Lovell, J. T.; Rhein, H.; Sreedasyam, A.; Bentley, N.; Klein, P. E.; Chatwin, W.; Padgitt-Cobb, L.; Harder, A.; Grabowski, P. P.; Carey, S. B.; Harkess, A.; Jenkins, J. W.; McLaughlin, C.; Plott, C.; Rifkin, J.; Song, J.; Webber, J.; Williams, M.; Grimwood, J.; Schmutz, J.; Wang, X.; Randall, J.

Many hermaphroditic species increase outcrossing rates by partitioning reproduction so that male and female organs mature at different times, a phenomenon known as dichogamy. Previous work has documented that dichogamy in pecan trees is governed by the Mendelian super-gene 'G-locus'; however, its non-recombinant sex chromosome-like architecture has impeded quantitative genetic exploration and candidate gene discovery. Here, we probe the genetic drivers of the G-locus through a pangenome-integrated quantitative genomics experiment. We first provide one of the clearest examples to date of mapping bias, where a linear reference-based GWAS discovered 66 off-target peaks while mapping with a pangenome graph reference resolved the known single Mendelian locus. Across six new genome assemblies, the fully haplotype phased G-locus QTL spanned 223-491kb and included 25 candidate gene families. The strongest candidate gene encoded a MATE efflux protein and had dominant allele-specific action during male flower developmental stages. Combined, these candidates and genomic resources provide a powerful foundation for breeding and optimal dichogamy phenotype engineering for future pecan orchards.