Combining Ability for Grain Yield Performance among CIMMYT Germplasm Adapted to the Mid-Altitude Conditions

The International Centre for Maize and Wheat Improvement (CIMMYT) develops maize (Zea mays L.) inbred lines and hybrids yearly that have several breeding and commercial attributes. However, no genetic analysis has been done on the recently developed inbred lines for yield performance under drought and low-N stress. The objectives of this study were to identify lines with positive general combining ability (GCA) effects for grain yield under stress environments and to identify the best single-cross hybrids with the highest specific combining ability (SCA) effects. Analysis of variance combined across sites showed significant mean squares for genotypes, locations and genotype by environment interaction (GEI) for grain yield. GCAlines, SCA and components of interaction effects were significant across sites. Additive genetic variance was more important than dominance variance in determining yield performance across locations indicating that selection based on grain yield under drought and low-N stress can be effective. Average grain yield across the eight locations ranged from 1.61 t ha to 10.63 t ha while narrow sense heritability for grain yield was 52.6% across sites and was slightly lower under managed drought and low-N stress. The testers CL115807 and CL106622 showed positive and significant GCA effects for yield performance under drought and low-N stress respectively. The best tester across all sites was CL115793 and line CZL0713 had consistently positive GCA effects for grain yield across sites. CML536 × CL115802 and CML312 × CL106508 were the best single crosses under low nitrogen stress sites while hybrid CML312 × C323-45 showed the highest positive SCA effects across sites. In conclusion, our results show that CIMMYT has new lines that have desirable adaptive attributes when grown under drought and low nitrogen stress environments in the mid-altitude region; hence these can be adopted for hybrid, synthetic and OPV formation.


Introduction
Drought and low-N stresses are factors that largely limit maize production in tropical environments (Bӓnziger et al., 1999;Edmeades et al., 1999;Bӓnziger & Diallo, 2001;Diallo et al., 2001).Yield declines are also being noticed in the productive mid-altitude eco-zones of central and southern Africa (Diallo et al., 2001).The mid-altitude zone falls within the altitudinal range of between 1000 and 1800 masl and it is characterized by rainfall of more than 500mm and mean temperature of 21.5 o C. The social-economic constraints, such as high inorganic fertilizer costs and lack of credit for small scale farmers (Diallo et al., 2001) are further worsening the bio-physical constraints that are hampering maize production.Hence development of maize hybrids that can adapt to these stresses is important (Hoisington, 2001;Betrán et al., 2003;Bӓnziger et al., 2004).
Selection for grain yield under severe drought stress has often been considered inefficient (Bolanos et al., 1992), because the estimate of heritability for grain yield has been found to decline as yield fell.But, these authors recommended the use of secondary traits as an option to increase selection efficiency under these conditions, since they have adaptive values, high heritability, and are easy to measure.Diallo et al. (2001) carried out a study to develop hybrid maize varieties that are tolerant to low-N and drought.In the study, drought tolerant inbred lines developed by CIMMYT-Harare in collaboration with CIMMYT-Mexico were crossed with two stress resistant testers (CML 202 and CML 206) during the 1997/98 minor season.The crosses were evaluated at seven sites, alongside local checks in 1999 under both stressed and unstressed conditions.The selected best single cross hybrids were crossed with other testers (CML 78 and CML 384) in 2000 and the resulting three-way hybrids were evaluated in 2001.Grain yield and secondary traits such as anthesis-silking interval, leaf senescence and number of ears plant -1 were used to select the most promising materials.The study identified eight drought and low-N tolerant three-way hybrids which yielded 24, 15 and 64% more than the best hybrid checks under optimum, low-N and drought stress conditions, respectively.Betrán et al. (2003) also did a study to evaluate a group of tropical white inbred lines for grain yield performance, combining abilities and stability under optimal, drought and low-N stress conditions.They found heterosis to increase with the intensity of drought stress.Significant interactions were observed for combining abilities under low-N and high N.The type of gene action appeared to be different under low N and high-N, with additive effects more important under drought and dominance effects more important under low-N.The importance of additive effects increased with intensity of drought stress.The results suggested the need to incorporate drought tolerance in both parental lines to achieve acceptable hybrid performance under severe drought.
CIMMYT-Zimbabwe recently developed maize inbred lines that can adapt to the mid-altitude conditions through the pedigree breeding method.The lines have desirable adaptive attributes under low-N, drought, and in agronomically favorable conditions (optimal environments).But the combining ability of the new lines with the old CIMMYT lines is not known, yet this information is important to the breeders as a decision making tool in hybrid development, population and tester formation.Hence, the objectives of this study were to identify lines with positive and significant GCA and SCA effects for grain yield under stress environments and to identify the best single-cross hybrids with the highest SCA effects.

Planting Materials and Evaluation Sites
The study was a combining ability trial of hybrids developed using the North Carolina Design II mating design, where 30 male lines were crossed with eight (8) female lines (Table 1).The male lines are the elite lines that were developed for adaptability in the sub-Saharan African conditions.The female lines are common CIMMYT lines and they were used as testers in this study.The hybrids were developed by crossing using hand pollination.The 240 hybrids developed were planted alongside 10 checks: two checks from Seed Co (SC 727 and SC 633) and eight checks developed by CIMMYT (CML444 × CML536, CML395 × CML444, CML312 × 442, CML539 × CML 442, CML539 × CML197, CML312 × CML444, CML444 × CML197 and CML444 × CML489).The evaluations were done at eight sites in the 2012-2013 summer and winter seasons in Zimbabwe (Table 2).The sites used represent the areas where maize is mostly grown in Zimbabwe.
Table 1.The pedigrees of the males and females that were crossed using an NCII design in the 2012 rainy season in Zimbabwe

Experiment Setup and Agronomic Management
The hybrids were grown in one-row plots at an inter-row spacing of 0.75 m and an intra-row spacing of 0.25 m at all sites.The experiments were grown using a 10 × 50 alpha (0, 1) lattice design (Patterson & Williams, 1996).Two hundred and fifty (250) entries, replicated twice with 50 blocks per replication were planted, but the hybrids were randomized differently at the eight sites.The trials were initially planted at two seeds per hole then later thinned to one plant per station at four weeks after crop emergence (4 WACE) to achieve a plant population of 53 000 plants/ha.
Maize fertilizer (7% N, 6% P, 6% K) was applied as a basal fertilizer at 400 kg ha -1 at the sites.But, the low-N sites did not receive any form of top-dressing fertilization.Basal fertilizer was broadcasted by a vicon and disked into the soil before planting.All sites, excluding the low-N sites received two applications of 200 kg ha -1 AN (ammonium nitrate) each as top dressing.The first split application was done at four (4) weeks after crop emergence and the second application was done at eight (8) weeks after crop emergence.These experiments only received four circles of irrigation and the total amount of water received by the plants ranged from between 200 to 250 mm.Irrigation was terminated at a time when the plants were left with about 50 days to shed pollen.
Karate (Labda cyhalomethrin) was mixed with 200 litres of water and was applied in the field at the rate of 100 ml per hectare before sowing to control ants, termites and other soil pests.Hand weeding was the predominant form of weed control at all sites, however selective application of paraquat at the rate of 1.5 l ha -1 was also done at all sites.Stalk borer (Buseola fusca) was controlled by applying Thiodan 1% granules at a rate of 3 kg ha -1 granules into the funnel of each plant at four weeks and eight weeks after crop emergence.

Data Collection and Statistical Analysis
The traits that were recorded in this study were: grain yield: GY (shelled grain weight per plot adjusted to 12.5% grain moisture and converted to tons per hectare), male flowering date: MF (measured as number of days after planting when 50% of the plants shed pollen), female flowering dates: FF (measured as number of days after planting when 50% of the plants produce silk pollen), anthesis-silking interval: ASI (calculated as the difference between MF and FF), and grain moisture: MOI (percent water content of grain as measured at harvest).
Line × tester analysis was performed for grain yield across all sites, under optimum, managed drought, and low-N sites to obtain probability estimates of SCA and GCA of the parents.This analysis was done using R.2.11.1, which is embedded in the CIMMYT Fieldbook software (Bӓnziger & Vivek, 2007) and the following model was used: where, i = 1, 2, 3, … 8, j = 1, 2, 3, … 30, k = 1, 2, and c ijkp represented the value of the progeny of a mating of the i th line, the j th tester, in the k th replication, and in the p th environment (site).The simple µ stands for grand mean, g i is the GCA effect common to all progeny of the i th line, g j is the GCA effect common to all progeny of the j th tester, s ij is the SCA effect specific to the progeny of mating the i th line and the j th tester, E p is the average effect of the p th environment, r k (E p ) is the effect of the k th replication that was nested within the p th environment, (gE) ip and (gE) jp are the interactions between the GCA effects and the environment, (sE) ijq is the interaction between the SCA effect and environment, and e ijkp is the random experimental error.This model was adopted from Lee et al. (2005).Genetic effects, broad and narrow-sense heritability were calculated using the AGD-R software (Rodríguez et al., 2015).
The GCA and the SCA estimates were calculated according to Beil and Atkins (1967).

Across Sites Analysis of Variance for Grain Yield
Results of the analysis of variance combined across sites revealed that entry mean squares were significant for grain yield.GCA lines , SCA effects, entry × site interaction and components of interaction effects were also significant across sites (Table 3).The highest yielding early maturing hybrid was CZL0713 × CL11579 (6.26 t ha -1 ).CML395 × CL106622 (6.03 t ha -1 ) was the best performer within the intermediate maturing hybrid group, while CML444 × CL115797 (6.35 t ha -1 ) showed highest performance among the late maturing hybrids, although its performance was surpassed by a commercial check hybrid SC 727 (6.65 t ha -1 ) (Table 4).General combining ability (GCA) for the lines (old CIMMYT lines) and specific combining ability (SCA) was significant (p < 0.001) across sites, under optimum, managed drought, and low-N environments.There were significant interactions (p < 0.001) between GCA lines , GCA testers and SCA with the environment (Table 3).Additive genetic variance was more important than dominance genetic variance, and at the same time, genotypic variance was more important than environmental variance for grain yield across sites, across managed drought, low nitrogen, and optimum sites.Grain yield performance was highly repeatable across all environments.Narrow-sense heritability slightly exceeded 50% across sites, but it was relatively low under managed drought sites (Table 3).

General Combining Abilities Effects for Grain Yield of Lines and Testers
The tester CL115807 showed positive and significant GCA effects for grain yield under managed drought sites.CL106622, showed positive and significant GCA effects under low-N environments.The best tester across the eight sites was CL115793 (Table 5).Line CZL0713 consistently showed positive GCA for grain yield in all the environments.However, this line only showed significant GCA effects for yield performance on the managed drought stress sites (Table 6).
Table 5.The GCA estimates for grain yield, their ranks and the significance levels across eight sites, three optimum sites, two drought sites, two low nitrogen sites and one random drought site of the male lines in hybrid combinations done using the NCII mating design in the 2012 winter season that were evaluated in 2012-13 summer and winter seasons Note.‫ﻟ‬ General combining ability (GCA); ‫ل‬ Tonnes per hectar.
Table 6.The GCA estimates for grain yield, their ranks and the significance levels across eight sites, three optimum sites, two drought sites, two low N sites and one random drought site of the female lines in hybrids combinations done using the NCII mating design in the 2012 winter season that were evaluated in 2012-13 summer and winter seasons Note.‫ﻟ‬ General combining ability (GCA); ‫ل‬ Tonnes per hectar.

Specific Combining Ability Effects for Grain Yield of Lines Versus Testers
The estimates of the SCA effects were summarized in Table 7. Line CML 537 showed the highest positive and significant SCA effects with the tester, CL106683 and, CML 312 showed the second highest positive and significant SCA effects with C323-45 under optimum environments.Under managed drought sites, the highest, positive and significant SCA effects were noted on combination between CML 312 and C323-45, followed by combination between CML 536 and CL106622.The best single-crosses under low nitrogen sites, with positive and significant SCA were CML 536 × CL115802 and CML 312 × C323-45.CML 312 × C323-45 is the single-cross hybrid that showed the highest positive and significant SCA across sites, with SCA effects of 1.462 t ha -1 (Table 7).
Table 7.The best 20 hybrid combinations that were crossed using the NCII mating design and evaluated in 2012-13 summer and winter seasons with the highest SCA effects for grain yield across the eight sites: three optimum sites, two drought sites, two low N sites and one random drought site Note.‫ﻟ‬ Specific combining ability (SCA); ‫ل‬ Tonnes per hectar.

Discussion
Adequate genetic variability is important to make selection progress in breeding programs targeting improved grain yield under both stress and non-stress environments.However, Weitholter et al. (2008) suggested that selection is mostly effective for qualitative traits that are highly heritable.The significant genotypic variation for grain yield performance observed in this study indicates that, good progress can be made in selecting for improved grain yield under drought and low-N stress environments.Studies done by Turi et al. (2007) and Salami et al. (2007) also showed significant genotypic variation for grain yield under both stressed and unstressed environments.Voichita et al. (2011) suggested that if genetic variation exists amongst maize inbred lines, the scenario suggests that those inbred lines could be useful in a maize-breeding program.
GEI mean squares for grain yield were significant and a study done by Beyene et al. (2011) also identified significant mean squares for GEI.The significant variation for GEI effects for grain yield suggest that these CIMMYT lines and hybrids should be tested in contrasting environments in multi-locational trials to identify the most stable line regarding drought and low-N tolerance.This point is supported by Anley et al. (2013) who coined that significant GEI interactions indicates inconsistency in genotypic performance across testing environments, and suggested that the genotypes have to be tested in several testing locations in order to select stable genotypes.
The highly significant site effect on grain yield under drought and low-N stress environments suggested that selection for improved maize grain yield has to be carried out for specific drought and low-N sites.These results are in-line with findings of Badu-Apraku et al. ( 2014) who reported highly significant location effect for maize grain yield measured under two Striga-free environments.Bӓnziger et al. (2004) suggested the need to do selections under carefully managed high priority abiotic stress sites, indicating that this initiative can significantly increase yields in a highly variable drought-prone environment.However, an experiment done by Miti et al. (2010) under low nitrogen soils showed a low heritability estimate (0.38) for grain yield indicating that selection based on grain yield under low-N was ineffective.
The results showed that the new CIMMYT lines have desirable adaptive attributes when grown under common environmental stresses prevalent in the mid-altitude climatic region: drought and low-N soils.This was confirmed by the importance of additive genetic effects for grain yield performance and the moderate heritability of this trait in these environments.Moderate to large genetic variance and also low to high heritability estimates indicates the presence of sufficient residual genetic variation in a population, which makes further improvement for the trait concerned possible (Badu-Apraku, 2007;Rajesh et al., 2013).These new CIMMYT lines can be used as parental lines in breeding and hybridization programs targeting production of maize that is adapted to drought and low-N stress environments.For example, CL115807 showed to be a good line to use when making hybrids that are tolerant under drought.On the other hand, CL106622 is also the best line to use in the formation of low-N tolerant hybrids.However for breeders who aim to develop varieties with stable performance across all environments, they can choose CL115793 that showed the highest positive GCA effects across environments.
In order to produce good hybrids and synthetics in maize breeding, there is always need to make good decisions in the choice of parents, since the parental attributes affects the performance of these breeding products.GCA is a very important factor when making these decisions (Pswarayi & Vivek, 2008;Singh et al., 2013).Hence, the identification of those many lines with positive GCA effects for GY across sites, under optimum, managed drought, low-N and random stress sites random stress sites are important.According to Anwar et al. (2011), this information will be useful to select best lines to use when making hybrids, synthetic and open pollinated varieties (OPV's) that are adapted to the targeted environments.The identified low-N and drought stress adapted genotypes will increase maize productivity under these stress environments, thereby lessening problems of hunger and poverty that are predominant in sub-Saharan Africa.
Pingali and Pandey ( 2001) indicated the importance of doubling maize production from the period of the year 2000 to the year 2020, in an endeavor to offsetting the high demands that are being caused by increasing population sizes, especially in developing countries and the increased demands for feed and biofuels globally.Hence, the new knowledge generated regarding the breeding value of the new CIMMYT lines will be useful in the development of desirable cultivars (Masny et al., 2008;Amiruzzaman et al., 2011;Singh et al., 2013).
The study also identified CML 312 as the best CIMMYT common line when breeding for drought and low-N adaptability.Its hybrid combinations with CL106851, CL106508 and C323-45 showed the best specific combining abilities in most stress environments.This information is important for maize breeders that are involved in breeding for stress adaptability in sub-Saharan Africa.

Conclusion
The study has demonstrated that the new CIMMYT mid-altitude adapted lines have some levels of low-N and drought stress adaptability.However, these new lines need to be evaluated together with the old lines as these also have the desirable adaptive attributes in the stressed environments.The best single crosses identified in this study can also be considered as single-cross testers or they can be further screened for grain yield stability across several environments and seasons, in an endeavor to facilitate their release as new hybrids.

Table 2 .
The eight sites used for the evaluation of the maize hybrids that were planted in the summer and winter seasons of 2012-2013

Table 3 .
Analysis of variance (ANOVA) for grain yield (t ha -1 ) across eight sites, two managed drought and low nitrogen sites, and three optimum sites for the North Carolina design II crosses of 30 testers and eight (8) lines following line × tester analysis procedures Note. ‫ل‬ Degrees of freedom (DF); ‫و‬ Mean of squares (Mean Sq).*, **, ***, ns Significant at 0.5, 0.01 and 0.001 significance levels and not significant, respectively.

Table 4 .
Grain yield performance of top four early, intermediate and late maturing hybrids developed using the NCII mating design in the 2012 winter season that were evaluated in 2012-13 summer and winter seasons, alongside 10 check hybrids