SSR-based genetic diversity assessment among Tunisian winter barley and relationship with morphological traits

Transcription

1 Euphytica 135: , Kluwer Academic Publishers. Printed in the Netherlands. 107 SSR-based genetic diversity assessment among Tunisian winter barley and relationship with morphological traits Sonia Hamza 1,, Wafa Ben Hamida 1, Ahmed Rebaï 2 & Moncef Harrabi 1 1 Laboratory of Genetics and plant breeding, Institut National Agronomique de Tunisie, Avenue Charles Nicolle, Tunis 1082, Tunisia; 2 Laboratory of plant protection and transformation, Centre de Biotechnologie de Sfax, BP K, Sfax 3038, Tunisia; ( author for correspondence; hamza.sonia@inat.agrinet.tn) Received 11 April 2003; accepted 11 October 2003 Key words: barley, genetic diversity, Hordeum vulgare, microsatellite, morphologicaltrait, SSR Summary For studying genetic diversity caused by selection for adaptation and end-use, 17 microsatellites (SSR), representative of the barley genome, were used in 26 barley (Hordeum vulgare L.) accessions and cultivars in Tunisia. The accessions/cultivars originate from different geographic regions and are of different end-use. For the 15 polymorphic SSR, the mean number of alleles per locus was 3.6 and the average polymorphism information content was Cluster analysis based on SSR data and on morphological data clearly differentiate the genotypes according to their type (local landraces vs. varieties), row-number and end-use. The correlation between both diversity measures was highly significant (r = 0.25, p <10 5 ) and the correspondence between the clustering based on SSR and morphological data was relatively good. Our results show the large genetic diversity of the Tunisian barley cultivars and the association of this diversity with adaptation traits. Introduction In Tunisia, barley is the most important cereal crop after wheat (Triticum aestivum) and occupies one 38% of the cereal cultivated area. Prior to national barley breeding programs, which started in the 1960 s, cultivated barley represented essentially winter-type landraces. Despite their low productivity, they still constitute 40% of the present barley cultivated area. The most important are seven geographical groups; Beldi (Sahel), Sahli (Moknine), Sfira (Gabes), Frigui (West South), Ardhaoui (Zarzis), Djerbi (Island of Djerba), and Aarbi (common barley) (El Felah & Mekni, 2000). The ability of local landraces to grow in drought stressed environments and to provide after transformation favourite tastes for preparing traditional dishes as well as their good feeding quality both as grain and straw, are the major reason for their maintenance as a crop and the conservation of their diversity. Since the creation of barley breeding centres in Tunisia, new cultivars were developed by (i) selection from local populations, (ii) introduction of new varieties and (iii) crossing and selection for yield. Several high-yielding winter barley cultivars of two and six-row type were introduced from Algeria, France, Denmark, Australia, USA and from the International Centre for Agricultural Research in the Dry Areas (ICARDA) in Syria. However, the cultivation of tworow barley for brewing were rapidly replaced by six-row cultivars, mainly because of farmers preference and agricultural practices, that were not relevant for two-row barley. Currently, the most interesting cultivars are Martin and Rihane-03 introduced, respectively, from Algeria in 1931 and from ICARDA in 1986, which occupy about 20% and 60% of barley cultivated area, respectively. Genetic and physiological characteristics of Tunisian germplasm assessed for morphological, physiological and agronomic traits showed that local germplasm is diverse and possesses sources for resistance/tolerance to biotic and abiotic stresses and high level of nitrogen content (El Felah & Mekni, 2000). Furthermore, biochemical characterisation of hordein patterns revealed

2 108 high level of polymorphism since 56 polymorphic patterns were scored from a collection of 174 samples (M. El Felah, personal communication). Notwithstanding the ability of this technique to differentiate between genotypes, hordein loci are all located on the short arm of chromosome 5 (Sogaard & Wettstein-Knowles, 1987). Molecular markers that reveal polymorphism at the DNA level have been shown to be a very powerful tool for genotype characterisation and estimation of genetic diversity. Among these, microsatellite or SSR (simple sequence repeats) markers were showed to have a high potential for identification and estimation of genetic diversity of barley germplasm collection (Saghai-Maroof et al., 1994; Russel et al., 1997b; Struss & Plieske, 1998; Pillen et al., 2000). Availability of SSR marker sequences for oligonucleotide synthesis, involvement of PCR amplification, the simplicity of protocol that produces reliable and highly detectable amplification products, their co-dominance and single localisation constitutes their advantage over AFLP, RFLP and RAPD markers. Moreover the fact that microsatellite sequences were shown to be more frequent in transcribed regions (Morgante et al., 2002) and that microsatellite markers are under the influence of natural selection in barley and wheat (Saghai-Maroof et al., 1994; Stachel et al., 2000) constitutes the advantage of these markers to assess the relationship between DNA polymorphism based on microsatellite markers and adaptation of barley accessions to particular environments. To date no information is available in genetic variation of Tunisian barley germplasm at the molecular level. The objectives of the present study were to use the microsatellite markers, described by Saghai- Maroof et al. (1994) and Liu et al. (1996) and which are distributed all over the barley genome, for characterisation of Tunisian barley germplasm including two-row and six-row winter barley accessions and estimation of the genetic diversity. The ability of microsatellite markers to identify barley genotypes was also tested using barley accessions collected from different areas of the country. Barley germplasm was also characterised for morphological and agronomic traits and relationship between genetic similarities based on SSR markers and euclidian distances based on some morphological and agronomic traits is described. Materials and methods Plant material and DNA isolation Twenty six Tunisian winter barley cultivars /landraces of diverse geographic origin, end-use and row type were used in this study (Table 1). A first set consisted of eleven Tunisian local barley (OLT) lines obtained by mass selection from local landrace and a local landrace Sfira. These landraces were named according to the region of origin (inside Tunisia) from which they were collected. These local landraces were retained because they showed yellow and large grain and resistance to Pyrenophora teres. Four accessions were collected in the northern Tunisia (Tunis, Ariana, Bizerte and Dinar). The other sets consisted, respectively, of two old local cultivars ( Martin, and Cowra ) that were used since the first half of the twentieth century, two fodder barley cultivars and three new commercial six-row varieties. OrgeBlanche is a white barley cultivar (also called fodder barley of Cap Bon) originated from Gabon and selected in Tunisia from a local landrace since The last three varieties are two-row barley commercial cultivars. Except the three two-row cultivars, all barley lines are of the six-row type and all of them (including the two-row cultivars) are of the winter or facultative-type. Plant DNA was isolated from fresh leaves (ten plants per genotype) by a modification of the method described by Saghai-Maroof et al., (1984). SSR markers and protocol Seventeen SSR markers from the commercially available Mappair primer sets developed by Research Genetics Inc. (Huntsville, USA) were selected on the basis of their repeat pattern (di, tri and tetranucleotide) and chromosomal locations (Table 3). They provided at least one marker per chromosome, however there were four markers on chromosomes 6 and 7. PCR amplification was performed with 20 ng DNA in 15 µl volume reaction containing 10 pmoles of each primer pair, 200 µm dntps,1.5µl 10X Taq polymerase buffer (Appligene) and 0.45 Unit Taq polymerase (Appligene). The reaction depending on the primer pair was run for 30 cycles (denaturing 94 C for 1 min, annealing at 50, 55, or 60 Cfor2min with extension at 72 C for 1 min 30 sec). These cycles were preceded by a denaturing step at 94 C for 3 min and ended by an extension step at 72 C for 5 min. The PCR amplifications were carried out

3 109 Table 1. Genotypes used in SSR and morphological analyses, their pedigree (when available), their breeding institute (for cultivars), end use and region of origin OLT landraces Origin of collection Breeder/ Country End use OLTGabes2 El Menzel (Oasis Gabes) Tunisia Food OLTGabes8 El Menzel (Oasis Gabes) Tunisia Food OLTGabes9 El Menzel (Oasis Gabes) Tunisia Food OLTGabes10 El Menzel (Oasis Gabes) Tunisia Food OLTMellita Mellita ( Kerkena ) Tunisia Food OLTDjerba Djerba Tunisia Food OLTTeboulba Teboulba (Mahdia) Tunisia Food OLTGuellala Seryndi (Oued Ezbib)-Guellala (Djerba) Tunisia Food OLTEnfidha B chachma -Enfida Tunisia Food OLTMahdia Mahdia Tunisia Food OLTMahres Mahres Tunisia Food Safra Gabes Tunisia Food Collected accessions Dinar East of Tunis Tunisia Bizerte Bizerte Tunisia Tunis Tunis Tunisia Ariana North of Tunis Tunisia Six-row varieties Pedigree Martin Unknown Algeria Food Cowra Unknown Australia Food OrgeBlanche Selected from a local landrace INRAT (Tunisia) Fodder Hor1259 L527/3/Harbing/As/Vaughn//Aths INRAT (Tunisia) Fodder Manel M126/CM67//As/Pro/3/Arizona5908/Aths//L640 INRAT (Tunisia) Food Mumtez L527/5/As54/Tra//2 Cer/TolI/3/Avt/TolI//Bz/4/vt/Pro//TolI INRAT (Tunisia) Food Rihane-03 As46//Avt/Aths ICARDA (Syria) Food Two-row varieties Pedigre Faiez Early Russian/Apam ICARDA (Syria) Brewing Roho unknown Riso laboratory (Denmark) Brewing Taj Bonus/C13576 West Institue /USA Brewing by GeneAmp PCR system 9700 (Perkin Elmer Applied Biosystems). The amplification products were resolved on 6% polyacrylamide gels (PAGE) followed by silver-staining according to the protocol described by Pillen et al. (2000). Morphological and agronomic traits Twelve traits (Table 2) were scored on the 26 barley lines. All traits, except the number of rows in the spike (NR), were evaluated using one-row plot of 1.5 m and three replicates. These include morphological traits (NL, LI, HMB, LB), earliness and maturity traits (FL, PhM, EP, DRG) and yield components (NGE, PMG). All traits were standardised before analysis. Data collection and analysis Marker polymorphism To measure the informativeness of the markers, the polymorphism information content (PIC) for each SSR was calculated according to the formula: PIC= 1- ( p i ý), where i is the total number of alleles detected for an SSR marker and p i is the frequency of the ith allele in the set of 26 barley cultivars investigated. Null allele is considered as one of the series of multiple al-

4 110 Table 2. Twelve morphological traits used to calculate morphological distances Code Trait NR Number of rows (2 or 6) NL Number of leaves on the master shoot Fl Flowering date in days from january 1 st (anthers visible on 50% of the spikes) PhM Physiological maturity in days from january 1 st EP Time to ear emergence in days from january 1 st (first spikelet visible in 50% of the spikes) LE Spike length in cm LI Internode length in cm HMB Height of main shoot in cm LB Length of barbs in cm DRG Kernel hardness NGE Average number of grains per ear PMG 1000-kernel weight Table 3. Characteristics of SSR markers used with the number of alleles and polymorphism information content calculated over a set of 26 barley genotypes Marker Repeat Chromosomal Number of PIC Number of pattern location alleles null alleles A HVM9 B (TCT) HVM11 (GGA) HVM20 (GA) HVM30 (CA) HVM31 (AC) HVM33 (CA) HVM40 (GA)6(GT) HVM43 (CA) HVM64 (GA)4(GT) HVM74 (GA) HVM77 (CA) HVCMA C (AT) HVBKASI (C)10,(A) WMS6 D (GA) HVLEU (ATTT) HVDHN7 (AAC) E HVDHN9 (AC) Average F A null allele was considered as one of the series of multiple allele and included in the computation of PIC value. B sequence according to Saghai-Maroof et al. (1994). C sequence according to Becker and Heun (1995). D sequence according to Röder et al. (1995); all other SSR loci are from Liu et al. (1996). E this value correspond to the number of genotype that did not show a second PCR product. F monomorphic loci are excluded in the computation of the average values.

5 111 lele and is included in the computation of PIC values. PIC is also an estimate of the discriminatory power of an SSR marker locus. Genetic similarity estimation and cluster analysis Each SSR band was scored as present (1), absent (0) or as a missing observation for the different cultivars. Genetic similartiy (gs) between two cultivars i and j was estimated following Nei and Li (1979) by the formula: gs ij =2N ij /(Ni+Nj), where N ij is the number of bands present in genotypes i and j, Ni (resp. Nj) is the number of bands present in genotype i (resp. j). Markers with missing observations for genotype i and/or j were not included in the calculation of gs ij. Based on the genetic similarity matrix (denoted GS), UPGMA cluster analysis were used to assess pattern of diversity among the barley entries. To test the goodness of fit of this clustering to the genetic similarity data, the cophenetic correlation, r c, that is the correlation between the matrix of gs ij and the matrix of cophenetic similarity value (computed from the tree matrix), was calculated. All calculations were performed using the NTSYS-pc version 2.1 software (Rohlf, 2000). Analysis of morphological traits A principal component analysis (PCA) was performed on observed morphological traits after standardisation All traits were standardised by subtracting the mean value and dividing by the standard deviation; this allows to remove scale effects before calculating Euclidian distances. Based on standardised trait values, euclidian distances (md ij ) between the lines were calculated. Morphological similarities (ms ij )werealso calculated as (1-md ij ). Matrix of these values is denoted MS. Using the matrix (denoted MD) of euclidian distances, an UPGMA cluster analysis was performed producing a second dendrogram depicting relationships among cultivars relative to their morphological characteristics. As for genetic similarity, the cophenetic correlation was calculated to measure the quality of the clustering with regard to the original data. Comparison of marker and morphological data Simple (r) and Spearman rank correlation (r s )coefficients between the 325 values of genetic similarities (gs ij ) and morphological similarities (ms ij )were calculated. P-values for these coefficients were calculated based on their respective asymptotic distributions (Kendall & Stuart, 1979). This is of course an approximation as the usual assumptions of independent samples of pairwise data does not hold. Such an approximation is expected to be valid as the number of pairwise distances (here 325) is large enough. Correspondence between the two similarity matrices GS and MS (matrix of ms ij values) as well as between their corresponding cophenetic matrices was tested with the Mantel Z statistic (Mantel, 1967). Significance of Z was determined by comparing the observed Z values with a critical Z value obtained by calculating Z for one matrix with 1000 permuted variants of the second matrix. All computations were performed with appropriate procedures of the NTSYS-pc version 2.1 software (Rohlf, 2000). Results Characteristics of SSR markers Seventeen microsatellite markers dispersed across the genome were used to test the genetic diversity of 26 landraces/cultivars. Fifteen SSR markers generated polymorphic patterns and two (HVDHN9 and HVM64) gave a monomorphic pattern, yielding a polymorphism rate of 88.2%. Null alleles resulting from absence of an amplification product were observed with forty-two marker-genotype combinations (Table 3). Null alleles were previously described by Donini et al. (1998) and Stachel et al. (2000) after DNA amplification of wheat and barley with SSR markers. Stachel et al. (2000) attributed this phenomenon to polymorphism in primer-binding site and considered that null allele represents one of a series of multiple alleles. Marker HVDHN7 resulted in the amplification of two PCR products within genotypes. The first PCR product was monomorphic and present in all accessions and the second was polymorphic and present only in three accessions. Therefore 23 null alleles with marker HVDHN7 were obtained and related to absence of the second PCR product (Table 3). A total of 55 alleles was detected by 15 markers. The number of alleles per locus varied from 2 to 7, with a mean of 3.6 alleles/locus (Table 3). The mean PIC for the 15 SSR markers was 0.45, with values ranging from about for HVM11 to 0.78 for HVM74.

6 112 Genetic diversity levels The average GS among the barley lines was 0.58 with values ranging from 0.31 between Roho and the accession collected from Bizerte and 0.94 between Cowra and Tunis. The average GS among the OLT accessions is significantly larger (0.70), ranging from 0.47 between OLTGabes2 and OLTGabes9 and 0.87 between OLTTeboulba and OLTGuellala. Based on GS values, we attempted the identification of the accessions collected from different areas in Tunisia by examining the most similar known cultivar (having the highest genetic similarity). The most similar cultivars/landraces to the accessions collected from the regions Tunis, Bizerte, Ariana and Dinar, were, respectively, Cowra, OLT Gabes8, and OrgeBlanche. Genetic diversity pattern UPGMA cluster analysis of SSR genetic similarity matrix resulted in the phenogram in Figure 1, which has quite a good fit to the GS matrix (r cs = 0.77). Four groups can be distinguished by truncating the dendrogram at gs value of The major group (denoted group I) consists of 20 genotypes and includes all OLT landraces, cultivars derived from local landraces ( OrgeBlanche ), the accessions collected from Tunis, Dinar, Ariana and Bizerte as well as introduced cultivars ( Rihane-03, Hor1259, Cowra ). In group I, a subgroup consisting of Hor1259 together with OrgeBlanche and Ariana corresponds to a homogeneous Group of fodder barley cultivars. Another group (Group III) includes all three tworow barley cultivars ( Faiez, Taj, Roho ). These cultivars have different origins and are very likely to be genetically unrelated (see Table1) which may explain the high diversity level within this group. Martin and Manel are clustered together and form a separate group (group II). The cultivar Mumtez is well separated from the other groups and forms Group IV. Morphological analyses In the principal component analysis (PCA), the first four principal components (having eigenvalues > 1) explained about 87% of the variation. The first two axes explaining about 58.4% of the variation, were, respectively, linked to variables related to yield and maturity (correlated negatively with EP, PhM and PMG and positively with HMB and NGE, (r> 0.7 ) and to earliness (positively correlated with NR and Fl, r>0.8). The third axis, explaining 17.9% was correlated positively with NF (r = 0.79) and negatively with LEN (r = 0.95). Based on the projection of varieties in the first principal plan (Figure 2A) one can distinguish four groups. The first group in the lower-left part of Figure 2A comprises Faiz, Roho and Taj, which are the tworow early barley commercial varieties introduced in the end of the seventies and mainly characterized by high PMG and small NGE. On the opposite side, a second group is composed by the accession collected from Ariana, OrgeBlanche and Hor1259, which have naked grain and are fodder barley characterised by their tall straw (high HMB), late ear emergence (high EP) and high leaf area (high NF). Note that these two groups are also clearly distinguished in the second principal plan of axes 1 and 3 (Figure 2B). The central group in Figure 2A could be split into two groups: first one composed of varieties Martin, Cowra, the accessions collected from Bizerte and Tunis, and Safra, and the second composed of all the remaining varieties, i.e., all the OLTs plus Manel, Rihane-03, Mumtez and the accession collected Dinar. However, when examining the second principal plan, one sees (Figure 2B) that Manel, Rihane-03, and Mumtez, which are sixrow high-yielding commercial cultivars could now be considered as a separate group whereas the group of the five cultivars/landraces Martin, Cowra, Safra, Tunis Bizerte and is now spread and partly mixed with the OLT group. This gives further homogeneity of OLT group, whereas the nine remaining genotypes ( Martin, Bizerte, Tunis, Cowra, Safra, Manel, Rihane-03, Mumtez and Dinar) may be considered as a heterogeneous group including both old cultivars used since the 1930s ( Cowra, Tunis, Martin, Safra, Dinar) and registered commercial varieties used since the end of the 1980s ( Manel, Rihane-03, Mumtez ). The mean MD among the barley lines was 0.416, with values ranging from 1.01 between OrgeBlanche and Roho and 0.10 between OLTGabes10 and OL- TEnfidha. The morphological-based dendrogram (Figure 3) has a good fit to the MD matrix (r cs = 0.79) and gave results in good agreement with the variety grouping obtained from the principal components analysis. In fact, when truncating the dendrogram at distance value of 0.5, we obtain almost the same grouping as that given by the PCA, which consists of the two-row barleys ( Roho Faiez and Taj ), the group of the three naked and fodder varieties (Ariana, Hor1259

7 113 Figure 1. Dendrogram resulting from an UPGMA cluster analysis of 26 barley genotypes and based on data of 15 microsatellite primer pairs. and OrgeBlanche ) and a large group composed of the OLTs and the remaining genotypes. This large group could however be split into smaller ones; if we truncate the dendrogram at 0.32, we get a large group of the OLTs plus Cowra, Martin and Dinar (three landraces), a group of newly introduced commercial cultivars ( Manel, Mumtez and Rihane-03 ), a group of Tunis and Safra and finally a group containing Bizerte alone. This depicts more precisely (than the PCA) the relationships between the landraces and cultivars. Comparison of SSR and morphological traits-based diversity estimates Correlations between GS and MS values were significant (r = 0.25, p<10 5 ; r s =0.22p<10 3 ). The Mantel Z test statistic was also significant between GS and MS matrices (Z = 2.43; p = 0.006) as well as between the cophenetic matrices of GS and MS (Z c = 2.88; p = 0.002). These results indicate a good correspondence between the two similarity measures. In fact, the clusters of local Tunisian barley lines (the OLT group), of naked fodder barley and of tworow barley ( Taj, Faiez and Roho ) are consistently found in both analyses, supporting the hypothesis of a common genetic origin for these lines. However, some discrepancies between the two dendrograms arise, es-

8 114 Figure 2. Principal component analysis (PCA) two-dimensional plots of 26 barley genotypes based on morphological trait values. A. The first principal plan (PC1-PC2) B. the second principal plan (PC1-PC3).

9 Figure 3. Dendrogram resulting from an UPGMA cluster analysis of 26 barley genotypes and based on euclidian distances calculated with 12 morphological traits. 115

10 116 pecially for the high-yielding commercial cultivars ( Manel, Mumtez and Rihane-03 ) that are quite spread out on the basis of SSR markers while constituting in one cluster on the basis of morphological similarity. Another discrepancy concerns the relation between Cowra and Tunis, which are very similar on the basis of SSR but relatively distant on the basis of morphological traits. Discussion Genetic diversity levels and patterns Barley microsatellite markers showed an average PIC value of 0.45, which confirms that SSR markers are highly informative. This value is higher than the PIC value (0.38) obtained by Pillen et al., (2000) who used 22 microsatellite markers and a set of 28 mainly German barley cultivars and two wild forms. This demonstrates that the genotypes used in this study are more diverse than German cultivars and could be related to different origin of the genotypes, local and introduced cultivars. In fact, utilisation of genotypes with the same origin imply a low genetic diversity, since these genotypes may have exchanged genetic material through breeding programs. However, a high level of PIC may also be due to the level of polymorphism of the microsatellite marker. The choice of highly polymorphic markers contribute to the enhancement of PIC values (Struss & Plieske, 1998). Using fewer genotypes (18) obtained from different parts of Europe and 13 microsatellite markers, a higher PIC value of 0.55 was obtained by Russell et al. (1997b). In this study, microsatellite markers were able to discriminate between all landraces/cultivars and allowed the identification of accessions based on the highest value of genetic similarity between these accessions and known cultivars. These relationships were supported by similarity of morphological traits of the spike. For example, the genotype Tunis and the cultivar Cowra which have high genetic similarity value, have also dense spike, and yellow awn and glume. It is recognised that several Cowra genotypes were introduced into Tunisia from Australia in 1930 and genotype Tunis is very likely to be one of them. The related genotypes OrgeBlanche and Ariana, show both lax spike with white awn and naked grain and share other morphological characteristics of fodder barley. Genotype Bizerte shows typical morphological traits of a local landrace e.g., lax spike. Genotype Dinar shares the same morphological traits of the spike as OrgeBlanche but does not have naked grain. UPGMA cluster analysis of SSR based genetic similarity matrix resulted on the classification of the introduced cultivars ( Rihane-03, Hor1259, Cowra ) and the OLTs within a same group. The position of cultivars Rihane-03 and Hor1259 within the OLT group is consistent with their origin; they are both derived from a cross involving Atlas (As) as a parent genome that is a local landrace collected from the Atlas mountains of North Africa. Several local accessions were collected by Australian breeders at the beginning of the nineteen century of which 55 were catalogued in ICARDA. The accessions were named Atlas and some of them, such as As 57 (Campos, 1977) and As 66 (Wilhelmi & Johnson, 1977), were used for breeding new cultivars. Therefore, the adaptive properties of cultivar Rihane-03 to drought environments and its popularity among several Tunisian farmers is related to the presence of As in its genome. Similarly, the position of cultivar Cowra (introduced from Australia) within the OLT group could be related to the utilisation of a local landrace for breeding Cowra cultivars. The microsatellite markers used were also able to discriminate between all the OLT landraces even when they were originated from the same region (four from Oasis Gabes) and have the same common name (Sfira). This high discrimination between closely related landraces is also the result of the large heterogeneity of local landraces, which are defined in Tunisia as geographically based populations. Each population has typical characteristics of the grain appreciated by farmers and local inhabitants for food and tasting properties (Medimagh, pers. Comm.). Grando et al. (2000) mentioned that considerable heterogeneity exists both between landraces collected in different farmers fields (if designated by the same name) and between individual plants within the same farmers field for several plant characteristics. Nevertheless, despite this heterogeneity, OLT landraces showed high genetic similarity values with an average GS value of 0.70, reflecting several common SSR alleles between landraces. Since chromosomal segments marked by SSR alleles were shown to be under the influence of natural selection (Saghai- Maroof et al., 1994; Stachel et al., 2000), SSR alleles within local landraces could have been selected through drought-stressed environments, thereby reducing genetic variability. In fact, Tunisian barley

11 117 landraces cultivated for several thousands of years under drought conditions had acquired adaptive characters to drought as fast grain filling, late flowering time, early physiological maturity and ability to profit from latest rains that occur in April-May (Mekni, pers. comm.). Using SSR markers, reduction of genetic variability was shown within wild barley grown in drought conditions indicating that stress conditions may lead to selection (Meyer et al., 2000). Two-row cultivars were clustered together and form a distinct group. These results are consistent with those obtained by several authors who used RFLP, RAPD and AFLP markers (Graner et al. 1994; Melchinger et al., 1994; Russell et al., 1997a). However, with SSR data, Struss and Plieske (1998) did not observe separation within winter barley and explained their results by the low similarity values generated by SSR markers. Relationship of SSR and morphological diversity patterns The relatively high correlation between morphological distances andgeneticsimilarity based on SSR markers is revealed by a similar grouping of varieties; in fact, three consistent groups were found in both classifications (SSR and morphological-based): the two-row group ( Faiez, Roho and Taj ), the naked fodder group (Ariana, Hor1259 and Orge blanche ) and the OLT group. The other varieties (including landraces and commercial cultivars) are so diverse with respect to their genetic background that no consistent relationship could be found. When these varieties were removed and the remaining were re-analysed, we obtained an even better correspondence between SSR and morphological similarities (Z = 4.47, p<0.0001; Z c = 4.85, p<0.0001). The good correspondence between both diversity measures is interesting as each similarity measure is based on two different characteristics of the plant and because diversity at molecular markers, which are a priori neutral, may not reflect diversity of quantitative traits (Karhu et al., 1996). It may be the result of an adequate representation of genetic relationships by the observed morphological traits and large variation of these traits among the set of lines used in our studies. The good correlation could be explained by the properties of SSR markers being associated with chromosomal regions selected by particular environment (Saghai-Maroof et al., 1994; Meyer et al., 2000; Stachel et al., 2000) and characterisation with some agronomic traits relevant to adaptation such as Fl, PhM and DRG. In fact, most of the variation between two-row varieties cultivated in favourable (humid) conditions and OLTs cultivated in drought conditions occurred for traits relevant to adaptation; the contrast between the means of these two groups of cultivars for Fl, PhM and DRG are respectively 6, 3 and 2.5 days. Considering that SSR used in our study could be linked to QTL influencing the adaptation traits, our results agreed with those observed by Hayes et al. (1997). These authors found associations between AFLP and QTL polymorphism for a wide range of malting quality and agronomic traits when AFLP marker linked to genomic regions with significant QTL effect were used to analyse polymorphism in barley. Similarly, using a set of 148 European two-row spring barley cultivars and 75 AFLP markers selected for polymorphism rate and linkage to QTL in other studies, Kraakman et al. (2001) found associations between markers and yield and stability for 11 markers. Beer et al. (1997) found association between RFLP markers and 10 QTLs influencing different morphological and yield traits in a germplasm pool of 64 oat cultivars and landraces. In this study the microsatellites markers proved to be valuable in genotyping barley accessions. The finding of relationship between SSR based polymorphism and morphological traits showed that SSR markers are powerful to examine functional diversity. A broader study with a large collection of well characterised accessions collected from several areas of Tunisia will be realised using EST-SSR markers to examine genetic diversity in relation to adaptive variation. Acknowledgements This study was funded by grants from the AUPELF- UREF (France) and the Secretary of State for Scientific Research and Technology (SERST, Tunisia). We would like to thank Dr M.S. Mekni and Dr El Felah for providing plant material and information on pedigree of barley cultivars. References Becker, J. & M. Heun, Barley microsatellites: allele variation and mapping. Plant Mol Biol 27: Beer, S.C., W. Siripoonwiwat, L.S. O Donoghue, E. Souza, D. Matthews & M.E. Sorrells, Association between molecular

12 118 markers and quantitative traits in an oat germplasm pool: Can we infer linkage? J Quant Trait Loci 3: Campos, E.R., Breeding for Wide Adaptability and Superior Agronomic Types in Barley. Fourth regional winter cereal workshop-barley. Jordanie Vol. II. pp Donini, P., P. Stephenson, G.J. Bryan & R.M.D. Koebner, The potential microsatellites for high throughput genetic diversity assessment in wheat and barley. Genet Res Crop Evol 45: El Felah, M. & M.S. Mekni, Estimated genetic variation diversity for abiotic stress tolerance (drought and heat) and related traits in advanced lines derived from local landraces improved barley cultivars. 8th International Barley Genetics Symposium. Adelaide, South Australia. Vol. III, pp Grando, S., S. Ceccarelli & B. Tekle, Diversity in barley landraces from the Near East. 8th International Barley Genetics Symposium. Adelaide, South Australia. Vol. II, pp Graner, A., W.F. Ludwig & A.E. Melchinger, Relationship among European barley germplasm. II comparison of RFLP and pedigree data. Crop Sci 34: Hayes, P.M., J. Cereno, H. Witsenboer, M. Kuiper, M. Zabeau, K. Sato, A. Kleinhofs, D. Kudrna, A. Kilian, M. Saghai-Maroof, D. Hofman & the North American Barley Mapping Project, Characterizing and exploiting genetic diversity and quantitative traits in barley (Hordeum vulgare) using AFLP markers. J Quant Trait Loci 3: Karhu, A., P. Hurme, K. Karjalainen, P. Karvonan, K. Kärkkainen, D. Neale & O. Savolainen, Do molecular markers reflect patterns of differentiation in adaptative traits of conifers? Theor Appl Genet 93: Kendall, M.G. & A. Stuart, The Advanced Theory of Statistics, Vol. 2. Third edn. Griffin, London. Kraakman, A.T.W., F.A. van Eeuwijk, C.J. Dourleijn & P. Stam, Fingerprinting of barley to study yield stability.. In: A. Gallais, C. Dillman & I. Goldgringer (Eds.), Quantitative Genetics and Breeding Methods: The Way Ahead, pp , Proceedings of the Eleventh Meeting of Eucarpia, section Biometrics in Plant Breeding. INRA Edition, Paris. Liu, Z.W., R.M. Biyashev & M.A. Saghai-Maroof, Development of simple sequence repeat DNA markers and their integration into a barely linkage map. Theor Appl Genet 93: Mantel, N.A., The detection of disease clustering and a generalised regression approach. Cancer Res 27: Melchinger, A.E., A. Graner, M. Singh & M. Messmer, Relationships among European barley germplasm. I: Genetic diversity among winter and spring cultivars revealed by RFLPs. Crop Sci 34: Meyer, R.C., C.A. Hackett, B.P. Forster, J.R. Russel, L.L. Handley, E. Nevo & W. Powell, Association of nuclear and chloroplast SSRs with adaptive characters in wild barley. 8th International Barley Genetics Symposium. Adelaide, South Australia. Vol. II, pp Morgante, M., M. Hanafey & W. Powell, Microsatellites are preferentially associated with non repetitive DNA in plant genomes. Nature Genet 3: Nei, M. & W.H. Li, Mathematical model for studying genetic variation in terms of restriction endonucleases, Proc Natl Acad Sci USA 76: Pillen, K., A. Binder, B. Kreuzkam, L. Ramsay, R. Waugh, J. Förster & J. Léon, Mapping new EMBL-derived barley microsatellites and their use in differentiating German barley cultivars. Theor Appl Genet 101: Rohlf, F.J., NTSYS-pc: Numerical Taxonomy and Multivariate Analysis System. Version 2.1 Exeter Publications, New York, USA. Röder, M.S., J. Plaschke, S.U. König, A. Börner, M.E. Sorrells, S.D. Tanksley & M.W. Ganal, Abundance, variability and chromosomal location microsatellites in wheat. Mol Gen Genet 246: Russell, J.R., J.D. Fuller, M. Macaulay, B.G. Hatz, A. Jahoor, W. Powell & R. Waugh, 1997a. Direct comparison of levels of genetic variation among barley accessions detected by RFLPs AFLPs, SSRs and RAPDs. Theor Appl Genet 95: Russell, J.R., J.D. Fuller, G. Young, B. Thomas, G. Taramino, M. Macaulay, R. Waugh & W. Powell, 1997b. Discriminating between barley genotypes using microsatellite markers. Genome 40: Saghai-Maroof, M.A., K.M. Soliman, R.A. Jorgensen & R.W. Allard, Ribosomal DNA spacer-length polymorphism in barley: Mendelian inheritance, chromosomal location, and population dynamics. Proc Natl Acad Sci USA 81: Saghai-Maroof, M.A., R.M. Biyashev, G.P. Yang, Q. Zhang & R.W. Allard, Extraordinarily polymorphic microsatellite DNA in barley: species diversity, chromosomal locations, and population dynamics. Proc Natl Acad Sci USA 91: Sogaard, B. & P. von Wettstein-Knowles, Barley: genes and chromosomes. In: S.W. Rasmussen (Ed.), Carlsberg Res Comm 52: Stachel, M., T. Lelly, H. Grausgruber & J. Vollmann, Application of microstallites in wheat (Triticum aestivum L.) for studying genetic differentiation caused by selection for adaptation and use. Theor Appl Genet 100: Struss, D. & J. Plieske, The use of microsatellite markers for detection of genetic diversity in barley populations. Theor Appl Genet 97: Wilhelmi, K.D. & V.A. Johnson, The international winter wheat performance nursery (IWWPN). Fourth regional winter cereal workshop-barley. Jordanie Vol. II, pp