Introduction

Maize (Zea mays L., 2n = 2x = 20) is the world's third most widely cultivated cereal crop, after wheat and rice. It is a vital food security crop in sub-Saharan Africa (SSA), constituting 85–95% of the region’s carbohydrate intake (Johnmark et al. 2022). In SSA, maize is a source of livelihood for more than 300 million Africans (Regassa et al. 2021; Kansiime et al. 2023). It is a raw material for manufacturing industrial products, including livestock feed (Dabija et al. 2021). Despite the importance of the cereal, one out of five people living in communal and small-scale maize farming systems are at risk of starvation (Arndt et al. 2023). This is because the crop is vulnerable to many stress factors, one major biotic stress being parasitic weeds of the genus Striga.

Striga, popularly known as witchweed, is endemic and widespread throughout SSA. The species S. asiatica (red flower type) is predominant in Southern Africa, and S. hermonthica (purple flower) is widely distributed in Western, Central, and Eastern Africa (Dossa et al. 2023b). Striga hermonthica has the largest geographical distribution in Africa, spreading from latitudes of 5° N and 20° S. The two holoparasites cause severe crop damage through stunting and leaf chlorosis, leading to yield losses ranging from 30 to 100% under severe infestation (Mutsvanga et al. 2022). Maize suffers yield losses caused by these parasites due to the paucity of Striga resistance sources in the maize gene pool, and only partial resistance has been reported. Over 50 million hectares of agricultural land under cereal cultivations, including maize, have been infested by Striga spp. (Dafaallah 2019; David et al. 2022). The annual losses caused by Striga have been estimated to USD 10 billion across SSA (Dafaallah 2019; Samejima and Sugimoto 2022). Under heavy, Striga infestations farmers are often forced to abandon their farms.

Cultural practices, crop protection chemicals, biological control, host plant resistance, and integrated Striga management approach are the main control strategies recommended for Striga management (Lobulu et al. 2021). Germination stimulants such as ethylene and ethephon, when applied on infested croplands before sowing, can deplete Striga seed banks by inducing suicidal germination without a host (Samejima et al. 2016). Imidazolinone herbicides are also effective against Striga in the field (Kanampiu et al. 2003). Maize seeds with imazapyr resistance can be coated with small doses of the herbicide, significantly reducing Striga emergence (David et al. 2011; Kamara et al. 2020). However, using imazapyr-coated maize in a smallholder maize production system is difficult since the chemical is toxic to other crops (Souto et al. 2020). Most of the proposed control methods are impractical to implement under communal and small-scale farming systems. Most farmers do not have enough land to crop rotate, nor do they have access to large quantities of organic matter for effective cultural control practices (Shayanowako et al. 2020). Furthermore, they do not have access to the money required to buy agrochemicals and sprayers. Hence, the development and use of Striga-resistant cultivars is the most feasible management option (Gasura et al. 2021; Dossa et al. 2023a).

Striga-resistant cultivars can reduce Striga seed production and the Striga seed bank in infested soils (Badu-Apraku et al. 2020a). Resistant cultivars can induce the germination of Striga seeds but prevent the parasite from attaching to the maize plant. Concerted efforts have been made by the International Institute of Tropical Agriculture (IITA) in Nigeria and national maize research programs to develop tropical maize genotypes with resistance to Sh (Badu-Apraku et al. 2020b; Yacoubou et al. 2021a). However, the performance of their germplasm against Sa is unknown. High-yielding and Striga-resistant sub-tropical maize varieties are yet to be developed (Shayanowako et al. 2018a). Therefore, the IITA genetic resources can serve as breeding parents with resistance to Sh, and need to be evaluated for their levels of resistance to Sa, and for yield improvements under Sa-infestation. Likewise, sub-tropical maize varieties should be screened for Striga resistance and yield performance under both Sa and Sh infestation. This can also benefit West and Central Africa with a diversity of resistance genes that could be useful for accumulating improved levels of partial resistance.

Striga emergence count, Striga damage rating, and grain yield under Striga infestation are the major selection indices used in resistance breeding (Menkir et al. 2007). The choice of the selection method to be used in the genetic improvement of maize depends on the type of gene action controlling Striga resistance in maize. Striga resistance in maize is quantitatively inherited with many minor genes with small additive effects and is significantly influenced by the environment (Lane et al. 1997; Badu-Apraku et al. 2020a), making breeding complex and challenging. Hybrid breeding, backcrossing, and recurrent selection are common methods widely used in incorporating Striga tolerance/resistance genes into well-adapted maize varieties. However, the initial steps of these methods include the collection and evaluation of maize germplasm with different genetic backgrounds to identify potential sources of resistance (Yacoubou et al. 2021b). Screening for Striga resistance in maize includes field, greenhouse, and laboratory conditions (Shayanowako et al. 2018b). However, the use of greenhouse conditions is the most efficient for managing the level of Striga infestation and environmental conditions (Kountche et al. 2019; Yacoubou et al. 2021b).

The existing Striga-resistant cultivars of maize in SSA are bred for Sh resistance, while no commercially grown maize varieties are resistant to Sa. In most of the East African countries, the two species occur in tandem (Gethi and Smith 2004). Germplasm with dual resistance to both parasites would be extremely valuable across the continent. This study aimed to evaluate 130 tropical and sub-tropical maize germplasms in a controlled environment for their reaction to Sh and Sa infestations and for resistance breeding. This study is one of the few attempts to report on the performance of tropical and sub-tropical maize germplasm under both Sa and Sh infestation to select inbred lines with dual resistance to the two dominant Striga species.

Materials and methods

Plant material and study sites

The study screened maize genotypes for Striga resistance at the University of Kwazulu-Natal Controlled Environment Facilities (UKZN-CEF) in two cropping seasons (December 2021–April 2022, and August 2022–December 2022). During the two summer seasons, average maximum temperatures are between 26 and 28 °C, while minimum temperature is 10 °C.The UKZN CEF is situated at the UKZN College of Agriculture, Engineering, and Science (29.62° S, 30.40° E). The study used 130 maize genotypes, comprising 74 acquired from the International Institute of Tropical Agriculture (IITA)/Nigeria, 45 from the International Maize and Wheat Improvement Centre, Zimbabwe (CIMMYT)/Zimbabwe, and 10 from the National Plant Genetic Resources Centre, South Africa (NPGRC)/South Africa (Table 1). The germplasm from IITA/Nigeria comprised 55 inbred lines (genotypes number 1–55 in Table 1), 4 single cross hybrids (genotypes number 127–130 in Table 1), and 15 open-pollinated varieties (OPV) (genotypes number 112–126 in Table 1). Out of the 55 inbred lines, 21 were generated from multi-parent crosses of elite Striga resistant lines, while the remaining were derived from synthetics IWD-SYN, Syn-Y-STR, and ACR97SYN, composites TZL CompI and TZE Comp5. The CIMMYT/Zimbabwe germplasm included 43 inbred lines (genotype number 56–98 in Table 1), and 1 OPV (genotype number 110 in Table 1). The NPGRC/South Africa germplasm included 5 OPVs (genotype 99–103 Table 1) and 6 hybrids (genotype 104–109 Table 1). The Nigerian accessions were tropical varieties developed for Sh resistance, generated from multi-parent populations (Simon et al. 2018; Gasura et al. 2019). CIMMYT provided sub-tropical maize germplasm. Their material was developed for drought tolerance. The OPVs and hybrids germplasms from NPGRC/South Africa and CIMMYT/Zimbabwe were used as local checks, while the OPVs and hybrids from IITA were Sh-resistant checks. The Sa and Sh seeds were collected from sorghum and maize-infested fields in Tanzania and Kenya, respectively.

Table 1 List and source of maize genotypes evaluated in the present study
Full size table

Experimental design and trial management

The 130 genotypes were evaluated under two Striga treatments using a 13 × 10 alpha lattice design with two replications in each Striga-infested environment. The maize genotypes were evaluated under Sa and Sh infestations. The experimental unit consisted of 4 plastic pots of 15-L capacity, filled with a composted pine bark potting mix for each Striga-infested environment. Two weeks before planting, each pot was infested with a scoop of sand mixed with 0.03 g of 2 years old Sa or Sh seed containing approximately 3000 Striga seeds. Standard agronomic practices recommended for maize production were followed. Hand weeding was routinely done to remove all weeds except Striga. Figure 1 shows the experimental setup with artificial infestations of maize with Sa (A, B, and C), and Sh (D and E) in the greenhouse condition at the University of Kwazulu-Natal, South Africa.

Fig. 1
figure 1

Photographs showing the experimental setup with artificial infestations of maize with Striga asiatica (A–C), and S. hermonthica (D, E) in the African Center for Crop Improvement greenhouse at the University of Kwazulu-Natal, South Africa

Full size image

Data collection

Data were collected on maize phenotypic traits and Striga parameters in the Sa and Sh-infested environments. The following phenotypic traits were evaluated on maize: Days to 50% silking (DS), recorded as the number of days taken by 50% of the plants to silk in each plot; Days to anthesis (DA), recorded as the number of days from planting until 50% of the plants have emerged silks and shed pollen, respectively; Anthesis-silking interval (ASI), measured as the difference between days to 50% silking and 50% anthesis; Plant height (PLHT) and ear height (EHT), measured as the distance from the base of the plant to the height of the first tassel branch and the node bearing the upper ear, respectively; Root lodging (RL) was recorded as a percentage of plants leaning more than 30° from the vertical; Stalk lodging (SLG) (percentage broken at or below the highest ear node); and Number of rotten ears (EROT). The number of ears per plant (EPP) was obtained by dividing the total number of ears per plot by the number of plants harvested. Husk cover (HUSK) was rated on a scale of 1–5, where 1 = husks tightly arranged and extended beyond the ear tip and 5 = ear tips exposed. Ear aspect (EASP) was recorded based on a scale of 1–9, where 1 = clean, uniform, large, and well-filled ears and 9 = ears with undesirable features. Grain yield per plant (GY/plant) was determined as the weight (g) of the grain from the ears of individual plants after shelling, adjusted to a constant moisture of 12.5%.

The Striga parameters were recorded, including the number of emerged Sa and Sh plants 8 and 10 weeks after planting, denoted as SEC8 and SEC10. A rating of host plant damage 8 and 10 weeks after planting, designated as SDR8 and SDR10, was done using a visual rating score of 1–9 where 1 = no damage, indicating normal plant growth and a high level of tolerance, and 9 = complete collapse or death of the maize plant, i.e., highly susceptible (Kim 1994).

Data-analysis

The collected data from Sa and Sh-infested environments were subjected to analysis of variance using a lattice procedure, using the package agricolae in RStudio version 2023. 06.1 (R Core Team 2023). The normality of the data was tested using kurtosis and skewness values, which were coumputed using Genstat version 23.1.0.651.The mean values of the test genotypes for the assessed traits were compared at the 5% significance level using Fisher’s least significance difference (LSD). Broad sense heritability (H2) (hereafter referred to as heritability) was computed using DeltaGen (Jahufer and Luo 2018) with the following formula:

$$({\text{H}}^2 ) = \frac{\sigma^2 g}{{\sigma^2 g + \frac{\sigma^2 s}{{ns }} + \frac{\sigma^2 r}{{nr}} + \frac{\sigma^2 b}{{nb}} + \frac{\sigma_\varepsilon^2 }{{ns + nr + nb}}}}$$

where \(\sigma^2 g,\sigma^2 s, \sigma^2 r, \sigma^2 b,\) and \(\sigma_\varepsilon^2\) are the variance components for genotypes, season, replication, block, and the pooled error, respectively, and ns, nr, and nb are the number of seasons, replications, and blocks, respectively.

Pearson’s correlation coefficients (r) were calculated separately for Sa and Sh-infested conditions using RStudio version 4.3.1 (Team, 2010). The rotated component matrix and principal component analysis biplots (PCA) were generated separately for the assessed traits under Sa, and Sh-infested conditions using the packages ggplot2, factoextra, and FactoMiner (Alboukadel 2017) in RStudio version 4.3.1. Cluster heatmap plots were generated based on the mean values of the traits recorded in both Sa and Sh environments to establish the Clustering of the genotypes using Deltagen (Jahufer and Luo 2018).

Results

Analysis of variance (ANOVA)

The analysis of variance revealed significant differences among the evaluated genotypes for all the recorded traits (P < 0.001) (Table 2) under both Sa and Sh-infested environments. Testing seasons exerted significant effects (P < 0.001) on all the traits under Sa-infested conditions except for EPP, PLHT, HUSK, and SEC10, and under Sh-conditions except for EPP, PLHT, EHT, and HUSK. Significant differences were recorded for all the assessed traits except for EPP due to the block nested to replication-by-season interaction effect under both Sa and Sh-infested environments.

Table 2 Analysis of variance for yield components and Striga parameters of 126 maize genotypes evaluated under Sa and Sh infestation conditions
Full size table

Mean performance and statistical summary

Under Sa-infested conditions

Table S1 summarizes the mean performance and the statistics of the 126 genotypes evaluated under Sa infestation. The kurtosis values ranged from − 7 to 7, except for EPP, HUSK, GY, and SEC10, while the skewness varied from − 2 to 2. The coefficient of variation (CV%) of all the traits under Sa infestation ranged from 4.89 to 426.82% (Table 3). Smaller variations were obtained for DA followed by DS, with CV values of 4.89% and 5.60%, respectively. The highest variation was exhibited by PLHT followed by ASI, with CV values of 426.82% and 268.88%, respectively. Inbred lines had a mean ASI of 2.77, while the OPV and hybrid checks had mean ASI values of 1.86 and 1.77, respectively. The EPP ranged from 1.00 to 2.00. The mean yield ranged from 0.00 to 277.50 g/plant for TZISTR1262 and CML540, respectively, with a mean of 62.77 g/plant for the inbred lines, while ranging from 00.00 to 214.00 g/plant for Hickory/1421 and N.Choice/1421, respectively for the hybrid checks, and from 35.00 to 169.50 g/plant (((IWD C3 SYN*2/(White DT STR Syn)) -DT C1 and NC.QPM/Z.DPLO respectively) for the OPVs checks. The top inbred lines and checks are shown in Table 3. The best-yielding genotypes were generally taller than the poorer-performing genotypes, with the longest cobs and moderate or high EASP. The greatest reduction in Striga emergence in SEC8 occurred with relatively high yielding genotypes, e.g., TZISTR1154 (2.00), TZISTR1263 (2.00), TZISTR1261 (2.50), TZISTR1015 (2.50), TZISTR1174 (3.00), TZSTRI113 (3.00), TZISTR1119 (3.00), TZISTR1205 (3.50), TZISTR1248 (3.50), the local OPVs M.Pearl/DT-STR (2.00), ZM1421/DT-STR (2.50), and the Striga-resistant checks DTSTR-Y SYN14 (2.50), and NC.QPM/DT-STR (3.00). The SEC10 mean was 45.50, 10.32, and 9.07 for the inbred lines, the OPVs, and the hybrids, respectively. Genotypes such as inbred line TZISTR1174 and Striga-resistant check OPV DTSTR-W SYN11 exhibited high numbers for SEC10 and are still relatively high yielding. The inbred lines exhibited an SDR8 mean value of 3.35, while the OPVs and hybrids showed an SDR8 of 3.14 and 2.36, respectively. SDR10 mean scores were 3.07, 4.30, and 2.71 for the inbred lines, the OPVs, and the hybrid checks, respectively. Genotypes with high yields generally showed moderate or large reductions in SDR8 and SDR10 values. The following high-yielding genotypes displayed relatively high and moderate reductions in scores for SDR8 and SDR10: CML566, CML440, TZISTR1248, ZM1421, and local check (B.King/1421). None of the genotypes exhibited SLG, RL, and EROT under Sa infestation. High broad-sense heritability was recorded for CL (0.97), EHT (0.96), EASP (0.96), DS (0.94), ASI (0.94), DA (0.90), and GY (0.88) under Sa-infested conditions. However, the low heritability of Sa resistance is worth noting, reflected in the Striga parameters SEC10, SDR8, and SDR10 under Sa-infested conditions. In contrast, under Sh-infested conditions, high heritability values were estimated for all the traits except for GY (0.02).

Table 3 Mean responses for 14 traits of 126 maize genotypes evaluated under Sa infestation, showing the top 10 inbred lines, the top 4 hybrids, and 6 OPVs
Full size table

Under Sh-infested conditions

Table S2 shows the mean performances and the summary statistics of the evaluated genotypes for all traits under Sh-infested conditions. The kurtosis values varied from − 7 to 7, except for ASI, EPP, PLHT, and HUSK, while the skewness ranged from − 2 to 2 except for EPP, PLHT, and HUSK. The CV% of the traits ranged from 7.15 to 597.49% (Table 4). The DA exhibited the lowest variation of 7.15%, while PLHT exhibited the highest variation of 597.49%, as was the case under Sa infested environment. The mean ASI was 1.52 for the inbred lines, 1.26, and 2.03 for the OPVs and the hybrids, respectively. The mean yield ranged from 10.05 to 151 g/plant, with a mean of 63.83 g/plant for the inbred lines, from 34.75 to 133.25 g/plant with a mean of 79.79 g/plant for the hybrids, and 33.60 to 144.25 g/plant with a mean of 70.81 g/plant for the OPVs. The sub-tropical inbred line CML304, which exhibited the lowest SDR10 under Sa, showed the highest GY under Sh. The lowest GY among the inbred lines was exhibited by the sub-tropical line HA04A-2107-36. The local check OPV ZM1423 exhibited the highest GY, whereas the Striga susceptible check TZBSTR showed the lowest GY. The top 10 inbred lines, as well as the top 10 check genotypes showing high GY under Sh, are presented in Table 4. The SEC8 mean was 63.89, 70.81, and 79.79, whereas the SEC10 mean was 3.25, 4.43, and 3.28 for the inbred lines, the OPVs, and the hybrids, respectively. The following high-yielding genotypes displayed the greatest reduction in SEC10 scores: TZDEEI50 (1.70) CML550 (2.63), TZISTR1001 (2.70), and B.King/1421 (2.63). The mean SDR8 was 5.25, 3.86, and 3.11 for the inbred lines, the OPVs, and the hybrids checks, respectively while the SDR10 means were 2.75, 3.05, and 3.11 in the same order. The high-yielding genotypes CML539 and ZM1421 resulted in the greatest reduction in SDR8. The genotype TZDEEI54 resulted in the greatest reduction of 0.75 in SDR10, while the least reductions were displayed by the low-yielding genotypes CKDHL0378 (5.75) and CZL1380 (5.50).

Table 4 Mean responses for 14 traits of 126 maize genotypes evaluated under Sh infestation, showing the top 10 inbred lines, the top 4 hybrids, and 6 OPVs
Full size table

Principal component and biplot analyses

The rotated components matrix showing the percentages of variances of different principal components (PC) and the respective loadings of recorded traits are shown in Table 5. The first four PCs under Sa-infested conditions had a cumulative variance of 50.80%. The first PC had the highest variation of 17.38% followed by PC2 with 13.32%. DS, DA, EHT, CL, GY, and SDR8 made the highest contributions to PC1, while EPP, EASP, SDR8, and SDR10 contributed strongly to PC2. The highest loadings for PC3 were DS, GY, and DA, followed by SEC10, while the highest loadings for PC4 were SEC8, and CL, followed by HUSK, EPP, and SEC10. Under Sh-infested conditions, DS, DA, EHT, SDR8, and SEC8 had high positive loadings into the first PC, explaining 17.06% of the total variance. PC2 was highly influenced by GY, SEC8, CL, and EHT, which had a high loading, explaining 31.82% of the total variance. SDR8, SDR10, DA, and DS had the highest loadings into PC3, while PC4 was most influenced by HUSK, PLHT, and ASI.

Table 5 Rotated component matrix for 10 yield components and 4 Striga parameters in 126 maize genotypes under Sa and Sh-infested conditions
Full size table

Biplots based on principal components are presented to decipher the performance of the maize germplasm with Sa and Sh infestations using Figs. 2 and 3, respectively. Smaller angles between vectors of recorded traits indicate a high correlation between the traits in discriminating genotypes. Genotypes plotted closer to and further along a vector line scored highly in that trait. Under Sa-infested conditions, the genotypes were evenly scattered across both PC1 and PC2 (Fig. 2). The angles between vectors of GY and those of PLHT and HUSK, were acute, indicating a strong positive correlation between the traits and with high-yielding genotypes TZSTRI115 (34), TZISTR1001 (36), CML451 (86), TZISTR1015 (98), TZBSTR (110), ZM1423 (104), TZISTR1205 (95), and NC.QPM/Z.DPLO (106). The same analysis can be made with all the Striga parameters SEC8, SEC10, SDR8, and SDR10, which showed small angles with each other and with ASI, pointing to a high positive correlation between the traits. Genotypes CML312 (49), CML571 (60), CML539 (55), CLHP0049 (83), CLHP00286 (85), CML304 (93), STR -SYN -Y2 (107), and SAMMMZ16 (115) showed more susceptibility to Sa because they were closely associated with vectors of SEC8, SEC10, SDR8, and SDR10. However, the angles formed between the vectors of GY, SEC8, SEC10, SDR8, and SDR10 were close to 90°, which means that the association between the traits is weak. EASP, DA, DS, and EPP showed a negative correlation with poor yielding genotypes including TZISTR1154 (1), TZISTR1275 (5), TZISTR1165 (9), TZISTR1159 (20), and TZSTRI109 (30).

Fig. 2
figure 2

Principal component of 126 maize population under Sa-infested condition. Genotypes are coded with numbers as recorded in Table S1. Dim = dimension, DA = days to 50% anthesis, DS = days to 50% silking, ASI = anthesis-silking interval, EPP = ear per plant, PLHT = Plant height, EHT = ear height, HUSK = husk cover, CL = cob length, EASP = ear aspect, GY = grain yield, SEC8 = Striga emergence counts 8 weeks after sowing, SEC10 = Striga emergence counts 10 weeks after sowing, SDR8 = Striga damage rating 8 weeks after sowing, and SDR10 = Striga damage rating 10 weeks after sowing

Full size image
Fig. 3
figure 3

Principal component of 126 maize population under Sh-infested condition. Genotypes are coded with numbers as recorded in Table S2. Dim = dimension, DA = days to 50% anthesis, DS = days to 50% silking, ASI = anthesis-silking interval, EPP = ear per plant, PLHT = Plant height, EHT = ear height, HUSK = husk cover, CL = cob length, EASP = ear aspect, GY = grain yield, SEC8 = Striga emergence counts 8 weeks after sowing, SEC10 = Striga emergence counts 10 weeks after sowing, SDR8 = Striga damage rating 8 weeks after sowing, and SDR10 = Striga damage rating 10 weeks after sowing

Full size image

Under Sh-infested conditions (Fig. 3), GY had a strong correlation with EHT, PLHT, and EPP and with the high-yielding genotypes, including CML304 (93) and ZM1423 (104), which showed a high score in these traits. The Striga parameters SEC8, SEC10, SDR8, and SDR10 positively correlated with EASP, and all plotted far from the vector of GY, indicating a negative correlation between the traits and low Striga emergence counts genotypes, including TZISTR1263 and Colorado/1421. The same summary is made with low SDR8 and SDR10 reduction genotypes CLHP0404, CKDHL0378, CZL1380, M.Pearl/DT-STR, and Colorado/1421, which were negatively correlated with GY.

Correlation of maize yield components and Striga parameters

Phenotypic correlation coefficients showing the relationship between GY and agronomic traits, within agronomic traits, within Striga parameters, and between GY and Striga parameters under both Sa and Sh infested conditions are shown in Fig. 4A, B, respectively. Under Sa conditions (Fig. 4A), GY had a positive and significant correlation with CL and PLHT, with correlation coefficients of r = 0.33 and r = 0.18, respectively. Negative and significant correlations were found between GY and EASP (r = − 0.50). The agronomic traits PLHT and EHT showed positive and significant correlations of r = 0.28 and r = 0.29 with CL, respectively. A positive and significant correlation was also recorded between DA and DS (r = 0.83). The correlation between EPP and EASP, DS and ASI, and PLHT and EHT were also positive and significant (r = 0.18, r = 0.30, and r = 0.54, respectively). However, negative and significant correlations were recorded between the following agronomic traits: DA and ASI (r = − 0.29), EPP and ASI (r = − 0.22), EHT and DS (r = − 0.18), CL and DA (r = − 0.20), CL and DS (r = − 0.21), PLHT and DA (− 0.18), PLHT and DS (r = − 0.20). ASI showed a positive and significant correlation between SDR8 (r = 0.18) and SDR10 (0.32). PLHT and SDR10 had a positive and significant correlation of r = 0.18. EPP exhibited a negative and significant correlation with SEC8, SDR8, and SDR10 (r = − 0.18, r = − 0.27, and r = − 0.24, respectively). SDR8 exhibited negative and significant correlations with EHT (r = − 0.32) and CL (r = − 0.22). The correlation between SDR8 and SDR10 was positive and significant (r = 0.54).

Fig. 4
figure 4

Correlation matrix plot between yield components and Striga parameters in a population of 126 maize genotypes under Sa (A) and Sh (B) infestation. The colour variation indicates the magnitude of correlations; traits with deep blue are strongly and positively correlated, while deep red are strongly negatively correlated. Note *, ** and *** denote the level of significance associations of traits at 0.05, 0.01, and < 0.001 probability values, respectively, and ns = non-significant. EHT = ear height, CL = cob length, GY = grain yield, DA = days to 50% anthesis, DS = days to 50% silking, SEC8 = Striga emergence counts 8 weeks after sowing, SEC10 = Striga emergence counts 10 weeks after sowing, EPP = ear per plant, EASP = ear aspect, HUSK = husk cover, SDR8 = Striga damage rating 8 weeks after sowing, ASI = anthesis-silking interval, SDR10 = Striga damage rating 10 weeks after sowing, PLHT = plant height

Full size image

Under Sh conditions (Fig. 4B), GY was positively correlated with EHT (r = 0.20), CL (r = 0.29), and PLHT (0.20), whereas with EASP and CL, GY exhibited a negative correlation with r = − 0.57 and r = − 0.21, respectively. Positive and significant correlations were revealed between the agronomic parameters DS and DA (r = 0.93), ASI and DS (r = 0.30), and PLHT and EHT (r = 0.61). However, negative and significant correlations were found between EHT and DS (r = − 0.19), PLHT and DA (r = − 0.26), and PLHT and DS (r = − 0.26). Positive and significant correlations were observed between maize agronomic traits and Striga parameters. These include the positive correlation recorded between DS and SEC8 (r = 0.21), EASP and SEC10 (r = 0.18), and ASI and SEC8 (r = 0.20). However, negative, and significant correlations were recorded for SDR8 with EPP (r = − 0.20) and EHT (r = − 0.20). A negative correlation was recorded for EHT with SEC8 (r = − 0.22) and SDR8 (r = − 0.36), and PLHT with SDR10 (r = − 0.22), and SDR8 (r = − 0.48). The Striga parameter SDR8 was positively correlated with SEC8 (r = 0.26) and SDR10 (r = 0.52).

Cluster analysis based on yield components and Striga parameters

Cluster heatmap analysis based on eight maize and Striga phenotypic traits is presented for both Sa (Fig. 5 and Table S3) and Sh-infested environments (Fig. 6 and Table S4). The heatmap shows clusters based on the mean performances of each trait from the lowest performance (blue colour) to the highest performance (red colour). In an Sa-infested environment, the heatmap revealed six clusters (Fig. 5) where genotypes in the first cluster (I) had the highest scores in EASP, ASI, and SDR8 reduction. For instance, the IITA lines TZISTR1154, TZISTR1225, TZISTR1018, TZISTR1178, TZISTR1163, and hybrid ZM1421, which exhibited high SDR8 reductions and high scores in EASP, are classified in that cluster. Genotypes in the second cluster (II) had low ASI and high SEC10 values, including some IITA lines TZISTR1166, TZISTR1244, TZSTRI101, TZSTRI104, some CMMYT lines CLHP0326, CLHP00378, CLHP0049, CLHP0022, and IITA OPVs DTSTR-W SYN13, and TZL COMP1-W C6*2/(White DT STR Syn-DT C1, which all exhibited low SEC10 reductions. Genotypes from the third cluster (III) had moderate EASP and moderate SEC8 values and included some CIMMYT lines CML312, CML444, CML442, CML571, CML390, IITA lines TZSTRI108, TZEEI14, TZEEI49, TZISTR1011, and TZDEEI50 and OPVs STR-SYN-Y2, and Z. Diplo.BC4C3-W-DT C1. Cluster IV comprised genotypes exhibiting the lowest SEC10 reduction, moderate ASI, moderate PLHT, and moderate SDR10 reduction. The genotype CIMMYT CML304, which showed a high SEC10 number, was clustered in that cluster together with the IITA OPVs (2*TZECOMP3DT/WhiteDTSTRSYN) C2, NC.QPM/DT-STR, ZM1421/DT-STR, and DTSTR-Y SYN14. Cluster V consisted of genotypes with high SDR10 reduction, moderate PLHT, and moderate SDR10, and comprised some good-yielding genotypes, including CKDHL0378, CML451, CLHP0350, CLHP0005, ZM1423/Z.DLO, TZISTR1174, (TZEOMP5C7/TZECOMP3DTC2) C2, and Colorado/1421. Cluster VI comprised genotypes that exhibited high GY, moderate PLHT, and high SDR10 reduction. The cluster had the high-yielding CIMMYT lines CML440, CML566, CML540, CML545, CLHP0156, and the IITA OPVs TZBSTR, Z.diplo-BC4-C3-W/DOGONA-1/Z.diplo-BC4-C3-W, and showed high GY under Sa infestation.

Fig. 5
figure 5

Six clusters (I–VI) heatmap plot showing a colour pattern of 126 maize genotypes based on 4 maize yield components and 4 Striga parameters recorded in Sa-infested environment. Blue: lowest performance; red: highest performance. The numbers at the right represent the genotype numbers as coded in Table S1. EASP = ear aspect, SEC8 = Striga emergence counts 8 weeks after sowing, SEC10 = Striga emergence counts 10 weeks after sowing, PLHT = Plant height, GY = grain yield, ASI = anthesis-silking interval, SDR10 = Striga damage rating 10 weeks after sowing, SDR8 = Striga damage rating 8 weeks after sowing

Full size image
Fig. 6
figure 6

Four clusters (I–IV) heatmap plot showing a colour pattern of 126 maize genotypes based on 4 maize yield components and 4 Striga parameters recorded in Sh-infested environment. Blue: lowest performance; red: highest performance. The numbers at the right represent the genotype numbers as coded in Table S2. EASP = ear aspect, SEC10 = Striga emergence counts 10 weeks after sowing, SDR10 = Striga damage rating 10 weeks after sowing, SDR8 = Striga damage rating 8 weeks after sowing, SEC8 = Striga emergence counts 8 weeks after sowing, ASI = anthesis-silking interval, PLHT = plant height, GY = grain yield

Full size image

Under Sh-infested conditions (Fig. 6 and Supplemental Table 4), the genotypes clustered into four, with Cluster I consisting of genotypes exhibiting low reductions in SDR8, SDR10, SEC8, and SEC10. This cluster included TZBSTR, CZL1380, and Colorado/1421, which were characterized by low GY, low SDR10, and lowest SEC10 reduction under Sh. Cluster II was composed of two types of genotypes. The first group comprised genotypes that showed low GY, low SEC8, and SEC10 reduction, and included the genotype HA04A-2107-36, which showed the lowest GY in Sh-infested environment. The second group comprised genotypes that showed average GY and moderate resistance to Sh. Cluster III genotypes had high values in GY, and EASP, and had moderate reduction in all Sh parameters. These included the CIMMYT lines CML540, CML566, CML304, CML550, CML539, CML440, CML545, and the IITA genotypes TZISTR25, TZISTR1174, TZISTR1119, TZISTR1166, TZSTRI113, B.King/1421, Shesha/1421, ZM1423, N.Choice/1421, DTSTR-W SYN13 all exhibited high mean yields under Sh infestation. Cluster IV comprised the NPGRC/SA OPV ZM1423/Z.DLO and the CIMMYT line CML571 showed exceptional scores in PLHT and GY.

Discussion

Breeding highly Striga-resistant maize genotypes adaptable to wide agro-ecological areas in SSA would be a sustainable solution to manage Striga infestations in maize crop. The mean values, genotypic and phenotypic variances, heritability, and the correlation coefficients of agronomic traits are key parameters used in breeding programs to select superior genotypes (Nzuve et al. 2014). The present study discerned genotypic differences for agronomic and Striga resistance traits, indicating that the target traits are amenable to selection. The skewness and kurtosis values for most of the evaluated traits ranged − 2 to 2, and − 7 to 7, respectively, suggesting a normal distribution of the data. The high variation of some parameters, such as PLHT, EPP, and EHT was expected due to the variability of the test populations, including inbred lines, OPVs, and hybrids. The genotype-by-season interaction effect was significant for any of the evaluated traits in both Striga species environments, suggesting the strong influence of environment on Striga traits. The significant effect of cropping season on some agronomic and Striga resistance traits suggested that the two seasons and associated growing conditions influenced traits responses. It shows that using two or more testing seasons supports the repeatability of the study for genotype comparison and selection. Nonetheless, in the current study there were statistically non-significant effects of the G × S interaction, suggesting that the test season/environment did not influence the ranking of genotypes for the studied traits.

The rotated component matrix showed that DA, DS, EHT, CL, GY, and SDR8 were the most discriminating traits under both Sa and Sh environments. This indicates that these traits are important selection indices. In addition to these traits, EPP, EASP, SDR8, and SDR10, which were loaded in the second PC in an Sa-infested environment, can also be considered during selection. Badu-Apraku et al. (2010) reported that SDR8, SDR10, EASP, and EPP were the best traits for selection under Striga-infested conditions. SDR was also reported to be the best selection criterion for improving GY performance in maize under Striga infestation by Dossa et al. (2023a). All the traits exhibited high heritability values in both the Sa and Sh environments, except for GY under Sh-infested conditions. This indicates a large contribution of the genotypes to the phenotypic variance (Ngugi 2013). Traits having high heritability are improved by selection based on phenotypic traits (Shekhawat and Singhania 2005; Boghara et al. 2016). However, it is important to note the low heritability of the Striga parameters SEC10, SDR8, and SDR10, with values of only 0.008, 0.11, and 0.16, respectively, under Sa infested conditions, and of GY, with a value of 0.02 under Sh infested conditions. Notably, maize selection parameters recorded high heritabilities compared to the heritability of Sa parameters. This suggests that the genotypic variance of the Striga resistance traits was low compared to the phenotypic variance, making direct selection difficult for these traits. This is also indicative of the complex nature of the Striga resistance traits. Badu-Apraku et al. (2007) reported in a previous study low heritability values for SDR and SEC.

The PC biplots confirmed the positive and significant correlation between GY and secondary traits, including PLHT, EHT, and CL, under both Sa and Sh-infested conditions, indicating the importance of these traits in improving maize GY under Striga infestation. Previous studies reported similar findings (Golam et al. 2011; Nzuve et al. 2014; Yahaya et al. 2021).

The ASI showed a positive correlation between SDR8 (r = 0.18) and SDR10 (0.32) under Sa-infested conditions, indicating that severe Striga damage prolongs the ASI. This confirms the stunted growth symptom of Striga-infested plants (Waweru et al. 2019). This is also supported by the positive correlation between EASP and SEC10, indicating that higher numbers of Striga plants increased the EASP score. The negative and significant associations between yield components, including EPP, CL, PLHT, and EHT, and Striga parameters, including SEC8, SEC10, SDR8, and SDR10, under both Sa and Sh infestation, were expected because when the parasitic weeds proliferate, damage to maize yield components increase (Shayanowako et al. 2020). Therefore, reducing Striga damage scores and the parasitic plant counts at 8 and 10 days after planting will have a positive impact on yield component and enhance GY under Striga infestation (Menkir and Meseka 2019; Badu-Apraku et al. 2021).

With Sa infestation, the yield of the evaluated inbred lines was higher than the yield of the OPVs checks. This was not expected, but it can be explained by the fact that most of these OPVs were bred for resistance or tolerance to Sh, with no screening to Sa. These results indicate that these OPVs that have been bred for Sh resistance are not resistant to Sa. In the present study, some genotypes, e.g. TZISTR1174, showed excellent GY performance and relatively high SEC under Sa-infested conditions compared to others and could be used as a source of resistance. Resistant genotypes support fewer Striga plants and produce a higher grain yield than susceptible ones. Conversely, in the case of tolerance, the host supports an equally severe level of the parasitic weed without losses in GY (Rodenburg et al. 2005). The following lines showed resistance to both Sa and Sh: sub-tropical lines CML440, CML566, CML540, CML539, CML451, CLHP0343, CLHP0326, and the tropical lines TZISTR1248, TZSTRI115, TZISTR25, TZISTR1205, TZSTRI113, TZISTR1119. The local hybrids B.King/1421, Shesha/1421, ZM1421, and the Striga-resistant check OPV DTSTR-W SYN13.

The phenotypic clustering analysis based on eight morphological traits clustered 126 maize genotypes into six clusters under Sa infestation and four clusters under Sh infestation. This indicates that the studied genotypes exhibited genetic divergence regarding morphological traits. The formation of different clusters using phenotypic traits in maize genotypes agreed with other reports (Saiyad and Kumar 2017; Han et al. 2018; Kasoma et al. 2020; Shayanowako et al. 2020). All clusters consisted of a mixture of genotypes from different sources (IITA, CIMMYT, and NPGRC). Cluster mean values of evaluated genotypes in Sa environment suggested that Cluster VI was the best with outstanding trait values, especially for GY and SDR. In a Sh-infested environment, cluster mean values showed that Cluster IV was best, with outstanding values for all the evaluated traits. Therefore, these clusters containing tropical and sub-tropical maize genotypes should be good sources for heterosis breeding in Striga resistance breeding programs.

Conclusions

This study identified IITA, CIMMYT, and NPGRC/SA conserved maize genetic resources that could be used as breeding parents to improve for both Sa, Sh, and dual resistance. The top five most resistant to Sa included tropical and sub-tropical maize inbred lines: CML540 (277.50 g/plant), CML566 (155.50 g/plant), TZISTR1001 (140.00 g/plant), TZISTR1205 (114.25 g/plant) and TZSTRI115 (112.50 g/plant). On the other hand, the top five yielding inbred lines under Sh were CML304 (151.00 g/plant), TZSTRI101 (144.00 g/plant), CLHP0404 (137.35 g/plant), TZISTR1119 (135.75 g/plant), and TZISTR25 (131.00 g/plant). The genotypes CML440, CML566, CML540, CML539, CLHP0343, CLHP0326, TZISTR1248, TZSTRI115, TZISTR25, TZISTR1205, TZSTRI113, and TZISTR1119, showed resistance to both Striga species, whereas TZISTR1174 showed tolerance to Sa. The identified tolerant and resistant inbred lines could be used to produce new hybrids combining tolerance and resistance to both Sa and Sh and market-preferred agronomic traits.

Supplementary Materials

The following supporting information can be downloaded: Table S1: Mean values for 10 yield components and 4 Striga parameters measured for 126 maize genotypes evaluated under Sa infestation. Table S2. Mean values for 10 yield components and 4 Striga parameters measured for 126 maize genotypes evaluated under Sh infestation. Table S3. Clusters and their member genotypes of 126 maize genotypes evaluated under Sa infestation. Table S4: Clusters and their member genotypes of 126 maize genotypes evaluated under Sh infestation.