Genotyping by sequencing and QTL mapping for Fusarium ear rot resistance and fumonisin B1 accumulation in maize.

Fusarium verticillioides is the causal agent of Fusarium ear rot (FER) in maize and contaminates the grains with fumonisins, a family of mycotoxins that affects feed and food. Quantitative genetic variations exists for resistance to FER and fumonisin B1 accumulation among maize genotypes. Both traits have moderate to high heritability and Marker Assisted Selection of resistant genotypes is an attractive approach to control this problem in maize crops. In order to genetically dissect FER responses in maize and identify molecular markers associated with resistance loci, a cross from the resistant CO441 and the susceptible CO354 parents was generated and F2:3 segregant maize families were evaluated for resistance to FER and fumonisin B1 accumulation in both 2011 and 2012. Phenotyping was conducted on artificially inoculated F3 ears at 15 days after pollination (DAP) with two side-needle inoculation methods. FER resistance was evaluated at maturity as a percentage of infected kernels on the ears. Fumonisin B1 accumulation in the grains was predicted by NIR spectroscopy. In parallel, a molecular linkage map was constructed for the CO441xCO354 progeny using a Genotyping-by-Sequencing (GBS) approach. GBS provides low cost, high-density information useful to develop highly saturated linkage maps and to add new value to traditional bi-parental mapping and breeding populations. In total 157 F3 DNA pools and the parents were restricted with ApeKI, 96-plex barcoded libraries were constructed according to the Elshire protocol (PLoS ONE 6(5):e19379) and sequenced on an Illumina HiSeq2000 instrument. Initial analyses identified a set of 16.236 SNP markers. Stringent criteria were applied for SNP calling and filtering included a minimum quality score of 20 (Phredscale) for reads bases and a minimum reads mapping quality of 30 (Phred-scale), absence of missing data in the reference samples, less than 30% missing data in the population for each SNP filters on segregation distortion and linkage disequilibrium. Finally a set of 339 SNPs were integrated with genotyping data for 72 SSRs to construct a linkage map. A total of 31 QTLs (four traits in two different years) were detected, using IM and MQM analysis, in five main chromosomal regions.