The genetic determinants of SHR resistance were investigated following a candidate gene approach intended to provide fine-scale resolution to QTL mapping efforts for this trait. Selection of candidate genes is one of the fundamental challenges of candidate gene AM, particularly when there is limited knowledge about the molecular mechanisms underlying the trait under study. For SHR, the most comprehensive experimental evidence comes from the study of gene expression changes in two genotypes of the oilseed B. napus infected with S. sclerotiourum using a whole genome microarray from A. thaliana. However, transferring this information to sunflower was hindered by ortholog identification between highly divergent species. Commonly, most researchers use pairwise distance comparisons algorithms, such as BLAST, COG (Clusters of Orthologous Groups), RBH (Reciprocal BLAST Hits), RSD (Reciprocal Smallest Distance Algorithm) and INPARANOID, to determine gene orthology . In contrast, although they have been shown to exhibit greater accuracy and lower error rates than pairwise comparison methods, phylogeny based approaches have been only partially exploited due to the complexity of the automation of sequence alignment and the choice of appropriate genes and species to be included in the analysis [42, 43]. Notwithstanding, it is only by phylogenetic reconstruction that ortholog and parolog relationships can be distinguished, especially for species with incomplete genome sequence data where the best hit of pairwise comparison methods is often not the nearest neighbor .
The phylogenetic approach devised here allowed identification of 17 genes from Asteraceae that are orthologs or paralogs to the A. thaliana loci detected as over-expressed at 24 hpi in the B. napus resistant genotype. Particularly, for At5g51550, At3g48310, At5g05940, At1g04450 and At1g13200, two to three loci were identified as paralogs to the A. thaliana sequence used as query, providing wider coverage of their putative functional spectrum (Table 1).
Two additional sources served to increase to 30 the number of candidate genes, an EST library obtained through suppressive subtractive hybridization from sunflower capitula of the genotype RHA801 infected with S. sclerotiorum at 48 hpi, and previous literature reports [36–38]. After identification of polymorphisms (SNPs and indels) within the core set of 10 inbred lines, a total of 21 candidate genes were selected to be further genotyped in the AMP. This number of candidate genes has proved adequate to find significant genotype-phenotype associations for traits with different degrees of complexity. Examples include the studies carried out in A. thaliana for flowering time , in potato for late blight  and in maize for aluminum tolerance . Moreover, in this study the authors used information sources similar to those described here to select the candidate genes, finding significant associations for six of them.
The AMP analyzed here is representative of the elite breeding pool used by INTA in the “Sunflower Breeding Program”. It encompasses germplasm from different geographical origins, with some of the lines being derived from introgressions with wild Helianthus species (Additional file 1). In agreement with the morphological diversity and intricate ancestry of the AMP, the SSR markers revealed high levels of genetic variability. Indeed, the mean number of alleles per locus was even higher than the estimate obtained for a set of compounds, populations and lines conserved at the INTA Germplasm Bank (6.6 vs. 5.71) .
As suggested by Stich et al. , marker-trait associations were assessed using MLM, which took into account trials, blocks, population structure and kinship relationships to control type I error rates. Both PCO and the Bayesian method (STRUCTURE software) were used to infer population structure for the 94 elite inbred lines. While PCO analysis showed no clear grouping pattern by drawing the two first Principal Coordinates, STRUCTURE suggested the presence of nine different gene pools (Additional file 2). The lack of defined groups in PCO is not unexpected given the restricted genetic of cultivated sunflower . However, either using P or Q matrixes in the association analysis did not have an impact on the P-values (Table 2). Moreover, varying the probability T for the calculation of the K matrix did not modify the association results.
A significant association was detected between a lower SRH incidence and the haplotype 3 of HaRIC_B (P < 0.01). The nature of the mutation found in the haplotype 3 of HaRIC_B, the highly specific expression pattern for this gene in sunflower and the experimental evidence on the role of this protein family in A. thaliana[48, 49] lend support to the biological significance of the detected association. HaRIC_B was selected from the study of Zhao et al.  following ortholog identification via phylogenetic analysis. Haplotype 3 showed an insertion of 311-bp at the 3’ end of exon two, which alters the RNA splicing and, consequently, leads to both the generation of a frame-shift and a premature stop codon in the mRNA. Only ca. 15 % of mutations that have been identified as being associated with genetic variation in plant quantitative traits involve changes in the amino acid composition of proteins. However, it has been shown that many of the associations that correspond to noncoding mutations located in introns, untranslated regions or intergenic regions show up as significant because they are in LD with untyped causal mutations that in turn are nonsynonymous substitutions .
Although HaRIC_B molecular function has been inferred by homology, the orthologous A. thaliana gene (AT1G04450, RIC3) has been experimentally described as a small binding-protein that interacts with ROP1 (Rho family GTPase). There are 11 Arabidopsis RIC proteins (for Rop-interactive CRIB motif–containing proteins) involved in pollen tube growth and other functions . Wu et al.  have shown that RIC3 transcripts are found only in the flowers and inflorescences of A. thaliana. In agreement with its proposed ortholog relationship, the expression pattern of HaRIC_B matches that found for RIC3, with transcripts being present in florets from R3 to R6 developmental stage (Figure 2D). It is noteworthy that R5.2 is the period of maximum susceptibility to S. sclerotiorum infection in the sunflower cultivation areas of South America, and florets are the main entry point for the pathogen .
Different studies suggest that the role of RIC3 involves elevation of cytosolic Ca2+ and regulation of (1) de-polymerization of actin filaments, (2) exo-cytosis, and (3) Ca2+ mediated signals [48, 51]. Interestingly, the Rop-interactive domain of HaRIC_B is located in exon two, which is missing in plants carrying haplotype 3. Thus, if a functional protein can be synthesized from it, its regulation by Rop GTPases would seemingly be not possible.
Recently, two necrosis and ethylene-induced peptides (NEPs) have been described in S. sclerotiorum (SsNep1 and SsNep2). SsNep2 expression is highly dependent on Ca2+ concentration, and compounds increasing calcium levels (i.e. caffeine and lanthanum chloride) greatly reduced S. sclerotiorum virulence and expression of SsNep2. Thus, consistent with its putative role of intracellular calcium elevation, the haplotype 3 of HaRIC_B might be participating in the defense against SHR by repressing expression of the necrosis factor SsNep2, through the de-regulation and elevation of cytosolic Ca2+ concentrations in the target organ for pathogen attack.
The power of association mapping greatly depends not only on the allele frequency distribution but also on the magnitude of the effect that can be ascribed to a locus, relative to other loci present in the population . Thus, the detection of association between HaRIC_B (haplotype 3) and a lower SHR incidence, despite the fact that haplotype 3 was in low frequency, suggests that it has a strong effect on the phenotype. Indeed, considering that resistance is the result of the interaction of multiple factors, having found a single haplotype that accounts for a SHR incidence reduction of about 20 % in average, emphasizes the importance of the finding. However, beyond the experimental evidence presented here, and the biological considerations that support the role of haplotype 3 on the resistance to SHR, validation of these results will require the re-evaluation of this candidate polymorphism in a larger AMP and different field testing environments. A wider association study is currently underway using large-scale gene sampling and high-throughput genotyping methods.
It has been shown that the causal polymorphism for a QTL can be distant from the functional gene under analysis, particularly in species with high levels of LD, such as sunflower [33–35, 54]. While it cannot be ruled out that the polymorphism responsible for a lower SHR incidence resides in a linked ungenotyped region, the evidences discussed here suggest that HaRIC_B can be considered a strong candidate to be directly involved in SHR resistance. To assess its relationships with QTL previously identified in biparental populations, HaRIC_B was genotyped in the sunflower RIL mapping population PAC2 X RHA266. It was mapped to LG11 between the markers E36M59_9 and E38M50_17 (data not shown), a region for which no QTL have been reported to date [3, 7, 9–14, 55]. This is not unexpected, as the parental lines used to investigate SHR resistance may not carry the haplotype 3 of HaRIC_B, especially considering its low frequency in the AMP. In this context, HaRIC_B may be thought of as a new and highly delimited QTL for SHR resistance.