McGilp, L.; Millas, R.; Mickelson, A.; Shannon, L. M.; Kimball, J.

Cultivated Northern Wild Rice (Zizania palustris L.) is an obligately outcrossing, self-incompatible cereal grown in aquatic paddies in the United States. Genetic improvement has relied primarily on phenotypic recurrent selection, and genomic approaches remain largely unexplored in this emerging crop. We applied a single-plant genome-wide association study (sp-GWAS) framework to dissect vegetative architecture traits in five open-pollinated cultivated populations evaluated across three years (n = 2,173 plants). Plant height (PH), basal stem width (BSW), primary stem width (PSW), flag leaf length (FLL), and flag leaf width (FLW) were analyzed using a mixed linear model accounting for population structure and kinship. Broad-sense heritability ranged from 0.03 to 0.34, and year effects explained up to 54% of phenotypic variance, indicating strong environmental influence. After filtering 73,363 SNPs, genome-wide linkage disequilibrium decayed rapidly (r2 = 0.1 at ~2.3 kb). A total of 124 significant SNPs (FDR < 0.01) were consolidated into 98 loci, of which 46 were associated with multiple traits and 11 were shared across four traits. Candidate genes near multi-trait loci included conserved regulatory classes implicated in grass architecture, including HLH/bHLH transcription factors. Diplotype analyses at candidate loci revealed both simple biallelic and complex multi-allelic haplotype structures, indicating that locus-level haplotype effects underlie several GWAS signals. Results demonstrate that sp-GWAS can detect statistically robust associations in a highly heterozygous, non-replicable crop system and suggest a polygenic, coordinated genetic architecture governing vegetative growth. These findings support genomic prediction and multi-trait selection strategies to accelerate improvement of cultivated Northern Wild Rice.