Gains in Genetic Enhancement of Early Maturing Maize Hybrids Developed during Three Breeding Periods under Striga-Infested and Striga-Free Environments

Striga hermonthica is a major maize production constraint in West and Central Africa (WCA). Fifty-four early maturing maize hybrids of three breeding periods: 2008–2011, 2012–2013, 2014–2015, were evaluated under Striga-infested and non-infested environments in WCA. The study aimed at assessing genetic improvement in grain yield of the hybrids, identifying traits associated with yield gain during the breeding periods, and grain yield and stability of the hybrids in Striga infested and non-infested environments. Annual increase in grain yield of 101 kg ha−1 (4.82 %) and 61 kg ha−1 (1.24%) were recorded in Striga-infested and non-infested environments, respectively. The gains in grain yield from period 1 to period 3 under Striga-infested environments were associated with reduced anthesis-silking interval, reduced Striga damage, number of emerged Striga plants, improved ear aspect, and increased ears per plant. Ear aspect, ears per plant, and Striga damage at 8 and 10 weeks after planting (WAP) were significantly correlated with yield in Striga-infested environments, whereas ears per plant and plant and ear aspects had significant correlations with yield in non-infested environments. Hybrids TZdEI 352 × TZEI 355, TZdEI 378 × TZdEI 173, and TZdEI 173 × TZdEI 352 were outstanding in grain yield and stability in Striga-infested environments, whereas TZEI 326 × TZdEI 352, TZEI 495 × ENT 13, and TZdEI 268 × TZdEI 131 were superior in non-stress environments. These hybrids should be further tested extensively and commercialized. Significant genetic gains have been made in breeding for resistance to Striga hermonthica in early maturing maize hybrids.


Introduction
Maize (Zea mays L.) is an important staple food crop in sub-Saharan Africa (SSA). Its prominence has increased in SSA owing to its use as a cheap energy source in both human and livestock diets. The high insolation, cold night, and minimal occurrence of pest and diseases that characterize the savanna agroecology of SSA make it an ideal environment for maize production [1]. The early maturing maize varieties that are often available in July during the food deficit period, when other food reserves have been exhausted due to the extended hunger period, have helped to alleviate starvation in the

Genetic Gains in Grain Yield of Early Maturing Hybrids in Striga-Infested and Non-Infested Environments
In Striga-infested environments, grain yield ranged between 2248 and 2917 kg ha −1 for hybrids of period 1 and period 3, respectively, resulting in the equivalent of 4.82% annual yield gain (Tables 2  and 3). Contrarily, in Striga-free environments, yield varied from 5016 kg ha −1 for hybrids of period 1 to 5442 kg ha −1 for those of period 3, which corresponded to an annual yield gain of 1.24%. Grain yield increased by 101 and 61 kg ha −1 year −1 in Striga-infested and Striga-free environments, respectively. The significant gain from selection for grain yield between periods 1 and 3 observed in Striga-infested environments was associated with reduced anthesis-silking interval, increased plant and ear heights, and improved ear aspect. Other characters included reduced Striga damage at 8 and 10 weeks after planting (WAP), fewer number of emerged Striga plants at 10 WAP, and increased ears per plant. Additionally, significant positive b estimates were observed for yield, plant, and ear heights, while significant negative b values were obtained for husk cover, ear aspect, days to 50% anthesis and silking, number of emerged Striga plants, and Striga damage at 8 and 10 WAP under Striga-infested conditions ( Table 3). Under non-infested conditions, however, only plant aspect and stalk lodging had significant b values.
Regression analysis of yield of the maize hybrids in Striga-free environments on yield under Striga-infested conditions, distinctly grouped hybrids into three breeding periods ( Figure 1). Despite overlaps in performance of the hybrids of the three periods, those of period 3 were the best in terms of grain yield in both research environments.

Interrelationships among Measured Traits
Of the possible 55 correlation coefficients recorded under artificial Striga environments, 47 were statistically significant while 16 out of the 28 correlation coefficients identified in Striga-free environments were significant (Figures 2 and 3). Under Striga infestation, grain yield displayed positive and significant associations with ears per plant and plant and ear heights, but negative correlations with anthesis-silking interval, husk cover, ear aspect, Striga damage, and number of emerged Striga plants at 8 and 10 WAP (Figure 2). Similarly, Striga damage at 8 and 10 WAP recorded positive and significant correlations with anthesis-silking interval, husk cover, ear aspect, and number of emerged Striga plants at 8 and 10 WAP. In Striga-free environments, grain yield displayed positive and significant correlations with ears per plant and plant and ear heights, while negative and significant correlations were found between yield and plant aspect, as well as ear aspect ( Figure 3). Additionally, plant aspect had positive and significant associations with ear aspect, husk cover, and anthesis-silking interval, but had significant and negative association with plant and ear heights in Striga-free environments.

Performance Assessment of Grain Yield Stability of Early Maturing Maize Hybrids Developed during Three Breeding Periods
Presented in Table 4 are the grain yield and other assayed agronomic characters of the 20 highest-yielding and 5 lowest-yielding maize hybrids identified by utilizing the IITA selection index

Interrelationships among Measured Traits
Of the possible 55 correlation coefficients recorded under artificial Striga environments, 47 were statistically significant while 16 out of the 28 correlation coefficients identified in Striga-free environments were significant (Figures 2 and 3). Under Striga infestation, grain yield displayed positive and significant associations with ears per plant and plant and ear heights, but negative correlations with anthesis-silking interval, husk cover, ear aspect, Striga damage, and number of emerged Striga plants at 8 and 10 WAP ( Figure 2). Similarly, Striga damage at 8 and 10 WAP recorded positive and significant correlations with anthesis-silking interval, husk cover, ear aspect, and number of emerged Striga plants at 8 and 10 WAP. In Striga-free environments, grain yield displayed positive and significant correlations with ears per plant and plant and ear heights, while negative and significant correlations were found between yield and plant aspect, as well as ear aspect ( Figure 3). Additionally, plant aspect had positive and significant associations with ear aspect, husk cover, and anthesis-silking interval, but had significant and negative association with plant and ear heights in Striga-free environments.  Agronomy 2020, 10, x FOR PEER REVIEW 10 of 20

Performance Assessment of Grain Yield Stability of Early Maturing Maize Hybrids Developed during Three Breeding Periods
Presented in Table 4 are the grain yield and other assayed agronomic characters of the 20 highest-yielding and 5 lowest-yielding maize hybrids identified by utilizing the IITA selection index for Striga resistance and the corresponding mean performance of different traits under optimal test environments. The values of the selection index ranged from −11.1 for the double-cross hybrid, (TZEI 59 × TZEI 108) × (TZEI 63 × TZEI 87) to 10.9 for TZdEI 352 × TZEI 383. Grain yield of the hybrids varied between 1426 kg ha −1 for (TZEI 59 × TZEI 108) × (TZEI 63 × TZEI 87) and 4186 kg ha −1 for TZdEI 352 × TZEI 355 across Striga-infested environments. Furthermore, grain yield of the hybrids across Striga-free environments ranged from 3506 kg ha −1 for TZEI 352 × TZdEI 352 to 6379 kg ha −1 for TZdEI 173 × TZdEI 280. Hybrids that possessed positive selection indices produced grain yields above 4200 and 2400 kg ha −1 in Striga-free and Striga-infested test environments, respectively.
The average reduction in grain yield caused by the parasitic weed was about 50%. Top performing hybrids identified using the Striga selection index were characterized by reduced yield losses under Striga infestation. The best five hybrids recorded yield losses varying from 24.5 to 37.6%. These losses were low compared to those of the five worst hybrids, which varied from 64 to 70%. The significant yield loss in the susceptible genotypes was associated with increased anthesis-silking interval, reduced ear and plant heights, fewer ears per plant, increased Striga damage, and number of emerged Striga plants at 8 and 10 WAP under artificial Striga infestation.
The significance of hybrid and hybrid × environment interaction for yield in the two research environments necessitated the use of the genotype main effect plus genotype × environment interaction (GGE) biplot analysis to partition the hybrid × environment interaction for better understanding of the yield performance and the stability of the hybrids in each test environment. The grain yield "stability vs. mean performance" GGE biplots of the hybrids under both research environments in West Africa between 2017 and 2019 are presented in Figures 4 and 5. Under Striga-infestation, the first (PRC1) and second (PRC2) principal component axes explained 61.2 and 11.6% of the overall variation, respectively; therefore, both principal component axes jointly explained about 73% of the overall variation in the yield of the hybrids. The PRC1 of the maize hybrids evaluated in four optimal test environments captured 38.7% of the overall variation, while PRC2 explained 35.9% of the overall variation. Thus, the two PRC axes captured about 75% of the overall variation in yield of the hybrids, an indication of adequate approximation of the environment-centered data. Furthermore, the average-tester coordinate (ATC; double-arrow line) y-axis of GGE biplot separates genotypes with yield above the mean on the right side of the line, distinguishing them from genotypes characterized by grain yield below the mean.  The mean performance of a hybrid is measured by the projection of the hybrid's marker on the abscissa, whereas the smaller the absolute length of a genotype on the ATC, the more stable it is [19]. Hybrids 37 (TZdEI 352 × TZEI 355), 26 (TZdEI 378 × TZdEI 173), and 19 (TZdEI 173 × TZdEI 352) were the top performing and most stable across Striga-infested research environments. These were therefore identified as ideal hybrids across test environments, while hybrids 2 [(TZEI 59 × TZEI 108) × (TZEI 63 × TZEI 87)] and 13 (TZEI 31 × TZEI 63), which recorded the lowest grain yield, were highly stable ( Figure 5). Additionally, hybrids 21 (TZdEI 173 × TZdEI 280) and 24 (TZdEI 173 × TZdEI 492) were productive but unstable across the Striga-infested environments, while hybrids 39 (TZEI 352 × TZdEI 352) and 40 (TZEI 355 × TZdEI 425), in addition to being low yielding, were among the least stable hybrids. Across Striga-free environments, hybrids 25 (TZdEI 268 × TZdEI 131), 41 (TZEI 326 × TZdEI 352), and 46 (TZEI 495 × ENT 13) displayed superior grain yield and had short projections onto the ATC y-axis (stable) across non-stress environments. Contrarily, hybrids 39 (TZEI 352 × TZdEI 352) and 13 (TZEI 31 × TZEI 63), which produced yield far below the average grain yield and had long projections onto the ATC y-axis, were the lowest yielding and most unstable. The average reduction in grain yield caused by the parasitic weed was about 50%. Top performing hybrids identified using the Striga selection index were characterized by reduced yield losses under Striga infestation. The best five hybrids recorded yield losses varying from 24.5 to 37.6%. These losses were low compared to those of the five worst hybrids, which varied from 64 to 70%. The significant yield loss in the susceptible genotypes was associated with increased anthesis-silking interval, reduced ear and plant heights, fewer ears per plant, increased Striga damage, and number of emerged Striga plants at 8 and 10 WAP under artificial Striga infestation.
The significance of hybrid and hybrid x environment interaction for yield in the two research environments necessitated the use of the genotype main effect plus genotype x environment interaction (GGE) biplot analysis to partition the hybrid x environment interaction for better understanding of the yield performance and the stability of the hybrids in each test environment. The grain yield "stability vs. mean performance" GGE biplots of the hybrids under both research environments in West Africa between 2017 and 2019 are presented in Figures 4 and 5. Under Striga-infestation, the first (PRC1) and second (PRC2) principal component axes explained 61.2 and 11.6% of the overall variation, respectively; therefore, both principal component axes jointly explained about 73% of the overall variation in the yield of the hybrids. The PRC1 of the maize hybrids evaluated in four optimal test environments captured 38.7% of the overall variation, while PRC2 explained 35.9% of the overall variation. Thus, the two PRC axes captured about 75% of the overall variation in yield of the hybrids, an indication of adequate approximation of the environment-centered data. Furthermore, the average-tester coordinate (ATC; double-arrow line) y-axis of GGE biplot separates genotypes with yield above the mean on the right side of the line, distinguishing them from genotypes characterized by grain yield below the mean. genetic gain in yield of the hybrids under Striga-infested environments was associated with increased plant and ear heights, decreased days to 50% anthesis and silking, as well as reduced anthesis-silking interval. Other traits associated with the genetic gain of the hybrids included improved ear aspect, increased ears per plant, reduced Striga damage, and decreased emerged Striga plants at 8 and 10 WAP. Regression of the yield of the early hybrids in non-infested environments over the yield in Striga-infested environments clearly classified hybrids of the three breeding periods into three groups with a few hybrids of different periods overlapping in performance. This result confirmed the superior performance of hybrids of period 3 over those of periods 1 and 2 in both research environments. This implied that significant progress has been achieved in developing productive hybrids possessing enhanced resistance to Striga hermonthica parasitism during the three breeding periods.
Information on trait association is important for designing effective breeding programs for maize genetic enhancement [22]. The significant and negative correlations obtained for grain yield and Striga damage and number of emerged Striga plants at 8 and 10 WAP implied that these characters were important for yield improvement under Striga infestation. This was a justification for the need to integrate the characters into the multiple-trait selection index for enhanced genetic gain from selection for grain yield under Striga-infested environments. Our findings corroborated

Discussion
The significance of environment, period, and hybrid effects recorded for most traits including grain yield under both research environments signified the uniqueness of the test environments and the significant variability among the hybrids of the three periods.
This facilitated the identification and selection of promising maize hybrids of early maturity in the test environments.
The significance of hybrid × environment interaction as well as period × environment interaction effects for most measured traits, including grain yield in both Striga-infested and optimal research environments, signified contrasting performance of the hybrids under the research conditions. This result emphasized the importance of testing genotypes in multiple environments, in years, and locations prior to recommendations for commercialization [20]. The high estimates of repeatability recorded for several traits in both research environments implied that the hybrids would be consistent in the expression of the measured traits in the research environments.
The significance of period and hybrid effects for most agronomic traits including grain yield necessitated the analysis of the genetic gains in order to evaluate the progress that was achieved in developing superior early maturing maize hybrids possessing durable Striga resistance. The 4.82% yield gain per year with an increase of 101 kg ha −1 in Striga-infested environments across the three breeding periods obtained in this study was considerably higher than the 3.28, 2.56, and 2.25% reported for a set of extra-early maturing OPVs under moisture deficit, Striga-infested, and optimal conditions, respectively [21]. Furthermore, the yield gain per year realized in the present study is also higher than the 1.93 and 1.0% reported for early maturing OPVs under Striga infestation and non-infested environments by Badu-Apraku et al. [2] and 3.2% yield gain per year in Striga-infested environments reported by Menkir and Meseka [18] for intermediate maturing hybrids. The implications of these results are that early maturing hybrids responded better to selection for improved resistance to Striga, as well as high grain yield relative to the extra-early and early varieties as well as intermediate hybrids. Also, the non-significant gains from selection for grain yield of the early maturing hybrids under optimal conditions confirmed that greater attention of the IITA maize breeders has been on improving Striga resistance under infested conditions rather than enhanced performance of the hybrids in Striga-free environments. In this study, the significant genetic gain in yield of the hybrids under Striga-infested environments was associated with increased plant and ear heights, decreased days to 50% anthesis and silking, as well as reduced anthesis-silking interval. Other traits associated with the genetic gain of the hybrids included improved ear aspect, increased ears per plant, reduced Striga damage, and decreased emerged Striga plants at 8 and 10 WAP.
Regression of the yield of the early hybrids in non-infested environments over the yield in Striga-infested environments clearly classified hybrids of the three breeding periods into three groups with a few hybrids of different periods overlapping in performance. This result confirmed the superior performance of hybrids of period 3 over those of periods 1 and 2 in both research environments. This implied that significant progress has been achieved in developing productive hybrids possessing enhanced resistance to Striga hermonthica parasitism during the three breeding periods.
Information on trait association is important for designing effective breeding programs for maize genetic enhancement [22]. The significant and negative correlations obtained for grain yield and Striga damage and number of emerged Striga plants at 8 and 10 WAP implied that these characters were important for yield improvement under Striga infestation. This was a justification for the need to integrate the characters into the multiple-trait selection index for enhanced genetic gain from selection for grain yield under Striga-infested environments. Our findings corroborated those of Badu-Apraku et al. [4], Kim et al. [6], and Karaya et al. [23]. Significant positive associations detected between grain yield and ears per plant and plant height under both research environments were earlier reported by Badu-Apraku et al. [17]. Hybrid TZdEI 352 × TZEI 355 developed in period 3 was the highest-yielding and most stable across Striga-infested environments, and one of the outstanding hybrids under optimal environments. This suggested that this hybrid is widely adapted to Striga endemic and optimal growing regions in WCA. Our results further justified the need for testing extensively outstanding hybrids at multiple locations and on-farm trials for commercialization in SSA. It is striking that of the 20 top performing hybrids identified using the Striga selection index, only hybrids TZdEI 352 × TZEI 355, TZdEI 378 × TZdEI 173, and TZdEI 173 × TZdEI 352 across Striga-infested environments and TZEI 326 × TZdEI 352, TZEI 495 × ENT 13, and TZdEI 268 × TZdEI 131 across Striga-free conditions were selected by the GGE biplot as productive and stable in performance. This was anticipated as GGE biplot analysis was conducted using the yield data only, compared to the Striga selection index which took into consideration yield and other important yield-related characters. The outstanding maize hybrids identified thus have the potential to combat hunger and alleviate poverty in Striga prone environments in SSA.

Development of Striga Resistant Early Maturing Maize Hybrids
A major objective of the IITA-MIP is to develop early maturing hybrids with outstanding grain yield under multiple stresses, viz., drought, low-soil nitrogen, and S. hermonthica parasitism. Towards this end, the IITA-MIP started a breeding program for Striga resistance in 1992 with the objective of combating the menace of Striga hermonthica in the savannas of SSA. By 1994, the IITA-MIP had developed several populations and OPVs with early maturity and had initiated inbred and a hybrid development program in maize. The early maturing inbreds were derived from several broad-based maize populations possessing resistance to maize streak virus (MSV) and resistance/tolerance to Striga formed from four diverse germplasm sources, which included TZE-W Pop × 1368 STR C 0 , TZE Comp 5-Y C 6 , TZE-Y Pop DT STR C 0 , and TZE-W Pop DT STR C 0 [24]. S 1 lines extracted from each population were evaluated under artificial Striga-infested conditions at Ferkessedougou and non-infested conditions at Sinematiali in Côte d'Ivoire in the 1997 cropping season. Superior S 1 lines from each of the populations were advanced through repeated cycles of selfing and selection under Striga-infested environments and managed drought. At the S 4 stage of inbreeding, 250−300 inbreds extracted from each of the populations were crossed to a tester that was the corresponding source population. The testcrosses as well as the S 4 lines were screened under artificial Striga infestation and non-infested conditions at Ferkessedougou and Sinematiali, respectively. The grain yield and the combining abilities of the lines for traits such as grain yield, Striga damage syndrome rating, Striga emergence counts, number of ears per plant, and plant and ear aspects across the two contrasting research environments served as criteria for selecting 90-100 S 4 lines for advancement to the S 6 stage. Selection for Striga resistance was based on an index of traits which included Striga damage, ears per plant, and grain yield. Through this program, several S 6 inbreds and OPVs were developed from the populations. Even though considerable progress had been made in developing Striga resistant OPVs, inbreds, hybrids, and several of the OPVs extracted from the populations still supported significant number of the parasitic weeds, which could flower and produce seeds, resulting in increased Striga seed bank in the soil. It was, therefore, desirable to enhance resistance levels of the populations. Therefore, in addition to the exploitation of the genetic variation available in domestic maize, Striga resistance alleles from the wild maize, perennial teosinte Zea diploperennis were introgressed into the breeding populations using the backcross breeding method. Badu-Apraku et al. [2] described in detail the strategies and procedures adopted in screening the early maize populations for Striga resistance. Additionally, the levels of tolerance to moisture stress in the populations were not very high. Consequently, a program was started in 2007 to enhance drought tolerance in the populations using the S 1 recurrent selection method. Also, Striga, low-N, and drought-tolerant lines identified in the program were employed as sources of drought tolerance alleles and were incorporated into each population. Subsequent genetic enhancement of the populations under managed drought employing the S 1 family selection scheme led to the development of a new generation of superior early maturing multiple-stress tolerant populations from which were derived inbreds, hybrids, and OPVs with combined high levels of low-N tolerance and improved levels of Striga resistance/tolerance, as well as drought tolerance. Multiple-stress tolerant inbreds selected based on outstanding performance were used in hybrid combinations to obtain the early hybrids used for the genetic gain study under Striga hermonthica infestation.
The hybrids were categorized into three breeding periods (2008-2010, 2011-2013, and 2014-2016) with each period of development comprising 18 hybrids. Pedigree information, period, and year of development of the hybrids are presented in Supplementary Table S1.

Trial Establishment and Agronomic Management
The set of 54 hybrids used in this study was evaluated in 2017, 2018, and 2019 under artificial Striga infestation in West Africa. In Nigeria, the hybrids were evaluated at Mokwa (9 • 18 N, 5 • 4 E, 457 m above sea level, 1.1 m annual precipitation) and Abuja (9 • 16 N, 7 • 20 E, 300 m above sea level, 1.5 m annual precipitation) from June to October, 2017-2019; both experimental sites are Striga endemic locations in the Southern Guinea Savanna of Nigeria. Additionally, the 54 hybrids were tested at Nyankpala (9 • 25 N, 0 • 58 W, 183 m above sea level, and 1000 mm annual precipitation) in Ghana and at Ina (9 • 30 N and 2 • 62 E, 119 m above sea level, 1500 mm annual precipitation) in the northern part of Benin Republic in 2017. The trial was laid in a 9 × 6 lattice design with three replicates while an experimental plot was 4 m long, spaced at 0.75 m between and 0.4 m within rows. Each plot at all locations was artificially infested with Striga seeds following the IITA Maize Breeding Unit infestation method [11]. The Striga seeds utilized for infestation in the present study were collected during the previous cropping season from neighboring sorghum farmlands near each experimental site. At about two weeks before planting, the Striga experimental fields were sterilized by injecting ethylene gas (a synthetic germination stimulant) directly into the soil to initiate self-destructive germination of Striga seeds in the soil. The suicidal germination strategy helped to reduce the existing Striga seed in the soil. Infestation was done by infusing an 8500 mg mixture of finely sieved sand and S. hermonthica seed inoculum (containing an estimated number of 5000 viable Striga seeds) in the same hill as the maize seeds. Fertilizer application on the maize plots was deferred until about 21 to 24 days after planting (DAP) when 30 kg ha −1 each of N, P, and K was applied. The decreased fertilizer dose and delay were to stress the maize plants to stimulate production of strigolactones, a hormone which facilitates the germination of seeds of the parasitic weed and the attachment of the emerging parasitic plants to the roots of maize plants in Striga infested plots [11]. Other weeds apart from Striga were controlled manually.

Trait Measurements
Data were recorded for grain yield in both Striga-infested and Striga-free trials. Yield (kg ha −1 ) was measured in both trials based on 80% (800 g grain per kilogram ear weight) shelling percentage and adjusted to 150 g grain kg −1 moisture content. Striga-infested trials were assessed for Striga damage and number of emerged Striga plants [25] at 8 and 10 WAP. Striga damage in each plot was rated on a scale of 1 to 9 (1 = no damage, indicating high resistance; 9 = severe damage or death of the maize plant, i.e., high susceptibility). Data on plant aspect were recorded only on the Striga-free plots on a scale of 1 to 9, where 1 = excellent plant type and 9 = poor plant type. Data on other measured traits, which included days to 50% anthesis, days to 50% silking, anthesis-silking interval, ear aspect, ears per plant, and plant and ear heights, as well as root and stalk lodging, were recorded as described in detail by Badu-Apraku et al. [17].

Analysis of Data
The data were analyzed for variances across the seven Striga-infested and three Striga-free research environments on plot means of each trait with PROC GLM in SAS 9.3 utilizing a RANDOM statement with the TEST option [26]. In the analyses, the test environments (location × year combinations), the breeding periods, replications, blocks, and hybrid × environment interactions for each experiment were considered as random factors and hybrids as fixed effects.
Repeatability (H) estimates of all measured traits were computed for each research environment as: where σ 2 g represents hybrid variance, σ 2 g×e the variance due hybrid × environment interaction, σ 2 e the error variance, e the number of test environments, and r the number of replications in a test environment.
The gain in yield of the 54 maize hybrids over the eight-year period of development was estimated by linear regression. The genetic gain representing the regression coefficient (b-value) was obtained by regressing hybrid means (dependent variable, y) of yield and other agronomic characters on the year of development (independent variable, x) under both infested and non-infested environments using SAS. Relative genetic gain per year was obtained by dividing the genetic gain (b value) by the intercept and multiplying by 100 [21]. Additionally, the relationship between grain yield under Striga infestation and optimal environments was determined for each breeding period. Employing the Microsoft Excel software for the regression analysis, the regression line was obtained.
The package "PerformanceAnalytics" in the R software [27] was used to compute the correlation coefficients between grain yield and other characters of the maize hybrids under Striga infestation and Striga-free test environments. The early maturing maize hybrids were characterized as either resistant or susceptible to Striga using a selection index that involved grain yield, ears per plant, Striga damage, and number of emerged Striga plants [14]. The means of the hybrids adjusted for block effects were standardized (using 1 and 0 as standard deviation and mean, respectively) to reduce the effects of varying scales. Hence, hybrids with Striga base index (BI) values greater than 0 were considered resistant to Striga whereas those with BI values less than 0 were rated as susceptible.
In order to identify outstanding hybrids in terms of high grain and stability under Striga infestation and Striga-free environments, yield across replications were analyzed using the genotype main effect plus genotype × environment interaction (GGE) biplot statistical tool to partition significant hybrid × environment interaction [21,28].

Conclusions
The annual yield gain of 4.82% of the hybrids studied under Striga-infestation revealed that considerable progress had been achieved in developing superior multiple stress tolerant early maturing maize hybrids for SSA. Improved ear aspect, reduced anthesis-silking interval, reduced Striga damage syndrome rating combined with fewer emerged Striga plants, and increased ears per plant were associated with the yield gain of the early hybrids. Ear aspect and ears per plant were identified as invaluable selection indices for achieving rapid gain in yield under Striga infestation and optimal research environments. The superior early maturing hybrids selected in this study should be extensively evaluated in on-farm trials and commercialized to combat food insecurity as well as contribute to alleviation of poverty in SSA.