Abstract
The predominant view regarding Asian rice domestication is that the initial origin of nonshattering involved a single gene of large effect, specifically, the sh4 locus via the evolutionary replacement of a dominant allele for shattering with a recessive allele for reduced shattering. Data have accumulated to challenge this hypothesis. Specifically, a few studies have reported occasional seed-shattering plants from populations of the wild progenitor of cultivated rice (Oryza rufipogon complex) being homozygous for the putative "nonshattering" sh4 alleles. We tested the sh4 hypothesis for the domestication of cultivated rice by obtaining genotypes and phenotypes for a diverse set of samples of wild, weedy, and cultivated rice accessions. The cultivars were fixed for the putative "nonshattering" allele and nonshattering phenotype, but wild rice accessions are highly polymorphic for the putative "nonshattering" allele (frequency ∼26%) with shattering phenotype. All weedy rice accessions are the "nonshattering" genotype at the sh4 locus but with shattering phenotype. These data challenge the widely accepted hypothesis that a single nucleotide mutation ("G"/"T") of the sh4 locus is the major driving force for rice domestication. Instead, we hypothesize that unidentified shattering loci are responsible for the initial domestication of cultivated rice through reduced seed shattering.