Combining Ability and Heterosis for Agronomic and Yield Traits in Indica and Japonica Rice Crosses

Understanding genetic variability and mode of gene action for agronomic and yield related traits is important in formulation of effective rice breeding program for genetic enhancement of grain yield. Combining ability analysis and heterosis was conducted to identify yield associated traits from nine male indicas and three female japonicas, together with their 27 F1 hybrids. Four parental lines, including Basmati 370, Basmati 217, K2-54 and Komboka showed good general combining ability in days to 50% flowering, days to maturity, number of tillers plant, number of spikelet’s panicle, number of panicles plant, number of filled grains panicle, and grain yield. While the combine K2-9 × Komboka, K2-9 × Basmati 370, K2-54 × Dourado Precoce and K2-54 × Basmati 217 showed specific good for grain yield. The hybrids K2-9 × Basmati 370, K2-8 × Basmati 217, K2-54 × Basmati 217 and K2-9 × Komboka showed 20% excess in standard check variety, suggesting that they could be good breeding donors.


Introduction
Rice is the second most valued crop after maize in the world today (Syed & Khaliq, 2008).The world rice requirement by 2050 will be 943.6 million tonnes which requires an annual increase of about 5.8 million tonnes from the present level of production (FAO, 2017).To achieve this target, it is necessary not only to utilize the existing upland and lowland land resources, but also develop upland rice varieties with high yield potential and durable resistance to both biotic and abiotic stresses (Dogara & Jumare, 2014).
Rice forms a larger part of the diet for both urban and rural populations in Kenya.Kenya produces about 140,000 metric tonnes against a demand of over 540,000 metric tonnes per year (MoA, 2014).Unfortunately, rice yield per hectare is low (< 3.6 t ha -1 ) because smallholder farmers rely on local cultivars with low yield potential, highly susceptible to bacterial leaf blight and rice blast (Kimani, 2010).Small holder farmers at Mwea irrigation scheme mainly grow Basmati rice under irrigated ecosystem.Basmati rice is mainly preferred by most consumers due to its aroma and good cooking qualities (Njiruh et al., 2013).However, yield of Basmati 370 is low (3.0 t ha -1 ).Upland rice varieties are not widely grown in Kenya despite their enormous potential in increasing national production (Kimani, 2010).This stagnation in productivity and associated high prices has led to decline in consumption of locally produced rice and increased imports from Pakistan, China, India and Vietnam (MoA, 2014).populations from which agronomically superior lines can be selected (Fasahat et al., 2016).Therefore, the knowledge of combining ability provides information on the nature and magnitude of gene effects that regulate grain yield and yield characters hence enabling the breeder to design an effective breeding method for genetic enhancement of grain yield and yield components (Dar et al., 2014).
Previous studies reported an increase in grain yield due to favourable heterosis for characters such as flag leaf area, number of spikelets per plant and number of filled grains per panicle (Vanaja & Babu, 2004).
Positive heterosis is desirable for grain yield and negative heterosis is preferred for earliness.Heterosis is divided into heterobeltiosis, mid-parent heterosis, and standard heterosis (Alam et al., 2004).Standard heterosis is more important of the two levels of heterosis because it is aimed at developing desirable hybrids superior to the existing high yielding commercial varieties (Chaudhary, 1984).The objective of this study was to determine the combining ability and heterosis for agronomic traits and grain yield in indica and japonica rice crosses as a criterion for developing superior rice varieties.

Study Location
The study was conducted at Industrial Crops Research Centre, Mwea.The research station is located in Mwea Division, Kirinyaga County, Kenya.The area receives an average rainfall of about 850 mm over two seasons, with an average temperature ranging from 15.6 o C to 28.6 o C. The research station lies on a latitude 0 o 37′ S and Longitude 37 o 20′ E on an elevation of 1159 m above sea level.The soil type is nitosol with a PH of about 5.65.Monocropping is mainly practiced at the location.Water for irrigation is conveyed using partially lined canal and distributed through earthen canals (KARI, 2000).

Rice Germplasm Used in the Study
The pedigree and origin of the rice germplasm used in the study are presented in Table 1.The male parents used in this study were indica type while the female parents were japonica type of rice.Four NERICAs, namely NERICA 1, 3, 10 and 11 were released in Kenya in 2006.NERICA 11 was the highest yielder (5,451 kg ha -1 ) followed by NERICA 1 (5,194 kg ha -1 ), NERICA 10 (4,663 kg ha -1 ) and NERICA 3 (4,270 kg ha -1 ) (WARDA, 2008).In addition, NERICA 1 is early maturing and well adapted to local conditions.

Development of F 1 Hybrids
The nine male indica and three female japonica parents were grown in a 5 × 5 m 2 hybridization nursery at the Industrial Crops Research Centre, Mwea during the short rain season between March and June 2016.The seeds were sown by direct seeding with two seeds per hill and the cultivars were replicated three times at an interval of 1, 14 and 21 days to synchronize flowering and facilitate generation of enough F 1 seeds.Di-ammonium phosphate fertilizer was applied at a rate of 50 kg N ha -1 and 40 kg P ha -1 during planting to enhance plant growth and root vigour.The parents were crossed following North Carolina II mating design according to Comstock and Robinson (1952) to generate 27 F 1 hybrids.Emasculation of the female was done using a vacuum emasculator (CAT No. 2027, Prasanthic Scientific, India) set at pressure of 15 kpa.The panicles of the male parents were cut using scissors before anther dehiscence and taken to the location of the female.The glassine bag was removed from the emasculated female and the pollinator panicles were gently shaken over the female panicle.The glassine bag was replaced over the female parent and the details of the male and female, emasculator and date of pollination were recorded on the tag, and attached to the female panicle (Coffman & Herrera, 1980).

Evaluation of F 1 Hybrids
The 27 F 1 hybrids and their parents were evaluated at Mwea Research Station during the cropping season of 2017.The trial was laid out in a randomized complete block design with three replications.The seeds were sown by direct seeding with an inter-row spacing of 20 cm and intra-row spacing of 20 cm with two seeds per hill.When the seedlings were 2-3 weeks old, gapping and thinning was carried out leaving one seedling per hill to ensure uniform plant density.Hand weeding was carried out at an interval of 20, 40 and 60 days after planting and DAP fertilizer (18:46:0) was applied at a rate of 50 kg N ha -1 .The crop was top dressed twice at tillering with NPK (17:17:17) fertilizer at a rate of 47 kg N ha -1 and at panicle initiation stage with calcium ammonium nitrates (CAN) fertilizer at a rate of 47 kg N ha -1 .The trial was irrigated twice every week while pesticide (Duduthrin 1.75 EC, Lambda-cyhalothrin 17.5 g/L) was applied every two weeks at a rate of 50 ml per 20 L of water to control stem borer, fall armyworm and leaf hoppers.

Determination of Agronomic and Yield Traits
Agronomic and yield data was collected according to the procedure outlined in Standard Evaluation System (SES) for rice (IRRI, 2013).The quantitative and qualitative data was collected at vegetative, flowering, maturity and harvesting stages as follows: (1) Days to 50% flowering: the number of days from sowing to when half of the plants in a plot and half of the tillers in these plants had flowered.
(2) Days to maturity: number of days from seedling emergence to when 85% of the spikelets were mature.
(3) Plant height: the height (cm) of the plant from the plant base to the tip of the highest panicle excluding the awn was measured using a meter ruler.
(4) Number of tillers plant -1 : the numbers of tillers hill -1 were determined by counting all emerging shoots.
(5) Flag leaf length: the length (cm) of the leaf immediately beneath the spike was measured from the collar to the nodal end of the leaf using a meter ruler.
(6) Panicle length: the length (cm) of the panicle from the neck node base to the tip of the top most grain in the panicle was measured using a meter ruler.
(7) Number of fertile tillers plant -1 : the number of tillers bearing panicles at physiological maturity.
(9) Number of filled grains panicle -1 : two plants were sampled per plot and the panicles harvested, threshed and dried to 14% moisture content.The filled and the unfilled grains were separated using the floatation method (IRRI, 2013).The grains the sunk to the bottom of the beaker and those that floated were dried separately and the total number of the filled and unfilled grains were counted.
(10) Grain yield: the mature panicles were harvested, threshed, cleaned and dried to 14% moisture content.The threshed grains from each plant were weighed separately using electronic weigh balance (Model SP 401, OHAUS Lab balances, Amazon UK), recorded in grams and converted to kg ha -1 .The grain yield plot -1 was calculated using the formula described by Gomez (1972): Where, w = weight of the grain from the harvested hills; n = number of harvested hills; N = total number of hills in a plot.
(11) Thousand grain weight: 1000 grains from each harvested plants were counted using electronic grain counter (Model AG64-100, Wagtech International, New York) and weighed after drying the grains to 14% moisture content.

Estimation of Combining Ability and Heterosis
The agronomic and yield data was subjected to analysis of variance using analysis of genetic designs in R (AGD-R) version 3.0 (Gregorio et al., 2015).Their means were separated using the least significant differences (LSD) at P ≤ 0.05.
The variation between families was divided into component differences among males, females and that due to male × female interaction.The expectations of males and females in this design were equivalent to the general combining ability (GCA) and the interaction between males and females were equivalent to the specific combining ability (SCA).F-tests were done to examine the differences between males and females.The GCA and the SCA effects for the parents and their hybrids were calculated based on the statistical model by Comstock and Robinson (1952): The ratio of GCA and SCA was computed according to Baker (1978): Where, σ 2 GCA denotes general combining ability variance and σ 2 SCA for specific combining ability variance.A ratio closer to one reveals GCA action while a ratio of less than one predicts SCA action.
The magnitude of heterosis in the F 1 hybrids was calculated in relation to the better parent rice variety as percent increase or decrease of the F 1 hybrids over the better parent.Better parent heterosis was computed according to Virmani et al. (1997): Where, F 1 = mean performance of a single cross and BP = mean performance of the better parent.

Analysis of Variance
General combining ability (GCA) was significant (P ≤ 0.05) for days to 50% flowering, days to maturity, number of tillers, number of spikelet's panicle -1 , number of productive tillers, number of filled grains panicle -1 and grain yield (Table 2).Specific combining ability (SCA) was significant (P ≤ 0.05) for plant height, panicle length, flag leaf length and 1000 grain weight.The significant variation in the GCA and SCA variances suggested the importance of both additive and non-additive gene action in the expression of the yield traits.
GCA/SCA predictability ratio for duration to 50% flowering, days to maturity, number of tillers, number of spikelet's panicle -1 , number of productive tillers, number of filled grains panicle -1 and grain yield ratio was close to one suggesting the influence of additive gene action in the expression of these yield characters.However, the ratio for plant height, panicle length, flag leaf length and 1000 grain weight had a predictability ratio of less than one indicating the influence of non-additive action in the control of these traits.Note.df = degrees of freedom, * Significance at 5% level of probability, ** Significance at 1% level of probability, σ 2 m = Male variance, σ 2 f = Female variance, σ 2 m × f = Interaction effect of male and female, σ 2 A = Additive variance, σ 2 D = Dominance variance.

General Combining Ability (GCA) Effects of the Parents
The general combining ability effects for agronomic and yield traits are presented in

Specific Combining Ability Effects of 27 F 1 Hybrids for Agronomic and Yield Traits
The specific combining ability effects for agronomic and yield related traits are presented in

Estimate of Genetic Components
The analysis of variance for combining ability showed that the rice genotypes differed significantly for all the agronomic and yield traits studied.Further analysis of GCA/SCA variances showed that the nature of gene action was additive due to high magnitude of fixable genetic component for days to 50% flowering, days to maturity, number of tillers plant -1 , number of spikelet's panicle -1 , number of effective tillers plant -1 , number of fertile grains panicle -1 and grain yield.Non-additive gene action was dominant in plant height, panicle length, flag leaf length and 1000 grain weight.These results are contrary to the findings of Sathya and Jebaraj (2015) who reported dominance of non-additive gene action for all the traits studied under aerobic condition.

Evaluation of Parents Based on GCA Effects
The combining ability analysis revealed significant GCA and SCA variance for most of the agronomic characters studied suggesting the importance of both additive and non-additive gene actions in the expression of the yield traits.Previous studies reported the significance of both additive and non-additive gene action in the expression of agronomic and yield traits (Thakare et al., 2009;Mirarab et al., 2011;Padmavathi et al., 2012;Dar et al., 2014 andMalemba et al., 2017).The rice parents such as Basmati 370, Basmati 217, K2-54 and Komboka had high and positive GCA effects for grain yield suggesting that these parental genotypes were good general combiners for grain yield.Previous studies also reported good general combiners for yield and yield traits in rice genotypes (Padmavathi et al., 2012;Raju et al., 2014;Sathya & Jebaraj, 2015).
The parental lines such as Basmati 370, Basmati 217, PAN 84 and Komboka exhibited high and positive GCA effects for panicle length, flag leaf length, number of tillers plant -1 , number of spikelet's panicle -1 , number of panicles plant -1 and number of filled grains panicle -1 suggesting their desirability for improvement of positive agronomic and yield traits because they contributed to high grain yield, in contrast parents with low and negative GCA estimates (NERICA 1, PAN 84, K 2 -9, Dourado precoce and K 2 -54) are preferred for improvement of negative traits of grain yield such as plant height, 50% days to flowering and days to maturity.Previous studies reported the significance of using parents with high and positive GCA effects for the improvement of positive agronomic traits (Mirarab et al., 2011;Raju et al., 2014;Malemba et al., 2017).This study revealed high GCA variances for most of the yield traits studied.Mirarab et al. (2011) conducted a study on heterosis, combining ability and genetic parameters of yield and yield components in rice.They reported that GCA was only significant for total number of kernels per panicle, number of filled grains and grain yield per plant implying that additive gene action was predominant in the control of these traits.Previous research also reported the predominance of additive gene action in the inheritance of days to maturity, number of productive tillers plant-1, number of panicles per plant and 1000 grain weight (Dar et al., 2014).Therefore, it is suggested that a breeding method that would take care of the fixable gene effects and as well as maintain heterozygosity for exploiting the dominance effects may be more efficient for the improvement of grain yield (Chakraborty et al., 2009).Hence, a simple selection procedure followed by pedigree breeding is sufficient to improve the traits influenced by additive gene action (Chakraborty et al., 2009).

Evaluation of F 1 Hybrids Based on SCA Effects
The hybrids K 2 -9 × Komboka, K 2 -9 × Basmati 370, K 2 -54 × Dourado precoce and K 2 -54 × Basmati 217 were good specific combiners for grain yield because they expressed high SCA effects for grain yield.Similar studies also reported superior SCA effects for drought tolerant and yield traits in rice varieties (Muhammad et al., 2010;Sathya & Jebaraj, 2015;Malemba et al., 2017).All the high yielding hybrids (K 2 -9 × Komboka, K 2 -9 × Basmati 370, K 2 -54 × Dourado precoce and K 2 -54 × Basmati 217) originated from high × high GCA combinations attributable to additive × additive type of gene actions thus the yield potential of these hybrids can be fixed in subsequent generations (Sandhyakishore et al., 2011).However, negative SCA effects for grain yield were observed in the crosses K 2 -9 × NERICA 1, K 2 -8 × Dourado precoce and K 2 -8 × Komboka suggesting a low yield potential.The hybrid K 2 -9 × NERICA 1 originated from high × low GCA combination while the rest of the crosses with low and negative SCA effects originated from low × high GCA combinations.The high yield potential observed in the hybrids with high × low GCA combinations could be attributed to interaction between positive alleles in the good combiner and negative alleles in the poor combiner (Hasan et al., 2015).This study revealed that hybrids (K 2 -54 × Basmati 217, K 2 -54 × Komboka, and K 2 -54 × Basmati 370) that originated from high general combiner parents with high and positive SCA estimates are expected to produce desirable transgressive segregants which can be recognized by carrying out pedigree breeding method (Muhammad et al., 2010).High SCA effects of hybrids (K 2 -9 × NERICA 1) that originated from high × low GCA combining parents would be unfixable in subsequent generations and hence cannot be exploited by pedigree selection procedure (Sathya & Jebaraj, 2015).But, in later generations these hybrids would produce desirable transgressive segregants upon modification of the conventional breeding methodology to accommodate both additive and non-additive genetic effects (Chakraborty et al., 2009).The present investigation revealed that parents with high GCA estimates were not always the best general combiners.Furthermore, the results also indicated that parents with high GCA effects were the best general combiners for only a specific trait but none of the parents or the specific crosses was best for all the characters studied.Similar findings were reported (Chakraborty et al., 2009;Malemba et al., 2017).
Positive heterosis for days to maturity was also reported (Nadali, 2010;Sunil Kumar et al., 2012).Development of early maturing and high yielding rice varieties are desired in rice breeding program (Rahimi et al., 2010).
Negative heterosis is desirable for duration to 50% flowering and days to maturity (Borah et al., 2017).The hybrids with negative heterosis over better parent were early maturing.Therefore, the early maturing hybrids suggested the possibility of developing early maturing lines.Similar findings were also reported (Rahimi et al., 2010).
In this study, high parent heterosis for effective tillers was observed in 16 F 1 hybrids suggesting that these F 1 hybrids had high number of effective tillers and thus possessed high yield potential.Previous studies reported high parent heterosis for number of tillers in F 1 rice hybrids (Mirarab et al., 2011;Dwivedi & Pandey, 2012;Borah et al., 2017).
The number of fertile grains panicle -1 is directly associated to grain yield and therefore a positive heterosis would be desirable in order to increase the productivity of rice varieties (Tiwari et al., 2011).In this study, 15 F 1 hybrids were superior in terms of filled grains per panicle over the better parent implying that these hybrids were high yielding because fertile grains per panicle has a direct association with grain yield.

Conclusions and Recommendations
The analysis of combining ability and heterosis for agronomic and yield related traits between indica and japonica rice crosses showed that the parental lines Basmati 370, Basmati 217, K 2 -54 and Komboka were good general combiners for grain yield and offered good scope for developing rice hybrids with high yield potential.
In contrast, their crosses K 2 -9 × Komboka, K 2 -9 × Basmati 370, K 2 -54 × Dourado precoce and K 2 -54 × Basmati 217 were good specific combiners and expressed better heterotic effects for grain yield.Days to 50% flowering, days to maturity, number of tillers plant -1 , number of spikelet's panicle -1 , number of effective tillers plant -1 , number of fertile grains panicle -1 and grain yield were governed by additive gene action suggesting that hybridization followed by selection in later generations may be recommended for improvement of these traits.Non-additive gene action was dominant in plant height, panicle length, flag leaf length and 1000 grain weight implying that the genetic potential of these traits would be unfixable in subsequent generations but modification of the pedigree breeding procedure to accommodate the non-additive genetic effects may be recommended.

Table 1 .
Characteristics of parental lines used for hybridization

Table 2 .
Analysis of variance for combining ability of 12 parents and their 27 F 1 hybrids for different yield traits

Table 3 .
General combining ability (GCA) effects of 12 rice parents for agronomic and yield components Note. * Significance at 5% level of probability, ** Significance at 1% level of probability.Table 4. Specific combining ability (SCA) effects of 27 F 1 hybrids for different agronomic and yield traits * Note.* Significance at 5% level of probability, ** Significance at 1% level of probability, *** Significance at P ≤ 0.001.

Table 5 .
Heterosis for agronomic and yield components in indica and japonica rice crosses