studied grapevine accessions was con structed on the 12 most discriminating descriptors as validated by PCA. Four groups were differentiated at the distance 17 of similarity using Euclidean distance (Fig. 4). The first group (G1) consisted of eight acces sions of different skin color but sharing same verai son date and the same sugar content at harvesting time. The second and the largest group (G2) consist ed of 13 accessions, sharing all the same bunch char acteristics in terms of density, juiciness in addition to the common sugar content at harvesting time. Moreover, these accessions started the stage of veraison almost at the same date around the 5 th of September. The third group (G3) contained 12 acces sions, characterized by medium green berries. A case of close similarity was found between three acces sions of different vernacular names 'Gnoubi', 'Gbaii' and 'Arasani'. The fourth group (G4) clustered 10 accessions of which eight accessions with dark red violet (e.g. 'Asali') to bleu black (e.g. 'Mariami') berries, and two accessions with green yellow berries ('Rins' and 'Samaani Abiad'). All these accessions have common leaf characteristics and intermediate sugar content at maturity time. The characters in bold are discriminant.

Discussion and Conclusions
In our study encompassing 43 local accessions, 51 bands were generated from nine primers, out of which 41 were polymorphic. Similarly, 55 polymor phic bands were obtained in Palestine by Basheer Salimia, (2015) in studying 36 grape accessions using 17 ISSR markers; while 69 polymorphic bands were obtained by Seyedimoradi et al. (2012) in studying 21 local Iranian grapevine cultivars using 10 ISSR primers. Our results related to polymorphism rate (5085%) and size of amplified bands (1251000bp) are close to the ones previously reported in Portugal (Castro et al., 2016), Turkey (Sabir et al., 2009), Egypt (Hassan et al., 2011, Iran (Seyedimoradi et al., 2012), andPalestine (BasheerSalimia, 2015). Larger frag ments (3001500 bp) were generated by the same ISSR markers with values up to 1500 bp in India (Dhanorkar et al., 2005) and 2500 bp in Turkey (Sabir et al., 2009). Moreover, our results indicated a dis crimination power value ranging from 0.3 (UBC868) and 0.9 (UBC864) and PIC value varying between 0.27 (i.e.UBC868) and 0.64 (i.e.UBC816) with an average of 0.49, which was similar to the results obtained in Iran by Seyedimoradi et al. (2012).
Our ISSR results revealed an important genetic diversity within the national collection of grapevine germplasm. About 41 different molecular profiles were clearly differentiated within the 43 accessions, with only one case of synonymy found between three accessions ('Arasani', 'Jbai' and 'Jnoubi'). These findings certainly confirm one more time the efficien cy of these ISSR markers in investigating the genetic variability of grapevine. It also indicates the efficiency of the initial survey and collection of local accessions in establishing this national collection of grapevine germplasm with only 5% synonymy. In Spain, the molecular characterization of grapevine accessions of the national gene bank at Alcalá de Henares allowed the differentiation of 177 accessions (30%) over 621 initially collected (Ortiz et al., 2004). Furthermore, the SSR analysis of the Eastern European cultivars led to the differentiation of only 659 unique profiles over 997 accessions studied (Maul et al., 2015). In Italy, Cipriani et al. (2010) reported 200 groups of syn onyms vs. 774 unique genotypes out of 1005 grapevine accessions studied. Also, Lopes et al. (1999), studying 49 supposed different cultivars from the Portuguese grapevine national collection of Terciera (Azores Island) through microsatellite mark ers, detected only 36 different profiles after deter mining the synonym cases.
Along with the genetic differentiation of the grapevine accessions, an ampelographic description was carried out for the 43 Lebanese accessions using 33 descriptors recommended by the OIV, and mostly related to various descriptors including yield compo nents like bunch and berries dimensions and weight (Kara, 1990;Ortiz et al., 2004;Santiago et al., 2005;OIV, 2007;Akram et al., 2019). Mature leaf descrip tors did not vary significantly among the Lebanese accessions. Most of these accessions had leaves with five lobes similarly to the results reported in other studies (Ecevit and Kelen, 1999;Chalak et al., 2016). On the other hand, our results revealed a large vari ability of bunch shape, weight, density, as well as for the berry shape, color and size, similarly to the results reported in previous works in the Eastern Mediterranean (Sabir et al., 2009;Biniari and Stavrakaki, 2016). On the other hand, the multivari ate analysis was efficient to analyze large data gener ated in our study by qualitative and quantitative descriptors to further identify patterns and relation ship among powerful statistical techniques. PCA allowed to extract the most discriminating descriptors, e.g. veraison date, maturity date, berry length, upper vein pigmentation, lower vein pigmen tation, bunch density, bunch weight, brix at harvest ing, juiciness of flesh, berry weight and firmness of flesh and skin color. Our results were in accordance with those previously obtained by Riachi (1998), Madi (2007 and Chalak et al. (2016) confirming one more the stability of the discriminating descriptors over years. This also in accordance with Leao et al. (2010Leao et al. ( , 2011 who reported the stability of the dis criminating characters for two consecutive years. Phenotypic clustering allowed differentiate the 43 accessions studied in four main groups. The syn onymy case first revealed between the three acces sions 'Gbaii', 'Gnoubi', and 'Arasani' by the ISSR clus tering was also confirmed by using ampelographic. This is the minimum rate for verifying synonyms and for clone selection processes (Cervera et al., 2002).
Furthermore, many cases of close phenotypic sim ilarity, first revealed by Madi (2007), were also recorded in this study. It is worthy to note that some of our accessions were also described in other coun tries. This is the case of 'Bzaz Al Anzi' reported in Egypt by Hassan et al. (2011) and presenting almost similar ampelographic characteristics of the bunch and the berries.
When comparing the two dendograms generated apart by the molecular and ameplographic descrip tors, the Mantel test indicated a weak correlation (r = 0.26, data not shown) between them. This reflects different structure for the accessions clustering whether it is generated by molecular or ampelo graphic descriptors. Exception is made to the case of synonymy between the three accessions 'Gnoubi', 'Gbaii', and 'Arasani', which was confirmed by both, ampelographic and molecular clustering. This dis crepancy between the two clustering types was also reported in other studies in grapevine collections (Knezović et al., 2017) and other fruit crops (Talebi et al., 2008;Zdunic et al., 2008).
On the other hand, considering some key ameplo graphic descriptors, it was possible to categorize the existing diversity of the collected Lebanese germplasm according to the phenological stage, and according to berry skin color. In five groups of matu rity time; with 27 accessions had green yellow berries, only one accession with red berries, two accessions with rose berries, five accessions with bleu black berries and eight accessions with dark red violet berries. Surprisingly, similar ratios between green and red accessions were also reported in Canary Islands and Madeira by Marsal et al. (2019).
Additionally, most of the accessions studied had crunchy berries with thick skin and very low juice content. This indicates the potentiality of using these accessions for table grapes rather than wine grapes. Only three accessions had thin skin, juicy flesh and high sugar content and, therefore, may be tested for fermentation. Such descriptive ampelographic char acterization of the berries is hopefully necessary for evaluating grapevine accessions conserved in field genebanks with respect to their usage as table grape or wine grape (Sabir et al., 2009;Ates et al., 2011;BasheerSalimia, 2015).
This diversity upon the 43 grapevine accessions conserved at the Lebanese national collection of grapevine as revealed by ISSR markers and ampelo graphic descriptors indicate the efficiency of sam pling/collecting strategy conducted in 1998 in the Bekaa and Chouf areas.
Nevertheless, additional grapevine varieties grow ing locally under different vernacular names, which do not appear in the national grapevine germplasm collection, were recently assessed in their growing site by (Chalak et al., 2016). Therefore, it is strongly recommended to further extend the survey and mis sion collection to cover multiple grapevine produc tion areas, particularly the North, the South and of Mount of Lebanon in order to enrich the national col lection. Combining both molecular makers and ampelographic descriptors as it was shown in this study, would be very useful in optimizing the collec tion of accessions by clarifying the mislabeling cases, understanding the genetic distances among acces sions, and setting up a descriptive assessment for these accessions. The comprehensive evaluation of the existing local gene pool of grapevine would fur ther allow the sustainable utilization of the valuable accessions directly for multiplication in certified nurs eries and also in further breeding programs.