34

But this was not considered as a proper method to determine percentage of functional pollens (Li et al. 1993). Our purpose is to develop a reliable method to determine pollen fertility and evaluate male and female gamete fertility more accurately.

Five F1s of Indica/Japonica crosses, i.e, 86-70/Hanjiou, Suweon//Maggel36/Binxu, TKM9/L183, Balila/IR36, Fubao/Suweon, one DH line 96-5 with low panicle fertility from F1 of Jingxil7/Zhaiyeqing, and one variety CPSL017 (control) were used. Their mature pollen fertility was examined by two methods: I2-KI staining and hae-matoxylin-staining. Seed-setting rate was also examined.

In the previous studies, after pollen grains were stained by 1 % I^2-KI solution, round, filled, and deep-color stained pollen was counted as fertile pollen. For haematoxylin staining (Zhao et al. 1994), we made some improvements: At first the starch in pollen grain was melted by HCl or other reagents. Then pollen grains were stained by haematoxylin solution. The pollen grain with one vegetative nucleus and two sperm (male) nuclei was considered as fertile (Fig. 1, left). Three other types of sterile pollens: 1) deformed pollen; 2) round pollen without any stained content; and 3) round pollen with only one vegetative-nucleus (Fig. 1, right).

The seed-setting rate and pollen fertility tested by I2-KI-stain was significantly different as shown in Table 1, though correlation coefficient was 0.760, reaching significant at 5% level. The seed-setting rate and pollen fertility tested by haematoxylin staining were not so different, and their correlation coefficient R was 0.984. The average difference between seed-setting rate and pollen fertility tested by haematoxylin staining was only 4.9%. While it was 28.2% between seed-setting rate and pollen fertility tested by I2-KI-stain (Table 1).

On the other hand we pollinated all 5 f1s with normal fertile pollens in order to provide enough pollens. But we could not find a significant increase in seed-setting rate. This suggests that the low seed-setting rate in F1s was not caused by their own low pollen fertility, but possibly caused by female gamete sterility. Here is the brief discussion: 1) Using I2-KI staining, only the content of starch and

its glutinous/nonglutinous character can be observed. It can be used for identifying pollen fertility in some kinds of male-sterile lines whose male sterility occurs at early stage (before 2 nuclei stage) causing little materials accumulated in pollen (Rice Heterotic Group 1978). In F1s between Indica and Japonica, their pollens possibly deteriorate at late stage (2 or 3 nuclei stage) and some starch on other genetic materials were accumulated, so we can't identify by I2-KI-staining whether or not they are functional pollen. 2) By haematoxylin staining we can see directly genetic materials in mature pollen grains. We think it may be an accurate method to identify sterile and fertile pollens of Indica-Japonica f1s. 3) Judging from the above results, we think that the seed-setting rate was determinated mainly by female gamete fertility and that the partial sterility appeared not only in male gamete but also in female gamete possibly with the similar degrees.

Pollen fertility (%)
Cross combination	Seed- setting (%)	I2-KI-stain	haematoxylin-stain
Balila/IR36	49.9	78.3	49.7
Fubao/Suweon	63.8	91.2	67.9
86-70/Hanjiu	3.6	6.9	5.0
TKM9/L183	32.0	82.0	39.8
Suweon// Manggel36/Binxu	33.6	84.9	48.4
96-5	10.5	49.4	19.0
CPSL017	97.8	92.3	92.5
Suweon// Manggel36/Binxu	41.2	69.3	46.0