4. A survey of within-population genetic diversity in land races and wild rices of tropical Asia

Populations of wild rices as well as cultivated land races are genetically heterogeneous in varying degrees. The within-population genetic variation is thought to represent evolutionary potentiality. In the common wild rice, perennial types carry more genetic variations in their populations than do annual types in quantitative characters and isozymes, whereas the annual populations tend to be more differentiated genetically from one to another (Morishima and Oka 1970; Oka 1988, p. 69-72). Some of land-race populations were found to be more heterogeneous than annual wild populations although the plants are predominantly self-pollinated.

To learn about the pattern of distribution of within-population genetic variations, the author has made a survey of variations in spikelet length and width using the data from samples collected from Nepal, India and Thailand in 1979 (cf. Morishinia et al. 1980). For estimating genetic variance for a quantitative trait, so-called environmental or error variance should be subtracted from the observed phenotypic variance. Spikelet length and width have relatively small error variances and are suited for this purpose. The error variance for spikelet size was estimated from variations within single panicles of a nearly homogeneous cultivar (Nipponbare) grown by a farmer near Mishima City, taking 10 spikelets from each of 40 panicles of different plants (n= 400, df= 360). The variance of spikelet length (mean - 7.63 mm) was 0.0475, that of spikelet width (mean - 3.62 mm) was 0.0110, and their covariance was 0.0049. It was assumed that these values were applicable to all samples collected overseas.

Thus, the data for 52 land-race and 26 wild-rice populations were examined. A sample consisted of about 30 to 120 mature spikelets, each taken from a plant randomly selected from the field. The integrated genetic variance of spikelet size was represented by the square root of generalized genetic variances for spikelet length (V/L/) and width (V/W/), (G)1/2 = (V/L/ x V/W/- C/OV/². Using the same data, the genetic correlation between length and width (r/g/) was also computed in each population.

The (G)1/2 values ranged from 0.02 to 0.26 in land races and 0.02 to 0.14 in wild populations. The rg values ranged from about 0.6 to -0.6 in both. The distributions of these values showing within-population genetic variations are summarized in Table 1. In the common wild rice, perennial populations tended to have larger (G)1/2 values than annual ones, as was found previously (Morishima and Oka 1970). In land races, populations from the hill area of Nepal tended to have larger (G)1/2 values than those from India (including samples from Jeypore Tract). Among Thailand samples, upland-rice populations had larger (G)1/2 values

The r/g/ values showed similar trends to those observed for (G)1/2. The (G)1/2 and r/g/ values were negatively associated (r= -0.42) among land-race populations, suggesting that in populations highly heterogeneous in spikelet size, spikelet length and width tend to be negatively correlated. Among wild-rice populations, this correlation was not significant.

During our study tour, we tried to judge from outward appearance whether Indica or Japonica land-race populations look like, but some populations appeared to be intermediate and were not distinguishable. The test of each plant for phenol reaction showed that some of the populations were mixtures of phenol-positive and negative plants. Such mixed populations generally had large (G)1/2 value and negative correlations between spikelet length and width.

The oldest rice grains excavated from remains in Zhejiang Province, China (about 7,000 BP) show considerably large (G)1/2 and negative r/g/ values (Table 2). Chinese experts presume that the oldest rice populations consisted of many Hsien (Indica) and fewer Keng (Japonica) types (You 1979). When compared with the present result of observation, it may be suggested that the incipient domesticates were in a situation not differentiated into the Indica and Japonica types completely.

Table 2. Length and width of spikelets (mm), their standard deviations, square root of generalized genetic variance (G)1/2, and genetic correlation (r/g/) of prehistoric rice grains excavated

Morishima, H. and H. I. Oka, 1970. A survey of genetic variations in the populations of wild- rice species and their cultivated relatives. Jpn. J. Genet. 45: 371-385.

Morishima, H., Y. Sano and H. I. Oka, 1980. Observations on wild and cultivated rices and companion weeds in the hilly areas of Nepal, India and Thailand: Report of study-tour in tropical Asia, 1979. Rep. Natl. Inst. Genetics, Misima, 97 pp.

Oka, H. I., 1988. Origin of Cultivated Rice. Elsevier/Jpn. Sci. Soc. Press, Amsterdam/Tokyo. 254pp.

You, X. L., 1979. Origin, differentiation and dissemination of cultivated rice in China, as suggested by rice grains excavated from Hemudu. Acta Agron. Sinica 5(3): 1-12. (in Chinese)