Growing multiple varieties of rice on neighbouring farms would risk cross pollination, causing genetic impurity of the varieties. Maintaining the recommended physical pacing of 110 m between neighbouring varietiesc is impractical for Asian farmers, who tend to grow many landraces on their small farms. Variety-specific flowering dates and short-lived pollen give a clue to avoiding cross-pollination. Cross-pollination can be avoided by maintaining a temporal difference between the flowering times of neighbouring cultivars.

An apparently uncontrollable source of varietal intermixing is cross-pollination, which occurs at a considerably low frequency between the cultivated rice and its wild relatives (especially O. rufipogon), and between landraces of the former. In order to prevent the risk of crosspollination, rice researchers recommend a spatial isolation of about 110 m from seed production plots to other rice varieties. Some authors recommend an
isolation distance of up to 200 m for male sterile (A line) multiplication. However, it may not be feasible for
small and marginal farmers of Asia to leave gaps of 110 m – or even 5 m – between plots of rice crops on their typically small farms growing two or more local rice landraces.

As a more practicable alternative, I
suggest here to maintain a temporal distance between cultivars in terms of flowering time. Some scholars recommend
keeping a gap of at least 30 days between the flowering stage of the parental lines in the seed-production field and that of other varieties grown within the area to avoid contamination by pollen5. However, I argue here that a temporal separation by 12 h between the onset of flowering of one cultivar and the beginning of the milking stage of its neighbour is sufficient to check out-crossing. of pollen release and stigma exposure in different rice landraces is not difficult because the onset of flowering in rice takes a characteristic, landrace-specific length of time after sowing. The date of
50% flowering (when half of the panicle
bears florets) is more or less landrace specific and relatively photoperiod-invariant. The length of time until flowering (DUF), measured in days between the sowing date and the date of first flowering of a given cultivar, does not vary more than 5 days beyond the mean number of days, regardless of variations in temperature, humidity and photoperiod. The duration of flowering stage (DF) from the onset of flowering to the ‘milk’ stage, is also relatively landrace-specific. Thus, each rice landrace can be identified by its characteristic DUF and DF, which may
serve as a helpful guide to design plantation of different landraces on a farm.

A cropping design based on flowering asynchrony between 360 rice landraces grown on adjacent plots of a small experimental farm of the Centre for Interdisciplinary Studies has proved effective in maintaining varietal purity over a period of five years. For each of these landraces, 18 morpho-logical characteristics were recorded for comparison from the year 2000 to 2005. Periodic examination of these 18 descriptors of each landrace revealed that none had deviated from the standard record of these characteristics.