Supporting Information

Files in this Data Supplement:

Fig. 5. Statistical power of transgressive tests. Each data point represents one simulation of 1,000 transcripts controlled by two loci with additive, opposing effects. The x axis gives the ratio of effect sizes of the loci, and the y axis reports the proportion of transcripts that pass a given test for transgressive segregation. Each curve represents one test. Red curve: For a transcript to be called significantly transgressive, >58 segregants must have phenotype values >1 s outside the parent means, where s is the pooled standard deviation among parental measurements. Green curve: >35 segregants must have phenotype values >1.5 s outside the parent means. Blue curve: >21 segregants must have phenotypes >2 s outside the parent means. Purple curve: >13 segregants must have phenotypes >2.5 s outside the parent means. Cyan curve: >8 segregants must have phenotypes >13 s outside the parent means. For all models, a = 5.

Fig. 6. Distribution of linkage statistics for real and simulated transcripts. Shown are the negative natural logarithms of P values from linkage tests on transcripts simulated under the best-fit additive model (red) and true transcripts with significant heritabilities (green). For traits linking to multiple markers, results are for the QTL with the most significant P value. The peak at –ln(P) = 37 is the effect of truncation for numerical reasons (see Materials and Methods). Density just to the left of this peak represents transcripts that link to QTLs with more significant P values than we expect from the 2-locus model but less than the 1-locus model, suggesting genetic control by a single major locus plus one or more minor loci.