Fusiform rust disease, caused by the endemic fungus Cronartium quercuum f. sp. fusiforme, is the most damaging disease affecting economically important pine species in the southeast United States. In this report, we detail the genomic localization and sequence-level discovery of candidate race-nonspecific broad-spectrum fusiform rust resistance genes in Pinus taeda L. Two full-sib families, each with ~1000 progeny, were challenged with a complex inoculum consisting of over 150 pathogen isolates. High-density linkage mapping revealed three QTL distributed on two linkage groups. The two QTL on linkage group 2 were additive with respect to their effects on the probability of disease outcome. All three QTL were validated using a population of 2057 cloned pine genotypes in a six-year-old multi-environmental field trial. As a complement to the QTL mapping approach, bulked segregant RNAseq analysis revealed a small number of candidate nucleotide binding leucine rich repeat genes harboring SNP significantly associated with disease resistance. The results of this study demonstrate that single qualitative resistance genes can confer effective resistance against genetically diverse mixtures of an endemic pathogen.

Linkage maps were produced using the two-way pseudo-testcross design for both families. Briefly, prior to linkage mapping, the genotype of each parent at each marker was ascertained. For each cross, a linkage map was produced using markers in either backcross configuration (AB:BB or BB:AB), resulting in separate maps for the maternal and paternal genomes. Markers in the intercross configuration (AB:AB) were used to generate the sex averaged map for each LG, but these were dropped from the dataset prior to QTL analysis since the linkage phase of heterozygous genotypes for which both parents are heterozygous is unknown in an outbred F1 cross. The consensus map combining the linkage maps from all four parents was generated through linear programming methods. Details of these procedures are presented in Supplementary Materials.

All QTL analysis was conducted using the R package R/QTL (Broman & Sen, 2009). Interval mapping was conducted via the Expectation Maximization algorithm, using a logistic regression model implemented in the scanone function of R/QTL. LOD thresholds for declaring QTL were determined using 1000 permutations of the phenotypic data. Peaks which surpassed the LOD threshold were declared as putative QTL, and their genomic positions under a conditional model (with more than one QTL) were determined using the refineqtl function. For identified QTL, genotype probabilities for each individual were estimated by calc.genoprob. Individuals were assigned to two genotype classes (Rr, rr) in family E9, and assigned to four genotype classes (rr/rr, Rr/rr, rr/Rr, Rr/Rr) in family E4. These genotype classes were used as categorical variables in generalized linear models.

The amount of variance explained by each QTL was estimated using the fitqtl function in R/QTL, which utilizes a simple transformation of the conditional LOD score. Additive effects for QTL haplotypes and specific contrasts were estimated using the GLIMMIX procedure of SAS/STAT software v9.1 (SAS Institute, Cary NC, 2013). In these models, the susceptible haplotype (either one- or two-QTL susceptible haplotype) was declared as the reference level.

The R/QTL datasets can be read directly into R. They are "format = csv".

To read the datasets into R/QTL, use the function "read.cross", specifying the full directory to the file, and using the "format=csv" argument.