Introduction

In Angola, whose population is estimated at 28,000,000, about 60% of the population (~ 16,800,000 people) consume an average of about 0.5 kg of maize flour per day (MINAGRIF 2018). Maize production in recent years, estimated at 2.5 million tons of grain per year, is insufficient, when consumption is closer to 5.0 million tons. To meet this demand, for human consumption only, Angola will have to produce at least 4 million tons of maize grain annually, which, at an average of 4 t ha−1, needs about 988,000 ha of land. However, maize productivity in the country is very low, averaging 0.7–1 t ha−1 (MINAGRIF 2020) and this scenario requires maize breeders to come up with tangible solutions that can promote food security at household and national level.

Maize is cultivated across the whole country, but production is common in the high and mid-altitude geographic zones, which receive abundant annual rainfall (800–1200 mm). These high rainfalls are associated with leaching as well as erosion of mineralized and applied nutrients and soil acidification, known to constrain maize productivity (Jandong et al. 2011). Soil acidity is a huge problem, limiting crop production on 30–40% of the world’s arable land, and causing yield losses of up to 70% of the world’s potentially arable land (Haug 1983) and up to 60% in many African countries (Dewi-Hayati et al. 2014; Tandzi et al. 2015, 2018), including Angola.

In order to minimize effects of soil acidity, farmers can use lime (Tandzi et al. 2015) and other agronomic technologies, including fallowing (Tonye et al. 1997; Mwangi et al. 2002). Liming is known to be expensive to resource poor farmers, who constitute the majority of food producers in African countries (Thé et al. 2005). Use of this technology requires a lot of skill, which the majority of farmers do not have, and is not an economically and environmentally sustainable solution (Tandzi et al. 2015; Ndeke and Tembo 2019). Additionally, liming is only effective in the topsoil and does not neutralize acidity in the subsoil, where it poses a severe problem to developing roots (Sierra et al. 2006). On the other hand, agronomic measures, for instance fallowing, cannot be viable options, especially when the focus is on the future, where land size suitable for agriculture is expected to dwindle (Tonye et al. 1997). Developing crop varieties adapted to acid soil conditions remains the most viable and sustainable means to improve maize productivity.

Genetic variability for tolerance to acid soil exists among maize genotypes, which can be exploited in developing high-yielding, acid-tolerant maize genotypes (Tandzi et al. 2018). For instance, a study in Cameroon showed that using adapted local inbred lines and crossing them with acid tolerant inbred lines from the International Maize and Wheat Improvement Centre (CIMMYT)—Colombia, could minimize grain yield losses caused by soil acidity (Tandzi et al. 2015; Petmi et al. 2016). Elsewhere, evaluation of maize single-cross hybrids on acidic soils in Indonesia showed that several hybrids, which were progeny of crosses between acid soil tolerant or moderately acid soil-tolerant inbred lines, yielded reasonably high (Dewi-Hayati et al. 2014). Although evidence suggests that maize cultivars bred to tolerate soil acidity can produce meaningful yields under acid conditions, this technology has not yet been explored in Angola.

It is also now widely believed that effective breeding programs are those that have well defined aims on a particular product profile. Looking at Angola, the country can be divided into two product profiles based on preferences of kernel colour of maize. Almost half of the population prefers yellow maize, while the other half prefers white kernel maize (MINAGRIF 2018). In this regard, breeding programs should also be designed to cater for the needs of these two consumer bases. With this in mind, exotic yellow acid donor inbred lines were sourced from CIMMYT—Colombia, which needed to be assessed for their combining ability with germplasm adapted to the mid-altitude climatic conditions, under acid and non-acid conditions in Angola. The majority of countries in eastern and southern Africa are classified in this mega-environment, hence germplasm selected anywhere within this climatic zone can potentially do well in any other location within the zone (Chapman et al. 2003). Additionally, stability of the corresponding single-cross hybrids between the yellow acid donor lines and the elite yellow mid-altitude adapted lines under acid and non-acid conditions remains unexplored. Assessing the combining ability (both the GCA and the SCA) between these two groups of parental lines helps in identifying acid tolerant donor lines that can be potential sources of acid tolerance genes in breeding programs in Angola and elsewhere within the mid-altitude climatic zone. Also, potential crosses that can be used as pedigree starting populations for developing new inbred lines adapted to soil acidity and other stress factors common in Angola, can be identified. On the other hand, stability analysis will help in assessing and identifying crosses that can be targeted for commercial release. Therefore, the specific objectives of this study were: (i) to identify acid tolerance yellow maize donor lines that can potentially improve adaptation of mid-altitude adapted germplasm under acid and non-acid conditions; and, (ii) to identify high yielding local (CIMMYT—Zimbabwe) × exotic (CIMMYT—Colombia) hybrids with stable grain yield performance under acid and non-acid conditions. It was hypothesised that the exotic yellow maize donor lines from CIMMYT—Colombia can contribute to improvement of maize productivity under both acid and non-acid soil conditions in Angola.

Materials and methods

Germplasm description and F1 formation

Ten elite yellow lines adapted to the mid-altitude climatic conditions were sourced from CIMMYT—Zimbabwe. Four were classified into heterotic group A and the other six into heterotic group B. These were crossed with three heterotic group A and one heterotic group B yellow donor lines sourced from the CIMMYT—Colombia breeding program (Table 1), using a line × tester (L × T) design. Crossing nurseries were planted in Muzarabani in Zimbabwe (latitude 16°19′60 S, longitude 31°10′0 E) during the winter season of the year 2014. Hand pollinations were performed. The L × T nursery yielded a total of 40 crosses, but four of them were discarded because they had insufficient seed for at least nine locations (Table 2).

Table 1 Mid-altitude adapted yellow inbred lines used as female parents in a line x tester design with the exotic acid tolerance donor lines as testers
Table 2 Line × tester crosses developed in a nursery established in Muzarabani during the 2014 winter season in Zimbabwe

Hybrid evaluation and description of the sites

The 36 crosses which had sufficient seed were evaluated alongside six acid tolerant commercial check hybrids at nine locations in Angola and Zimbabwe (Table 3), making up a total of 42 hybrids per trial. Trials were established during the 2014–16 cropping seasons under acid and non-acid soil conditions. Sites with sandy soils, but with a historic record of receiving normal to above normal rains, and those known to be acid (such as low P and low pH) were chosen as acid soil sites. Optimal and random stress sites were classified as non-acid. Briefly, low P sites were those where non-leguminous crops were repeatedly grown without use of phosphate fertilizers, and crop residues were removed from the field immediately after harvesting. Optimal sites were those where the crop was subjected to all the recommended agronomic measures including fertilization and supplemental irrigation during water-deficit periods. These sites also occurred naturally in environments where the climatic conditions are suitable for maize production. The random stress sites were those where chances of mid-season drought were close to 100% during the rainy season and if drought occurred, no supplemental irrigation was given. These sites represented the real conditions to which maize is subjected in the small-scale farming sector, where most of the crop production is done in many countries in Africa.

Table 3 Climatic, geographical and soil chemical characteristics of the top 30 cm of the soil profile at nine sites in Angola and Zimbabwe, used to evaluate 42 hybrids during the 2014–16 cropping seasons

Soil sampling was done at the beginning of the experiments. Samples were collected using the Horneck et al. (2011) method; from the top 30 cm of the soil profile at all the nine sites. Six soil samples were collected per site and were bulked. A representative sub-sample from the bulked four samples collected in Angola was taken and submitted for chemical analysis at the Chianga Experimental Station soil laboratory of the Agricultural Research Institute (IIA). Similarly, representative samples from five bulk samples gathered in Zimbabwe were analysed. Samples were analysed for pH and available P, K, Ca and Mg (Table 3).

Experimental design and trial management

The 42 yellow hybrids were laid out at each of the nine sites in Angola and Zimbabwe using an alpha (0, 1) lattice design (Patterson and Williams 1976) with two replications at each site. Each replication accommodated seven incomplete blocks, with each incomplete block containing six hybrids. Each hybrid was planted in a single row plot of 4 m length, having a uniform inter- and intra-row spacing of 0.75 m and 0.25 m, respectively. Two seeds were hand-planted on each hill, and the trials were later thinned to one plant on each planting station, 3–5 weeks after crop emergence, in order to have an optimum plant population of 53,333 plants per hectare. Border rows were planted to avoid border effects. A total of 400 kg ha−1 of Compound D (N12 P24 K12) was applied as basal dressing and 250 kg ha−1 of urea (NH2; N = 46%) was split-applied as top-dressing fertilizer at all sites in Angola. In Zimbabwe, the same quantity of 400 kg ha−1 of compound D (N7 P14 K7) was applied as basal dressing at most of the sites, except for the low P site. Ammonium nitrate (N = 34.5%) was split-applied as top-dressing at a rate of 400 kg ha−1 at the sites used in Zimbabwe.

Data collection and exploitation

Data collection followed the CIMMYT (1985) standard procedures. Grain yield (GYD), was measured on a whole plot basis. Shelled grain weight per plot was adjusted to 12.5% grain moisture and converted to ton per hectare using the following formula:

$${\text{GYD }}\left( {{\text{t ha}}^{{ - {1}}} } \right) \, = \, \left[ {{\text{Grain weight }}\left( {{\text{kg plot}}^{{ - {1}}} } \right) \, \times { 1}0 \, \times \left( {{1}00 - {\text{MC}}} \right)/\left( {{1}00 - {15}} \right)/\left( {\text{plot area}} \right)} \right]$$
(1)

where MC = grain moisture content; and plot area = row length × 0.75 (4 × 0.75 = 3 m).

Across-site ANOVA was performed using the ‘aov’ function in the Agricolae R package. The treatments (crosses/hybrids) were considered as fixed. The model for combined ANOVA was:

$${\text{Y}}_{{{\text{ij}}\left( {\text{k}} \right)({\text{l}})}} = {\text{ b}}_{{\text{j}}} \left( {{\text{r}}_{{\text{k}}} } \right)\left( {{\text{E}}_{{\text{l}}} } \right) \, + {\text{ r}}_{{\text{k}}} \left( {{\text{E}}_{{\text{l}}} } \right) \, + {\text{ g}}_{{\text{i}}} + {\text{ E}}_{{\text{l}}} + {\text{ gE}}_{{({\text{il}})}} + {\text{ e}}_{{{\text{ij}}({\text{k}})}}$$
(2)

where Yij(k)(l) is the response of the ith genotype in the jth incomplete block nested within the kth replication nested in the lth environment; bjr(k)E(l) is the effect of the jth incomplete block nested in the kth replication also nested in the lth environment and j = 1, 2, 3, 4; rk(El) is the effect of the kth replication nested in the lth environment and k = 1, 2, 3; gi is the effect of the ith genotype and I = 1, 2, 3,…10; El is the effect of the lth environment and l = 1, 2, 3,…6; gE(il) is the interaction effect of the ith genotype and the lth environment; and eij(k)(l) is the random error term.

Broad-sense heritability estimates, best linear unbiased estimators (BLUEs), as well as genetic correlations between grain yield and the other agronomic traits were calculated using the Multi-Environment Trial Analysis with R (META-R) version 5.0 (Alvarado et al. 2015). Mean comparisons were performed using fisher’s protected least significance difference (LSD) (Little and Hills 1978) at 5% significance level. L × T crosses with superior grain yield performance, but harbouring other desirable agronomic traits that are of importance in the sub-tropical regions, were visualised on scatter plots using the ‘ggplot’ function in the ggplot2 R package (Wickham 2016). Stability of the top performing crosses, selected within the A- and B-heterotic groups, was assessed using ranking, and Genotype–Genotype × Environment (GGE) Biplot in the GenStat software, 17th edition (Payne et al. 2009).

Preliminary data checking and individual site ANOVA were performed using CIMMYT Fieldbook software (Bänziger and Vivek 2007). L × T analysis was performed for grain yield across the acid and non-acid sites, as well as for acid and non-acid sites, separately. The L × T procedures in the R software v3.0.1 (RDevelopmentCoreTeam 2013), embedded in the CIMMYT Fieldbook software were followed. Briefly, the procedure uses functions in the lme4 (Bates et al. 2015, 2019), lattice (Deepayan 2018) and matrix R packages (R Development Core Team, 2013), to estimate GCA and SCA effects for lines and testers. The model for the combined sites L × T was as follows:

$${\text{Y}}_{{{\text{ijkp}}}} = \, \mu \, + {\text{ g}}_{{\text{i}}} + {\text{ g}}_{{\text{j}}} + {\text{ s}}_{{{\text{ij}}}} + {\text{ E}}_{{\text{p}}} + {\text{ r}}_{{\text{k}}} \left( {{\text{E}}_{{\text{p}}} } \right) \, + \, \left( {{\text{gE}}} \right)_{{{\text{ip}}}} + \, \left( {{\text{gE}}} \right)_{{{\text{jp}}}} + \, \left( {{\text{sE}}} \right)_{{{\text{ijq}}}} + {\text{ e}}_{{{\text{ijkp}}}}$$
(3)

where, i = 1, 2, 3, …10, j = 1, 2, 3,4, k = 1, 2, and Yijkp represented the value of the progeny of a mating of the ith CIMMYT—Zimbabwe elite yellow inbred line (i.e., line), the jth CIMMYT—Colombia yellow acid donor inbred line (i.e. tester), in the kth replication, and in the pth environment (site). The µ represents the grand mean, gi is the GCA effect common to all progeny of the ith line, gj is the GCA effect common to all progeny of the jth tester, sij is the SCA effect specific to the progeny of mating the ith line and the jth tester, Ep is the average effect of the pth environment, rk (Ep) is the effect of the kth replication that was nested within the pth 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 eijkp is the random experimental error. This model was adopted from Lee et al. (2005).

The BLUEs were calculated following the procedures of Puntanen and Styan (2011) and the broad-sense heritability (H2) estimates were calculated using the Multi-Environment Trial Analysis with R (META-R) software v5.0 (Alvarado et al. 2015). The following model was used to calculate H2:

$${\text{H}}^{2 }=\left(\frac{{\upsigma }^{2}\text{g}}{\frac{{\upsigma }^{2}\text{g}}{\text{re}}+\frac{{\upsigma }^{2}\text{ge}}{\text{e}}+{\upsigma }^{2}\text{p }}\right)*100$$
(4)

where; \(\sigma\)2g is genotypic variance, \(\sigma\)2ge is genotype × environment variance, \(\sigma\)2p is phenotypic variance, e represents sites and r represents the replications.

Mean comparisons were performed using the Fisher’s Protected LSD (Little and Hills 1978) at 5% significance level. To identify the best yielding and stable crosses across the acid sites and across the non-acid sites, a stability coefficient method known as superiority performance, which calculates cultivar superiority indices according to Lin and Binns (1988), was performed in GenStat software, 17th edition (Payne et al. 2009). GCA and SCA of CIMMYT Zimbabwe and CIMMYT Colombia inbred lines involved in the highest grain yielding crosses under acid, non-acid and across acid and non-acid sites, were visualized using a scatter plot. The most stable, but high yielding L × T crosses under acid and non-acid soil conditions were also visualized using a scatter plot. The scatter plots were graphed using the ‘ggplot’ function in the ggplot2 R package (Wickham 2016).

Results

Hybrid grain yield performance under acid and non-acid soil conditions

Significant genotype effects (P < 0.05) on GYD were observed at CIMMYT—Zimbabwe, Chibero and Chianga (optimal management) and Alto-Kapaka (random stress). Genotype effects were also significant for anthesis date (AD), anthesis-silking interval (ASI) and number of ears per plant (EPP) at CIMMYT—Zimbabwe (optimal and low P management) and Chianga (low P management). Heritability values of ≥ 30% for GYD were observed at most sites (Table 4).

Table 4 Individual site analysis of variance for grain yield, anthesis date, anthesis-silking interval and number of ears per plant performance of the CIMMYT—Zimbabwe elite lines × CIMMYT—Colombia acid tolerance donor line hybrids evaluated across nine sites in Zimbabwe and Angola during the 2014–16 cropping seasons

Significant line effects (p < 0.05) were observed on GYD, AD and EPP and this trend was consistent across all soil conditions. Tester effects on AD and EPP were significant across acid and non-acid soil conditions. Across all soil conditions, tester effects were significant on AD, ASI and EPP. Line × tester effects on GYD were significant across non-acid and across all soil conditions. Broad- and narrow-sense heritability estimates for GYD were > 50% across all soil conditions (Table 5).

Table 5 Grain yield, anthesis date, anthesis-silking interval and number of ears per plant combined analysis of variance of CIMMYT—Zimbabwe elite lines CIMMYT—Colombia acid tolerance donor line crosses, evaluated under acid and non-acid soil conditions in Angola and Zimbabwe

3.2 Grain yield performance of the CIMMYT—Zimbabwe elite lines × CIMMYT—Colombia acid tolerance donor line F1s in comparison with commercial checks under acid and non-acid soil conditions

Comparing the five highest yielding experimental hybrids with the highest yielding commercial check hybrids showed the potential of the CIMMYT—Zimbabwe elite lines and the CIMMYT—Colombia acid tolerance donor lines to promote maize productivity under acid and non-acid soil conditions in Angola (Table 6; Fig. 1). The best five experimental hybrids yielded more than the five highest yielding commercial checks under both the acid (average GYDExperimental = 4.81 t ha−1 > GYDChecks = 4.61 t ha−1) and the non-acid soil conditions (average GYDExperimental = 6.61 t ha−1 > GYDChecks = 6.46 t ha−1). The five experimental hybrids also flowered earlier than the highest yielding commercial checks under both acid (average ADExperimental = 70 days < ADChecks = 71.8 days) and non-acid (average ADExperimental = 66.69 days < ADChecks = 68.86 days) conditions. Similar trends were observed for the hybrids selected across the acid and non-acid sites (average GYDExperimental = 5.40 t ha−1 > GYDChecks = 5.31 t ha−1 and average ADExperimental = 68.92 days < ADChecks = 70.51 days). CIMMYT—Zimbabwe elite lines, ZY2 (GCAacid = 0.17 t ha−1; GCAnon-acid = 0.64 t ha−1; GCAacid+non-acid = 0.40 t ha−1), ZY1 (GCAacid = 0.02 t ha−1; GCAnon-acid = 0.64 t ha−1; GCAacid+non-acid = 0.32 t ha−1) and ZY3 (GCAacid = 0.63 t ha−1; GCAnon-acid = 0.68 t ha−1; GCAacid+non-acid = 0.66 t ha−1), were involved in the three highest yielding experimental hybrids under all conditions. The inbred line ZY3 consistently showed the highest positive GCA effects for GYD and was involved as a parent in more than one cross among the five highest yielding experimental hybrids under all conditions (Table 6; Fig. 1) and was ranked the best line under acid conditions and across acid and non-acid conditions (Table S1).

Table 6 Grain yield, anthesis date, anthesis-silking interval and number of ears per plant performance of the top five crosses and their lines and testers compared to the top five commercial check hybrids under acid and non-acid soil conditions and across the two conditions
Fig. 1
figure 1

Best linear unbiased estimates for combining ability for grain yield performance (t ha−1) of the top five crosses under acid and non-acid soil conditions and across the two stress conditions. (Color figure online)

The CIMMYT—Colombia donor lines CY3 (GCAacid = 0.143 t ha−1; GCAnon-acid = −0.032 t ha−1; GCAacid+non-acid = 0.019 t ha−1) and CY1 (GCAacid = 0.17 t ha−1; GCAnon-acid = 0.053 t ha−1; GCAacid+non-acid = 0.114 t ha−1) were parents in the two highest grain yielding hybrids under all soil conditions (Table 6). Interestingly, the yellow acid donor line, CY1, was a parent in three of the five highest yielding hybrids under the acid soils and across the acid and non-acid conditions (Table 6; Fig. 1) and was also ranked the best tester under the acid conditions and across all soil conditions (Table S2). The CIMMYT—Zimbabwe elite inbred line ZY8 was ranked the highest under non-acid conditions, whilst the CIMMYT—Colombia acid donor line, CY4, was ranked the best tester under non-acid conditions (Table S1 and S2).

The best specific combiners for GYD performance under acid soil conditions were ZY10 × CY3 (Entry 35; SCA = 0.802 t ha−1, BLUEGYD = 3.78 t ha−1), ZY1 × CY1 (Entry 16; SCA = 0.65 t ha−1, BLUEGYD = 4.86 t ha−1) and ZY4 × CY4 (Entry 25; SCA = 0.55 t ha−1, BLUEGYD = 4.32 t ha−1). Under non-acid soil conditions, the best yellow specific combiners for GYD performance were ZY6 × CY3 (Entry 36; SCA = 3.29 t ha−1, BLUEGYD = 4.15 t ha−1), ZY2 × CY1 (Entry 17; SCA = 1.44 t ha−1, BLUEGYD = 7.01 t ha−1) and ZY10 × CY3 (Entry 35; SCA = 1.05 t ha−1, BLUEGYD = 4.76 t ha−1). Across acid and non-acid conditions, the best specific combiners for GYD were ZY6 × CY3 (Entry 36; SCA = 1.00 t ha−1, BLUEGYD = 4.23 t ha−1), ZY10 × CY3 (Entry 35; SCA = 0.93 t ha−1, BLUEGYD = 4.19 t ha−1) and ZY7 × CY1 (Entry 12; SCA = 0.59 t ha−1, BLUEGYD = 5.27 t ha−1) (Tables S3–S5).

Grain yield stability of the five highest grain yielding experimental hybrids and checks under acid and non-acid conditions

A check hybrid identified as Check 5 (Entry 41; BLUE_GYD = 5.85 t ha−1) was ranked as the highest yielding and stable genotype under acid conditions (Table 7; Fig. 2). Experimental hybrids CH142442 (Cross = ZY7 × CY1; GYD = 4.74 t ha−1), CH142464 (Cross = ZY2 × CY3; GYD = 5.05 tha−1), and CH142444 (Cross = ZY3 × CY1; GYD = 4.65 t ha−1) were more stable and high yielding than most of the other checks. Under non-acid conditions, the experimental genotype CH142447 (Cross = ZY2 × CY1; GYD = 7.01 t ha−1) was the most stable and was slightly outperformed by the genotype Check 3 (Entry 38; GYD = 7.15 t ha−1), in terms of GYD performance (Fig. 2; Table 7).

Table 7 Mean grain yield and cultivar stability indices of the top five crosses and commercial check hybrids under acid and non-acid soil conditions
Fig. 2
figure 2

Cultivar superiority indices (CSI) of the top five crosses and commercial check hybrids under acid and non-acid soil conditions. (Color figure online)

Discussion

Low pH is the most common abiotic stress constraining maize productivity in Angola, where both yellow and white maize is essential for food and feed requirements. Here, four yellow acid soil tolerance donor lines from CIMMYT—Colombia were crossed with ten yellow elite lines adapted to the mid-altitude climatic conditions from CIMMYT—Zimbabwe, to identify donor lines, which can be potential sources of acid tolerance genes in breeding programs in Angola. L × T analysis revealed significant (p < 0.05) line and line × tester effects on GYD across the acid and non-acid environments. Combining ability analysis identified the CIMMYT—Colombia acid tolerance donor lines CY3 and CY1 as the most ideal sources of acid tolerance genes in Angolan breeding programs as they showed the highest positive GCA effects on grain yield. In addition, crosses involving these two donor lines were also identified to be highest in yielding ability and stability under both acid and non-acid conditions. The crosses between the donor and the CIMMYT—Zimbabwe elite lines also flowered earlier than the commercial hybrids.

The significant line as well as line × tester effects on GYD is consistent with previous studies by Yadav et al. (2002), Rafique et al. (2004), Seanki et al. (2005), Akbar et al. (2006), Dagne et al. (2007), Nesir (2007), Ali et al. (2010), Vashistha et al. (2013), Abdel Moneam et al. (2014), Sudika et al. (2015) and Andayani et al. (2018). The findings suggest sufficient genetic variability in the studied germplasm for grain yield and other related agronomic traits such as anthesis dates, which can be exploited for selecting productive and adaptive maize germplasm in Angola.

Comparing the highest yielding five experimental hybrids with the highest yielding commercial check hybrids under acid, non-acid and across all soil conditions, showed that the top five yielding experimental hybrids had high average mean GYD compared to the top five yielding check hybrids and they also had lower average flowering dates compared to the checks (Table 6). This translated to good yield potential of the inbred lines used in hybrid combinations, which can be exploited in high yielding hybrid formation in different environments in Angola. Early flowering can be beneficial especially under the current climatic scenarios where seasons are becoming shorter. Usually, maize that flowers early also matures early and early-maturing varieties cope better with drought stress which predominantly occurs towards the end of the cropping season in sub-Saharan Africa. The experimental hybrid CH142464 (ZY2 × CY3) with 5.05 t ha−1 was the best hybrid combination under acid soil conditions, while under non-acid soil conditions, hybrid CH142447 (ZY2 × CY1) with 7.01 t ha−1 showed superiority amongst the top five yielding experimental hybrids. When combining the acid and non-acid soil conditions, the experimental hybrid CH142461 (ZY3 × CY3) with 5.54 t ha−1 was the best of the top five yielding experimental hybrids. These highest yielding crosses have to be assessed further for adaptability as well as seed productivity in order to generate more information that can support their release as new varieties in Angola.

The CIMMYT—Zimbabwe elite lines involved in the highest yielding hybrids had positive GCA effects for GYD under acid and non-acid soil conditions. According to Egesel et al. (2003), Bhatnagar et al. (2004), Fan et al. (2007) and Bello and Olaoye (2009), positive GCA effects for grain yield indicates high genetic variability among the lines. The CIMMYT—Zimbabwe elite lines ZY1, ZY2 and ZY3, were involved in the three highest yielding experimental hybrids and line ZY3 with the highest positive GCA effect for GYD, was the best line and involved as parent in more than one cross among the top five yielding experimental hybrids. Having significantly positive GCA line effects for grain yield and for at least two yield component traits (Fan et al. 2007), these lines could be used directly in yellow maize hybrid development programs in Angola.

On the other hand, the CIMMYT—Colombia acid tolerance donor lines (testers) CY1 and CY3 were identified as the best lines due to their parentage in the two highest yielding experimental hybrids, and tester CY1 ranked the highest under acid, non-acid and across acid and non-acid soil and was a parent in three of the five highest yielding experimental hybrids (Table 6; Fig. 1). This indicated that these two testers had good potential and could be used in high yielding hybrid formation under acid soil conditions in Angola.

The CIMMYT—Zimbabwe elite inbred line, ZY8 was the best line under non-acid conditions, whilst the CIMMYT—Colombia yellow acid donor line, CY4, was the best tester under non-acid conditions. These two inbred lines can be used specifically in high yielding hybrid formation under non-acid soil conditions in Angola.

Under acid soil conditions, hybrids ZY10 × CY3, ZY1 × CY1 and ZY4 × CY4 were identified as the best combiners. Exploiting their potential, these inbred lines and their combinations could significantly contribute to maize breeding programs for high yield and tolerance to acid soils in Angola. Hybrids ZY6 × CY3, ZY2 × CY1 and ZY10 × CY3 were the best combiners under non-acid soil conditions. These crosses are suitable for developing high yielding yellow maize hybrids for other environments than acid soil. Finally, across acid and non-acid soil conditions, two of the three crosses selected in non-acid soil (ZY6 × CY3 and ZY10 × CY3) had the same tester (CY3). Of these, cross ZY10 × CY3 was the best hybrid in this study, because it was top yielding under acid and non-acid soil conditions. CY3 was the best tester, which was a parent in the best hybrids more than once, followed by CY1, ZY6 and ZY10 which were the best lines for the same reason.

Assessing genotype grain yield stability is one of the key attributes for variety recommendation. High yield stability usually refers to a genotype’s ability to maintain a constant yield across different environments (Falconer 1990; Dyke et al. 1995). Cultivar superiority indices were used in this study to identify the most stable of the five highest yielding experimental hybrids against the five highest grain yielding checks.

Under acid soil conditions, except for hybrid check 5 which was ranked as the highest yielding and stable genotype, the experimental hybrids CH142442 (ZY7 × CY1), CH142464 (ZY2 × CY3) and CH142444 (ZY3 × CY1) were more stable and high yielding than the other checks. Under non-acid conditions, hybrid CH142447 (ZY2 × CY1) was the most stable genotype and was (marginally) outperformed by check 3 in terms of GYD. The highest yielding and stable experimental hybrids identified in this study could be tested in large-scale trials across environments for adaptation in diverse agro-ecological regions. Improved hybrids can stabilize the production level of the maize crop and it could improve the national production and productivity since in Angola, production and productivity levels are very low.

In conclusion, the data demonstrated the potential of tropically-adapted exotic yellow maize acid tolerance donor lines to contribute in increasing productivity and adaptability of sub-tropical yellow maize under low-pH as well as optimal conditions. The donor lines can be used directly as parental lines in hybrid development or the acid tolerance genes which they harbor can either be integrated or introgressed in the sub-tropical maize populations in order to develop new inbred lines adapted to sub-tropical climatic conditions. These inbred lines, in the long-run, will be very useful to develop hybrids, synthetics and open pollinated varieties productive and adapted in sub-Saharan Africa.