Abstract: Natural quantitative variation in developmental processes, such as cell proliferation, must be driven by allelic variation. Yet, the genotype-phenotype relationships underlying such trait variation are understudied due to their inherent complexity. Taking advantage of the simple Caenorhabditis elegans germline stem cell system, we used a quantitative genetic approach to characterize natural differences in germ stem cell niche activity of two distinct wild isolates, measured as differences in germline progenitor zone (PZ) size. Using quantitative trait locus (QTL) analysis, we detected multiple candidate loci, including two large-effect loci on chromosomes II and V. Resolving the chromosome V QTL, we discovered that the isolate with a smaller PZ exhibits a unique 148 base pair deletion in the promoter region of the Notch ligand, lag-2, a central signal promoting germ stem cell fate and proliferation. As predicted, introducing this deletion into the isolate with a large PZ resulted in significantly smaller PZ size. Unexpectedly, re-introducing the deleted ancestral sequence in the isolate with a smaller PZ caused PZ size to reduce even further. We show that these seemingly contradictory phenotypic effects are due to, partly antagonistic, epistatic interactions among the lag-2 promoter, the chromosome II QTL, and additional loci elsewhere in the genome. While we identified an obvious causal polymorphism explaining natural variation in germ stem cell niche activity, the effects of this variant became unpredictable due to higher-order epistasis. Studying the genetic architecture of quantitative developmental systems without taking into account its natural variation may be misleading, emphasizing the need for a better integration of developmental and quantitative genetics.