Swinnen, G.; Afonso, S.; Lacchini, E.; Stolz, S.; Lize, E.; Soyk, S.

Phenotypic diversity arises in large part from genetic variants at multiple interacting loci, many of which alter gene dosage rather than abolish gene function. Dosage-sensitive variants, which often produce nonlinear phenotypic outcomes, can be exploited to fine-tune quantitative traits for crop improvement using genome editing. However, the phenotypic outcomes of individual variants can differ substantially across genetic backgrounds, as segregating alleles may modulate allelic effects in unexpected ways. Yet, how genetic background shapes gene dosage effects remains underexplored. Here, we show that MADS-box gene dosage effects, which can be used to tune tomato inflorescence architecture for optimal fruit yield, differ profoundly between distinct genetic backgrounds. We mapped the genetic basis of this background dependency and identified the cryptic modifier locus suppressor of branching 2 (sb2), which contains the conserved floral identity gene ANANTHA. We show that natural variation at sb2 modulates how inflorescence architecture responds to MADS-box dosage effects from natural and engineered loss-of-function mutations. Our findings illustrate how cryptic genetic variants can reshape gene dosage relationships and underscore the importance of characterizing such hidden variation for predictive engineering of quantitative traits using genome editing.