Rice awn reduction: not a domestication trait

In a recent paper in Economic Botany, Svizzero, Ray and Chakraborty (2019) ask whether awn reduction is really part of the domestication syndrome of rice or instead a crop improvement trait. It is true that I once listed this as part of the domestication syndrome traits in cereals (Fuller 2007): awn/appendage reduction, but I don't think I really pressed for this to a be central part of domestication. It was only meant to identify a tendency that recurs across most cereals. I would now agree that this post-domestication and varietal improvement trait. That this should be regarded as post-domestication/improvement trait is clear from the fact that it affects only some lineages of rice (most temperate japonica, many indica, but few tropical japonica or circum-aus, and mostly not Oryza glaberrima). Many cultivars have awns of varying length. The authors usefully illustrate diagrammatically how economy (foraging, pre-domestication cultivation, agriculture), harvesting methods and the evolution of some morphological traits (like the awn) are not synchronous changes. These are part of what makes domestication a protracted and entangled process.

In the case of rice, the presence of awns may have actually been necessary during the early stages of cultivation! This is the implication of recent work by the Kobe rice genetics group (Ishii, Ishikawa and colleagues), who have recently published data identify three interacting loci in rice (Amarasinghe et al 2020 in Rice) . They have worked the slow processingf of backcrossing indica genetic components into a population of wild rice, and are able to document the additive effects of 3 loci on reducing the awn length. The reason that awns are important in wild rice has to do with seed dispersal-- like in other cereals awns help to propel spikelets across and into the soil, and in the case of rice they likely also play a role in dispersal across water. But the reason they would have been important to early rice gatherers and the first rice cultivators is that awns mean that grains, even after they have shattered, to get stuck within the rice panicle, making them more readily gatherable by people. This goes hand in hand with a key domestication trait, previously identified by work at Kobe, namely the closed panicle morphology (controlled by the OsLG1 mutation, also called SPR3, published in Nature 2013). This closed panicle trait would make the more compact panicles easier to harvest, especially with long awns that allow shattered spikelets to get caught. Indeed, the data in Ishii et al (2013), allows us to estimate an increased in return rate on harvested rice, with just this mutation of about +50%. This actually implies that awns were selected for during earliest phases of cultivation, but would have lost selection advantages (in terms of harvest) once non-shattering became dominant.  Thus, as concluded by Amarasinghe et al 2020,  after domestication it became reasonable to selected for reduced awns. Without awns harvested spikelets take or less space and plant would not have needed the same degree of metabolic investment in producing awns. The reduction of awns is also common in many varieties of wheats, barley, oat, but I suspect it was not actually part of the initial domestication in these either. Thus the conclusion of Svizzero et al (2019) may well have relevance to many cereals beyond rice.

However, Swizzero et al are wrong that this trait can not be studied archaeologically, even if it really hasn;t been so far.  We are able nor to recover lemma apex remains and determine these as awned or awnless (as in the example at left from work by Cristina Castillo at Khao Sam Kaeo, Thailand). In recent years work by Castillo, myself, and others has increasingly recovered these from flotation samples. They are lightly smaller than spikelet bases, less common than spikelet bases but nevertheless can be reognized as either complete apices (with a beak like end) or torn (like the one at left) indicating that they were awned. Thus there is potential to study this archaeologically. The example shown here is part of a dataset that indicates the predominance of awned japonica rice in Thailand upto at least the Iron Age. In waterlogged material from the Lower Yangtze it is clear that early rice, even non-shattering ones are awned. More work to be done on recognizing when and where awnless rices became common. Reduced awn or awnless rices appear to be present at Roman era sites on the Red Sea soast of Egypt (at Berenike and Quseir al-Qadim), at which time they are presumably being imported from somewhere in southern or western India (published in the books by Rene Cappers, and Marijke van der Veen).

Rice domestication was slow, and post-domestication varietal diversification, including the reduction of awns in some varieties, took even longer. And there is much we still don't know about this history-- but a history of varietal and geographical diversification starting to be unravelled in part through genomic history (but that should be a post for another time).