Reduction in seed shattering was an important phenotypic change during cereal domestication. Here we show that a simple morphological change in rice panicle shape, controlled by the SPR3 locus, has a large impact on seed-shedding and pollinating behaviors. In the wild genetic background of rice, we found that plants with a cultivated-like type of closed panicle had significantly reduced seed shedding through seed retention. In addition, the long awns in closed panicles disturbed the free exposure of anthers and stigmas on the flowering spikelets, resulting in a significant reduction of the outcrossing rate. We localized the SPR3 locus to a 9.3-kb genomic region, and our complementation tests suggest that this region regulates the liguleless gene (OsLG1). Sequencing analysis identified reduced nucleotide diversity and a selective sweep at theSPR3 locus in cultivated rice. Our results suggest that a closed panicle was a selected trait during rice domestication.
This is further important work from the Kobe University rice research group (Ishii, Ishikawa, and colleagues) which derives from careful experimental growing on wild rice and breeding of crop traits in wild rices. A few years ago they showed that sh4 alone was not apparently an effective mutation for non-shattering, as normally inferred, but required additional interacting mutations. They have identified a key trait (if not the key trait), SPR3, which delays shattering, and together with sh4 essentially leads to the key domestication trait of non-shattering. Of particular interest it is has the side effect of decreasing cross pollination, and thus pushing rice towards selfing, another key change with domestication. The main feaure of SPR3 seems to be towards a closed (and less branching) panicle, which we would expect to have the effect of making it easier to gather a larger proportion of grains when harvesting using hunter-gatherer methods like basket beating. Ethnographically wild rice gathering (including of Zizania amongst the Ojibwa) often features tieing panicles into a knot after flowering when grains are green so as to catch early shattering grains. Essentially this mutation achieves a similar end and we might therefore see this as likely quite early in the domestication process. I suspect this goes some way towards helping explain how rice domesticiation worked in the absence of sickles (which are 3rd Millennium BC in the Yangtze), a clear contrast from wheat and barley. It is also worth noting that long awns play a role in retaining mature grains, which goes some way to explain why selection for awnlessness is so clearly an incomplete and post-domestication development.