Northern leaf blight also known as Northern corn leaf blight (NCLB) or Turcicum leaf blight (TLB), has persisted for decades as a major foliar disease in maize producing regions of the World (Drechsler, 1923; Savary et al., 2019). Following its first discovery in Parma Italy, NLB was only well documented in the United States of America (USA) in 1878 and only emerged as an outbreak in 1889 (Drechsler, 1923). The disease has since appeared in the Americas, and Asia (Shi et al., 2017; Bashir et al., 2018; Mueller et al., 2020; Navarro et al., 2021). In Africa, NLB was first reported in Uganda in 1924 and has since been reported in several sub-Saharan African countries ( Figure 1 ; Table 1 ) (Emechebe, 1975; Adipala et al., 1995; Abebe and Singburaudom, 2006; Haasbroek et al., 2014; Nieuwoudt et al., 2018; Berger et al., 2020).

The causal pathogen of NLB is Exserohilum turcicum (Pass.) K.J. Leonard & Suggs, which is the asexual form of this hemibiotrophic Dothideomycete (Leonard and Suggs, 1974). Researchers have defined physiological races of E. turcicum based on six maize qualitative resistance genes namely; Ht1, Ht2, Ht3, Htn1, Htn2, and Htm1, that E. turcicum is able to overcome (Jordan et al., 1983; Jindal et al., 2019; Navarro et al., 2021; Muñoz-Zavala et al., 2023). The race groups are determined based on the screening of a differential set of maize lines, each with a different resistance gene (Leonard et al., 1989). Routine screening of E. turcicum races is carried out using maize differential genotypes in some maize producing countries (Weems and Bradley, 2018; Jindal et al., 2019; Turgay et al., 2020; Navarro et al., 2021; Muñoz-Zavala et al., 2023). However, maize resistance responses in these germplasm sets are highly variable and dependent on growth room/glasshouse conditions, and genetic background effects (Weems and Bradley, 2018; Jindal et al., 2019). To date, there are 29 described race groups based on the combinations of maize resistance genes that can be overcome, with race 0 defined as being able to overcome all known resistance genes ( Table 2 – see references within). Race 0 has been described from countries in all the continents except South America ( Table 2 ). Twelve of the race combinations have been reported from African countries ( Table 2 ), although screening has only been done on E. turcicum isolates from South Africa, Kenya, Uganda, and Zambia (Craven and Fourie, 2011).

Research on maize resistance genes to E. turcicum has revealed that the Htn1 gene encodes ZmWAK-RLK1, a wall-associated receptor-like kinase (Hurni et al., 2015). Further research provided evidence that maize Ht2 and Ht3 genes encoded the same ZmWAK-RLK which corresponded to a different allele of ZmWAK-RLK1 (Yang et al., 2021). This is consistent with previous work that Htn1, Ht2, and Ht3 map to chromosome 8 (Yang et al., 2021). This brings into question the validity of using Ht genes only to allocate E. turcicum race classes when screening with differential maize panels. The authors propose that responses may be confounded by “modifier genes” as a result of (i) different genetic backgrounds, and (ii) variation in the size of the introgression surrounding an Ht gene in each differential maize line (Yang et al., 2021). This may explain why some isolates of E. turcicum gave different responses and were thus classified as race 2 or race 3 on the “differential” maize lines carrying Ht2 and Ht3 genes. In addition, it is well known amongst maize pathologists that responses to the pathogen are very sensitive to environmental conditions, which confounds reproducibility in E. turcicum race screening using differential panels (Weems and Bradley, 2018).

Globally, GLS is the second most economically important foliar disease of maize after NLB, and is the most important foliar disease in the USA and Canada (Mueller et al., 2016, Mueller et al., 2020). Gray leaf spot disease was first reported in 1925 (Tehon and Daniels, 1925), and only became economically important in the late 1970s in the USA (Latterell and Rossi, 1983) and has since been reported in the Americas (Zhu et al., 2002; Juliatti et al., 2009; Neves et al., 2015; Mueller et al., 2020) and Asia (Manandhar et al., 2011; Liu and Xu, 2013). In Africa, GLS was first reported in 1988 in South Africa (Ward et al., 1997) and has since been reported in sub-Saharan Africa ( Figure 1 ; Table 1 ).

Cercospora zeae-maydis Tehon & E.Y Daniels (Tehon and Daniels, 1925), and Cercospora zeina Crous & U. Braun (Crous and Braun, 2003) cause GLS. Cercospora zeae-maydis is predominant in the Americas and Asia, whereas C. zeina is found in Africa, Brazil, some parts of Asia, and the Eastern corn belt of the USA (Wang et al., 1998; Goodwin et al., 2001; Okori et al., 2003; Meisel et al., 2009; Neves et al., 2015; Duan et al., 2022). Although other Cercospora spp. have been associated with GLS, namely Cercospora spp. CPC 12062, a single isolate from South Africa (Crous et al., 2006) and Cercospora sorghi var. maydis Ellis & Everh. reported in Kenya (Kinyua et al., 2010) and Brazil (Neves et al., 2015), their role in pathogenicity has not been determined. The rest of this review will, therefore, focus on C. zeina, which is the predominant pathogen in Africa.

Southern corn leaf blight, also known as Maydis leaf blight, was first reported in the USA in 1923 (Drechsler, 1925) and became a serious concern in the 1970s (Smith et al., 1970). Since then, SCLB reports have emerged from Western Europe, Asia and Africa (Munjal and Kapoor, 1960; Ullstrup, 1972; Fisher et al., 1976; Gregory et al., 1979; Carson et al., 2004; Singh and Srivastava, 2012; Manzar et al., 2022). The first report of SCLB in Africa was an outbreak in 1974 in South Africa, which resulted in the withdrawal of Texas male-sterile cytoplasm (T-cms) maize germplasm from the country’s breeding programs (Levings, 1990; Rong and Baxter, 2006) ( Table 1 ). Since then, no reports on SCLB have emerged from the country and on the rest of the continent until two decades later in Kenya (Mwangi, 1998) and recently on seeds in Nigeria (Biemond et al., 2013), thus indicating that SCLB can be a seedborne disease. These reports indicate that SCLB is present in Africa ( Figure 1 ), currently at levels where its severity and occurrence are still insignificant. However there is the potential of SCLB becoming a severe and serious phytosanitary threat to maize production in Africa.

Bipolaris maydis (Y. Nisik. & C. Miyake) Shoemaker is the causative pathogen of SCLB (Smith et al., 1970). Previously known as Cochliobolus heterostrophus, B. maydis has been adopted as the most widely accepted species name (Rossman et al., 2013). Four physiological races of B. maydis (races O, T, C, and S) are known globally (Smith et al., 1970; Wei et al., 1988; Levings, 1993; Manamgoda et al., 2014), while races C and S only exist in China (Wei et al., 1988; Wang et al., 2010; Zhao et al., 2012; Bruns, 2017) ( Table 2 ).

Overall, the USA was the first to report NLB, GLS, and SCLB in the early 20th-century (Drechsler, 1925; Tehon and Daniels, 1925). This could have been a result of the USA’ system of Land Grant Universities with farmer extension services, vigilant crop disease diagnosis activities, and adoption of hybrid maize breeding (Duvick, 2001). At that time, maize production expanded and the planting of monocultures increased, creating a high risk of disease if susceptible genotypes were planted (Dodd, 2000; Duvick, 2001).

Gray leaf spot differs from NLB and SCLB, with no known physiological races among its populations. Physiological races of E. turcicum and B. maydis follow the “gene for gene” hypothesis. Races are classified based on a pathogen’s ability to overcome resistance genes in maize inbred lines, often by loss of an avirulence gene recognized by a specific maize resistance gene (Leonard et al., 1989). However, the GLS-maize pathosystem has no known virulence genes or major resistance genes, so there are no known physiological races.

Upon infection of maize leaves, E. turcicum causes greyish lesions that start as chlorotic flecks and later mature into elliptical or cigar-shaped lesions from 2.5 to 30 cm in length (White, 1999) ( Figures 2 , 3 ). Disease establishment occurs within 6–18 h post-infection, and mature lesions develop within two weeks of host-pathogen interaction under favorable environmental conditions (Levy and Cohen, 1983; Bentolila et al., 1991; White, 1999; Kotze et al., 2019) ( Table 1 ; Figure 2 ). The pathogen invades the host through the epidermis and blocks the vascular tissues (Kotze et al., 2019). This causes plant lodging and a reduction in photosynthetic leaf area, leading to 30%–91% yield losses in cases of severe infections during silking and grain filling (Tilahun et al., 2012; Nwanosike et al., 2015; Jindal et al., 2019; Berger et al., 2020). The NLB disease is a splash- and wind-borne polycyclic disease that spreads via conidia from infected debris left in the fields ( Figure 2 ), and from secondary infections over long distances across fields (Schwartz and David, 2005).

Cercospora zeina invades the host leaf tissues intracellularly resulting in irregular chlorotic lesions that after 14 days post infection, mature into grey to tan linear rectangular lesions that run parallel with leaf veins (Latterell and Rossi, 1983; Ward et al., 1999) ( Figures 2 , 3 ). Extensive disease development results in the coalescence of the lesions, blighting, necrosis of the leaf tissue, reduced photosynthetic area and plant lodging (Paul and Munkvold, 2005; Lennon et al., 2016). The calculations made based on spore size (40 - 165 µm × 4 - 9 µm), wind speed (varies per location) and the height of vertical mixing of the atmosphere above the crop, estimate flight distances of spores to range between 0.1 - 40 km as wind speed increases from 1 to 10 m/s (Ward et al., 1999). The spores have also been reported to spread to a distance of 80 - 160 km annually, making it a fast-spreading disease (Manandhar et al., 2011). Yield losses due to GLS have been estimated to be 20–80% (Latterell and Rossi, 1983; Ward et al., 1999; Manandhar et al., 2011).

Irrespective of race, B. maydis infections in maize generally take between 12 to 18 h for fungal penetration, and 2 to 3 days to form mature lesions (Singh and Srivastava, 2012) ( Figure 2 ; Table 2 ). The race O causes small diamond-shaped lesions that elongate into rectangular lesions limited within veins to a length of 20-30 mm that later coalesce, resulting in the entire leaf blighting (Jeffers, 2004; Singh and Srivastava, 2012). The race T causes oval-shaped yellow to brown lesions that are larger than race O (Jeffers, 2004; Singh and Srivastava, 2012), and produces a T-cms-specific polyketide toxin (T toxin) that is specific to T-cms maize genotypes (Condon et al., 2018). This race, whose origin is still a mystery, was implicated in a serious epidemic in the USA in 1970 (Bruns, 2017). The SCLB disease thrives in hot and humid agroecosystems (Warren, 1975) ( Table 1 ). Yield losses of 10–40% due to SCLB infections have been reported, depending on the physiological race, environment and the maize hybrid grown (Fisher et al., 2012; Bruns, 2017).

All three fungal pathogens infect the same plant parts (leaves). They have similar growth requirements of moderate temperatures between 20°C and 30°C with relative humidity above 90% favoring disease establishment. Yield losses from individual pathogens can be 10-80%. Therefore, there is a need to determine the impact of combined infections by measuring the percentage of co-occurrence of these three diseases on a plant, field, and larger spatial scale to model their combined potential yield losses. This will facilitate the development of management strategies that target both single and co-infections.

Traditionally, plant disease diagnosis is performed through conventional visual field inspection of infected plant tissues (symptoms) using experienced technical human resources (Bock et al., 2010). Two standard scales (1–5 and 1–9, where 1 = resistant and 5 or 9 = susceptible) are being used to rate the severity of NLB (Abebe et al., 2008; Asea et al., 2009; Vivek et al., 2010; Kumar et al., 2011), GLS (Bubeck et al., 1993; Munkvold et al., 2001; Danson et al., 2008; Chung et al., 2011; Berger et al., 2014; Benson et al., 2015), and SCLB (Zwonitzer et al., 2009; Chung et al., 2010, Chung et al., 2011; Singh and Srivastava, 2012) ( Table 2 ). Some plant pathologists have preferred using the 1 to 9 scale in the field and later converted it to a scale of 1 to 5 using the following formula: 0.5 * (disease score (1 to 9 scale) + 1) = disease score (1 to 5 scale) (Vivek et al., 2010). These scales are used to assess disease severity over time which can be expressed as area under disease progress curve (AUDPC) or disease index (Ma et al., 2022). While this traditional method has been refined over time, it is plagued by the inherent subjectiveness of disease estimates and is time-consuming (Nutter et al., 1993; Bock et al., 2008; Poland and Nelson, 2011). Sometimes, morphological traits are misleading because of the similarities between disease symptoms. For instance, race O lesions of SCLB are sometimes mistaken for GLS, whereas the initial symptoms of NLB, GLS, and SCLB (chlorotic spots) can potentially lead to misidentification ( Figure 3 ).

Digital imaging techniques based on standard RGB images or hyperspectral images captured manually with cameras, mobile phones, or captured automatically with drones or by satellites hold great promise for crop disease diagnostics (Mohanty et al., 2016; DeChant et al., 2017). These methods involve training computer models with datasets of disease images which have been pre-classified by plant pathologists. The models are then tested on new sets of disease images to evaluate the accuracy of diagnosis. Examples of machine learning methods that have been used for plant disease diagnosis are spectral angle mapper (SAM), partial least squares regression (PLSR), support vector machines (SVMs), and convolutional neural networks (CNNs) (Xie et al., 2012; Stewart and McDonald, 2014; Mutka and Bart, 2015; Pauli et al., 2016; MuLaosmanovic et al., 2020).

This is a very active area of research that is also being applied to maize foliar diseases such as NLB and GLS with accuracies of greater than 90%, but it is still in its infancy since most models are being trained on images with only one disease symptom type (Zhang, 2013; Xu et al., 2015; Mohanty et al., 2016; Qi et al., 2016; Singh et al., 2016; DeChant et al., 2017; Zhang et al., 2018; Craze et al., 2022; Pan et al., 2022). Attempts are underway to diagnose individual foliar diseases with more field-realistic images with multiple disease symptom types; for example a neural network model was developed to identify GLS symptoms on maize leaves which had mixed symptoms of NLB, common rust, and white spot disease (Craze et al., 2022).

These high-throughput diagnostic methods are a foundation for understanding pathogen ecology, epidemiology, and biology. Their integration into management programs for several plant diseases has the potential to foster a more targeted approach for the prevention of epidemics.

For years, pathogen identification has relied on conventional techniques such as culturing, re-inoculation, microscopy, and biochemical assays (Sharma and Sharma, 2016). Morphological methods, which depend on visible signs of post-fungal infections, such as symptoms and fungal propagules, can be used to distinguish between E. turcicum and B. maydis, based on a hilum. Bipolaris maydis has a subtle hilum (Alcorn, 1988), while E. turcicum has a truncated, prominent hilum with a bubble (Leonard, 1974) ( Figure 4 ). Cercosporoid fungi, however, are mainly distinguished based on conidia, hila, and pigmentation of their asexual structures (Crous and Braun, 2003; Crous et al., 2006; Nsibo et al., 2021). Cercospora zeina conidia are characterized by their septate, hyaline, thin walls, smooth apex, and thick darkened and refractive hila ( Figure 4 ). These characteristics are similar to those of C. zeae-maydis. However, they differ in conidia shape, conidiophore length, and growth rate (Crous et al., 2006). Furthermore, C. zeae-maydis produces a photoactive phytotoxin, cercosporin, in vitro, whereas C. zeina does not (Goodwin et al., 2001; Crous et al., 2006; Swart et al., 2017).

Disease diagnosis of NLB, GLS, and SCLB based on morphological differences of the fungal morphology is possible. However, this often requires isolation and culturing of the fungal pathogen, which is time-consuming. This makes these approaches inadequate for accurate and timely species-level identification (McCartney et al., 2003; Pryce et al., 2003).

More advanced methods of identification, such as PCR-based amplification of nucleic acids and sequencing, are increasingly being employed for E. turcicum, C. zeina, and B. maydis. These methods can be more sensitive, are highly specific, faster, and require limited prior knowledge of the pathogen or expertise in plant pathology (McCartney et al., 2003; Ward et al., 2004). PCR amplification followed by sequencing of a fragment of the nuclear ribosomal DNAs (rDNAs), particularly the internal transcribed spacer (ITS), nested between conserved sequences of the 18S, 5.8S, and 28S rRNA gene regions, has been extensively employed for fungal species identification. The ITS marker is used as a universal barcode and is applicable for E. turcicum (Goh et al., 1998; Weikert-Oliveira et al., 2002; Ramathani et al., 2011; Haasbroek et al., 2014; Hernández-Restrepo et al., 2018), C. zeina (Dunkle and Levy, 2000; Crous et al., 2006; Meisel et al., 2009; Korsman et al., 2012; Liu and Xu, 2013; Bakhshi et al., 2015; Neves et al., 2015) and B. maydis (Goh et al., 1998; Emami and Hack, 2002; Manamgoda et al., 2012; Gogoi et al., 2014) ( Table 3 ). The ITS region has multiple (identical) copies in the genome and it’s PCR products are small (less than 1 Kb) which allow for easy PCR amplification, even in dilute or partially degraded DNA (White et al., 1973; Gardes et al., 1991; Lee and Taylor, 1992; Schoch et al., 2012).

Other available DNA targets for pathogen identification include regions of the translation elongation factor 1-α, calmodulin, β-tubulin, glyceraldehyde-3-phosphate dehydrogenase and mating type genes (Carbone and Kohn, 1999; James et al., 2006; Walker et al., 2012). Most of these gene regions have been employed for the identification of E. turcicum (Ramathani et al., 2011; Haasbroek et al., 2014; Hernández-Restrepo et al., 2018), C. zeina (Meisel et al., 2009; Bakhshi et al., 2015; Muller et al., 2016; Nsibo et al., 2019, Nsibo et al., 2021), and B. maydis (Leonard, 1974; Turgeon et al., 1995; Manamgoda et al., 2012; Tan et al., 2016) in Africa and around the world ( Table 3 ).

Various species-specific PCR diagnostic tools that do not require sequencing have been developed ( Table 4 ). For E. turcicum, mating-type genes are currently the only available species-specific diagnostic method. The amplification of PCR products of 608 bp and 393 bp using either a MAT1-1F or MAT1-2F primer together with MAT_CommonR primer indicates the presence of MAT1-1 or MAT1-2, respectively (Henegariu et al., 1997; Haasbroek et al., 2014) ( Table 4 ).

A species-specific PCR diagnostic that can differentiate three maize Cercospora species (C. zeina, C. zeae-maydis, and Cercospora sp.) was based on the histone H3 gene region (Crous et al. (2006). A multiplex PCR was used where universal primers CylH3F and CylH3R amplify a 389-bp fragment common to all three species. This universal primer pair is multiplexed with species-specific primers CzeaeHIST, CzeinaHIST, or CmaizeHIST in three separate PCR reactions for each unknown sample. Each reaction produces the common 389-bp fragment, and one of the three reactions will give a species-diagnostic 284-bp fragment (Crous et al., 2006) ( Table 4 ). Limitations of this approach is the need to do three PCR reactions per sample, and the requirement for highly optimized PCR conditions to ensure only the correct species-specific primer binds to the target.

A cytochrome P450 reductase (cpr1) has been used to distinguish C. zeina and C. zeae-maydis from other maize pathogens. The CPR1_F and CPR1-R primers amplify a 164bp product from C. zeina and C. zeae-maydis but not from other maize pathogens (Korsman et al., 2012). Furthermore, the assay can also be used to differentiate C. zeina and C. zeae-maydis based on melting temperature differences between the products that can be measured after a real-time PCR reaction (Korsman et al., 2012).

Rapid identification of C. zeina or C. zeae-maydis is routinely carried out using primers in the cercosporin biosynthesis CTB7 gene region, since different sizes are produced for C. zeina compared to C. zeae-maydis (Swart et al., 2017; Nsibo et al., 2019, Nsibo et al., 2021). Cercospora zeina mating type markers amplify fragments that differentiate C. zeina MAT1-1 from MAT1-2 strains, with no amplification from species like C. zeae-maydis (Muller et al., 2016; Nsibo et al., 2019, Nsibo et al., 2021) ( Table 4 ). For B. maydis, a multiplex mating-type PCR assay was optimized using primers MAT113, MAT123, and MATcon5 to amplify 702-bp (MAT1-1) and 547-bp (MAT1-2) fragments unique to B. maydis (Gafur et al., 1997) ( Table 4 ).

Recently, high throughput diagnostics are being employed for the early detection of crop diseases. For example, nano-material-enabled sensors including carbon-based, metal- and metal oxide-based nanomaterials, are currently being used in the early detection of plant diseases based on the changes in the physiology of plants (Li et al., 2021). In addition, the RNA programmable nuclease of CRISPR/Cas is a nucleic acid detection tool that is currently being employed for crop disease diagnosis (Zhang et al., 2020; Wheatley et al., 2021; Liang et al., 2024). These methods are yet to be optimized for the detection of E. turcicum, B. maydis and C. zeina.

Similar management strategies, including the use of tillage practices, rotation with non-host crops, and manipulation of environmental factors, are being used against NLB, GLS, and SCLB to reduce the amount of initial inoculum of the causal pathogens in the field (Ward and Nowell, 1998; Hooda et al., 2017). Deep tillage ensures the burial and destruction of pathogen inoculum in the soil (Payne and Waldron, 1983; Huff et al., 1988). Rotations for at least two years with non-host crops reduce fungal inoculum, especially in seasons with low disease incidence (Sharma and Payak, 1990; Ward and Nowell, 1998). In addition, manipulation of favorable environmental conditions (temperature, relative humidity, and leaf wetness) for pathogen development, especially early in the growing season, is crucial for hindering early season disease development (Ward and Nowell, 1998). Although these cultural practices are useful in managing these diseases and may be effective in low-risk areas, they are less effective when the disease is well-established (Ward et al., 1997; Lipps et al., 1998; Ward and Nowell, 1998).

Broad-spectrum fungicides, specifically demethylation inhibitor (DMI), quinone outside inhibitor (QoI), and succinate dehydrogenase inhibitor (SDHI), are effective against NLB, GLS, and SCLB, especially in susceptible hybrids (Reddy et al., 2013; Weems and Bradley, 2017; Dai et al., 2018; Neves and Bradley, 2019; Sun et al., 2023). Furthermore, Iturin A2, a Bacillus subtilis compound, was developed into a fungicide effective against B. maydis and other fungi (Gong et al., 2006; Kim et al., 2010; Ye et al., 2012). Iturin A2 could potentially treat other maize pathogens like E. turcicum and C. zeina and should be tested. Despite fungicide effectiveness, resistance has developed in other cereal pathogens like Zymoseptoria tritici, Pyrenophora teres f. teres, and Magnaporthe oryzae (Bohnert et al., 2018; Ellwood et al., 2019; Garnault et al., 2019). A few fungicide sensitivity studies have been conducted on E. turcicum and B. maydis to DMI, QoI, and SDHI fungicides. To date, all have revealed high sensitivities to fungicides, with no resistance buildup yet (Chapara et al., 2012; Weems and Bradley, 2017; Yuli et al., 2017; Hou et al., 2018). However, Africa lacks baseline sensitivity studies on E. turcicum, C. zeina, and B. maydis, despite increasing fungicide demands and use in large-scale field plantings. These studies are needed before fungicide resistance monitoring in maize foliar pathogens is initiated in Africa. Chemical control is also too expensive for many smallholder farmers. Therefore, affordable and long-lasting strategies such as host resistance through breeding need to be integrated and utilized.

Host plant resistance is the most economical, eco-friendly, and adjustable approach for maize disease management. Nelson et al. (2018) pointed out that effective resistance depends on the effect and strength of resistance genes in the host. Major genes provide complete or near-complete resistance, while quantitative resistance involves multiple minor genes with small additive effects (St. Clair, 2010).

Resistance to E. turcicum in maize is both qualitative and quantitative and can be used either separately or in combination with qualitative resistance following a gene-to-gene model (Welz and Geiger, 2000; Ogliari et al., 2005) ( Table 2 ). Qualitative resistance is mediated by Helminthosporium turcicum (Ht) resistance genes (Welz and Geiger, 2000). The four well-known Ht genes include Ht1, Ht2, Ht3, and Htn1, where the functions of Ht1, Ht2 and Ht3 have yet to be characterized (Van Staden et al., 2001; Yin et al., 2003; Ogliari et al., 2005). The Htn1 gene is highly conserved in E. turcicum hosts, particularly maize, sorghum, rice, and foxtail millet (Setaria italica), and encodes a wall-associated receptor-like kinase that confers resistance against race 12 (Hurni et al. (2015). Other resistance genes include HtP against races 123x and 23rx (Ogliari et al., 2005) and the recessive genes ht4 and rt that confer resistance to a wide range of E. turcicum races (Ogliari et al., 2005).

For GLS, a qualitative resistance locus is yet to be characterized. To date, GLS resistance is qualitatively inherited. Few major resistance quantitative trait loci including Qgls8 derived from teosinte (Zhang et al., 2017), and gRgls1 and qRgls2 from maize (Zhang et al., 2012) have been precisely defined.

Qualitative genes can confer resistance to B. maydis races. For instance, rhm gene mainly protects maize against race O and, to a lesser extent, race T strains (Zaitlin et al., 1993; Chang and Peterson, 1995). Chang and Peterson (1995) proposed a two-gene model in which two homozygous recessive genes, rhm1 and rhm2 which, in combination, increased host resistance to B. maydis. This was confirmed by the reduced lesion size as compared to the control experiment (Chang and Peterson, 1995) or the effect of an individual gene (Simmons et al., 2001). Qualitative resistance is effective against race O, while the best defense against race T is to avoid T-cms maize germplasm in breeding programs (Leonard, 1977a). Resistance to C and S races is so far unknown (Rong and Baxter, 2006).

Quantitative disease resistance is known to reduce disease severity and incidence, rather than completely eliminate the disease (Young, 1996; Poland et al., 2009). In recent years, QTL mapping studies have characterized several traits of crops, including resistance to several plant pathogens (Bernardo, 2008; Xu and Crouch, 2008).

Quantitative trait loci for resistance against NLB span the entire maize genome and have been identified in several mapping populations (Welz and Geiger, 2000; Wisser et al., 2006; Chen et al., 2016; Wang et al., 2018; Wende et al., 2018; Rashid et al., 2020). Using techniques such as genome-wide nested association mapping, QTLs with several potential candidate genes have been characterized and confirmed to confer resistance against NLB (Poland et al., 2011; Rashid et al., 2020). Although many QTLs are known to confer resistance to a broad spectrum of E. turcicum races, some QTLs are known to confer race-specific resistance to NLB (Chung et al., 2010, Chung et al., 2011).

Hot spots of QTLs conferring resistance to GLS span discrete regions of chromosomes 1, 2, 3, 4, 5, and 7 (Lehmensiek et al., 2001; Berger et al., 2014). Most notably, a candidate gene encoding a maize caffeoyl-CoA O-methyltransferase that confers quantitative resistance to GLS and SCLB has been cloned, implicating lignin and the phenylpropanoid pathway in maize defense against foliar diseases (Yang et al., 2017).

Many of these QTLs are derived from bi-parental crosses between susceptible and resistant genotypes tested under different disease pressures, germplasm backgrounds, and environmental conditions (Clements et al., 2000; Lehmensiek et al., 2001; Balint-Kurti et al., 2008; Berger et al., 2014). A majority of the QTLs are environment-specific; however, many QTLs expressed in several environments have also been characterized (Berger et al., 2014). These can be introgressed into maize genotypes grown in different environments. Molecular breeding to develop GLS resistant maize for small-holder farmers in Africa has been reported (Kibe et al., 2020). Recently, advanced populations of maize developed by CIMMYT were used in a combination with linkage and association mapping with genome wide SNP markers to identify QTLs for GLS and NLB resistance in East Africa (Omondi et al., 2023).

Using recombinant inbred lines (RILs), Carson et al. (2004) identified 11 QTLs spanning chromosomes 1, 2, 3, 4, 7, and 10, which are associated with SCLB resistance. Additional SCLB resistance QTLs have been characterized from different maize genotypes at different maturity stages (Balint-Kurti and Carson, 2006; Zwonitzer et al., 2009; Negeri et al., 2011).

Thus, qualitative, and quantitative resistance breeding are crucial for managing NLB, GLS, and SCLB. Maize geneticists have made great progress in identifying the genomic loci associated with resistance to one or more of these three diseases. To identify these loci tools such as the nested association mapping (NAM) panel, and the availability of genome wide SNP markers (Benson et al., 2015) have been employed. Importantly, some loci appear to confer multiple disease resistance, and recent advances have validated some QTL in independent maize populations (Lopez-Zuniga et al., 2019). A limitation is that most of these studies have been carried out in the USA and Europe often with inbred lines adapted to these temperate climates, with disease scoring carried out against local populations of each pathogen (Technow et al., 2013). Molecular marker-assisted breeding tools can now be employed to introgress these resistance alleles into germplasm adapted to maize-producing regions in Africa to determine whether crop protection is conferred against pathogen populations on the continent.

Microsatellite markers have replaced earlier techniques such as Random Amplified Polymorphic DNA (RAPD) and Amplified Fragment Length Polymorphism (AFLP) markers to study the global population structure of E. turcicum. Reports from Asia, Europe, the Americas, and Africa show that E. turcicum is genetically and genotypically diverse, with higher diversity in Asia and Africa (Borchardt et al., 1998b; Ferguson and Carson, 2004; Dong et al., 2008; Haasbroek et al., 2014; Tang et al., 2015; Nieuwoudt et al., 2018). European E. turcicum populations are characterized by low genetic diversity (Borchardt et al., 1998a; Turgay et al., 2021) and are partially differentiated due to the Alps (Borchardt et al., 1998a). African geographic boundaries, like mountains and the large lakes of the Rift Valley, may affect E. turcicum population structure, although this has not yet been investigated. Sexual recombination is a major evolutionary factor in E. turcicum’s global population structure, even in Europe, where sexual occurrences are rare, based on the frequency distribution of mating types and a lack of observed sexual structures in nature or in the laboratory (Fan et al., 2007; Turgay et al., 2021). Mating-type genes were found to be equally distributed and frequent in several countries where mating-type studies have been conducted, except in Europe, indicating sexual recombination (Haasbroek et al., 2014; Human et al., 2016; Weems and Bradley, 2018). Even though S. turcica, the sexual stage of E. turcicum is very rare in nature, with the only existing report being from Thailand (Bunkoed et al., 2014), it has been induced under laboratory conditions using Sach’s medium with barley culm (Moghaddam and Pataky, 1994; Fan et al., 2007).

Population genetic analysis allows for the study of physiological race distribution, potential race re-emergence, and the identification of alternative hosts. However, limited knowledge exists on the population genetic diversity and race diversity of E. turcicum in Africa, with the exception of populations in Kenya and South Africa. Therefore, understanding the population structure of E. turcicum in maize-growing African countries is needed.

Prior to the classification of the GLS causal pathogens into two distinct species (Crous et al., 2006), all studies conducted on GLS referred to the disease as being caused by C. zeae-maydis (Latterell and Rossi, 1983; Lipps, 1998; Ward et al., 1999; Okori et al., 2003). As more studies based on taxonomy, molecular and phylogenetic tools emerged, it became evident that there were two sibling species, C. zeina (formerly known as C. zeae-maydis type II) and C. zeae-maydis. Initial molecular studies of Cercospora isolated from maize in Africa indicated the presence C. zeae-maydis type II (C. zeina) using AFLP and RFLP markers with genetic relatedness to Type II, and not Type 1 isolates in the Americas (Wang et al., 1998; Dunkle and Levy, 2000; Okori et al., 2003; Liu and Xu, 2013; Muller et al., 2016). Subsequent surveys in sub-Saharan Africa confirmed that C. zeina was the only causal pathogen in Africa since to date, no isolates of C. zeae-maydis have been found in a collection of more than 1000 isolates from East and Southern Africa (Nsibo et al., 2021).

Populations studies have shown that C. zeina is a highly diverse pathogen in Africa with a partially defined population structure within and among countries (Okori et al., 2003, Okori et al., 2015; Muller et al., 2016; Nsibo et al., 2019, Nsibo et al., 2021). Given that maize is non-native to Africa, the dominance of C. zeina on the continent is attributed to more than one introduction, followed by several sexual recombination events, intra-continent gene flow, migration, and local adaptations (Nsibo et al., 2021). This work was further refined by genome sequencing of 30 isolates of C. zeina from two countries in East Africa (Kenya, Uganda) and three countries in Southern Africa (Zambia, Zimbabwe and South Africa) (Welgemoed et al., 2023). This showed population differentiation but no major differences in diversity indices between regions, indicating two possible introductions over the approximately 500-year time period since maize entered the continent (Welgemoed et al., 2023). The study of C. zeina populations from other continents and the search for alternate hosts is underway to explore alternative hypotheses, including the hypothesis that C. zeina in Africa was derived from a host jump from an unidentified grass species onto maize (Dunkle and Levy, 2000; Crous et al., 2006).

Although sexual structures of C. zeina have not been observed under field and laboratory conditions, cryptic sexual recombination has been suggested based on the presence of mating-type genes with equal distribution and frequency, in addition to the low levels of linkage disequilibrium among some populations (Groenewald et al., 2006; Muller et al., 2016; Nsibo et al., 2021). Possible explanations for the failure of laboratory experiments to induce or discover the sexual stage of C. zeina may include the absence of environmental parameters that the pathogen encounters in nature to trigger sexual recombination. It could also be a failure to systematically monitor the development of an ascocarp in this presumably asexual pathogen (Dyer et al., 1992) or fertility decline in the pathogen (Dyer and Paoletti, 2005).

There is clear evidence that C. zeina is a well-established pathogen in Africa with the potential to threaten food production on the continent if not monitored to determine its diversity and migration patterns and deploy more effective management strategies.

There is limited information regarding the genetic diversity of B. maydis. RAPD markers have been used to understand the genetic structure of B. maydis populations, especially in India, where most reports have emerged. In India, B. maydis has been reported to be highly diverse, with little to no population differentiation (Karimi, 2003; Jahani et al., 2011; Gogoi et al., 2014), suggesting that gene flow plays a major evolutionary role in the population structure of the pathogen. Furthermore, the physiological race O is the most predominant race in India (Gogoi et al., 2014; Pal et al., 2015), with high genetic variability among isolates of the same race (Gafur et al., 2002; Pal et al., 2015). Sexual recombination is another major evolutionary factor driving observed genetic diversity (Gafur et al., 1997, Gafur et al., 2002). The availability of the B. maydis genome (Condon et al., 2013) offers a unique opportunity to develop more robust molecular markers, such as microsatellite and single-nucleotide polymorphism (SNP) markers, that can be exploited to enable comprehensive studies of the pathogen from all the countries where the disease exists.

Bipolaris maydis is a potential threat to maize production in Africa although it has only been reported in Kenya (Mwangi, 1998) and South Africa (Rong and Baxter, 2006). The risk of its spreading to other countries is heightened by the fact that B. maydis is both an air- and seed-borne pathogen (Aylor and Lukens, 1974; Manoj and Agarwal, 1998; Biemond et al., 2013). Due to increasing anthropogenic activities and global trade, unreported incidences of the pathogen in the rest of Africa are possible. Therefore, countries where SCLB has not yet been reported must be vigilant through the establishment of phytosanitary regulations and bodies that test and ensure the movement of healthy seeds across geographical boundaries. Methods such as roguing, seed dressing, and proper storage to minimize contamination have been suggested as alternative ways to ensure seed health (Biemond et al., 2013).