African Archaeobotany watch: the small matter of tef

For those who like to generalize, it is often suggested that early farmers chose those grasses in their environment that had the largest seeds (e.g. in Diamond's Guns, Germs and Steel), but if you want an exception, look to the Ethiopian "millet" tef [it is as much a "millet" as Finger millet, as both are Eragrostidae not Panicoids, the true "millet-grasses"]. It has by far the smallest grains of any cereal little different in size from poppy seeds when judged by the naked eye [it's length is around 1-1.5 mm and its breadth about 0.5 mm]. While domesticated cereals tend to evolve bigger grains than their wild ancestors, in Eragrostic tef this is only very marginally the case. In a recent exhaustive study published in Economic Botany Dec. 2008, the archaeobotanist Cathy D'Andrea lays out what we know, including much important new archaeobotanical data of her own. The article includes extensive data from carbonization experiments on modern wild and domesticated Eragrostis, a review of the Axumite and pre-Axumite agriculture of Ethiopia (with the continuing problems of whether indigenous crops pre-dated the introduced wheat, barley and flax; and how much earlier than 1000 BC agriculture got started in Ethiopia). From the point of view of domestication process, I note a few points from her discussion.

• Tef has only partly lost grain shattering; while it is apparently reduced it remains a problem with mature crops (much as is true of sesame). The fact that this 'semi-domesticated' state has persisted for at least 2500-3000 years should be a warning to those pondering domestication processes: non-shattering need not evolve (either rapdily or in absolute terms) for a crop to be successfully cultivated and a persistant part of subsistence. Is there is model here for wheat, barley, rice or other millets during the enigmatic era of 'pre domestication cultivation'?
  
  
• The lack of clear early selection for grain size increase also attracts discussion. In part this might be accounted for by increase in grain number. Yield is a combination of both factors. She suggests that lack of intensive tillage may have operated against selection here. As a proponent for a tillage-grain size increase link I, of course, am prone to agree (see my Ann Bot paper). But importantly she attributes this to a concern for lodging, in which tall plants fall over under their own weight, or the weight of their grain, reducing yields and harvestibility. Apparently lodging is a big problem in tef, and made worse when soils are heavily tilled. This raises interesting comparative questions about other crops, such as to what extent similar concerns might have played a role early in cultivation. Most domesticated cereals have rather thicker culms than their wild progenitors; they are also often taller and straighter. Wild rices, and I reckon primitive cultivars, were spreading and thin-stalked. Operating against the development of erect forms may have been concerns with lodging, at least until thicker stalks evolved, and in some contexts this may have also operated against the trend towards larger grains, and helped to contribute to the vast variability in grain size one encounters across archaeological rices.
  
  
• D'Andrea points to the domesticated Digitaria spp. of west Africa (iburu and fonio) as possibly following a similar domestication pathway with similar constraints. Is there is more generalizable pathway here, perhaps applicable to Brachiaria spp. and Paspalum scrobiculatum as well, in which think culms that threaten lodging operate againt major increases in grain size and grain density, but not overall grain number which is dispersed across the panicle? This might also be relevant for some Panicum spp. (sumatrense, sonoram; but I guess not P. miliaceum)?

Much food for thought and grist for the mill in this paper, in addition to being a nice contribution to the under-archaeobotanized East Africa.