Genetic Analysis of New Maize Hybrids for Yield and Resistance to Aflatoxin Accumulation

Maize (Zea mays L.) is the most important cereal crop in sub-Saharan Africa providing food for many of its inhabitants. However, production is hampered by low yields and aflatoxin contamination, among others. The toxin contaminates maize before harvest and during storage. Ghana lacks regulatory infrastructure for monitoring and detecting aflatoxin in grains prior to market, and most of the local maize varieties have been found to be susceptible to aflatoxin accumulation. Host resistance is envisaged as a key approach in addressing the aflatoxin menace. Thus, this study was conducted to identify new hybrids that combine high yield with resistance to aflatoxin accumulation as well as study the mode of gene controlling tolerance to aflatoxin accumulation. Sixteen aflatoxin resistant inbreds from exotic sources were crossed as males to 10 local germplasm in a North Carolina II design to generate 160 new hybrids. These were planted together with 9 checks using a 13 × 13 alpha lattice with three replications. The new hybrids were evaluated across six environments in two seasons. Five plants each per hybrid were inoculated with a local strain of Aspergillus flavus at a concentration of 9 × 10conidia/ml. Ninety-six out of the 169 crosses were analyzed statistically. Significant effect of environment and genotypes for all traits especially, for aflatoxin accumulation resistance and yield were observed. The general combining ability effect of males for all traits were found significant (P < 0.05) whereas that of the females were not significant for all traits. Inbreds with consistent significant negative GCA effect for aflatoxin reduction were identified (MP715, TZI8, MP719). Furthermore, the underlying genetic control for the aflatoxin accumulation resistance trait was found to be via both GCA and SCA effects. Heritability estimates were moderate, suggesting permissible transfer of traits during selection to create high yielding aflatoxin resistant hybrids for consumers.

Current average yield of maize in Ghana is around 1.94-2.73 metric tons/ha (FAOSTAT, 2019; https://knoema.com/atlas/Ghana/topics/Agriculture/Crops-Production-Yield/Maize-yield). Further improvement to raise average yields to at least 4 metric tons/ha is needed to meet projected demand. Along with increases in yield, levels of aflatoxin accumulation have also been increasing, necessitating a program for improvement to resistance against aflatoxins. Previous varietal development in Ghana did not focus on aflatoxin resistance, as the serious nature of the problem was not understood. However, current surveys by Perrone et al. (2014) on harvested and consumed maize in Ghana revealed the presence of high levels of aflatoxin, well above approved recommended limits of the USA (20 ppb) and that of European Union (4 ppb). Availability of aflatoxin accumulation resistant hybrids will reduce aflatoxin contamination in maize and invariably help address in part the reported cases of serological contamination of over 98% of West African populations who consume contaminated maize and its products (Wild et al., 1990;Jolly et al., 2006;Agbetiameh et al., 2020).
To capitalize on heterosis in hybrid maize breeding requires the development of inbred parents with superior specific combining ability (SCA) for yield and disease resistance. Aflatoxin resistant inbred breeding lines have been developed by CIMMYT, IITA and the USDA, ARS, in Mississippi, USA; many of these lines originate from the Mexican landrace Tuxpêno (Warburton et al., 2014). These lines contain useful genetic variation for aflatoxin accumulation and A. flavus resistance breeding and can be used in testcrosses to identify productive and resistant hybrids.
Genetic variation associated with aflatoxin accumulation resistance is highly quantitative in nature, with many Quantitative Trait Loci (QTL) that has major effects on resistance. These QTL may act additively and may display high general combining ability (GCA) (Widstrom et al., 1984;Betran et al., 2002;Williams et al., 2008;Womack et al., 2020), which facilitate resistance breeding. Maize inbred lines with resistance to aflatoxin accumulation were identified in multi-environment field trials by Warburton et al. (2013). How these inbreds perform in Ghana, and when crossed to locally adapted inbred lines; and if these inbreds can produce high yielding hybrids for local cultivation, is unknown. Utilization of these resistant lines in Ghana for hybrid development could be direct, as parents of new hybrids, or may require transferring resistance genes into elite locally adapted genotypes. Thus, this study was conducted to identify new hybrid crosses that combine high yield with resistance to aflatoxin accumulation; to determine the combining abilities (GCA and SCA) and mode of gene action (additive, dominant, over dominant or epistatic) for genetic variation controlling tolerance to aflatoxin accumulation; and assess the performance of the hybrids across different growing environments in Ghana.

Plant Materials
The male germplasm used consisted of 16 inbred lines found to be resistant to aflatoxin accumulation over several locations and years by Warburton et al. (2013). They were obtained from the USDA ARS Corn Host Plant Resistance Research Unit, CIMMYT, and IITA and 10 additional genotypes that are well adapted to the Ghanaian environment and preferred by farmers (Table 1). The North Carolina design II was used to generate 160 hybrids at the CSIR-Crops Research Institute at Fumesua in 2016. The hybrids were evaluated with nine local checks; namely, CRI-ETUBI (hybrid) and CRI-MAMABA (hybrid), CRI-OMANKWA (OPV), CRI-TINTIM (hybrid), CRI-ABROHEMA (OPV), CRI-OBAATANPA (OPV), CRI-ABONTEM (OPV), CRI-AHOODZEN (OPV) and CRI-AFRIYIE (hybrid).

Experimental Sites and Field Layout
The experiments were carried out in three locations all with a bimodal rainfall pattern that normally runs from March to July for the major season, and from September to November for the minor season. Fumesua is in the semi-deciduous forest zone at 286 m above sea level between 6.712 N and 1.523 W. Rainfall averages 1500 mm in total, and temperatures range from 21-31 °C. Soils around Fumesua are classified as Asuansi series, a ferric acrisol. Akomadan is within the forest savanna transition zone at 7.396 N, 1.973 W with temperature range of 20-35 °C, with a rainfall ranging from 700 mm and 1200 mm. Wenchi is located at 7.733 N, 2.100 W in the transitional savanna zone with temperature range of 21.2-31 °C, with an annual rainfall of between about 1,140-1,270 mm. The soil types are the savannah ochrosol with some lithosols (MoFA, 2020).
The experiments were planted in a 13 × 13 alpha lattice design during the major and minor seasons of 2017 in all three locations in three replications. Single row plots of 5 m spaced at 0.70 m apart with 0.4 m within plants in each row. Standard field agronomic practices were followed as needed.

A. flavus Inoculation and Aflatoxin Measurements
Pure cultures of A. flavus were obtained from local sources. L-type A. flavus as described by (Cotty, 1989;Klich & Pitt, 1988) were maintained on potato dextrose agar (PDA) as described by Jha (1995). This toxigenic isolate was used to prepare inoculum as described by Windham et al. (2003). Briefly, the isolate was multiplied on sterile corn cob grit in 500-ml flasks each containing 50 g of grits and 100 ml of sterile distilled water. This was incubated at 28 °C for 3 weeks. Conidia in each flask was then washed from the grits using 500 ml of sterile distilled water containing 20 drops of Tween 20 per liter and then filtered through four layers of sterile cheesecloth. The concentrations of conidia determined with a hemocytometer and adjusted with sterile distilled water to 9 × 10 7 conidia per ml.
The side needle technique described by Scott and Zummo (1994) was used to inoculate 5 plants/plot 14 days after mid-silk by inserting the needle under the husks on the upper 1/3 of the ear and 3.4 ml of a spore suspension of 9 × 10 7 conidia/ml was injected over the kernels. Approximately 60 days after mid-silk, cobs were harvested, dried shelled, and 50 g of samples ground using appropriate reagents for analysis using high-performance liquid chromatography (HPLC) to determine aflatoxin concentration levels as described by Wacco et al. (2014).

Data Collection
In addition to aflatoxin levels, data were collected on 15 traits: days to 50% silking; days to 50% pollen shed; anthesis-silking interval; ear aspect; plant aspect; root lodging; stalk lodging; ear number per plant; husk cover; stay-green; m a i z e blight resistance; maize streak resistance; grain moisture content; field weight and grain yield (kg/ha), which was determined at 15% moisture content according to Barreto and Raun (1988)

Statistical Analysis
Analysis of Variance (ANOVA) was performed on plot means for grain yield and all other agronomic traits across environments using the PROC GLM procedure of the SAS software package, version 9.4 (SAS Institute, 2012). Data on aflatoxin contamination was natural log transformed and data for ear rot, stalk and root lodging percentages were transformed using arcsine. Genotype or entry means were adjusted for block effects and analysed according to alpha lattice design (Cochran & Cox, 1960). Each environment was defined as season × site × aflatoxin inoculation treatment. Effects of environment were considered as random while genotypes were classified as fixed.
The hybrids component of variation was divided into variation due to male, female, and female × male interaction. Male and female main effects represent general combining ability (GCA) while female × male interaction represents specific combining ability (SCA) effect (Hallauer & Miranda, 1988). The F tests for male, female, and female × male mean squares were computed using the mean squares for their respective interaction with environment. The mean square attributable to environment × female × male was tested using the pooled error mean squares. The general linear model for the NC II mating design used was: Where, Y ijk = observed trait value; µ = Mean effect; M i = Effect of the i th male; F j = Effect of the j th female; MF ij = Effect of interaction between i th male and j th female; R k = Effect of K th replication; d ε ijk = experimental error.
Restricted maximum likelihood (REML) estimates of the inbreds' and hybrids' genetic and phenotypic variances were obtained with SAS PROC variance components and were used to compute broad-sense heritability (H 2 ) for each trait as; Where, σ 2 G is variance for differences in phenotype due to genotype; σ 2 E is error variance, σ 2 GE is genotype × environment interaction variance; r is number of replications; e is number of environments (Fehr, 1991). Correlation coefficients were computed between grain yield and other agronomic traits using the PROC CORR from SAS.
Estimates of GCA and SCA were calculated and their significance determined by t tests. Male and female general combining ability was estimated as: Where, GCA m and GCA f = General combining ability of male and female parents respectively; X f and X m = Mean of male and female parents respectively; µ = Overall mean of crosses in the trial.
Specific combining ability (SCA) was estimated as: Where, SCA x = specific combining ability of the cross x; X X = Observed mean value of the cross; E(X X ) = Expected mean value of the cross based on the GCA of both parent; µ = Overall mean of crosses.
between sum of square due to GCA to the total genotypic sum of squares (GCA and SCA sum of square) (Beck et al., 1990;Pswarayi & Vivek, 2008). Estimates of GCA of each male were obtained based on its performance in F 1 hybrid combinations with all possible females. GCA and SCA effects were determined for each agronomic trait across all environments.
The mid-parent heterosis (MPH) and better parent heterosis (BPH) values for a cross was computed for aflatoxin levels and yield/ha as: Where, F 1 = Mean of the hybrid; MP = the mean of the parents that constituted the hybrids; BP = the mean of the better parent; MPH and BPH were averaged across aflatoxin contaminated environment.
A modified base index based on the selection index proposed by Badu-Apraku et al. (2011) was formulated to rank the hybrids as follows: Where, GY = Grain yield; SG = Stay green; CA = Cob aspects; PA = Plant Aspects; ln(Y + 1) = Aflatoxin levels; ER = Ear rot.

Combined Genetic Analysis of Traits Across Six Environments
Statistical analysis of agronomic and aflatoxin resistance traits across the six environments for 96 hybrids are presented in Table 2. The analysis shows significant effects of environment and genotypes for all traits and the effect of replication within each environment. The general combining ability effect of lines used as males for all traits was significant (P < 0.05) whereas those used as females was significant for some of the traits (Table 2).

General Combining Ability Estimates of Inbred Lines for Traits
The general combining ability estimates for grain yield was positive and significant only for inbred line NC298 used as a male (1176.82), which is indicative of a good combiner (Table 3). Two inbreds, MP715 and Tzi8, displayed significant negative GCA for aflatoxin accumulation resistance (where negative is the desired outcome) when used as males, but all exotic lines except CML5, CML343 and Hi27 displayed negative but non-significant GCA for aflatoxin reduction when used as males (Table 3). General combining ability estimates for the other traits for all inbred lines are indicated in Table 3. Note. *, *** = significant at 0.01 and 0.05 respectively, ns = not significant. H 2 = broad sense heritability; Env = environment; Rep = replication; GCA = General combining ability; SCA = Specific combining ability. Note. *, *** = significant at 0.01 and 0.05 respectively, ns = not significant. H 2 = broad sense heritability.  Note. *, **, Significant at 0.05 and 0.01probability levels. SE±: standard error; -: Missing data.
Further analysis revealed significant genotype by environment interaction for all traits except days to 50% silking, which suggest that most of the hybrids had similar silking period across sites. Environmental effect on male GCA was significant for all traits as was that of female GCA except for ear rot. Interaction between environments and hybrids was significant for grain yield, anthesis silking interval and plant aspect, but did not significantly influence the rest of the traits ( Table 2).
The mean grain yields obtained across the six environments ranged from 4732.12 kg/ha (TZEEI-6 × MP705) to 7588.96 kg/ha (TZEEI-15 × MP715) (Table 4). Aflatoxin accumulation in the hybrids ranged from 10.80 ppb (ENTRY-5 × Ki3) to 17.20 ppb (ENTRY-5 × TZI8) among the best fifteen aflatoxin resistant hybrids while the worst hybrid recorded 348.00 ppb (TZEEI-15 × Ki3) ( Table 4). The various ranges of means for selected traits studied can be found in Table 4. To select hybrids that combined all suitable agronomic traits with reduced aflatoxin accumulation across the six environments, the top twenty hybrids were identified using a base index value (Table 5). Significant (P < 0.05) differences were observed among the genotypic performance for all traits evaluated. Entry-5 × Tzi8 ranked highest with a score of 5.89 and TZEEI-15 × MP705 being the least within the top 25 hybrids, with a score of -12.9. It must be noted that some of the hybrid checks outperformed treatments based on the base index rankings (Table 5). Note. SED = standard error of difference; (ppb) = parts per billion.

Relative Contributions of Combining Ability Effects
Heritability estimates of traits largely depend on the relative contributions of all genetic effects compared to non-genetic effects, and narrow-sense heritability specifically depends on additive vs. non-additive genetic effects. The relative proportion and importance of GCA effects was determined as the ratio of GCA effects to total genetic effects using the sum of squares. A close to one means greater predictability based on GCA and a better ability of those plants to form hybrids with many other lines. In contrast, more SCA indicates only specific hybrid combinations will be successful (Baker, 1978). The total contribution of GCA [GCA m plus GCA f )] to the total variation among hybrids across the six environments varied from 65% for husk cover to 96% for days to 50% pollen and silking. In general, high GCA for genotypic sum of squares were observed for most of the traits evaluated, particularly for days to 50% silking and pollen, grain yield, leaf blight and maize streak virus disease resistance. For grain yield, it was evident that GCA sum of squares for male and female were relatively similar although the contribution from the female was slightly higher (Figures 1a and 1b). The general combining ability sum of squares for the female was twice that of the male for traits such as anthesis silking interval and three times more for days to 50% silking and pollen, stay green, leaf blight and maize streak virus disease resistance (Figures 1a and b). Meanwhile, traits such as cob aspect, plant aspect, leaf rust and ear rot were dominated by higher GCA for genotypic sum of squares for males (Figures 1a and 1b).
On the whole, the genotypic sum of squares for SCA of all traits was lower than GCA, and SCA varied from as low as 4% for days to 50% pollen and silking to a maximum of ~73% for aflatoxin accumulation resistance Mid-parent heterosis for grain yield of best performing hybrids varied from 325.41% for ENTRY-85 × CML158Q to 691.27% for TZEEI-6 × CML11 with a mean of 464.08. Best-parent heterosis also ranged from a minimum of 233.93 to a maximum of 670.37% for ENTRY-85 × CML158Q and TZEEI-6 × CML11, respectively. Forty percent (40%) of the hybrids produced a higher best-parent heterosis than the mean of 376.32 (Table 6). In general, the high levels of heterosis observed gives an indication of the extent of genetic diversity existing among the genotypes that can be exploited for breeding gain. Additional analysis performed to determine the relationships between aflatoxin accumulation and other agronomic traits revealed a range of weak to moderately strong correlations (Table 7). Aflatoxin accumulation levels were found to correlate negatively and significant (P < 0.05) with grain yield, cob aspect and plant aspect, while it correlated positively but not significant with leaf blight, ear rot and insect damage (Table 7).

Discussion
The presence of significant differential environment effects on majority of the traits measured in the current study indicates that the growing environment of germplasms will affect how traits are expressed. Aflatoxin contaminations happen when there are environmental situations which affect genotypes especially if these genotypes are susceptible. Significant genotypic difference shows that there was adequate genetic variability among the germplasm to allow for genetic enhancement through selection (Badu-Apraku et al., 2011b;Okoth et al., 2017). The significant GCA and SCA detected for aflatoxin accumulation resistance and most of the other traits represents an opportunity for improvement via selection and offer opportunity for rapid improvement for hybrid variety development.
The significant GCA × environment interaction observed in this study for aflatoxin accumulation resistance and several other traits has also been reported by Badu-Apraku et al. (2011), Mageto et al. (2017) who highlighted the need to evaluate hybrids and their parental lines in several contrasting environments in order to identify the most promising hybrid combinations for specific traits for each environment. This suggests that more hybrids, each adapted to its own environment, should be offered to farmers, along with recommendations on where each hybrid is expected to perform, rather than a single high yielding hybrid that is expected to do well in all environments, and is unlikely to be found. Farmer variety trials could increase information and adoption of hybrids and could be the next step in the process of creating and disseminating high yielding, aflatoxin accumulation resistant lines for Ghana.
Mid-parent and high-parent heterosis for aflatoxin accumulation resistance among the selected hybrids were -36.39 and -32.40 while that for grain yield were 464.08 and 376.32 respectively. This observation was similar in magnitude to that reported by Saleh et al. (2002) in a study involving tropical single cross, double cross and three-way cross maize hybrids but differed in magnitude from the findings of Betran et al. (2003) who reported average high parent heterosis more than 1225%. In addition, heritability estimates for grain yield was higher (53.93%) than for all other traits except aflatoxin accumulation resistance (Table 2). In addition, several traits were negatively correlated to both grain yield and aflatoxin accumulation resistance, which will complicate selection of superior parental lines. Use of a base index for selecting the best performing genotypes may allow gain from selection to be achieved over time for all traits (Menkir et al., 2006;Badu-Apraku et al., 2011a). The use of a base index for identification of parents in this work has been very useful in identifying which of the hybrids combined aflatoxin accumulation resistance and good agronomic characteristics with high yields, thus identifying potential parents of good performing hybrids for consumers to manage the issue of aflatoxin contamination in maize within a limited time.
Nine hybrids were identified across the six environments with consistently good agronomic traits and reduced aflatoxin accumulation based on the base index. They include TZEEI-15 × CML343, TZEEI-6 × CML247, ENTRY-85 × Tzi8, ENTRY-5 × TZI8, TZEEI-6 × CML11, ENTRY-85 × CML-158Q, ENTRY-5 × CML-287, TZEEI-6 × CML343 and TZEEI-15 × MP715. In addition, while some of the local hybrid checks performed better than some of the crosses, most of the hybrids outperformed the checks (Table 5). The range of aflatoxin accumulation resistance measured in the selected hybrids is comparable to the range recorded by William and Windham (2015) and Brown et al. (2016), but the upper range for the present study is lower than the previous studies. This is quite significant, because the pathogen pressure applied in this study was very high. Natural infection by A. flavus would have been lower, and aflatoxin levels correspondingly lower too. Another objective of this study was the GCA effect of individual inbreds on control of aflatoxin resistance and other agronomic traits. Mp715 and Tzi8 consistently produced negative GCA results for aflatoxin accumulation resistance, while Ki3, CML247 and Tzi8 produced positive GCA for grain yield. Generally, most of the exotic inbreds used as males consistently produced negative GCA for aflatoxin accumulation resistance, confirming their initial resistance, and their ability to pass that resistance on to hybrids for controlling aflatoxin accumulation. Mid-and high-parent heterosis levels for aflatoxin accumulation resistance observed in this study indicate hybrids ENTRY-5 × Ki3 and TZEEI-15x MP715 were simultaneously good for both reduced aflatoxin and high grain yield.

Conclusions
This study evaluated 96 hybrids under different environments to provide information on the heterosis and mode of gene action conditioning aflatoxin resistance and other important agronomic traits. The study identified some high yielding and aflatoxin resistant hybrids across the environments. Inbred lines with consistent positive GCA for grain yield and negative GCA for aflatoxin accumulation control were identified and can be used for aflatoxin resistance breeding. Hybrids with good performance in specific environments can be recommended for on -farm trials by farmers in those environments. Heritability estimates realised for most traits were moderate to high, suggesting potential gains can be made in trait levels via selection. Most of the hybrids measured out-performed the leading checks.