Combining ability of extra‐early biofortified maize inbreds under Striga infestation and low soil nitrogen

Abstract Striga hermonthica (Del.) Benth parasitism, low soil N, and nutritional deficiencies of normal‐endosperm maize (Zea mays L.) threaten maize yield and exacerbate nutritional problems in sub‐Sahara Africa (SSA). This study was conducted (a) to evaluate genetic variation among extra‐early maturing maize hybrids with provitamin A and quality protein characteristics, (b) to investigate gene action governing the inheritance of Striga resistance, grain yield, low N tolerance, and other measured traits under low‐N, high‐N, and Striga‐infested environments, and (c) to identify hybrids with high yield and stability across environments. One hundred and fifty hybrids developed using North Carolina Design II were evaluated with six checks under low‐N, high‐N, and Striga‐infested environments in Nigeria. Mean squares for hybrids were highly significant (P < .01) for grain yield and other traits across environments. Only general combining ability (GCA) for female and/or male mean squares were significant for measured traits under low N. In addition to significant GCA effects for most traits, specific combining ability was significant (P < .05) for Striga emergence count under Striga infestation, and ear height and ears per plant under high N, indicating that additive and nonadditive genetic effects controlled the inheritance of few traits under Striga and high N, whereas additive genetic effect governed the inheritance of the traits under low N. Hybrids TZEEIORQ 55 × TZEEIORQ 26, TZEEIORQ 49 × TZEEIORQ 75, and TZEEIORQ 52 × TZEEIORQ 43 were high yielding and stable across environments and have potential for improving nutrition and maize yields in SSA.


INTRODUCTION
Maize (Zea mays L.) is an important cereal crop in West and Central Africa (WCA) as well as the dominant staple food and crop in eastern and southern Africa (Edmonds et al., 2009). The average yield of maize in sub-Sahara Africa (SSA) is 2.4 t ha −1 (FAOSTAT, 2017). This is considerably lower than the world average yield of 5.6 t ha −1 (FAOSTAT, 2017). Millions of people (especially the poor in rural areas) in SSA subsist on normal-endosperm maize, deficient in provitamin A (Safawo et al., 2010) and the basic amino acids tryptophan and lysine (Le, Chua, & Le, 2016), resulting in malnutrition and food insecurity in the subregion. The two-to more than fourfold projected increase in human population in SSA between 2010 and 2050 and the consequent increase in demand for maize and other major cereals may worsen the food insecurity problem in the region, unless urgent measures are taken (van Ittersum et al., 2016).
The parasitic weed, Striga, and low N are among the key stresses that constrain maize yield in SSA (Edmonds et al., 2009;Menkir, Franco, Adepoju, & Bossey, 2012). The biotic stress [Striga hermonthica (Del.) Benth] could cause 100% maize yield loss (Fajemisin, 2014), whereas the abiotic stress (low N) may reduce maize yield by close to 50% (Amegbor, Badu-Apraku, & Annor, 2017). Although there are different methods of Striga control (Emechebe et al., 2004;Teka, 2014), resistance and tolerance of host plants is the cheapest and most sustainable strategy to alleviate adverse effects of the weed on maize (Badu-Apraku et al., 2004). Maize hybrids and varieties that are resistant to Striga are the pivot of an integrated Striga control strategy (Kim & Adetimirin, 1997). Severe effects of Striga on maize are commonly observed in areas with poor soil fertility and low soil N (Sauerborn, Kranz, & Mercer-Quarshie, 2003). Nitrogen is needed for optimal growth and productivity of maize, but it is traditionally low in tropical soils (Abe, Adetimirin, Menkir, Moose, & Olaniyan, 2013;Betrán, Beck, Banziger, & Edmeades, 2003). Improvement in soil N through application of inorganic fertilizer during maize production in SSA is rarely done, and when carried out, application is at rates considerably lower than recommended rates due to prohibitive prices of fertilizer to resource-constrained farm families (Amegbor et al., 2017). In effect, maize is usually grown under N stress resulting in low yield of the crop. Although a few varieties and hybrids of maize with Striga resistance and tolerance, low N tolerance, and improved provitamin A content have been developed and commercialized in WCA (Badu-Apraku et al., 2016;Menkir, Maziya-Dixon, Mengesha, Rocheford, & Alamu, 2017), no maize hybrid with Striga resistance and tolerance, tolerance to low N, extra-earliness, and high provitamin A, lysine, and tryp-tophan contents is available. Such hybrids, if developed and commercialized, will help to jointly address Striga and low N constraints, as well as mitigate the adverse effects of vitamin A deficiency and quality protein malnutrition in WCA.
Combining ability is a crucial test that is carried out in hybrid breeding. It can provide information regarding the genetic effect controlling traits of inbreds (Machida, Derera, Tongoona, & Macrobert, 2010). The test helps in selection of promising lines that could serve as parents of hybrids in maize hybrid programs (Abera, Hussein, Derera, Worku, & Laing, 2016;Hallauer, Carena, & Miranda Filho, 2010). Different reports on gene action controlling Striga resistance and tolerance, grain yield, low N tolerance, and other traits of maize lines are available. Additive genetic effect was reported to be more important than nonadditive effect in modulating Striga resistance (Amegbor et al., 2017;Gethi & Smith, 2004), whereas nonadditive gene action played greater role in genetic control of Striga resistance trait (Akaogu et al., 2019;Kim, 1994). Also, dissimilar reports are available on the genetic effect controlling maize grain yield in low-N environments. Nonadditive genetic effect regulated grain yield under low N, whereas additive gene action controlled grain yield in high-N environments (Betrán et al., 2003;Makumbi, Betrán, Banziger, & Ribaut, 2011). In contrast, other authors reported that additive gene action controlled grain yield in low-N environments, whereas nonadditive gene action regulated yield in high-N environments (Below, Brandua, Lambert, & Teyker, 1997).
In order to breed extra-early maize hybrids with tolerance and resistance to many stresses and improved nutritional qualities in SSA, the International Institute of Tropical Agriculture (IITA) began a breeding program in 2011 to develop first-generation extra-early inbred lines from the provitamin A Striga-resistant quality protein maize variety 2009 TZEE-OR2 STR QPM. Seventy-six extra-early maturing provitamin A quality protein maize (PVA-QPM) inbreds were extracted from the population. Although limited information is available on the per se performance of the novel inbreds under Striga, low N, and heat and drought stress, no information is available on the performance of the inbreds in hybrid combinations evaluated in low-N, Striga, and high-N environments. Equally, the genetic effects regulating Striga resistance, low N tolerance, grain yield, and other measured agronomic characters of the inbreds are yet to be investigated. Extra-earlymaturing PVA-QPM hybrids that combine high yield and stability across low N and Striga with good performance under high N are currently not available in WCA. Therefore, this study sought (a) to examine genetic variation among single-cross extra-early maize hybrids possessing PVA-QPM characteristics, (b) to investigate genetic effects for Striga resistance, grain yield, low N tolerance, and other characters of extra-early PVA-QPM hybrids under low N, high N, and Striga conditions, and (c) to identify hybrids that combine high yield and stability across environments.

Genetic materials and mating design used
Thirty inbred lines were selected from the lines extracted from the extra-early-maturing PVA-QPM variety 2009 TZEE-OR2 STR QPM developed by IITA, Nigeria. Selection of the lines was based on their varying responses in Striga-infested and low-N environments. In addition, kernels with deep orange color and appropriate modification for QPM characteristics (using a light box), indicating the presence of opaque-2 recessive alleles (Krivanek, De Groote, Gunaratna, Diallo, & Friesen, 2007), were selected. Pedigree information and reactions of the 30 selected inbreds to Striga and low N are shown in Table 1. The 30 inbreds were categorized into six unique sets. Each set comprised five PVA-QPM inbreds. Of the five lines in each set, one was susceptible to Striga hermonthica and/or low N (Table 1). Five inbreds in each group were used as female parents in one set and crossed with another five lines as male parents, in a separate set, in the North Carolina Design II (NCD II) (Comstock & Robinson, 1948). One hundred and fifty extra-early PVA-QPM hybrids were generated. These hybrids and six Striga and low-N-tolerant extra-early maturing normal endosperm yellow hybrids

Field evaluations and management
The 150 NCD II crosses and the six hybrid checks were studied under low-and high-N conditions at Ile-Ife (7 • 28′ N, 4 • 33′ E; 244 m altitude; 1,350 mm annual rainfall) in 2016. Similarly, the hybrids were examined under low-and high-N environments at Mokwa (9 • 18′ N, 5 • 4′ E; 457 m altitude; 1,100 mm annual rainfall) in 2016 and 2017, and under high-N conditions at Abuja (9 • 15′ N, 7 • 20′ E; 300 m altitude; 1,700 mm annual rainfall) in 2017. These translate to three low-N and four high-N environments for evaluation of the hybrids and the checks. A 12 × 13 simple lattice design was used in all the evaluations. Continuous maize planting and removal of the stover after every harvest for many years were used to deplete the low-N fields of N. The total soil P, N, and K were quantified using colorimetric and Kjeldahl digestion method (Bremner & Mulvaney, 1982). At Ile-Ife, the soil had 0.84 g kg −1 of N, 2.05 mg kg −1 of P, and 0.36 cmol kg −1 of K, whereas the soil at Mokwa had 0.85 g kg −1 of N, 6.32 mg kg −1 of P, and 0.20 cmol kg −1 of K. Based on the results of the soil tests, urea was applied to bring the available N in the high-N plots to 90 kg N ha −1 , whereas the low-N plots were brought up to a total of 30 kg N ha −1 . Nitrogen application was done at 2 wk after planting (WAP) and 4 WAP in equalsplit doses. Whereas 15 kg N ha −1 was applied to the low-N field at 2 WAP, high-N plots received 45 kg N ha −1 from urea. Also, at 2 WAP, low-and high-N fields received 60 kg P ha −1 and 60 kg K ha −1 as single superphosphate (P 2 O 5 ) and muriate of potash (K 2 O), respectively. At 4 WAP, lowand high-N plots were appropriately top-dressed with urea to achieve the two N levels used in the study. Plots consisted of single rows, each 4 m in length. Spacing between rows was 0.75 m, and spacing within rows was 0.40 m. At planting, three seeds were sown per hole. Two WAP, seedlings were thinned to two plants per stand in order to achieve 66,666 plants ha −1 . Weeds were controlled using Primextra and Gramoxone at the rate of 5 L ha −1 each of atrazine (2-chloro-4-ethylamino-6-isopropylamino-1, 3, 5-triazine) and paraquat (N,N′-bipyridinium dichloride) as pre-and postemergence herbicides, respectively. The chemical weed control was augmented with manual weeding as the need arose. In addition, the 150 hybrids along with the six checks were assessed under artificial infestation with Striga hermonthica at Mokwa and Abuja in 2016 and 2017. Apart from the plot length, which was 3 m under Striga, the experimental design, number of replications, and interand intra-row spacing were the same as for low-N and high-N experiments. Prior to sowing of seeds, artificial Striga infestation of the field was carried out according to Kim (1991). Each planting hole was infested with 8.5 g of the sand-Striga mixture (containing ∼5,000 germinable Striga seeds) prior to planting (Badu-Apraku et al., 2016). Three seeds were sown per infested hole. At 2 WAP, seedlings were thinned to two plants per stand resulting in 66,666 plants ha −1 . Under Striga, fertilizer (15-15-15 N-P-K) application was carried out 4 WAP at the rate of 30 kg ha −1 for N, P 2 O 5 , and K 2 O. Weeds were removed by hand, except Striga plants. In all the experiments, fall armyworms (Spodoptera frugiperda) were controlled using Ampligo (a.i. 100 g L −1 chlorantraniliprole + 50 g L −1 lamda-cyhalothrin) at 300 ml ha −1 .

Data collection
This was carried out on plot basis under low-N, high-N, and Striga-infested environments. Number of days to 50% silking (DS) was the number of days when 50% of the plants in each plot had emerged silks. Number of days to 50% anthesis (DA) was the total number of days that 50% of maize plants in each plot had shed pollen. The difference between DS and DA represented the anthesis-silking interval (ASI). Plant and ear heights were measured as the distance between the base of the plant and the first branch of the tassel, and the distance between the base of the plant and the node carrying the uppermost ear, respectively. Ears per plant (EPP) was estimated as the ratio of the total number of ears harvested per plot to number of plants harvested per plot. Ear aspect (EASP) was assessed on a scale of 1-9, where 1 = large, uniform, clean, and well-filled ears, and 9 = small, variable, rotten, and partially filled ears. Husk cover was rated on a scale of 1-5, where 1 = husks firmly arranged with ear tip covered, and 5 = husks loosely arranged with ear tip exposed. Under low-N conditions, stay-green characteristic (STGR) was rated on a scale of 1-9, where 1 = almost 100% of the leaves were green, and 9 = almost 100% of the leaves were dead. In low-and high-N environments, plant aspect (PASP) was visually rated on a scale of 1-9 based on plant type, where 1 = excellent plant type, and 9 = poor plant type (Badu-Apraku et al., 2016). Additional data collected under artificial Striga infestation included Striga (host) damage rating and Striga emergence count at 8 and 10 WAP. Striga (host) damage was visually scored on a scale of 1-9, where 1 = normal plant growth with no visible symptoms, and 9 = total scorching of all leaves, causing premature death of host plant with no ear formation, whereas Striga emergence count was recorded as the number of Striga plants that emerged per plot at 8 and 10 WAP (Adetimirin, Aken'Ova, & Kim, 2000). Under low N, grain yield (kg ha −1 ) was estimated from grain weight and grain moisture content and thereafter adjusted to 15%. However, under Striga and high N, grain yield (kg ha −1 ) was determined from field weight of cobs, assuming 80% shelling percentage, and subsequently adjusted to 15% moisture content.

2.4
Chemical analyses of seed samples of extra-early PVA-QPM hybrids for carotenoid, lysine, and tryptophan contents Seed samples of crosses used for carotenoid, lysine, and tryptophan analyses were obtained through selfing the first and last two plants in each plot of the Design II crosses and checks (Suwarno, Pixley, Palacios-Rojas, Kaeppler, & Babu, 2014). The samples were obtained from plants under high N at Ile-Ife and Mokwa in 2016. Self-pollinated ears per plot, for each location, were separately harvested, dried, and shelled (Azmach, Gedil, Menkir, & Spillane, 2013). The processed seed samples were then stored at 4 • C (for ∼5 mo) in the cold storage equipment of IITA. Subsequently, random samples of 20-30 maize kernels from the top-yielding and stable PVA-QPM hybrids along with the best check, obtained from composite grains of the hybrid trials of 2016 at Mokwa and Ile-Ife, were drawn from the storage. The samples were dispatched to the International Maize and Wheat Improvement Center (CIM-MYT) for carotenoid, lysine and tryptophan analyses. The maize kernels of the selected hybrids, at CIMMYT, were frozen at −80 • C and ground to fine powder (0.5 µm). The institute's laboratory protocols for carotenoid analysesnamely, extraction, separation, and quantification by high performance liquid chromatography (HPLC)-were used (Galicia, Nurit, Rosales, & Palacios-Rojas, 2009). The following carotenoids were determined from each sample: lutein, zeaxanthin, beta-carotene (all-trans, 9-cis, and 13cis isomers), and beta-cryptoxanthin. Total provitamin A content of each hybrid was computed according to Suwarno et al. (2014) as: total provitamin A = 0.5(betacryptoxanthin) + beta-carotene (all-trans + 9-cis + 13cis isomers). Quantification of the percentages of tryptophan and lysine in whole grain of the hybrids were carried out according to Nurit, Tiessen, Pixley, and Palacios-Rojas (2009). Each whole grain sample was ground and defatted using a Kjeldahl apparatus, and an enzyme, papain, was added to hydrolyze the protein. A mixture of glacial acetic acid and H 2 SO 4 was added to induce a purple color, and the concentration of the color induced was determined with a spectrophotometer at 560 nm. The reading from the spectrophotometer was converted to percentage tryptophan: Percentage of tryptophan = Corrected OD at 560nm × Factor where corrected optical density (OD) = OD 560-nm sample − OD 560-nm average of papain blanks , and Factor = Hydrolysate volume Standard curve slope × Sample weight

Data analysis
Prior to statistical analyses, Striga (host) damage at 8 and 10 WAP (SDR1 and SDR2, respectively) and Striga emergence counts at 8 and 10 WAP (ESP1 and ESP2, respectively) were log transformed to achieve variance homogeneity (Badu-Apraku et al., 2010). In this study, each year-location combination was a test environment (Ukalski & Klisz, 2016). Analyses of variance for each environment and across environments were carried out on plot means. In the combined ANOVA across research environments, replicates, environments, and incomplete blocks within replications were considered random factors, whereas genotype was considered a fixed effect (Suwarno et al., 2014). The ANOVA for the 150 NCD II crosses were pooled over sets for each environment (Hallauer et al., 2010) and across research environments using SAS version 9.4 (SAS Institute, 2012). The hybrid component of the source of variation was divided into variation due to males (sets), females (sets), and female × male (sets) interaction. The F tests for male (sets), female (sets), and male × female (sets) mean squares were conducted using male (sets) × environment, female (sets) × environment, and male × female (sets) × environment mean squares, respectively. Mean squares of male (sets) × environment, female (sets) × environment, and male × female (sets) × environment were tested using pooled error mean square (table not shown). The general combining ability (GCA) effects for female and male within sets (GCA f and GCA m ) and specific combining ability (SCA) for each trait were estimated according to Kearsey and Pooni (1996) as shown below: where GCA f and GCA m are the GCA effects of female and male parents respectively; X f and X m are the average performance of a line when it was used as a female and male in crosses, respectively, and µ is the overall mean of crosses in the set. Specific combining ability (SCA) effect for the crosses was estimated as

Crop Science
where SCA X is SCA of the cross X; XX is the observed mean value of the cross; E (XX) is the expected mean value of the cross (on the basis of GCA of parents); and µ is the overall mean of crosses.
Standard errors for testing significance of GCA m and GCA f estimates, for the trait of hybrid, were computed from the mean squares of GCA m × environment and GCA f × environment, respectively, as shown below: where MS m × e and MS f × e are the mean squares of the interaction between male and environment, as well as female × environment, respectively; f, m, r, and e are the number of females, males, replicates, and environments, respectively. Performance across low-N, Striga, and high-N environments of extra-early PVA-QPM hybrids was examined using a multiple-trait base index. The base index combined grain yield performance across environments, as well as other important agronomic traits under each environment as seen in Badu-Apraku et al. (2016) and shown below: where yield and EPP are the grain yield and EPP across environments, respectively; EASP is the ear aspect across environments; STGR is the stay-green characteristic the in low-N environment; PASP is the plant aspect across low-and high-N conditions; SDR1 and SDR2 are the Striga (host) damage rating at 8 and 10 WAP in the Striga environment, respectively; and ESP1 and ESP2 are the number of emerged Striga plants at 8 and 10 WAP in the Striga environment, respectively. The means, adjusted for block effect, of each genotype for each measured variable was standardized to reduce the effects of the varying scales used. A positive multiple trait base index value showed tolerance of the genotype to both low N and Striga, whereas a negative multiple trait index value indicated susceptibility of the genotype to the stresses (Badu-Apraku et al., 2015).
Grain yield of the 20 top-performing and five worst extra-early PVA-QPM hybrids (as indicated by the multiple trait base index) along with six checks were subjected to genotype main effect plus genotype × environment interaction (GGE) biplot analysis using genotype x environment analysis with R for Windows (GEA-R) software (Pacheco et al., 2016). The "mean versus stability" view of the GGE biplot was used to identify hybrids with high yield and stability across Striga, low-N, and high-N environments. The model used is as shown below: − =λ 1 ξ 1 η 1 +λ 2 ξ 2 η 2 +Σ where Y ij is the average yield of genotype i in environment j; Y j is the average yield across all genotypes in environment j; λ 1 and λ 2 are the singular values for principal components PC1 and PC2, respectively; ξ i1 and ξ i2 are the PC1 and PC2 scores, respectively, for genotype i; η j1 and η j2 are the PC1 and PC2 scores, respectively, for environment j; and Σ ij is the residual of the model associated with genotype i in environment j. The data used for the analysis were not transformed (transform = 0) or standardized (scale = 0) but were environment centered (centering = 2) (Yan, 2001).

ANOVA for grain yield and other important agronomic traits under low N, high N, Striga, and across environments
Results of ANOVA for each environment (low N, high N, and Striga) and across environments indicated that hybrid had significant (P < .01) mean squares for grain yield and other agronomic traits, except EPP under low N (Table 2) and ear height (EHT) under Striga (Table 3). Mean squares of environment and hybrid × environment interaction were significant (P < .01) for all traits determined under each environment and across 11 research environments except hybrid × environment effect for yield, EPP, EASP, and PASP under low N (Table 2) and hybrid × environment mean square for EHT under Striga (Table 3).
Under low N, GCA m and GCA f were significant for DS, DA, and plant height (PLHT); only GCA m was significant for EHT, whereas only GCA f was significant for ASI and STGR (Table 2). Specific combining ability was not significant for any characters determined under low N (Table 2). Under high N, five traits (DS, DA, ASI, EASP, and PLHT)

Crop Science
TA B L E 2 Mean squares obtained from the combined ANOVA for grain yield and other agronomic traits of extra-early provitamin A quality protein maize hybrids evaluated across low-and high-N environments in 2016 and 2017 at Ile-Ife, Mokwa, and Abuja in Nigeria  Yield, grain yield; EPP, number of ears per plant; DS, days to 50% silking; ASI, anthesis-silking interval; EASP, ear aspect, where 1 = large, uniform, clean, and well-filled ears, and 9 = small, variable, rotten, and partially filled ears; PLHT, plant height; EHT, ear height; HUSK, husk cover, where 1 = husks firmly arranged with ear tip covered, and 5 = husks loosely arranged with ear tip exposed.; SDR1, Striga High N ---- for six of the eight traits, whereas GCA f was significant for seven traits (yield, EPP, DS, ASI, EASP, PLHT, and EHT). The SCA was significant for only three of the eight traitsnamely, yield, DS, and EHT (Table 3). Although GCA f × environment and GCA m × environment interaction effects were not significant for grain yield and EASP under low N, they were significant for grain yield and several other traits in each of high N (Table 2), Striga, and across environments ( Table 3). The SCA × environment effect was significant for two traits (ASI and STGR) under low N, for three traits (grain yield, DA, and EASP) under high N (Table 2), but it was not significant for any of the traits determined under Striga (Table 3).

Percentages of hybrid sum of squares attributed to pooled GCA and SCA, as well as maternal effects for grain yield and other agronomic traits of the extra-early PVA-QPM under low N, high N, Striga, and across environments
Although the percentage contributions of GCA and SCA to hybrid sum of squares revealed greater contribution from SCA to the total variation than GCA for grain yield, EPP, EASP, PASP, and STGR under low N (Table 4), SCA mean squares were not significant for these traits and other traits under the stress. The contribution of SCA Crop Science sum of squares to hybrid sum of squares was greater than GCA sum of squares for EPP and PASP under high N (Table 4), but SCA mean square was not significant for PASP. Under Striga, GCA and SCA made comparable contribution to hybrid sum of squares for grain yield and EASP, but GCA made greater contribution to Striga-related traits ( Table 4).
The ratio of GCA f sum of squares to GCA m sum of squares for grain yield ranged from 1.2 to 2.1 under low N, high N, Striga, and across environments (Table 4). For EPP, GCA f was lower than GCA m only under low N. Under high N, Striga, and across environments, the ratio of GCA f to GCA m for EPP ranged from 1.2 to 1.9 (Table 4). Stay-green characteristic was determined only under low N; for this, the ratio of GCA f to GCA m sum of squares was 1.9. The ratio of GCA f and GCA m for EASP under low N, PASP under high N, and ESP2 was 1.0. The ratio of GCA f to GCA m sum of squares was 1.5 for SDR1.

GCA for male and female effects for grain yield, Striga resistance indicator traits, and stay-green characteristic
Under low-N conditions, TZEEIORQ 53 and TZEEIORQ 5 of the 30 inbreds evaluated showed significant and positive GCA f effect for grain yield, whereas only TZEEIORQ 64 showed significant positive GCA m effect for this trait (Table 5). Under the high-N environment, inbreds TZEEIORQ 27, TZEEIORQ 53, TZEEIORQ 56, and TZEEIORQ 5 showed significant and positive GCA f effect for grain yield. Under Striga infestation, TZEEIORQ 61 and TZEEIORQ 35 exhibited significant and positive GCA f effect for grain yield (Table 5). Across environments, inbreds TZEEIORQ 27, TZEEIORQ 53, TZEEIORQ 61, TZEEIORQ 5, and TZEEIORQ 35 exhibited significant and positive GCA f effect for grain yield, whereas only TZEEIORQ 64 displayed significant and positive GCA m effect for the trait across environments ( Table 5). The GCA f and GCA m effects for STGR were significant and negative for the inbred TZEEIORQ 52. Significant and negative GCA f effect only was detected for STGR in inbreds TZEEIORQ 33 and TZEEIORQ 61. Under Striga infestation, 5 of the 30 lines showed significant and negative GCA m or GCA f for SDR2. These were TZEEIORQ 53, TZEEIORQ 33, TZEEIORQ 61, TZEEIORQ 69, and TZEEIORQ 35 (Table 5). Of these, only TZEEIORQ 53, TZEEIORQ 61, and TZEEIORQ 69 had significant and negative GCA m and/or GCA f for ESP2.

Multiple trait base index values for the highest-and lowest-yielding extra-early PVA-QPM hybrids along with the best check
Grain yield of extra-early PVA-QPM hybrids across research environments (low N, high N, and Striga) varied from 2,767 kg ha −1 for TZEEIORQ 30 × TZEEIORQ 11 to 4,950 kg ha −1 for TZEEIORQ 25 × TZEEIORQ 64 with an average of 3,937 kg ha −1 (Table 6). Of the 15 extra-early PVA-QPM hybrids with the best base index values along with the best check, Hybrid TZEEIORQ 33 × TZEEIORQ 75 had the best multiple base index value (8.97), whereas the multiple trait base index value was lowest for the best extraearly yellow check, TZEEI 79 × TZEEI 9 (1.12) ( Table 6).

DISCUSSION
The study addressed the development of maize hybrids with potential to alleviate the nutritional problems under the most important biotic (Striga hermonthica) and abiotic (low soil N) constraints to increased maize production and productivity in WCA. Hybrids specifically developed for cultivation in areas infested with Striga and with low N are also required to show good performance under optimal growing conditions. This will ensure adaptability to the varied conditions under which maize is cultivated in WCA. This is the first study in WCA aimed at addressing protein quality, provitamin A content, resistance to Striga, and tolerance to N stress of extra-early maize. The highly significant differences detected for grain yield, as well as other agronomic characters, of extra-early PVA-QPM hybrids under each environment and across research environments indicated that substantial genetic variability existed among extra-early PVA-QPM hybrids studied and that considerable progress could be made from selection for important agronomic traits under the stress environments, as well as nonstress environments. Bhatnagar, Betrán, and Rooney (2004), Langa (2005), and Tilahun et al. (2017) also reported significant differences for grain yield and other traits of QPM lines studied in hybrid combinations, whereas Mushongi (2010) reported significant differences among hybrid maize genotypes evaluated in low-and high-N conditions in Tanzania. The significant and large environmental variation observed for all the characters determined across environments showed the uniqueness of the environments. This underscores the need for the multi-environment testing of breeding materials and experimental varieties when pursuing low N tolerance, Striga resistance, and good performance in high-N environments. Similar results were reported by Badu-Apraku et al. (2013) with normalendosperm extra-early-maturing yellow inbred lines evaluated under different environments. The lack of significant hybrid × environment interaction mean squares observed for yield, EPP, EASP, and PASP under low N, and EHT under Striga environments, indicated that the response patterns of the hybrids were similar for the measured traits Crop Science TA B L E 7 Provitamin A, tryptophan, and lysine contents of the composite seeds obtained under high N conditions at Ile-Ife and Mokwa in 2016 from the 15 highest-yielding and most stable provitamin A quality protein maize hybrids and the best check across six environments (two environments each of low N, Striga, and high N) et al. (2013) in hybrid combinations, they interacted differently with the Striga environment. This may be because the inbred lines were extracted from a germplasm pool that combines Striga resistance with provitamin A and quality protein characteristics. The significant GCA m and/or GCA f mean squares detected for many characters under each and across research environments showed that additive genetic effect was more important than nonadditive genetic effect in the genetic control of the traits of extra-early PVA-QPM studied under low N, high N, Striga, and across environments. The mean squares of many traits with higher percentage contribution of SCA sums of squares to the hybrid sum of squares (especially under low N) were not significant, suggesting that nonadditive gene action was less important in regulating those traits. However, the significant SCA Crop Science mean squares observed for EPP and EHT under high N; ESP1 and ESP2 under Striga; and grain yield, DS, and EHT across environments suggested that nonadditive genetic effect was important in the inheritance of the traits under the environments. These results indicated that recurrent selection and hybridization would be useful for genetic improvement of Striga tolerance or resistance, low N tolerance, and good performance under high N in a population derived from the extra-early PVA-QPM materials studied. This result is in agreement with the report that additive and nonadditive genetic effects played important roles in the inheritance of grain yield and many other traits of early-maturing provitamin A maize inbreds across Striga-infested and optimal growing conditions in Nigeria (Laban, Badu-Apraku, & Diakaridia, 2017). Significant GCA m × environment and GCA f × environment effects detected for grain yield and a host of other traits in each environment and across environments indicated that the additive genetic effects for these traits interacted with the environment in their expression. These results justify a multilocation improvement strategy in the development of hybrids with resistance or tolerance to Striga and low N, as well as expression of good performance under an optimal N regime.
The NCD II yields two estimates of GCA (GCA m and GCA f ), a comparison of which provides information on maternal effects. Higher GCA f to GCA m indicated that the cytoplasm had genetic factors that influenced the expression of a trait, in addition to nuclear genes. In this study, the ratio of GCA f effect to GCA m effect was >1 for grain yield and some other measured traits under low N, high N, Striga, and across environments, suggesting that maternal effects contributed to the expression of the respective traits of the crosses in each environment and across the environments. The preponderance of GCA f sum of squares to GCA m sum of squares obtained for STGR under low N in this study was a clear indication of a role for maternal effect. Maternal effect was not detected for EASP under low N, PASP under high N, and ESP2. However, the ratio of GCA f sum of squares to GCA m sum of squares was >1 for SDR1. Given the suggestive role of maternal effect on Striga damage, a careful choice of female parent is required to exploit favorable cytoplasmic factors for this trait. Maternal effects for grain yield of early-maturing maize under well-watered conditions in Nigeria reported by Oyekunle and Badu-Apraku (2013) are similar to the results obtained under high N in the present study. Therefore, it could be hypothesized that cytoplasmic factors tend to positively influence grain yield under optimal environmental conditions. However, the observation in this study differs from the findings of the earlier authors who reported similar contributions of GCA m and GCA f for grain yield and some traits of early-maturing normal-endosperm maize assessed under low-N, Striga, and optimal environmental conditions (Ifie et al., 2015). The differences in the results of this study and that of Ifie et al. (2015) could be attributed to dissimilarities in the genetic materials used and the level of severity of the stresses achieved in the various studies.
The positive and significant GCA m or GCA f effects exhibited for grain yield by inbred lines TZEEIORQ 53, TZEEIORQ 5, and TZEEIORQ 64 under low N and across environments, as well as the significant and positive GCA f effect shown for the trait by TZEEIORQ 61 and TZEEIORQ 35 under Striga and across environments, indicated that the lines have potential to contribute favorable alleles for yield in their progenies if used as female or male parent under the stresses and across environments. The significant and negative GCA effect detected in inbred lines TZEEIORQ 52, TZEEIORQ 33, and TZEEIORQ 61 for STGR under low N suggested that progenies of crosses involving these lines are expected to have delayed leaf senescence under N stress. This finding is similar to the reports of Ifie et al. (2015), who obtained negative and significant GCA f effect for STGR in few of the early-maturing inbred lines studied under low soil N. Desirable inbred lines under Striga, in addition to having the capability to reduce host damage and number of emerged Striga plants in crosses to other inbred lines, are required to also exhibit high yield in crosses. The significant negative GCA for Striga damage score and Striga emergence count, and the significant and positive GCA for grain yield of TZEEIORQ 61, indicated that the inbred line has potential for use as a parent when developing Striga-resistant or -tolerant PVA-QPM hybrids. Three of the 15 top-performing hybrids across environments had inbred line TZEEIORQ 61 as the female parent, suggesting that the line transmitted favorable genes for grain yield, Striga tolerance and resistance, and STGR to its progenies. Amegbor et al. (2017) reported significant GCA effects for Striga emergence count, Striga (host) damage, and STGR in a study of extra-early maize lines in Nigeria.
High grain yield performance and stability of a genotype is desirable for increased crop productivity (Laban et al., 2017). The hybrids in the present study were ranked along the AEA, with the direction of the arrow indicating a higher mean performance. The long projections of hybrids on either direction away from the biplot origin, on the average tester coordinate (ATC) axis, indicated greater genotype × environment interaction and increased instability. Although Hybrids 12, 4, and 18 were high yielding, they had long projections onto the ATC (i.e., they were relatively unstable). Hybrids 7 (TZEEIORQ 55 × TZEEIORQ 26), 13 (TZEEIORQ 49 × TZEEIORQ 75), and 14 (TZEEIORQ 52 × TZEEIORQ 43) combined long projections on AEA with short projection on ATC, indicating that they combined high grain yield with stability across environments.

Crop Science
The per se provitamin A contents of all the 15 extraearly PVA-QPM hybrids selected for provitamin A, lysine, and tryptophan analyses in this study were higher than the provitamin A level of the hybrid with the highest provitamin A content reported by Halilu, Ado, Aba, and Usman (2016) but lower than the provitamin A levels of the best 11 hybrids selected by Suwarno et al. (2014), and the provitamin A concentrations of the extra-early PVA hybrids TZEEIOR 197 × TZEEIOR 205 (20.1 µg −1 ) and TZEEIOR 202 × TZEEIOR 205 (22.7 µg g −1 ) reported by Badu-Apraku et al. (2019), indicating the possibility of further improving the provitamin A levels of the extra-early PVA-QPM hybrids developed in this study.