Tropical Agricultural Research Vol. 9 1997,35-48 Genetic Analysis of New Plant Type in Rice {Oryza sativa L.) with a View to Exploit Yield Potential A.P. Bentota, D. Senadhira1 and M.J. Lawrence2 Regional Agriculture Research and Development Centre Bombuwela Kalutara, Sri Lanka ABSTRACT. The genetical architectures of twelve quantitative characters of interest have been investigated using the basic generations and F}families from two crosses, chosen at random from IRRJ's New Plant Type (NPT) programme (japonica x iavanica) with the objective of increasing the genetic yield potential to 13-15 t/ha of direct seeded irrigated crop in lowland tropics. The experimental plants were raised in a completely randomized experiment for each cross. The parents of both crosses differed significantly for most of the characters understudy, except for proportion of filled spikelets (PFS) and grain yield (GY) in Cross 1, and grain weight (GW) in Cross 2. Only three of these characters, panicle number (PN), GY and dry matter (DM), in Cross 1 displayed heterosis. The genes controlling the characters in Cross 1 displayed mainly additive or additive and dominance effects, except for two related characters, tiller number (TN), and PN, which displayed 'j' type epistasis. In contrast, those controlling all but two of the characters in the second cross displayed epistasis, which for most was of the duplicate type. The results show that characters in both crosses were heritable except for harvest index (HI) in Cross 1 and heritabilities varied from low to high Genetic correlations between characters in Cross 1 were very different from those in Cross 2, which suggest that the chief cause of these correlations is the linkage disequilibrium of linked genes, rather than pleiotropy. An assessment of the potential of these crosses showed that, while it should be relatively easy to achieve the NPT targets for six characters in both crosses, this is unlikely for PFS, GYandHl with one cycle of inbreeding. Plant Breeding, Genetics and Biochemistry Division, International Rice Research Institute, P.O. Box 933,1099 Manila, Philippines. Wolfson Laboratory for Plant Molecular Biology, School of Biological Sciences, University of Birmingham, BIS 2TT, United Kingdom. Bentota, Senadhira & Lawrence INTRODUCTION » \ • The belief among rice breeders that tropical cultivars of the indica type have reached a yield plateau of 10 t/ha, has caused the breeders at International Rice Research Institute (IRRI) to switch their programme to the breeding of a radically different ideotype, known as New Plant Type (NPT) for direct seeded, irrigated rice crop which would yield 13-15 t/ha, in lowland tropics. The chief characteristics of the NPT, were determined considering several different perspectives (Vergara,1988; Janoria, 1989; Dingkuhn et al., 1991), such that the plant should produce 3-4 tillers, all of which produce panicles containing 200-250 grains on very sturdy stems, should bear dark green leaves, should be about 90 cm in height and 100-130 days growth duration and should have multiple pest and disease resistance and a harvest index of 0.6. Some of these characteristics lie well outside the normal range of the variation found in indica cultivars which in any case, are believed to have insufficient genetic variation for yield. Therefore the NPT breeding programme has been initiated from the crosses between japonica and javanica (according to Glaszmann (1987), refers to as tropical japonica) varieties (IRRI, 1989). However knowledge on genetical architecture of the characters of interest in the NPT programme is not well known. This paper reports a genetic analysis of two of the japonica * javanica crosses chosen at random from the IRRI's NPT programme. MATERIALS AND METHODS Two of the crosses chosen for the investigation were the Jinmibyeo x Gaok (Cross 1) and Sangnambatbyeo * Kemandi Pance (Cross 2). Jinmibyeo and Sangnambatbyeo are the two Korean varieties that belong to japonica type. These female varieties were chosen because of their short stature, high yielding ability under temperate conditions, dense spikelets per panicle and cross- compatibility with javanica types. Male parents Gaok and Kemandi Pance are two traditional Bulu (Javanica) varieties from Indonesia that were chosen because of their large panicles, low number of tillers, sturdy stems and dark green and thick leaves. Three groups of families namely basic generations (P,, P* F„ F2, BC,, BCj), F3 families and back crosses to F2 individuals (L,= F2 * P,, L2 = F2 * P2, Lj = F2 * F,) of the same age were raised at IRRI, Phillipines in the dry season of 1994, in two adjacent blocks in a completely randomized design, one block per cross. In each cross, 20 plants were raised from each of the non- segregating families, 56 from F2 population, 50 from each of the first back 36 Genetic Analysis of New Plant Type in Rice (Oryza sativa L.) crosses and 30 F} families with 8 plants from each family. Eight plants were also raised in each of 35 * 3F2 back cross families for Cross 1 and 32 * 3F2 such families for Cross 2. Two recently improved IRRI varieties, IR72 and IR74 were also included in the two experiments. This paper reports the genetic analysis based on basic generations and F3 families. All of the cross seeds were produced using modified vacuum (Herera and Coffman, 1974) emasculation method followed by hand pollination the following day. The F, families, IR72 and IR74 in the experiment were from naturally self pollinated seeds. All the experimental seeds were pre-germinated in petridishes and were sown in 5 * 5 * 5 cm small plastic pots filled with sterilized soil, one seed being sown in each pot. The pots of each experiment, one for each cross were arranged in a pair of completely randomized block designs (CRBD) in the green house. Experiment 1 (Cross 1) consisted of 1336 individually randomized plants and experiment 2 (Cross 2) with 1264 plants. After three weeks, plants were transplanted in the field in the same CRBD at a spacing of 30 * 20 cm at IRRI, Philippines in the dry season of 1994. The experiments were managed in the usual way except for the special efforts made to keep them free from pests and diseases as much as possible. Each of the plants included in the experiment 1 and experiment 2 were scored for the following twelve characters: Days to heading (DH) - Number of days from the date of sowing to the first panicle emerged from the flag leaf sheath; Days to maturity (DTM) - Number of days from sowing to the point at which panicles of the plant turn brown in colour; Tiller number (TN) - Number of tillers on the date of heading; Culm length (CL) - Length (cm) from the base of the plant to the neck node of the main panicle; Panicle number (PN) - Number of panicles at maturity in each plant; Panicle length (PL) - Average length (cm) of the panicles in each plant; Spikelet number (SN) - Average number of spikelets, in each plant; Proportion of filled spikelets (PFS); 37 Bentota, Senadhira & Lawrence Grain weight (G W) - Weight of 100 seed (g) at 14% moisture; Grain yield (GY) - Total weight (g) of the grain produced by a plant at 14% moisture; Dry matter (DM) - Weight (g) of the above ground parts (straw and grains) after oven drying at 70°C for 72 hours; Harvest index (HI) - The ratio between grain yield at 0% moisture level to the total dry matter of the plant. RESULTS AND DISCUSSION Mean of the basic and F 3 generation The parents differed significantly for the majority of characters scored in both crosses except for proportion of filled spikelets (PFS), and grain yield (GY) in Cross 1 and for hundred grain weight (GW) in Cross 2 (Table 1). Tropical japonica parent was chosen because of low number of tillers, larger panicles and higher number of spikelets per panicle. As expected, Gaok, the tropical japonica variety was the lower scoring parent in the first pair of characters and the highest scoring parent for the latter pair in Cross 1. The Kemandy Pance, tropical japonica variety of the second cross was also chosen for the same but it did not show expected performances of lower number of tillers and panicles though it has larger panicles with more number of spikelets per panicle. Some of these differences between the parents, such as those for spikelet number (SN) in both crosses and, in particular for GY and dry matter (DM) and harvest index (HI) in Cross 2, are much larger than those commonly found in indica crosses (Perera et al., 1997), which may be due to the fact that parents are from very different provenance. F, means of the characters PN, GY and DM in Cross 1 are significantly higher than the better parent i.e., japonica parent, indicating that heterosis is present for PN, G Y and DM characters of Cross 1. There is very little evidence that heterosis is present in Cross 2. As the type of heterosis present indicates that the genes controlling the characters are dispersed between their parents of the cross, it is a good indicator for breeders to judge the potential of the cross in advance. 38 Genetic Analysis of New Plant Type in Rice (Oryza saliva L.) Table 1. M e a n performance of the families of the basic and the F } generations of each cross. Character P, P 2 F, F, BC, BC, Cross 1 DH 78.1- 63.8 69.6 70.3 70.5 74.3 67.0 DTM 110.42 94.7 103.1 105.1 103.5 109.3 101.6 TN 8.6' 4.9 9.8 8.6 7.3 8.6 8.5 CL 115.71 72.2 95.6 95.9 91.9 108.1 84.6 PN 9.0' 5.2 10.9 9.2 8.1 9.4 9.8 PL 3I.4 J 20.0 26.1 25.6 25.2 28.1 22.8 SN 211.4* 120.0 171.4 169.5 151.6 184.0 154.5 PFS 84.9' 80.9 89.9 85.4 85.4 87.8 81.9 GW 2.93 2.1 2.6 2.6 2.6 2.8 2.4 GY 22.6' 19.5 41.7 30.2 25.2 35.8 29.8 DM 47.6 : 30.9 66.2 54.6 44.3 66.3 47.1 HI 56.3' 41.8 54.5 Cross 2 48.3 51.2 54.9 47.3 DH 95.24 65.4 71.7 65.4 77.0 80.9 65.4 DTM 132.9* 94.3 97.2 104.8 113.0 117.2 97.9 TN 10.6* 5.3 7.9 8.3 8.9 10.2 7.1 CL 101.44 81.6 94.0 100.9 98.9 112.3 96.3 PN 12.4* 5.3 7.9 8.4 9.3 12.2 7.4 PL 26.74 18.7 21.8 23.0 24.4 27.3 21.0 SN 173.0* 116.5 146.3 116.3 170.5 179.2 124.2 PFS 82.8' 71.9 85.1 59.7 72.8 71.2 42.4 GW 2.41' 2.4 2.4 2.4 2.4 2.5 2.6 GY 35.1* 11.6 24.2 16.3 26.0 22.4 15.8 DM 100.8* 19.5 40.7 38.8 56.9 78.0 32.3 HI 55.3' 29.9 53.2 38.0 43.6 44.1 24.5 Superscript of the P indicate the parental variety involved where 1 = Jinmibyeo, 2 = Gaok, 3 - Sangnambatbyeo and 4 = Kemandi Pance. 39 Bentota, Senadhira & Lawrence The component of basic generation means The component of means were separated by weighted least squares procedure (Mather and Jinks, 1982). The satisfactory models (Table 2) in which all components were significantly different from zero and the x2 testing, the goodness of fit between the observed and expected values of the generation means, which were not significant could be found for all characters in Cross 1. Simple additive model (m and [d]) or additive and dominance model (m, [d] and [h]) was sufficient for all characters except for related characters tiller number (TN) and panicle number (PN) for which a fj] component (additive * dominance non-allelic digenic interaction) was also required. The mean component [h] was greater than [d] for TN, PN, proportion of filled spikelets (PFS), grain yield (GY) and dry matter (DM), suggesting that the genes controlling these characters were dispersed between parents. In the Cross 2, the inadequacy of a simple m and [d] or m, [d] and [h] model for most of the characters, except for TN and days to maturity (DTM), indicates the presence of epistasis which, for six of them, appears to be of the duplicate type as indicated by the opposite signs of [h] and [I] (Table 2). Average panicle length (PL) required a six parameter model, hence, test of adequacy of the model was not possible. Furthermore, it was initially not possible to find a satisfactory model for either spikelet number (SN) or DM. The inspection of Table 1 reveals that there were marked differences between generations for PFS. 40 When proportion of filled spikelets (PFS) in generations was considered, there is a drastic drop in filled grains particularly in F 2 and B 2 in Cross 2. Usually indica cultivars consist of spikelets that are 80% filled and the objective of the NPT progamme is to produce plants consisting of spikelets that have 100% filled grains. This criteria will have to be considered seriously when advancing crosses further in the breeding programme. Though we expect the F 2 mean to be greater than F 3 mean, according to the usual genetical expectations (Mather and Jinks, 1982) in the absence of negative heterosis, which is marginally (though not. significantly) the case with grain weight (GW) in Cross 2, inspection of Table I reveals that days to heading (DH), GW, and PFS in Cross 1 and most of the characters except culm length (CL) and panicle number (PN) in Cross 2 had greater F 3 mean than F2 mean. This kind of effect may be due to the higher seed quality of F 3 produced from undipped spikelets when compared with the rest of the seeds. Genetic Analysis of New Plant Type in Rice (Oryza saliva L.) Table 2. Components of mean by weighted least squares procedure. Character m [I] 0] [1] M Cross I D M 70.5 7.0 D T M 102.8 8.0 4.0 T N 6.82 1.86 3.29 -1.70 C L 95.8 22.7 PN 7.21 1.91 4.08 -1.99 PL 25.6 5.6 S N 168.6 40.8 PFS 81.5 3.9 8.3 G W 2.56 0.39 G Y 21.2 2.2 20.4 D M 40.3 10.7 27.3 H I 48.0 7.4 5.5 Cross 2 D M 50.1 15.1 39.6 30.2 -18.0 D T M 113.8 19.3 -6.0 T N 8.34 2.97 C L 91.5 9.9 44.7 5.7 -42.2 PN 12.06 3.75 -4.06 -3.03 PL 18.2 4.0 15.6 4.5 2.2 -12.1 S N 132.9 19.1 129.9 115.7 -1.3 PFS 77.3 5.40 -86.3 22.8 94.1 G W 1.65 2.15 0.74 -1.43 G Y 8.4 10.6 16.5 14.4 D M 55.5 34.7 35.4 44.4 -0.5 HI 42.6 12.7 -38.9 7.0 49.4 41 Bentota, Senadhira & Lawrence To overcome this problem, the generation mean was regressed onto PFS and a model was found with an additional parameter, e, which can be regarded as a dummy variable for SN and DM- This had the desired effect with both SN and DM, though the fl testing for the adequacy of the model with respect to the latter was still just significant. It is surprising that this has not been the case with the other related characters, such as GY and HI. The phenomenon that [h] > [d] for days to heading (DH), CL, PN, panicle length (PL), SN, PFS, GW, GY, DM and HI suggest that the genes controlling these characters are dispersed between parents, which suggest that the recombinant inbred lines that can be extracted from these crosses should show transgressive variation. Table 3. Analysis of variance of F 3 families. Character MSb MSw MSb MSw Cross 1 Cross 2 df 29 197-201 29 206-209 DH 51.0*** 13.3 343.3*** 79.0 DTM 106.1** 50.7 735.3*** 134.3 TN 8.4* 5.0 28.3*** 7 2 CL 469*** n o 758*** 123 PN 16.9*** 8.6 38.5*** 10.7 PL 29.9*** 4.4 29.9*** 5.6 SN 4250*** 759 6123*** 1490 PFS 240*** 136 1000*** • 204 GW 0.43*** 0.15 0.14*** 0.05 GY 154ns 109 359*** . 133 DM 606** 328 3145*** 645 HI 183ns 137 846*** 149 ns denotes non significance ** denotes significance at 0.1% * denotes significance at 1% *** denotes significance at 0.01% 42 Genetic Analysis of New Plant Type in Rice (Oryza saliva L.) Analysis of F3 families One way analysis of variance using data from F3 families for both crosses shows that there are significant differences between families for all the characters except grain yield (GY) and harvest index in (HI) Cross 1 (Table 3). It is clear, therefore, that apart from these two, all other characters are heritable. The failure to detect heritable variation for HI for this cross, even though the parents differed significantly for the character, could be due to the fact that these values are derived as the ratio of two primary characters (HI = GY/DM x 100), which could be subjected to a high sampling variance. The estimates of heritabilities of these characters varied from low to high (Table 4). Table 4. Narrow sense heritabilities. Character Cross 1 Cross 2 DH 0.40 0.47 DTM 0.11 0.60 TN 0.08 0.27 CL 0.26 0.40 PN 0.18 0.20 PL 0.65 0.46 SN 0.40 0.50 PFS 0.07 0.15 GW 0.36 0.14 GY 0.05 0.11 DM 0.12 0.77 HI 0.04 0.39 43 Bentota, Senadhira & Lawrence Potential of the crosses The proportions of superior recombinant inbred lines that can be extracted with the desired NPT target or the desired interval for each cross was predicted according to Jinks and Pooni (1976) and is tabulated in Table S. The experiment was conducted in 30 * 20 cm spacing, under transplanted condition, a target of 8-10 productive tillers (for TN and PN) has been used as a reasonable compromise for the original target of 3-4 tillers. At this density, for a crop to achieve a yield of 13 t/ha, each plant would have to produce 78 g of grain. The standards for PL, GW and DM were not specified for NPT; these have been taken, therefore, as equivalent to the score of the best parent (>P|). Table 5. Percentage of predicted recombinant inbred lines. Character Target in D P% m D P% Cross 1 Cross 2 DH 70-80 70.5 9.96 56 77 66.4 45 DTM 100-130 103.5 14.65 82 113 151.0 77 TN 8-10 7.26 0.90 22 8.85 5.32 34 CL 90-100 91.9 95.12 37 98.9 159.8 30 PN 8-10 8.10 2.26 42 9.27 7.11 29 PL i 31.4 25.5 6.88 1 24.4 6.21 17 SN 200-2S0 151.6 942 6 170.5 1182 9 PFS 1 85.4 28 0 72.8 203.1 3 GW i 2.89 2.56 0.076 18 2.39 0.024 45 GY i 78 25.2 13.4 0 26 57.1 0 DM i 47.6 44.3 75.0 49 56.9 633.8 4 HI i 0.6 51.2 43.6 176.7 11 44 Genetic Analysis of New Plant Type in Rice (Otyia saliva L.) Bentota, Senadhira & Lawrence The evidence presented by Perera et al. (1992) and Fahim (1995) in rice suggested that the actual performances agreed best with the F 3 predictions which are more relevant to plant breeders. Therefore the predictions (Table 5) have been made for each character in each cross by using estimates of the mid parent (m) and additive genetic variance (D) from F3 families. These results indicate that it could be fairly easy to extract recombinant inbred lines from this cross for which the performance achieved the NPT targets for days to heading (DH), days to maturity (DTM), tiller number (TN), culm length (CL), panicle number (PN) and spikelet number (SN) but that it would be more difficult to adopt the same for proportion of filled spikelets (PFS) and nearly impossible, with one cycle of inbreeding, to achieve the targets for grain yield (GY) and harvest index (HI). Several hundred lines would have to be raised in order to find crosses for which the performance exceed the better parent for panicle length (PL). Since there is no evidence that HI is heritable in Cross 1, it appears impossible to achieve the desired target for this character. The results in Cross 2 indicate that it should be relatively easy to extract recombinant inbred lines from this cross whose performance achieve NPT targets for DH, DTM, TN, CL, and PN but is difficult with PFS and nearly impossible to achieve the NPT target for GY with one cycle of inbreeding. Therefore, neither of these crosses appear to offer much promise with regards to the yield improvement and proportion of filled spikelets. Apait from special cases such as breeding for some pest and disease resistance, plant breeders are, often concerned with improving their crops for several characters simultaneously. A knowledge of the genetical correlation between characters can, however, indicate, in a quantitative sense, at least, how likely a breeder is to meet the desired target for several characters at a time. The absolute values of genetic correlations greater than 0.5 of which the values which are significantly different from zero are presented in Figure 1 for both the crosses. Some of these correlations such as those between DH, DTM, TN arid PN, for which r > 0.75 in both crosses involve characters which have an obvious pleiotropic relationship. Since, however, the pattern of correlation in Cross 1 is very different from that of Cross 2, the chief cause of genetic correlation in these crosses appear to be due to the linkage disequilibrium of genes that are linked in their inheritance. 46 Genttic Analysis of New Plant Type in Rice (Oryza saliva L.) 47 " C O N C L U S I O N S The genetical architecture o f the N P T characters across the two crosses w e r e dissimilar . T h e characters o f Cross 1 displayed main ly addi t ive or additive and dominance effects while the characters o f Cross 2 display epistasis o f dupl ica te kind. There was significant genetical variat ion exis t ing in all characters in both crosses except G Y and HI in Cross 1. Heritabil i t ies o f these characters varied form low to high. From the investigation o f the potent ial o f c rosses , it is not possible to achieve the N P T target for G Y and P F S in ei ther c ross wi th only one cycle of inbreeding, though it is easy to mee t the N P T target for D H , D T M , TN, C L and PN in both crosses . T h e genet ic correlat ion in these crosses appear to be due to the linkage disequilibrium of genes that are l inked in their inheri tance. T h e mos t str iking feature o f the results obtained from the second cross was that there was a marked loss o f fertility in te rms o f proportion of filled spikelets, in the F 2 and back cross generations, which is also a major problem with inbred lines developed by N P T p rog ramme . A C K N O W L E D G E M E N T S The work reported in this paper was supported by the D O A S L - S A R E C project. The authors acknowledge Drs. M.J. Kearsey, H.S. Pooni and P .S . Vine for their advice on the analysis and interpretation o f the data. R E F E R E N C E S Oingkuhn. M„ Penning de Vrics, F.W.T., De Datta, S.K. and Van Laar, H.H. (1991). Concepts for a new plant type for direct seeded flooded Rice in the Tropics. In: Direct seeded Flooded Rice in the Tropics. IRRI, Los Banos, Philippines, pp. 17-38. Fahim, M. (199S). Quantitative inheritance of agronomic characters and comparison of different breeding methods in rice. Ph.D. thesis, University of Birmingham, U.K. pp. 89. Glaszmann, J.C. (1987). Isozymes and classification of Asian rice varieties. Theor. Appl. Genet. 74:21-30. Herera, R.M. and Coffinan, W.R. (1974). Emasculation of rice by vaccum extraction. Proc. Crop Sci. Soc, Philippines. 5: 12-14. IRRI. (1989). Work plan for 1990-1994. IRRI, Los Banos, Philippines, pp. 16. Janoria, M.P. (1989). A basic plant ideotype for rice. Int. Rice Res. Newsletter. 14:12-13. Bentota, Senadhira & Lawrence Jinks, J.L. and Pooni, H.S. (1976). Predicting the properties of recombinant inbred lines derived by single seed descent. Heredity. 36:253-266. Mather, K. and Jinks, J.L. (1982). Biomedical Genetics. Chapman and Hall, London (3rd Edition). Mode, C.J. and Robinson, H.F. (1959). Pleiotropism and the genetic variance and covariance. Biometrics. IS: 581-537. Perera, A.L.T., Fahim, M., Sriyoheswaran, S., Dhanapala, M.P., Senadhira, D. and Lawrence, M.J. (1997). Quantitative genetics of rice. 1. No evidence of a shortage of exploitable genetical variation in modern indica cultivars. Field Crop Res. (submitted). Perera, A.L.T., Palihawadana, S. and Lawrence, M.J. (1992). Testing the predicted performance of recombinant inbred lines in rice using Triple Test Cross (TTC) design, Basic generations and F, families. Int. Rice Res. Newsletter. 17(6): 5. Vergara, B.S. (1988). Raising the yield potential of rice. Philipp. Tech. J. 13:3-9. 4 48 X