Tree Genetics & Genomes

, Volume 9, Issue 3, pp 805–812

A newly identified locus controls complete resistance to Microcyclus ulei in the Fx2784 rubber clone


  • Dominique Garcia
  • Carlos Mattos
    • Plantações Michelin da Bahia
  • Olivier Fouet
  • Fabien Doaré
    • UPR BioagresseursCIRAD
  • Virgile Condina
    • UPR BioagresseursCIRAD
  • Marc Seguin
Original Paper

DOI: 10.1007/s11295-013-0599-7

Cite this article as:
Le Guen, V., Garcia, D., Mattos, C. et al. Tree Genetics & Genomes (2013) 9: 805. doi:10.1007/s11295-013-0599-7


Using cultivars which are genetically resistant to South American leaf blight (SALB) caused by the fungus Microcyclus ulei is the only way to plant rubber trees in disease-affected areas. Numerous field observations led to the hypothesis that the resistance of the cultivar Fx2784 to SALB is likely to be monogenic. In this study, we investigated this hypothesis by examining the distribution of the trait in a cross between the resistant cultivar and a susceptible one. The individuals resulting from this cross were planted in field trials in French Guiana and Brazil. The resistance of all the trees was assessed by field observations. Bulk segregant analysis (BSA) using microsatellite markers was performed in French Guiana to determine which markers were genetically linked to resistance, and the results were validated by field observations in Brazil. In both locations, a 1:1 segregation of the resistance trait was observed, thus reinforcing the monogenic hypothesis. BSA showed tight linkage between resistance and the microsatellite markers located in linkage group 2 in the Hevea genome and enabled to pinpoint the resistance locus. The location was confirmed by observations on the trees planted in Brazil. This result should facilitate the use of Fx2784 resistance in future breeding programs for SALB resistance. This is the third major locus conferring resistance to SALB identified in rubber tree (Hevea spp.). These three loci are genetically independent, a favorable situation for genetic improvement of SALB resistance.


Hevea brasiliensisMicrocyclus uleiComplete resistance


The development of rubber tree cultivation belongs to the long list of agricultural success stories. Still not mastered at the beginning of the twentieth century, rubber tree cultivation now covers millions of hectares worldwide, principally located in Southeast Asian countries. Starting with few plantlets of Hevea brasiliensis originated from the Brazilian Amazon forest, worldwide natural rubber production has reached 10 million tons in 2008, a consequence of both an amazing increase of the area under cultivation and of the continuous improvement of yields. The gain in rubber productivity was made possible by the adoption of rational agricultural and tapping techniques, as well as the breeding of high-yielding and well-adapted cultivars. However, South American countries did not take advantage of this new crop opportunity because of the presence of the South American leaf blight (SALB) of rubber tree in the Neotropical region. The Ascomycota fungus Microcyclus ulei is responsible for this devastating disease, lethal to susceptible cultivars. Though this pathogen was first described more than a century ago (Ule 1905), its negative impact on rubber tree cultivation in South America still remains high and continues to prevent development of this crop on most of the areas infected by the disease.

Attempts by Ford Company to establish large-scale rubber tree plantations in the Brazilian state of Pará in the 1930s failed because of severe SALB attacks. Nevertheless, these unsuccessful attempts provided useful information regarding the SALB resistance of many rubber cultivars. The most important was that all cultivars created in Southeast Asia were highly susceptible to SALB. Conversely, some totally or partially resistant genotypes were found among native rubber trees originated from the Amazon forest (Townsend 1960). Particularly, accessions from the wild species Hevea benthamiana or Hevea pauciflora appeared very interesting for their complete resistance, although their poor latex production compared to that of the domesticated species H. brasiliensis was generally unacceptable. Attempts were made to combine the high latex productivity of Asiatic cultivars with the good SALB resistance of these Amazonian genotypes by crosses between them (Gonçalves 1968). However, the lack of knowledge on the genetic basis of these resistances did not enable their successful transfer to the progenies of these crosses, and none of the newly created cultivars appeared durably resistant to SALB (Bos and McIndoe 1965; Darmono and Chee 1985).

The first investigations aimed at identifying quantitative trait loci (QTLs) or genes for SALB resistance were carried out on a segregating progeny population from the cross between the resistant cultivar RO38 (also known as FX3899) and a susceptible one. Several resistance QTLs and a major resistance gene were identified under controlled conditions of SALB inoculation (Lespinasse et al. 2000a) and confirmed under natural infection conditions in French Guiana (Le Guen et al. 2003). Nevertheless, it has been shown that some strains of M. ulei were able to overcome all genes and QTLs of resistance coming from RO38 (Le Guen et al. 2007). In another cultivar originated from the Madre de Dios region in Peru, one strong resistance QTL and one major qualitative resistance locus were detected, responsible for the 30-year long resistance in a highly SALB-infected area (Le Guen et al. 2011a).

Among others, the clone Fx2784 has been mentioned for more than 30 years (Chee 1976) for being totally resistant or showing varying degrees of resistance (Chee and Holliday 1986). Recently, rubber tree breeders’ attention was drawn to this clone not only because of its total resistance to SALB attacks in some highly infected areas (in French Guiana and in the Brazilian state of Mato Grosso) but also for its extreme susceptibility in other locations (in the Brazilian state of Bahia). Inoculations with 50 different isolates from Bahia under controlled conditions revealed that this cultivar was totally resistant to 46 of them but highly susceptible to four isolates that were probably of the same physiological strain of the fungus (Mattos et al. 2003). Such behavior strongly suggests the existence of a major gene responsible for the resistance to numerous strains of M. ulei, but whose efficiency may have been overcome by specific races of the pathogen. However, until now, no exclusively monogenic SALB resistance has been evidenced in rubber, although this type of resistance would be very useful for the breeding of new cultivars. The methodology known as bulk segregant analysis (BSA) applied to the identification of markers linked to disease resistance genes has been described by Michelmore et al. (1991) and proven to be very efficient in numerous studies, inclusive of tree species. It was, for example, successfully used in apple tree for detecting apple scab resistance genes (Gygax et al. 2004) and in poplar for identifying markers linked to rust resistance (Tabor et al. 2000). The aim of the present study was to demonstrate the monogenic resistance of the Fx2784 cultivar using molecular markers in a BSA strategy. Comparison of the location of the newly identified major resistance gene with already characterized resistance QTLs in other genotypes will help design efficient breeding strategies for rubber trees.

Material and methods

Mapping population

Fx2784 was crossed with a susceptible rubber clone, and the resulting progeny population was submitted to SALB infection to observe the segregation of the resistance trait among the individuals. The susceptible cultivar used as female parent in the cross with Fx2784 was PB260, an Asiatic high-yielding clone which was already used in two previous QTL mapping studies for SALB resistance (Le Guen et al. 2011a; Lespinasse et al. 2000a). The progeny genotypes from the cross PB260 × Fx2784 were divided in two batches that were spread out among two different locations: one on the Combi experimental station of CIRAD in Sinnamary (French Guiana) and the other on the Plantações Edouard Michelin estate from the Michelin Company in the Brazilian state of Mato Grosso. The first location is hereafter referred to as “Combi”, and the second as “PEM”.

Experimental planting designs

In both locations, Combi and PEM, were planted field experiments using progeny individuals from the cross PB260 × Fx2784. In Combi, the germinated seeds were grown in a nursery, and as soon as sufficient development was reached by each seedling, they were multiplied by grafting onto rootstocks. Each genotype thus originated a clone of grafted plantlets. A total of 118 progeny clones and the parental genotypes PB260 and FX2784 were planted in one unique block, with elementary plots of four trees of the same clone. Previous experiments with other progeny populations segregating for SALB resistance in the same location showed that the block effect on symptom expression was non-significant (Le Guen et al. 2003, 2011a).

In PEM, seedlings from the same cross of PB260 × Fx2784 were planted in the field without preliminary multiplication by grafting, as one of the progenies tested in a seedling evaluation trial, along with two other progenies. An amount of 294 seedlings representing as many different genotypes of the cross PB260 × Fx2784 was planted in a single plot without replication. The parental cultivars could not be included in the PEM experiment but were present as accessions of an ex situ collection close to the area of experimentation.

The genotypes under observation in the two field experiments were all genetically distinct sib progeny individuals. There was no replication of progeny individuals between sites so that a total of 388 distinct genotypes was planted in the overall experimental design.

SALB resistance evaluation

The evaluation of SALB resistance was made under field infestation conditions according to already described observation methodology (Le Guen et al. 2003, 2011a). Under these conditions, the plants were submitted to a natural inoculum, variable in its genetic composition and density. All observations were carried out on young trees from 6 to 24 months after planting. At this stage, trees produce a new foliar flush every 4 to 6 weeks, each flush consisting of a whorl of 8 to 20 trifoliate leaves. A successful inoculation by M. ulei can only occur on young leaves until 2 to 3 days after budding of new leaves. Although the SALB pressure was generally very high in both experimental sites, the inoculum was not uniform across time and field area so that some plants may punctually have escaped from natural infection because of the non-concomitance of susceptible young leaves and sufficient inoculum pressure. To overcome this potential bias, numerous rounds of observations were carried out in each location. The Combi experiment was observed seven times during a 9-month period (minimum of 14 days and maximum of 104 days between two consecutive observation rounds), and the PEM experiment was observed nine times during a 17-month period (minimum of 1 month and maximum of 9 months between two consecutive observation rounds). Two types of symptoms were observed for each round of observations: reaction type (RT) and stromata density (SD). RT consisted of a visual assessment of the type of lesion caused by the inoculum (i.e., necrotic, chlorotic, or sporulating lesion) on a semi-quantitative 0–6 scale. Score 0 corresponded to a symptomless tree, scores 1 and 2 corresponded respectively to necrotic and chlorotic lesions without emission of spores, scores 3 to 5 corresponded to sporulating lesions on the lower surface of the leaflet with increasing levels of sporulation, and score 6 corresponded to highly sporulating lesions on both sides of the leaflets. Because of the absence of emission of any kind of spores, scores 0 to 2 were considered as resistance interactions. On the other hand, scores 3 to 6 were characteristic of susceptible interactions with increasing levels of spore emission. SD was an assessment of the density of stromata, corresponding to the sexual phase of the fungus, on a 0–4 scale (0 for absence of stromata and 4 for high density of stromata). Stromata are durable symptoms that can be observed as long as mature leaves are attached to the tree. For each observation round, this trait was evaluated on the last mature flush of the young tree. A score of SD = 0 was considered as resistance interaction, whereas scores 1 to 4 were considered as increasing levels of susceptible interactions. In the Combi experimental site, leaves with symptoms difficult to assess on mere visual inspection were brought to the laboratory for examination under a stereo microscope (×40). Such complementary assessment was not possible in PEM. Each genotype was finally declared resistant only if its maximum scores were both RT < 3 and SD = 0 over all observation rounds. Otherwise (i.e., RT > 2 or SD > 0) it was considered susceptible.


Rubber tree total genomic DNA was extracted from fresh leaves following a previously described method (Le Guen et al. 2009). Two types of genotyping procedures were used, depending on the nature of the analyzed samples: bulk of resistant vs susceptible DNA or individual DNA.

In Combi, a resistant bulk and a susceptible bulk were constituted by grouping together DNA from ten genotypes which were considered resistant or susceptible. The final concentration of each individual DNA was adjusted to 2.5 ng μL−1 for each bulk. Previous experiments with other mapping populations enabled us to build up genetic linkage maps of rubber based on microsatellite markers (Le Guen et al. 2011a; Prapan et al. 2006). From these results, 80 microsatellite markers were chosen for being approximately distributed every 30 cM across the genome. The two bulks along with the resistant parent Fx2784 were screened with these 80 markers using PCR conditions already described (Lespinasse et al. 2000b). PCR products were labeled by binding of α-[33P]-CTP during amplification cycles and visualized following electrophoresis on polyacrylamide gels by autoradiography. The use of radiolabeled technique enabled us to visualize variations of signal intensity on the autoradiography and then to estimate the differences of allele frequency in resistant and susceptible bulks.

Once microsatellite markers with potential linkage to the resistance trait were identified, these markers, along with other markers located in the same linkage group, were used to genotype all individual DNA from the progeny planted in the Combi and PEM trials. PCR was done using 25 ng of template DNA and the forward primer labeled with a fluorochrome. PCR products were visualized using a Li-Cor 4300 DNA analyzer according to previously described technique (Le Guen et al. 2011b).

Statistical analysis

Linkage analysis and map construction were done using the JoinMap software, version 4.0 (Van Ooijen 2006). The progeny population was analyzed as a cross between two heterogeneously heterozygous and homozygous diploid parents, with unknown linkage phases (CP population). The resistant trait was converted into a binary variable as described above, and the genotypes coded as in a nn × np segregation type (nn for susceptible and np for resistant). Linkage between markers and the resistant trait was searched for only in the resistant parent. The segregation ratio of the markers was tested with a χ2 test. The grouping of markers was based on independence LOD scores calculated from the recombination frequencies. These frequencies were converted for markers of the same group into map distance using the Kosambi mapping function. Two distinct linkage maps were established for the two populations, and an integrated map was generated by merging the genotyping data of all markers for the 388 full-sib genotypes that have been planted and observed in both experiments.


Segregation of SALB resistance in the progeny

Both experiments were of relatively small size (1,920 m2 for the Combi experiment and 1,030 m2 for the PEM experiment). We draw the maps of stromata annotation registered per elementary plot (Combi) or per genotype (PEM). On these maps, no geographical bias appears within the two experiments (see Figs. 1 and 2 in the Supplementary material). It is then possible to infer that even if the expression of symptoms may have been subjected to uncontrolled localized and temporary environmental effects, every tree should have been evenly infected by natural inoculum throughout the experiment duration. The susceptible parent (PB260) showed abundant symptoms of heavy infestations in both locations throughout all experimentation, and the resistant parent FX2784 never exhibited any symptom of SALB attack.

In the Combi experiment, the percentages of trees with stromata (the easiest symptom to record) were respectively 22.7, 26.3, 26.5, 10.4, 9.8, 15.2, and 22.3 % from the first to the seventh observation round. We have computed from these data the probability for a susceptible genotype to never show stromata during the seven observation rounds (whether due to absence of observable leaves or because it always escaped from the disease). This probability was P = 0.0025 for a genotype with four living trees, P = 0.011 for three living trees, and P = 0.050 for two living trees. These probabilities were remote enough to consider that all susceptible genotypes were correctly scored and conversely that genotypes without symptoms were protected against disease by their genetic constitution.

The score distributions for the two observed symptoms are presented in Fig. 1 for the Combi trial and in Fig. 2 for the PEM trial. In Combi, where each clone was represented by four grafted trees, data were consistent between RT and SD notations (Fig. 1a, b). The synthesis between both trait notations gave a segregation of 56 resistant and 62 susceptible genotypes (Fig. 1c). These observed frequencies are compatible with the null hypothesis of a 1:1 segregation of the susceptibility among the progeny (chi-square = 0.305 for 1 degree of freedom P < 0.5).
Fig. 1

Notation of resistance amongst the progeny individuals of the cross PB260 × Fx2784 planted in Combi. a Distribution of the maximum score of reaction type registered per genotype. b Distribution of the maximum score of stromata density registered per genotype. c Synthesis of resistant (R) and susceptible (S) genotypes
Fig. 2

Notation of resistance amongst the progeny individuals of the cross PB260 × Fx2784 planted in PEM. a Distribution of the maximum score of reaction type registered per genotype. b Distribution of the maximum score of stromata density registered per genotype. c Synthesis of resistant (R) and susceptible (S) genotypes

In PEM, a slightly significant excess of resistant genotypes was registered for RT trait (Fig. 2a) and a large excess of susceptible genotypes for SD trait (Fig. 2b). Because the scores were inconsistent between the two traits, the synthesis of these two datasets gave an increased unbalance between resistant and susceptible genotypes (Fig. 2c). To overcome this difficulty, we decided to only take into account for further mapping study the genotypes with consistent and reliable scores for both traits. We then considered resistant the genotypes with maximum scores of SD = 0 and RT < 3, and susceptible the genotypes with maximum scores of SD > 1 and RT > 3. With this restriction, only 144 individuals out of the 294 could be assessed for their resistance. A total of 77 individuals were considered resistant and 67 susceptible. The 1:1 segregation hypothesis for the resistance character was also confirmed by this result (chi-square = 1.389 for 1 degree of freedom P < 0.8).

BSA and resistance locus identification on the Combi population

Genotyping of the resistant and susceptible bulks with the 80 microsatellite markers revealed a difference of band intensity between the two bulks in one of the segregating alleles for ten markers. These ten microsatellite markers were used to individually genotype the 118 progeny individuals. Genotyping data were analyzed jointly with the resistance trait which, because of its 1:1 segregation pattern, could be considered as a Mendelian marker. The linkage analysis showed a tight linkage between the resistance trait and five of the markers, all located in a linkage group previously named g2 (Lespinasse et al. 2000b). For the remaining five markers picked up by BSA, the genotyping of the whole population did not enable us to establish a significant link between the allelic composition and the phenotypic trait. The differences of band intensity between the two bulks should have been merely the consequence of random allele sampling among resistant vs susceptible bulks.

These genotyping data were completed with two other markers located in g2 and polymorphic in Fx2784. The final mapping of this linkage group g2, based on the study of progeny individuals from Combi, is presented in Fig. 3a. The resistant trait could be localized in this linkage group between the markers A2368 and A2510. We named this new marker M2fx (for Microcylus, linkage group 2, Fx2784).
Fig. 3

Localization of the major resistance gene M2fx on the genetic map of linkage group g2 for the rubber clone Fx2784. a Genetic map established from the 118 genotypes planted in Combi (French Guiana). b Genetic map established from the 294 genotypes planted in PEM (Brazil). c Synthetic map with all genotypes planted in Combi and PEM. Distances in centiMorgan from the beginning of the linkage group are indicated on the left side for each linkage group. On the right side are indicated the corresponding microsatellite markers

Validation on the PEM population

The 294 progeny individuals planted in the PEM trial were genotyped with 11 microsatellite markers of linkage group g2. The phenotype of resistance was also scored for the 144 individuals that were unambiguously characterized, and combined with genotyping data for linkage analysis. The resulting linkage map of group g2 is presented in Fig. 3b. The linkage group was 78.5 cM long, with the resistant marker M2fx located between the same two SSR markers. An additional marker, A2734, was linked to A2368 and A2510 with an intermediate position between both markers, thus reducing the size of the interval within which M2fx was located. In the map obtained with the PEM population, this resistance trait is then localized between A2734 and A2510 in an interval of 11.1 cM (Fig. 3b). The length of this interval was reduced to 10.3 cM in the synthetic map obtained after merging both Combi and PEM maps (Fig. 3c).


Our purpose in the present study was to establish the monogenic nature of the resistance of the clone Fx2784 to SALB and possibly to localize the genetic factor responsible for this resistance in the rubber genome. Prior to our work, this statement was a hypothesis based on observations of resistance against various strains of M. ulei, a hypothesis that was never demonstrated. To confirm this hypothesis, we carried out a QTL analysis of the resistance trait for progeny individuals resulting from the cross between FX2784 and the susceptible cultivar PB260. The only QTL for SALB resistance ever characterized in PB260 was a very weak one and was efficient against a unique strain of the fungus under controlled conditions of infestation (Lespinasse et al. 2001a). Like almost all other Asiatic clones when cultivated in SALB-affected areas, PB260 exhibits numerous lesions, abundant sporulation, leaflet deformation, and allows formation of the sexual stage of the fungus (presence of stromata on the upper side of the mature leaflets). On adult trees, this can lead to the fall of majority of the foliage and to the death of the plants.

Our results indicated that the pattern of segregation of this trait in a progeny population was 1:1, a strong indication of the resistance character being under Mendelian monogenic control. Once coded as a genetic marker and its linkage with molecular markers analyzed, this genetic factor was localized in a linkage group previously named g2. These two results were first obtained in part of the progeny population planted in French Guiana and submitted to natural SALB infestation. The same results were subsequently confirmed in another fraction of the same progeny planted in the Brazilian state of Mato Grosso, more than 2,000 km far from French Guiana. This second part of the experimentation in Brazil did not benefit from the same experimental conditions as in French Guiana (absence of replication of individual genotypes, no possibility to confirm the notation under stereo microscope), which could explain the inconsistence between RT and SD traits. Even though, a proper assessment of resistance phenotype and elimination of ambiguous observations enabled the same estimate position of this genetic resistance factor to be obtained. This is a good indication of the reliability of our observation of symptoms, with a low incidence of environmental conditions, particularly of grafting and rootstock heterogeneity, in the expression of SALB resistance.

However, a precise localization of this Mendelian resistance trait would necessitate the observation of trees systematically inoculated under controlled conditions. Replications of these controlled inoculations using different M. ulei strains in a larger population size should enable M2fx to be confidently located on the FX2784 linkage map.

In their comprehensive study of M. ulei isolates from the southeast Bahia state (Brazil), Mattos et al. (2003) showed that the clone Fx2784 was totally resistant to 46 isolates and highly susceptible to 4 isolates. Furthermore, the aggressiveness of these latter four isolates was low on other rubber clones, inclusive of Asiatic clones known for their absence of SALB resistance. This suggested that these four isolates belonged to a new and highly specialized race of the pathogen. We can now infer that this specialization enables isolates from this new race to overcome the resistance conferred by M2fx. Our results indicate that this newly acquired virulence seems sufficient for this race to efficiently grow on the clone Fx2784 so that the genetic load represented by high aggressiveness towards other cultivars became unnecessary and dropped for not having more adaptive value. This bypassing of M2fx resistance factor apparently did not occur yet in Combi and neither in PEM, which may be due to the smaller areas planted with Fx2784 in these two places. It still remains that the resistance to SALB conferred by M2fx exhibits efficiency against a large spectrum of races: 35 of the 36 physiological races identified by Mattos et al. (2003) were not virulent on Fx2784, to which an unknown number of other races that compound the natural inoculum in Combi and PEM can be added. Although being bypassed by at least one race of the pathogen, this locus of resistance remains interesting for breeding. The cumulative effect of M2fx with other major resistance genes or QTLs in the same cultivar should give a cross protection to a larger variety of races of the pathogen.

Previous QTL mapping studies already identified major QTLs or genes for SALB resistance in the genome of rubber. Such major resistance loci were identified in the linkage group g13 in the RO38 genome (Lespinasse et al. 2000a) and in the linkage groups g13 and g15 in the MDF180 genome (Le Guen et al. 2011a). Some QTLs with minor effect were localized in the linkage group g2 in the RO38 genome, but none of them are at the same position as M2fx (Le Guen et al. 2003, 2007; Lespinasse et al. 2000a). In the study from Mattos et al. (2003), these two genotypes, RO38 and MDF180, have shown a high level of partial resistance against the four M. ulei strains highly virulent on the FX2784 genotype. This feature suggests that the genotypes RO38, MDF180, and FX784 have non-redundant resistance loci which are efficient against different physiological races of the pathogen. It would then be of utmost interest for rubber breeders to combine all major resistance loci identified in Fx2784, RO38, and MDF180 in an improved, resistant population using a gene pyramiding strategy. Even if some of the major genes harbored by these genotypes have already been bypassed by few strains, they may continue participating to their basic resistance. It has been often hypothesized that defeated resistance genes might conserve a residual effect contributing to the overall level of partial resistance (Durel et al. 2003; Nelson 1978; Pedersen 1988) and that quantitative resistance can increase the durability of qualitative resistance (Brun et al. 2010). As already suggested by Palloix et al. (2009) for some annual crops, such “poly-resistant” population combining quantitative and qualitative resistances would be subsequently used by crossing with high-yielding genotypes in view of producing performing cultivars with sustainable SALB resistance.

The identification of the gene or the regulation site underlying the M2fx locus would help in the comprehension of the molecular mechanism of Fx2784 resistance and how it was overcome by some races of the pathogen. However, because of the big size of the Hevea genome, where 1 cM is approximately equivalent to 0.7 million nucleotides (Lespinasse et al. 2000b), and the absence of publicly available genome sequence, this identification would be very difficult through a direct genetic approach. A study of differentially expressed genes between Fx2784 and a susceptible cultivar when submitted to SALB infection, as recently undertaken by Garcia et al. (2011), could help in determining the nature of this gene or regulator by genetic mapping of the candidate resistance genes associated with the detected ESTs.

In conclusion, it is worth mentioning that high-yield rubber cultivars that are resistant to SALB are very rare because of our lack of knowledge on the genetic determinism of the different sources of SALB resistance. Resistance to a disease can only be efficiently used in a breeding program if breeders know how the resistance is inherited and whether it is linked or not with other desirable or unfavorable traits. By locating the resistance locus in the Fx2784 cultivar and showing that it is different from all already known resistance loci, we provide a new opportunity to use this cultivar in short-term breeding programs for the creation of new clones as well as in long-term strategies of gene pyramiding for improved, resistant populations. An assessment of rubber yield in the progeny we analyzed in the present work would quickly provide valuable information towards these goals.


This work was performed as part of the Cirad–Michelin collaborative project “Genesalb,” supported by a grant from the French national research agency (Agence National de la Recherche; contract ANR/Génoplante n° GPLA07017C).

Supplementary material

11295_2013_599_MOESM1_ESM.docx (4.1 mb)
ESM 1(DOCX 4185 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2013