Crop Science 40:640-646 (2000)
© 2000 Crop Science Society of America
CROP BREEDING, GENETICS & CYTOLOGY
Identification of Hybrid Sterility Gene Loci in Two Cytoplasmic Male Sterile Lines in Rice
P.S. Devananda,
M. Rangaswamyb and
H. Ikehashia
a Graduate School of Agric., Kyoto Univ., Kyoto 606-01, Japan
b Centre for Plant Breeding and Genetics, Tamil Nadu Agric. Univ., Coimbatore 641 003, India
ikehashi{at}kaias.kyoto-u.ac.jp
 |
ABSTRACT
|
|---|
In hybrid rice (Oryza sativa L.) breeding, spikelet sterility is a barrier to attaining enhanced heterosis in crosses between different groups or subspecies of rice. Our objective was to identify the hybrid sterility gene loci (HSGLi) in two indica cytoplasmic male sterile A lines (`IR58025A' and `IR62829A') and their corresponding maintainer B lines (`IR58025B' and `IR62829B'). To detect the HSGL, three testers, `IR36' (indica cultivar), `Dular' (indica wide compatibility variety [WCV]) and `Nekken 2' (japonica WCV) were crossed to the two groups in the form of IR36/WCV//B and A//IR36/WCV. The level of spikelet fertility in the F1 progeny of three-way crosses was differentiated by morphological and isozyme marker genes at the known HSGLi, S5, S7, S8, and S15 on chromosomes 6, 7, 6, and 12, respectively, and by two new HSGLi, on chromosomes 3 and 11. The levels of pollen fertility were differentiated by specific marker alleles among segregating genotypes in the cytoplasmic male sterile (cms) background. The presence of marker alleles from IR36 was associated with high pollen fertility, while marker alleles from Dular and Nekken 2 were associated with low pollen fertility due to the effect of HSGLi in the cms background. The effect of dominant and recessive restorers from IR36 was not estimated due to lack of appropriate markers. To alleviate problems of hybrid sterility, the level of sterility should be tested in crosses between maintainers and restorers before the former are used for backcrossing to create new cms lines.
Abbreviations: cms, cytoplasmic male sterile • HSGLi, hybrid sterility gene loci • Rfs/rfs, fertility-restoring gene • WA, wild abortive • WCG, wide compatibility gene • WCV, wide compatibility variety
|
INTRODUCTION
|
|---|
RICE IS THE STAPLE FOOD in densely populated Asia. The yield increase from semidwarf pureline cultivars has been stagnant (Food and Agriculture Organization of the United Nations, 1995), while hybrid cultivars have yielded more than the best pureline cultivars by 15 to 20% (Yuan et al., 1994). Hybrid cultivars are cultivated on more than 17 million hectares in China, and extend to other countries like India and Vietnam (Anonymous, 1996).
Spikelet sterility of panicles in hybrids between indica and japonica rice is a barrier in attaining potential levels of heterosis. A "single-locus allelic interaction" model involving three alleles at locus S5 on chromosome 6, that is, S5i (indica), S5j (japonica) and S5n (neutral), has been tested to analyze hybrid sterility (Ikehashi and Araki, 1986). According to the model, the S5i/S5j heterozygote produces semisterile panicles due to partial abortion of female gametes carrying S5j. The donor of S5n is called a WCV, as both heterozygotes of S5n/S5i and S5n/S5j do not show female gamete abortion. Of about one thousand indica and japonica cultivars from China, less than 3% showed hybrid sterility in their crosses to such WCVs as Dular and Nekken 2 (Wan and Ikehashi, 1995). S5n has been incorporated into cms lines to overcome hybrid sterility (Ikehashi, 1991; Yuan, 1992; Zou et al., 1992), but some hybrids have shown semisterility despite the presence of the S5n allele (Ikehashi and Araki, 1987). In further analyses, HSGLi, that is S16 on chromosome 3, S9 on chromosome 4, S8 on chromosome 1, S7 on chromosome 7, and S15 on chromosome 12, were identified in hybrids between different varietal groups (Ikehashi and Wan, 1996).
Though HSGLi can be identified for both male and female gametes, the reported HSGLi are due to abortion of female gametes only. Identification of HSGLi for male gametes is hindered by the technical difficulty of evaluating pollen fertility due to its high variability in hybrids. Female gamete abortion is easily detected as empty spikelets, and viable female gametes can be fertilized by viable pollen even if ratios of viable pollen were decreased in hybrids.
About 95% of the existing cms lines were developed on the basis of the wild abortive (WA) cytoplasm, which was incorporated from the cytoplasm of O. sativa f. spontanea in China (Yuan and Virmani, 1988). In hybrid rice breeding, a large number of promising breeding lines are pollinated to cms lines. In such test crosses, parents that result in progeny showing nearly complete pollen sterility are classified as potential maintainers and used for further backcrossing to cms lines, while those resulting in progeny showing a high level of pollen fertility are classified as potential restorers and tested further for combining ability. With such screening, hybrid sterility genes of potential maintainers are not detectable due to the low pollen fertility resulting from lack of Rf under the cms background. Therefore, many new cms lines, developed by incorporating their nuclear genes into the standard cms background, may show hybrid sterility in crosses to potential restorers. Such hybrid sterility may narrow the range of combinations of cms lines and potential restorers. To measure the extent of such hybrid sterility, it is necessary to examine the HSGLi in major cms and maintainer lines.
The objective of our study was to identify HSGLi in two major pairs of cms and maintainer lines. As mentioned above, HSGLi are easily detectable for female gametes and are so far identified with a limited number of markers. We attempted to clarify the effects of HSGLi on pollen fertility under cms backgrounds. Despite the use of limited markers, a series of HSGLi were identified on chromosomes 6, 7, and 12, together with new HSGLi on chromosomes 3 and 11. Improved methods for cms line development are proposed.
|
Materials and methods
|
|---|
Plant Materials
Hybrid sterility gene loci were evaluated for both pollen and spikelet fertility in the major cms lines, IR58025A and IR62829A, and for spikelet fertility in their respective maintainer lines, IR58025B and IR62829B. Male sterility of the cms lines is controlled by recessive alleles (rfs) of the fertility restoring genes under the WA cytoplasm. The maintainer lines belong to the indica subspecies and are the male-fertile isonuclear lines with normal cytoplasm for the cms lines. The two cms and maintainer lines are most widely used for commercial hybrid seed production in tropical countries (Anonymous, 1996) and are used further to develop new cms lines. Three testers, IR36, Dular, and Nekken 2, were used as an indica cultivar, an indica WCV, and an japonica WCV, respectively (Table 1)
.
Fourteen single crosses and eight three-way crosses were made at Kyoto Univ. and Tamil Nadu Agric. Univ. from 1996 to 1997. The single crosses were made in the form of (A or B)/(WCVs or indica tester). The three-way crosses A(P1)//IR36(P2)/WCV(P3) and IR36(P2)/WCV(P3)//B(P1) were made after confirming that the hybrid, P1/P2 or P2/P1 was fertile and the hybrid P1/P3 or P3/P1 was semisterile. The F1 plants were transplanted from the nursery 
30 d after seeding at Kyoto Univ. in 1998. Single plants were spaced 10 cm apart and the rows were 30 cm apart.
Determination of Pollen and Spikelet Fertility
To determine the pollen fertility of the tested plants, 10 to 20 spikelets were collected from each plant at flowering before anthesis, and fixed in 70% ethanol. Five anthers were sampled at random from the collected spikelets. Pollen grains were stained in 1% iodine potassium iodide (IKI) solution, and observed under the microscope at 40x magnification. From a total of 10 fields, 100 to 200 random pollen grains were scanned on each slide and classified as sterile or fertile based on their staining behavior (Chaudhary et al., 1981). All round and dark brown stained pollen were scored as normal fertile and irregular-shaped, yellowish or unstained pollen grains were scored as sterile. To determine spikelet fertility, three to four panicles per plant were sampled at maturity 30 d after heading and the number of fertile (filled grain) and sterile (unfilled or chaffy grain) spikelets on the upper half of each panicle were counted. Using both pollen and spikelet fertility, each plant was classified as fertile (>70%) or sterile (<70%) (Wan and Ikehashi, 1995).
Marker Genes for Hybrid Sterility Loci
Isozyme analysis was conducted for the F1 plants resulting from eight three-way crosses, in which the total number of progeny per cross ranged from 60 to 90. The following enzymes were examined: acid phosphatase (E.C. 3. 1. 3. 2.), catalase (E.C. 1. 11. 1. 6.), esterase (E.C. 3. 1. 1.-), malate dehydrogenase (E.C. 1. 1. 1. 37), phosphogluconate dehydrogenase (E.C. 1. 1. 1. 44), phosphoglucose isomerase (E.C. 5. 3. 1. 9), peroxidase (E.C. 1. 11. 1. 7), and shikimate dehydrogenase (E.C. 1. 1. 1. 25.), respectively. Using 1 M sucrose and 0.05 M 2-mer-captoethanol in 0.2 M Tris-HCl buffer adjusted to pH 8.5, enzymes were extracted from lateral buds of 45-d-old seedlings during tillering. Imbibed filter paper wicks were immersed in the extract and then inserted into a horizontal starch gel for electrophoresis at 4°C. Two morphological marker genes, apiculus color (C) on chromosome 6 and red pericarp (Rc) on chromosome 7, were visually scored. To detect the HSGLi, at least one marker for each HSGLi was selected according to Ikehashi and Wan (1996) as shown in Table 1. The isozymes were analyzed according to Glaszmann et al. (1988). Gene symbols for the individual loci followed the standard system by Morishima and Glaszmann (1991). For each cultivar, the marker genes and isozyme genotypes are summarized in Table 1.
Analysis of Hybrid Sterility Gene Loci in Three-Way Crosses
According to the single locus allelic interaction model, the progeny of three-way cross, P1//P2/P3 or P2 /P3//P1 should segregate fertile plants from P1/P2 or P2/P1 and semisterile ones from P1/P3 or P3/P1 at a ratio of 1:1. If any genetic marker cosegregates with the semisterility, a HSGL will be indicated by such a marker. When such cosegregating marker genes were detected, a few plants showed semisterility with a marker allele of high fertility, or some showed normal fertility with a marker allele of low fertility. Such plants were assumed recombinants between the marker genes and the gene for semisterility.
A total of four three-way crosses with maintainer lines were examined to confirm the segregation of spikelet fertility. In the three-way crosses that included the cms line as a female parent, pollen fertility was determined in addition to spikelet fertility as described above.
|
Results
|
|---|
Hybrid Sterility in Single Crosses
The F1 of IR58025A/IR36 and IR58025B/IR36 showed normal pollen and spikelet fertility, indicating pollen fertility restoration with the cms (Table 2)
. The F1 of IR58025A/Dular showed pollen sterility, while IR58025B/Dular showed normal pollen fertility, indicating incomplete pollen fertility restoration. The F1 of IR58025B/Dular showed normal spikelet fertility, while IR58025A/Dular showed clear spikelet sterility. Lower spikelet fertility in the latter cross was due to lowered pollen fertility as the nuclear genes for spikelet fertility are not different in the cms and maintainer lines. The F1 of IR58025A/Nekken 2 and IR58025B/Nekken 2 showed nearly complete pollen sterility with cms and low pollen fertility without cms, respectively. Both F1 hybrids showed a low level of spikelet fertility.
The F1 of IR62829A/IR36 and IR62829B/IR36 showed normal pollen and spikelet fertility, indicating pollen fertility restoration. The F1 of IR62829A/Dular and IR62829B/Dular showed similar levels of low pollen and spikelet fertility, indicating no effect of cms. The low fertility levels were probably due to the presence of hybrid sterility genes. The level of pollen fertility in the F1 of IR62829A/Nekken 2 was lower than that of IR62829B/Nekken 2, indicating the effect of cms. Spikelet fertility in IR62829B/Nekken 2 was low due to hybrid sterility, while that in IR62829A/Nekken 2 was much lower with cms influence. The F1 of IR36/Dular showed normal pollen and spikelet fertility, while IR36/Nekken 2 showed normal spikelet fertility but low pollen fertility.
Bagging of panicles was not conducted, leaving open the possibility that female plants were pollinated by neighboring plants. But any effect of outcrossing may have been negligible because the outcrossing rate is very low in rice (Virmani, 1994).
Isozyme Marker Associations for Hybrid Sterility
When markers for one HSGL showed similar results, only one of them was reported in Tables 3 and 4
. In the F1 of IR36/Dular//IR58025B, most plants showed normal spikelet fertility (data not shown) as expected by the F1 of IR58025B/IR36 and IR58025B/Dular, which showed normal spikelet fertility (Table 2). Thus, further genetic analysis by isozyme markers was not conducted.
View this table:
[in this window]
[in a new window]
|
Table 3 Distribution for spikelet fertility classified by marker genotypes for F1 hybrids from the general cross IR36/WCV//Maintainer. Numbers underlined are assumed recombinants
|
|
View this table:
[in this window]
[in a new window]
|
Table 4 Distribution for pollen fertility and mean spikelet fertility classified by marker genotypes for F1 hybrids from the general cross, cms//IR36/WCV
|
|
In the F1 of IR36/Nekken 2//IR58025B, the level of spikelet fertility was significantly differentiated by marker genotypes at three loci (i.e., at Est-9, a marker for the S7 locus on chromosome 7; at Cat-1, a marker for S8 on chromosome 6; and at Pgi-1 on chromosome 3 [Table 3]). The results implied that the spikelet sterility in the F1 of IR58025B/Nekken 2 was caused by S7 and S8 among the tested loci and by a new locus near Pgi-1. In the F1 of IR36/Dular//IR 62829B, the level of spikelet fertility was significantly differentiated by marker genotypes at Pox-2, a marker for the S15 locus on chromosome 12, and at Pgd-1 on chromosome 11 (Table 3). Thus, spikelet sterility in the F1 of IR62829B/Dular was caused by S15 and a new locus near Pgd-1. In the F1 of IR36/Nekken 2//IR62829B, the level of spikelet fertility was significantly differentiated by the same marker genotypes as indicated for the F1 of IR36/Nekken 2//IR58025B (Table 3). Spikelet sterility in the F1 of IR62829B/Nekken 2 was caused by S7, S8, and a new locus near Pgi-1.
Pollen and spikelet fertility were determined for three-way crosses made with the A line, IR36 as the restorer, and the WCVs, in the form of A//IR36/WCV. In the F1 of IR58025A//IR36/Dular and IR62829A//IR36/Dular, pollen and spikelet fertility was significantly differentiated by marker genotypes at Pox-2 and Pgd-1 (Table 4). In the F1 of IR58025A//IR36/Nekken 2 and IR62829A//IR36/Nekken 2, pollen and spikelet fertility was significantly differentiated by marker genotypes at C and Est-2 on chromosome 6, and at Est-9, Cat-1, and Pgi-1 (Table 4).
Further, we examined the relationship between pollen and spikelet fertility by selecting at least one isozyme marker for each HSGLi. Low levels of spikelet fertility were associated with low pollen fertility in the hybrids between a cms line and WCVs. The high level of spikelet fertility is due to the pollen fertility by the presence of neutral alleles of IR36 (Fig. 1 and 2)
. In all four three-way crosses, the marker genotypes that contained the allele from IR36 showed high levels of pollen and spikelet fertility and vice versa (Table 4, Fig. 1 and 2). The lowered spikelet fertility was considered to be caused by HSGLi and the low pollen fertility under the cms background. The relative magnitudes of female and male sterility could not be determined in this experiment.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 1 Relationship between pollen fertility and spikelet fertility in the F1 of (A, B) IR58025A//IR36/Dular and (C, D, E, F) IR58025A//IR36/Nekken 2 with regard to isozyme markers. (-) and (+) indicate the absence and presence of IR36 allele, respectively
|
|
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 2 Relationship between pollen fertility and spikelet fertility in the F1 of (A, B) IR62829A//IR36/Dular and (C, D, E, F) IR62829A//IR36/Nekken 2 with regard to isozyme markers. (-) and (+) indicate the absence and presence of IR36 allele, respectively
|
|
|
Discussion
|
|---|
In crosses without the cms background, pollen fertility was not tested. It was considered that three to four fertile pollen grains are practically sufficient to fertilize viable female gametes (Virmani, 1994). HSGLi were identified regardless of pollen fertility level in the past experiments. In this experiment, several known HSGLi were confirmed in the major inbred lines. The spikelet sterility associated with S7 and S8 confirms earlier findings of Wan et al. (1993), who found spikelet sterility in the hybrid between a Korean indica cultivar, Yeong Pung, and Nekken 2 to be independently controlled by S7 and S8. Spikelet sterility in the F1 of IR62829B/Dular was caused by the S15 locus. Wan et al. (1996) also found sterility of the hybrid between IR2061-628-1 and Dular to be controlled by the S15 locus. In addition to the known HSGLi, our results indicate two new HSGLi near Pgi-1 and Pgd-1, which were tentatively named S19(t) and S18(t), respectively.
Pollen fertility was evaluated in three-way crosses with the cms line as the female parent because a high level of variation was expected for this trait. In fact, pollen fertility was differentiated by many marker gene loci in the cms background. In all cases, marker genotypes containing the allele from IR36 showed a high level of pollen fertility. Because the major restorer genes (Rfs) have been reported on chromosomes 1 (Zhang and Huang, 1996), 7 (Bharaj et al., 1995), and 10 (Bharaj et al., 1995; Zhang and Huang, 1996), any effects of Rfs from IR36 in this experiment was considered to be distributed at random into two contrasted marker genotypes, except those at Est-9 on chromosome 7. Therefore, pollen fertility may have been differentiated by a set of HSGLi for pollen sterility under the cms background.
A set of HSGLi for pollen sterility were suggested near markers C and Est-2 on chromosome 6, Cat-1 on chromosome 6, and Pgd-1 on chromosome 11. Such HSGLi for pollen sterility have been reported using isozyme markers Est-9 on chromosome 7 (Lin and Ikehashi, 1993), Sdh-1 on chromosome 12 (Wan et al., 1997), and Pgi-1 (Lu et al., 1998, personal communication). Though pollen and spikelet sterility were reported as independent in previous studies, simultaneous effects of the two were observed in this study. The possibility that functions of some HSGLi might have been modified under the cms background remains to be clarified by using an adequate number of markers such as restriction fragment length polymorphisms (RFLP).
These results, which were obtained with limited isozyme markers (Table 1), are considered to reveal a part of an existing set of HSGLi. Nevertheless, it is suggested that those cultivars which showed a very low level of pollen fertility in crosses to the indica cms lines have been retained as potential maintainers. This explains the basis for our previous finding that the majority of potential maintainers exhibited japonica type alleles, while most of the potential restorers exhibited indica type alleles in terms of isozymes (Devanand et al., 1999). If such potential maintainers are used in backcross breeding programs, the resulting cms lines would show again hybrid sterility when pollinated by indica type restorer lines.
To alleviate problems of hybrid sterility in hybrid rice breeding programs, it is recommended to incorporate a full set of wide compatibility genes into the genetic background of cms or maintainer lines. The S5n allele has been incorporated into cms lines to overcome hybrid sterility (Ikehashi, 1991; Yuan, 1992; Zou et al., 1992). Because the incorporation of wide compatibility genes into the maintainer and then to cms backgrounds may take a long time, at present at least, it seems necessary to test the level of sterility in hybrids between representative restorers and potential maintainers before the latter are used for backcrossing to cms lines.
|
ACKNOWLEDGMENTS
|
|---|
The authors extend their gratitude to Dr. P. Vaidyanathan, Dr. K. Thiyagarajan and Mrs. P. Jayamani (Paddy Breeding Station, Coimbatore, India) for providing part of the genetic materials. This research was supported by a grant from the Ministry of Education, Government of Japan.
Received for publication March 9, 1999.
|
REFERENCES
|
|---|
- Anonymous Annual report of `Hybrid Rice Scheme' (ICAR/UNDP/FAO). Rajendranagar, Hyderabad, India: Directorate of Rice Research, 1996.
- Bharaj T.S., Virmani S.S., Khush G.S. Chromosomal location of fertility restoring genes for `wild abortive' cytoplasmic male sterility using primary trisomics in rice. Euphytica 1995;83:169-173.
- Chaudhary R.C., Virmani S.S., Khush G.S. Patterns of pollen abortion in some cytoplasmic-genetic male sterile lines of rice. Oryza 1981;18:140-142.
- Devanand P.S., Wan J., Rangaswamy M., Ikehashi H. Isozyme divergence between maintainers and restorers in hybrid rice breeding programs in India. Crop Sci. 1999;39:831-835.[Abstract/Free Full Text]
- Food and Agriculture Organization of the United Nations. 1995. Report of the International Rice Commission. p. 1–33. Food and Agriculture Organization of the United Nations, Rome, Italy.
- Glaszmann J.C., de los Reyes B.G., Khush G.S. Electrophoretic variation of isozymes in plumules of rice (Oryza sativa L.) — A key to the identification of 76 alleles at 24 loci. IRRI Research Paper Series no. 134. Manila, Philippines: IRRI, 1988.
- Ikehashi H. Genetics of hybrid sterility in wide hybridization in rice. In: Bajaj Y.P.S., ed. Biotechnology in agriculture and forestry. Berlin, Heidelberg: Springer-Verlag, 1991:113-127.
- Ikehashi, H., and H. Araki. 1986. Genetics of F 1 sterility in remote crosses of rice. p. 119–130. In Rice genetics. IRRI, Manila, Philippines.
- Ikehashi, H., and H. Araki. 1987. Screening and genetic analysis of wide-compatibility in F1 hybrids of distant crosses in rice (Oryza sativa L.). p. 1–79. Technical Bulletin Tropical Agriculture Research Center, no. 23. Tropical Agriculture Research Center, Ministry of Agriculture, Forestry and Fisheries, Tsukuba, Ibaraki, Japan.
- Ikehashi H., Wan J. Differentiation of alleles at seven loci for hybrid sterility in cultivated rice (Oryza sativa L.). In: Khush G.S., ed. Rice genetics III. Manila, Philippines: IRRI, 1996:404-408.
- Lin S.Y., Ikehashi H. A gamete abortion locus detected by segregation distortion of isozyme locus Est-9 in wide crosses of rice (Oryza sativa L.). Euphytica 1993;67:35-40.
- Morishima H., Glaszmann J.C. Current status of isozyme gene symbols. Rice Genet. Newsl. 1991;7:50-75.
- Virmani S.S. Outcrossing mechanisms and hybrid seed production practices in rice. In: Virmani S.S., ed. Heterosis and hybrid rice breeding. Berlin, Heidelberg, Ger-many: Springer-Verlag, 1994:79-96.
- Wan J., Ikehashi H. Identification of a new locus S-16 causing hybrid sterility in native rice varieties (Oryza sativa L.) from Tai-hu Lake region and Yunnan province, China. Breed. Sci. 1995;45:461-470.
- Wan J., Ikehashi H., Horisue N. Analysis of genetic loci for pollen sterility in remote crosses of cultivated rice (Oryza sativa L.). (In Japanese). Breed. Sci. 1997;47(supple. 2):188.
- Wan J., Yanagihara S., Kato H., Ikehashi H. Multiple alleles at a new locus causing hybrid sterility between a Korean indica variety and a javanica variety in rice (Oryza sativa L.). Jpn J. Breed. 1993;43:507-516.
- Wan J., Yamaguchi Y., Kato H., Ikehashi H. Two new loci for hybrid sterility in cultivated rice (Oryza sativa L.). Theor. Appl. Genet. 1996;92:183-190.[ISI]
- Yuan L.P. The strategy of the development of hybrid rice breeding. In: Yuan L.P., ed. Current status of two line hybrid rice research. (In Chinese with English summary). Beijing: Agricultural Publishing Ltd, 1992:1-5.
- Yuan, L.P., and S.S. Virmani. 1988. Status of hybrid rice research and development. p. 7–24. In W.H. Smith (ed.) Proc. Int. Symp. on Hybrid Rice. Changsha, Hunan, China. 6–10 Oct. 1986. IRRI, Manila, Philippines.
- Yuan L.P., Yang Z.Y., Yang J.B. Hybrid rice in China. In: Virmani S.S., ed. Hybrid rice technology: New developments and future prospects. Manila, Philippines: IRRI, 1994:143-147.
- Zhang, Q., and N. Huang. 1996. Maping and molecular marker-based genetic analysis of fertility related genes for efficient hybrid rice breeding. p. 13. In K. Muralidharan (ed.) Proc. Int. Symp. Hybrid Rice, 3rd. Hyderabad, India. 14–16 Nov. 1996. IRRI, Manila, Philippines.
- Zou J., Nie Y., Pan Q., Fu C. The tentative utilization of wide-compatibility strain 02428 in indica/japonica hybrid rice. In: Yuan L.P., ed. Current status of two-line hybrid rice research. (In Chinese with English summary). Bejing: Agricultural Publishing Ltd, 1992:333-339.
TOP
ABSTRACT
INTRODUCTION
