Abstract
Recent work has provided strong empirical support for the classic polygenic model for trait variation. Population-based findings suggest that most regions of genome harbor variation affecting most traits. This view is hard to reconcile with the experience of researchers who define gene functions using mutagenesis, comparing mutants one at a time to the wild type. Here, we use the approach of experimental genetics to show that indeed, most genomic regions carry variants with detectable effects on complex traits. We used high-throughput phenotyping to characterize demography as a multivariate trait in growing populations of Caenorhabditis elegans sensitized by nickel stress. We show that demography under these conditions is genetically complex in a panel of recombinant inbred lines. We then focused on a 1.4-Mb region of the X chromosome. When we compared two near isogenic lines (NILs) that differ only at this region, they were phenotypically indistinguishable. When we used additional NILs to subdivide the region into fifteen intervals, each encompassing ~0.001 of the genome, we found that eleven of intervals have significant effects. These effects are often similar in magnitude to those of genome-wide significant QTLs mapped in the recombinant inbred lines but are antagonized by the effects of variants in adjacent intervals. Contrary to the expectation of small additive effects, our findings point to large-effect variants whose effects are masked by epistasis or linkage disequilibrium between alleles of opposing effect.
