TYPE Original Research
PUBLISHED 07 November 2023
DOI 10.3389/fgene.2023.1282673
Combination of linkage and
OPEN ACCESS association mapping with
EDITED BY
Krishnanand P. Kulkarni, genomic prediction to infer QTL
Delaware State University, United States
REVIEWED BY regions associated with gray leaf
Qin Yang,
Northwest A&F University, China
Zhenhua Wang, spot and northern corn leaf blight
Northeast Agricultural University, China
*CORRESPONDENCE resistance in tropical maize
Mathews M. Dida,
mitodida@yahoo.com
Manje Gowda, Dennis O. Omondi1,2, Mathews M. Dida1*, Dave K. Berger3,
m.gowda@cgiar.org
Yoseph Beyene4, David L. Nsibo3, Collins Juma4,2,
RECEIVED 24 August 2023
ACCEPTED 18 October 2023 Suresh L. Mahabaleswara4 and Manje Gowda4*
PUBLISHED 07 November 2023
1Department of Crops and Soil Sciences, School of Agriculture, Food Security and Environmental
CITATION
Sciences, Maseno University, Kisumu, Kenya, 2Crop Science Division Bayer East Africa Limited, Nairobi,
Omondi DO, Dida MM, Berger DK,
Kenya, 3Department of Plant and Soil Sciences, Forestry and Agricultural Biotechnology Institute (FABI),
Beyene Y, Nsibo DL, Juma C,
University of Pretoria, Pretoria, South Africa, 4The Global Maize Program, International Maize and Wheat
Mahabaleswara SL and Gowda M (2023),
Improvement Center (CIMMYT), Nairobi, Kenya
Combination of linkage and association
mappingwith genomic prediction to infer
QTL regions associated with gray leaf
spot and northern corn leaf blight
resistance in tropical maize.
Front. Genet. 14:1282673. Among the diseases threateningmaize production in Africa are gray leaf spot (GLS)
doi: 10.3389/fgene.2023.1282673 caused by Cercospora zeina and northern corn leaf blight (NCLB) caused by
COPYRIGHT Exserohilum turcicum. The two pathogens, which have high genetic diversity,
© 2023 Omondi, Dida, Berger, Beyene, reduce the photosynthesizing ability of susceptible genotypes and, hence, reduce
Nsibo, Juma, Mahabaleswara andGowda. the grain yield. To identify population-based quantitative trait loci (QTLs) for GLS
This is an open-access article distributed
under the terms of the Creative and NCLB resistance, a biparental population of 230 lines derived from the tropical
Commons Attribution License (CC BY). maize parents CML511 and CML546 and an association mapping panel of
The use, distribution or reproduction in 239 tropical and sub-tropical inbred lines were phenotyped across multi-
other forums is permitted, provided the
original author(s) and the copyright environments in western Kenya. Based on 1,264 high-quality polymorphic
owner(s) are credited and that the original single-nucleotide polymorphisms (SNPs) in the biparental population, we
publication in this journal is cited, in identified 10 and 18 QTLs, which explained 64.2% and 64.9% of the total
accordance with accepted academic
practice. No use, distribution or phenotypic variance for GLS and NCLB resistance, respectively. A major QTL
reproduction is permitted which does not for GLS, qGLS1_186 accounted for 15.2% of the phenotypic variance, while
comply with these terms. qNCLB3_50 explained the most phenotypic variance at 8.8% for NCLB
resistance. Association mapping with 230,743 markers revealed 11 and 16 SNPs
significantly associated with GLS and NCLB resistance, respectively. Several of the
SNPs detected in the association panel were co-localized with QTLs identified in
the biparental population, suggesting some consistent genomic regions across
genetic backgrounds. These would be more relevant to use in field breeding to
improve resistance to both diseases. Genomic prediction models trained on the
biparental population data yielded average prediction accuracies of 0.66–0.75 for
the disease traits when validated in the same population. Applying these prediction
models to the association panel produced accuracies of 0.49 and 0.75 for GLS and
NCLB, respectively. This research conducted in maize fields relevant to farmers in
western Kenya has combined linkage and association mapping to identify new
QTLs and confirm previous QTLs for GLS and NCLB resistance. Overall, our
Frontiers in Genetics 01 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
findings imply that genetic gain can be improved inmaize breeding for resistance to
multiple diseases including GLS and NCLB by using genomic selection.
KEYWORDS
maize, gray leaf spot, northern corn leaf blight, quantitative trait loci, associationmapping,
genome-wide association study
1 Introduction with the traits of interest. Previous QTL studies have mapped
QTL for resistance to GLS and NCLB on all 10 maize
Despite its importance, maize production in Kenya is still low chromosomes (Berger et al., 2014; Chen et al., 2015). A
with an estimated average production of 1.8 t/ha-1 among number of these QTLs have been fine-mapped with others
smallholder farmers when compared to the country’s potential cloned and the molecular mechanisms underlying such QTL
production yield of 6 t/ha-1 (Munialo et al., 2019). This is partly characterized. Additionally, QTL mapping offers the advantage
due to the threat of highly destructive and virulent fungal pathogens of mapping as early as at the F2 populations; however, this is
limiting crop production (Beyene et al., 2019). In this context, characterized by the limited number of recombination events
northern corn leaf blight (NCLB), also known as northern leaf captured and sizeable confidence interval (Challa and Neelapu,
blight (NLB) or Turcicum leaf blight (TLB), caused by Exserohilum 2018; Rashid et al., 2020).
turcicum (Pass.) (Leonard and Suggs, 1974), and gray leaf spot (GLS) Genome-wide association studies (GWAS) attempt to overcome
caused by Cercospora zeina Crous & U Braun (Crous et al., 2006) on the drawbacks of QTL mapping as they utilize age-old
the African continent (Nsibo et al., 2019; Nsibo et al., 2021) are the recombination events in a large array of unrelated individuals
most lethal and economically significant foliar diseases of maize leading to high-speed decay of linkage disequilibrium (Xiao et al.,
(Beyene et al., 2019; Sserumaga et al., 2020). 2017; Kolkman et al., 2020). GWAS studies dig into the entire
The two diseases reduce the photosynthetic potential of a plant genome of different varieties (considering the SNPs present in the
and eventually decrease the grain yield (Saito et al., 2018). GLS is genotypic data) to establish the link between genotypic variations
characterized by tan-to-gray rectangular lesions that are limited and the corresponding trait (Challa and Neelapu, 2018). Kibe et al.
within the leaf veins (Korsman et al., 2012). It is associated with yield (2020a) combined the use of linkage mapping and GWAS to detect
losses of approximately 30%–50%, particularly when using the significant SNPs and QTL conditioning resistance to GLS in an
susceptible lines (Kinyua et al., 2010). On the other hand, NCLB ImprovedMaize for African Soils (IMAS) diversity panel and a set of
is characterized by long elliptical cigar-shaped lesions on leaves that DH populations in Kenya. Several putative candidate genes involved
are gray–green (Welz, 1998). NCLB or TLB has been reported to in the transportation channel were identified to have a role in plant
cause yield reductions of 36%–72% in susceptible maize genotypes defense. In the present study, we attempted to validate some of the
(Berger et al., 2020). The high genetic diversity reported for C. zeina GLS resistance QTLs reported in the study by Kibe et al. (2020a) by
and E. turcicum in Kenya (Borchardt et al., 1998; Nsibo et al., 2021) using a common tropical parent CML511.
could lead to recombining pathogen populations, hence posing a Genomic prediction (GP) is another promising genomic tool
greater risk to the vulnerable susceptible lines (McDonald and that has been applied successfully in plant breeding programs
Linde, 2002). Therefore, there is a need to continuously discover (Crossa et al., 2017). Previous studies indicated the potential of
new sources of resistance. GP to increase genetic gain and reduce the time taken in breeding
The doubled haploid (DH) technology offers the fastest programs significantly (Beyene et al., 2019; Kibe et al., 2020a;
alternative to achieve 100% homozygosity (attained in two Kibe et al., 2020b). In contrast to genetic mapping which
generations) which is essential for a mapping population and identifies significant marker–trait associations, GP uses all
population improvement (Prasanna et al., 2021). To complement markers available to estimate their effects, thus providing a
the DH technology, genotyping by sequencing platforms such as powerful approach to account for any effects that might have
Diversity Arrays Technology (DArT) offers a high-throughput been missed by either genetic or association mapping (Beyene
platform for genotyping single-nucleotide polymorphism (SNP) et al., 2019). GP exhibits a close relationship to GWAS owing to
markers (Kilian et al., 2012; Sansaloni et al., 2020). The DArTseq the large genomic and phenotypic datasets used by the methods
platform is purposefully a powerful tool for genome-wide discovery (Beyene et al., 2021).
of SNP markers without prior sequence information (Wenzl et al., However, this does not mean the complete withdrawal of genetic
2004). In addition, it generates high-density linkage maps, and it is mapping but rather the incorporation of the two in genetic studies as
also cost-competitive (Jaccoud et al., 2001; Sánchez-Sevilla et al., complementary approaches since each provides considerable
2015). advantages. With this background, the objectives of this study
Complex traits such as GLS and NCLB resistance are were as follows: (1) to phenotypically characterize an elite
controlled by polygenic genes with minor effects that are tropical DH population and 240 tropical and sub-tropical maize
distributed throughout the genome (Welz and Geiger, 2000; inbred lines panel for their responses to GLS and NCLB, including
Wisser et al., 2006; Poland et al., 2011; Van Inghelandt et al., correlation with other agronomic traits; (2) to identify population-
2012; Chen et al., 2015; Ding et al., 2015). Mapping of the based common QTL regions and significant SNPs using GWAS and
quantitative trait loci (QTLs) based on linkage analysis is a linkage mapping; and (3) to assess the usefulness of GP in breeding
powerful tool for identifying the genomic regions associated for GLS and NCLB resistance in tropical maize.
Frontiers in Genetics 02 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
2 Materials and methods covering >10% of the leaf area, 7 for GLS and NCLB lesions
dominating the leaf area on all the leaves with 50% of the maize
2.1 Study sites and genetic material leaf surface diseased, 8 for GLS and NCLB lesions prevalent on all
the leaves of the maize plant with 80% of the maize leaf surface
This study used (i) a biparental DH population derived from the diseased, and 9 for GLS and NCLB lesions prevalent on all the
tropical×tropical germplasm CML511×CML546 inbred lines and leaves of the maize plant with the maize plant exhibiting a gray
(ii) an association panel made up of a collection of 239 tropical and appearance with >80% of the maize leaf area diseased. For both
sub-tropical maize inbred lines with early and intermediate maturity GLS and NCLB, the DS scores over five intervals were used to
in Eastern Africa, representing some of the genetic diversity (for low calculate the area under the disease progress curve (AUDPC,
N, drought, and biotic stresses, Beyene et al., 2021; Prasanna et al., Shaner, 1977). For the GWAS panel, both DS data were recorded
2021). The association panel was evaluated in three locations in at the dough stage of the plants. For DH population, data were
western Kenya, at Kitale (1.0191° N and 35.0023° E, 1900 masl); also collected on days to anthesis (AD, the number of days from
Shikutsa (0°16′57.83″N and 34°45′6.71″E, 1561 masl); and planting to when 50% of the plants in a plot were shedding
Kakamega (0°17′3.19″N and 34°45′8.24″E, 1535 masl). The pollen), days to silking (SD, the number of days from planting to
biparental population was evaluated across different ecologies in when 50% of maize crops in a plot were showing silk), plant
western Kenya; Maseno University field demonstration site in 2018 height (PH, cm), and ear height (EH, cm). Maize development
(0°00′18.2″S and 34°35′43.5″E, 1500masl), Maseno 2019 stages were recorded using the scale of Purdue University (http://
(0°00′07.0″S and 34°35′41.9″E, 1503 masl), and farmer’s field in extension.entm.purdue.edu/fieldcropsipm/corn-stages.php).
Kabianga 2018 (0°25′24.1″S and 35°07′31.7″E, 1780 masl).
2.4 Statistical analysis of the phenotypic data
2.2 Experimental design
Multi-environment trial analysis with R for windows
Two hundred and thirty (230) entries (228 DH lines and two (META-R) version 6.0 (Alvarado et al., 2015) was used to
parents) of the biparental population were planted in a 5 × 46 alpha obtain the best linear unbiased estimations (BLUEs) and best
lattice design, randomized, and replicated three times at each site by linear unbiased predictions (BLUPs). In addition to BLUEs and
using the CIMMYT’s field book (Vivek et al., 2007). The association BLUPs, META-R was also used to compute the genetic
panel was also planted in 5 × 48 alpha lattice design, randomized, correlations among all the variables and among environments,
and replicated two times in each of three environments. least significant difference (LSD), grand mean, variance
Experimental plots consisted of 3 m long single rows with the components, coefficients of variation (CV), and broad-sense
rows spaced at 0.6 m apart. Adjacent plots were planted 0.75 m heritability for all the variables. Analysis of the phenotypic
apart with an alley of 1.2 m at the end of each plot. Each plot was data for both biparental population and association panel was
planted with 13 hills, with two seeds getting planted per hill. conducted both within and across environments.
Thinning was later conducted to one plant per hill. Border rows The BLUEs and the BLUPs were calculated for DS of GLS and
of susceptible genotypes were also planted to act as spreaders of the NCLB, PH, EH, AD, SD, and the AUDPC which were the response
pathogen. The experimental plots were managed using standard variables. The columns in the input files were selected to be the
agronomic practices. factor names with the environment, replicate, block, and genotype as
the independent variables. For analysis across environments using a
lattice design, the following linear mixed model was used.
2.3 Phenotypic evaluation and data Yijkl = µ +Envi + Repj(Envi) +Blockk(Envi Repj) +Genl +
collection Envi ×Genl + εijkl.
From the aforementioned equation, Yijkl represents the
GLS and NCLB disease severity (DS) were scored on a per- performance of the trait of interest, µ corresponds to the all-
row basis using an ordinal 1–9 scale adapted from the work of inclusive mean, Envi represents the effect of the ith environment,
Berger et al. (2014) for GLS and Berger et al. (2020) for NCLB. For Repj(Envi) represents the effect of the jth replication within the
DH population, DS ratings for GLS and NCLB were taken once ith environment, Blockk(Envi Repj) represents the effect of the
per week for at least 5 weeks starting on average at 15 days after kth incomplete block within the jth replication in the ith
flowering (R2; reproductive stage two). All the data were environment, Genl represents the effect of the lth genotype,
collected using the CIMMYT’s field book (Vivek et al., 2007). Envi ×Genl represents the environment by genotype
The DS scale was used as follows: score 1 for no GLS or NCLB interaction, and εijk is the error variance. When calculating the
lesions visible on the entire plant, 2 indicated close inspection of BLUEs, genotypes and covariates were considered fixed effects of
each leaf is necessary to find lesions, 3 indicated lesions are more the model while other terms were included as random effects of
easily seen but are majorly restricted to leaves lying below the the model. The covariate was considered as fixed effect of the
ears, 4 indicated individual lesions are just becoming visible on model while all other terms were included in the random effects
the ear leaf and the leaves above the ears, 5 indicated lesions are of the model to estimate the BLUPs. Heritability for the different
more visible on the leaves above the ears, with the infections traits was calculated as the ratio of the estimated genotypic
capturing <10% of the top leaves, 6 indicated lesions are more variance to the estimated phenotypic variance (Knapp et al.,
easily seen on the leaves above the ear leaf with infections 1985).
Frontiers in Genetics 03 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
2.5 Genotyping and QTL mapping by Berger et al. (2014). The individual QTLs were assigned names
according to the QTL, trait name, chromosome, and marker
The CML511×CML546 DH and parental lines were grown in a position, as described in the work of Kibe et al. (2020a).
greenhouse. Maize leaf tissue samples were collected from young,
healthy seedlings at the V3 stage (3–4 weeks old), stored at −80°C,
and later freeze-dried for 72 h. High-quality genomic DNA was 2.6 Genotyping and association mapping
isolated from freeze-dried tissues using the standard CIMMYT
laboratory protocol (CIMMYT, 2005). The DH lines were The DNA of all 239 inbred lines of the association panel was
genotyped using DArTSeq in Canberra, Australia. Approximately extracted from seedlings at the 3–4 leaf stage and genotyped using
15,000 SNPs were used for further quality checks (Murithi et al., the genotype-by-sequencing (GBS) platform at the Institute for
2021). Trait Analysis by aSSociation Evolution and Linkage Genomic Diversity, Cornell University, Ithaca, United States,
(TaSSEL) (Bradbury et al., 2007) was used to summarize using high-density markers, as per the procedure described by
genotype data by site, determine the allele frequencies, and Elshire et al. (2011). SNP calling and imputation were conducted
implement quality screening. All SNP variants that were at Cornell University. For SNP calling, raw data in a FASTQ file
monomorphic between the two parents, had heterozygosity together with the barcode information and Tags On Physical Map
of >0.05 and a minor allele frequency of <0.05, were filtered, and (TOPM) data, which had SNP position information, were used. We
1,264 high-quality SNPs were retained for QTL mapping. used TOPM data from AllZeaGBSv2.7 downloaded from Panzea
Redundant markers were removed using the BIN tool in QTL (https://www.panzea.org/), which contained information for
IciMapping v.4.2 (Meng et al., 2015). In the parameter setting 955,690 SNPs mapped with B73 AGPv2 coordinates. SNP calling
window for general information, eight functionalities were used was then performed using the TASSEL-GBS pipeline (Glaubitz et al.,
to define the mapping population. In the indicator row, 1 was 2014; Wang et al., 2020). Using TASSEL ver5.2 (Bradbury et al.,
selected to denote QTL mapping in actual populations and 3 as the 2007), SNPs with a heterozygosity of >5%, MAF of >0.05, and
population type as this study used a DH population. Kosambi was minimum count of 90% were excluded by filtering from raw GBS
set as the mapping function, marker position as the marker space datasets, and 230,743 high-quality SNPs were retained for further
type, 10 as the number of chromosomes, 230 as the size of the analysis in the association panel. To explore the population structure
mapping population, and 6 as the number of traits. and the ultimate number of subpopulations, principal component
The number of markers in each chromosome was specified in analysis (PCA) as described by Price et al. (2006) was conducted in
the chromosome information part. The scores for all the DArTseq TASSEL using SNPs across all panels. The first three principal
markers were transformed into genotype codes following the scores components were instrumental to visualize the existing
of the parents (2 denoted the marker type of the first parent, population stratification within the association panel, and this
0 denoted the second parent, 1 for the F1 marker type, was clearly displayed in a 3D plot. The PCA plots of the
and −1 for the missing markers) (Meng et al., 2015). The genetic association panel were computed using 230,743 SNPs; we then
linkage maps were constructed using the MAP functionality of QTL plotted the variance (y-axis) against the principal components
IciMapping v.4.2 (http://www.isbreeding.net). Three steps were (x-axis) to estimate the number of clusters within the population
followed in linkage map construction: grouping, ordering, and (Sanchez et al., 2018). The data point at which the increase in the
rippling. The logarithm of odds score was set at 3.0 for grouping. number of principal components did not result in an increase in
Ordering was performed using the ordering instruction with the variance (leveling off) indicated the number of subgroups within the
nnTwo Opt algorithm. The sum of adjacent recombination panel. To estimate the amount of genetic relatedness among
frequencies (SARF) as the criterion and window size of 5 as the individuals, a kinship matrix was explored.
amplitude were used to ripple the marker sequence and to fine-tune GWAS was performed with different models to compare and
the chromosome orders. All the outputting functionalities were choose the appropriate model with relatively less false positives. To
checked, and the map was drawn using the MAP functionality detect marker–trait associations, GWAS was performed using the
(Meng et al., 2015). following models: (1) mixed linear models (MLMs; PCA + K + G)
In the phenotypic data, the BLUPs for the different traits were that incorporated PCA, kinship (K), and genotypic data as
used as the input files for QTL identification across environments covariates; (2) the general linear model (GLM; PCA + G) which
(Littell et al., 2007). The input file was loaded onto the project of incorporated the genotype data (G) and the PCA (Q) that both acted
IciMapping v.4.2 (Meng et al., 2015). In the parameter setting as fixed effects to correct for the population structure; and (3) Fixed
window ICIM-ADD, other parameters, such as step in scanning and random model Circulating Probability Unification (FarmCPU),
represented by cM and stepwise regression of phenotype on marker in which the kinship (random) and the three-component analysis
variables, were defined. An LOD threshold of 3.0 and (fixed) were identified as covariates (Lipka et al., 2012). Single-locus
1000 permutations at α = 0.01 were set to declare the significant GWAS models such as the GLM are characterized by high false
QTL. The percentage of total phenotypic variance explained by an positive rates, as a complementary model, and the MLM utilizes the
individual QTL was determined using stepwise regression. To Bonferroni correction to overcome the challenge of false positive
ascertain the actual locations of the QTL for all the traits on the rates and identify the loci of interest (Khan et al., 2021). The software
chromosomes, the physical position of the identified QTL was TASSEL (Bradbury et al., 2007) was instrumental to run the GLM +
assigned based on the known physical position of the linked PCA and MLM. The − log 10 p values for all the analyzed SNPs for
markers and also available at the maize genetics and genomics both GLS-DS and NCLB-DS were used to construct the Manhattan
database (http://www.maizegdb.org/data_center/map), as described plots. Q–Q plots were plotted from the estimated -log10 (p) from the
Frontiers in Genetics 04 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
association panel for GLS-DS and NCLB_DS traits. Analysis of the evaluations show that CML511 is resistant and CML546 is
association panel was conducted in TASSEL based on moderately susceptible to GLS. A large portion of the biparental
230,743 filtered SNPs. The R package ‘FarmCPU’ with the population was extensively blighted by NCLB. Transgressive
Genome Association and Prediction Integrated Tool (GAPIT) segregation was observed in the population for GLS, NCLB, and
was used for GWAS analysis (Tang et al., 2016). The false AD, as some of the genotypes were more resistant or susceptible
discovery rate (FDR) was calculated for significant associations compared to the parental lines in the biparental population
using the Benjamini and Hochberg (1995) correction method, (Figure 1).
with 8 × 10−5 as the threshold. To summarize GWAS results per The frequency distribution of GLS DS scores was fairly skewed
chromosome, Manhattan scatter plots were generated by plotting toward resistance in the biparental population (Figure 1A). The
genomic positions of the SNPs against their negative log base 10 of frequency distribution of the DS scores for NCLB in the biparental
the p-values obtained from the GWAS model, with the F-test for the population followed an approximately normal distribution, as
null hypothesis on the y-axis. shown in Figure 1A. The wide segregation of DS and AUDPC
SNPs detected in the association panel were examined as scores for NCLB provided more evidence for the quantitative
polymorphisms in linkage disequilibrium with putative candidate control of resistance. The DH population also exhibited
genes from the B73 reference gene set (https://phytozome-next.jgi. continuous distribution for the days to anthesis, days to silking,
doe.gov/jbrowse/index.html?data; Zeamays Zm-B73-REFERENCE- plant height, and ear height (Figure 1A). The GLS and NCLB DS
NAM-5.0.55) (Goodstein et al., 2012). Putative candidate genes were scores for the maize association panel ranged from 1.5 to 6
selected by delving into the information from Gene Ontology, Kyoto (Figure 1B), which were similar to the biparental DH population
Encyclopedia of Genes and Genomes (KEGG), and protein families scores, although the association panel trended toward higher GLS
(Pfam) (Ashburner et al., 2000; Kanehisa and Goto, 2000; Bateman and lower NCLB scores. On the other hand, the use of the nine-point
et al., 2004). rating scale revealed extensive phenotypic variation in resistance to
Genomic prediction was carried out with ridge regression BLUP GLS and NCLB across the association panel, with the panel
(Zhao et al., 2012) within a biparental DH population for GLS, harboring more resistant lines (Figure 1B). The association panel
NCLB, and agronomic traits at five-fold cross-validation. The was also characterized by shorter days to anthesis compared to the
BLUEs across environments were used for the analysis. The same biparental population (Figure 1B).
set of 1,264 high-quality uniformly distributed SNPs with no
missing values and MAF>0.05 were used. For the maize
association panel, quality screening criteria of SNPs with 3.2 Correlations between environments and
MAF >0.10 and no missing values were applied, and finally, variables
8,365 SNPs from the 230,743 SNPs were retained for the
analyses. The prediction was ‘within population’, where training In the DH population, the correlation between environments for
and validation sets were derived from within the biparental GLS DS was positive and highly significant (p < 0.001)
population. For each trait, 100 iterations were carried out for the (Supplementary Table S2). A moderately high correlation was
sampling of the training and validation sets. The prediction accuracy observed between environments for NCLB DS scores
was calculated as the correlation between the observed phenotypes (Supplementary Table S2). The correlation across environments
and genomic estimated breeding values (GEBVs) divided by the for AD and SD was also highly significant at p < 0.001
square root of heritability (Dekkers, 2007). (Supplementary Table S2).
The analyses of variance revealed significant genotypic and
genotype × environment interaction variances for GLS, NCLB
3 Results DS, and AUDPC values as well as other agronomic traits
(Table 1). Heritability estimates on an entry mean basis were
3.1 Phenotypic data 0.81, 0.81, 0.79, and 0.80 (Table 1) for GLS DS, AUDPC for
GLS, NCLB DS, and AUDPC for NCLB, respectively in the DH
As expected for western Kenya, there was high natural disease population. However, the heritability estimates for DS in the
pressure for both NCLB and GLS for all field trials of the biparental association mapping panel were lower (0.35 for GLS and 0.64 for
CML511×CML546 DH population (Figure 1A), as well as the NCLB).
association panel (Figure 1B). DS scores for both NCLB and GLS Interestingly, the correlation analyses in the DH population
were highest at the final disease rating time point in all field trials, showed low positive significant correlation between GLS (and
which corresponded in most cases to the last disease rating time AUDPC_GLS) and NCLB (Figure 2), indicating that there are
point (Supplementary Table S1). A significant difference in different genomic loci that explain the variance for each disease.
resistance to NCLB was reported for the two parents The agronomic traits for reproductive traits, namely, flowering time
CML511 and CML546 (p-value = 0.011831, α < 0.05). Our data (AD) and days to silking (SD), were significantly negatively
show that CML511 is moderately susceptible and CML546 is more correlated with DS and AUDPC for both GLS and NCLB
resistant to NCLB. On the other hand, the two parents differed diseases (Figure 3). This indicated that maize lines with earlier
slightly but not significantly in resistance to GLS (p-value = maturity had higher DS. As expected, ear height (EH) was highly
0.200588, α < 0.05). For GLS, CML511 had an average score of correlated with plant height (PH). GLS DS and AUDPC values were
2.47 at the final rating, while CML546 had an average score of 4.0 at weakly correlated with PH and EH, whereas NCLB DS and AUDPC
the final average DS score (Supplementary Figure S1). These values were positively and significantly correlated with PH and EH
Frontiers in Genetics 05 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
FIGURE 1
Frequency distributions for GLS, NCLB disease, and other agronomic traits, namely, anthesis date, silking date, plant height, and ear height, evaluated
across the three locations in western Kenya. (A). Biparental CML511×CML546 DH population of 230 lines. (B). Association panel of 239 sub-tropical and
tropical maize lines across the three locations. DS scores were for the last rating time point.
TABLE 1 Estimates of means, components of genotypic (σ2G), genotype × environment interaction (σ2GxE), error variances (σ2e), and heritability (h2) for the
biparental CML511×CML546 DH population and the association panel evaluated across three environments each for GLS, NCLB, and other agronomic traits.
AD SD PH EH GLS GLS NCLB NCLB GLSa NCLBa
DS AUDPC DS AUDPC
Mean 81.74 82.69 127.89 55.02 2.36 52.27 4.58 108.46 3.11 3.62
σ2G 11.07** 14.37** 282.87** 67.79** 0.94** 382.44** 0.42** 311.71** 0.03* 0.07**
σ2GxE 2.37** 3.63** 16.66** 5.94** 0.49** 213.07** 0.19* 156.61** 0.09** 0.04**
σ2e 11.59 12.6 219.57 101.49 0.52 175.63 0.45 216.91 0.16 0.17
h2 0.84 0.85 0.90 0.84 0.81 0.81 0.79 0.80 0.35 0.64
LSD5% 2.69 3.04 11.21 7.07 0.84 17.31 0.6 15.95 0.33 0.52
CV 4.17 4.29 11.59 18.31 30.56 25.36 14.73 13.58 34.09 18.06
AD, days to anthesis; SD, days to silking; PH, plant height; EH, ear height; DS, disease severity on a scale of 1–9; GLS, gray leaf spot; AUDPC, area under the disease progress curve; NCLB;
northern corn leaf blight; CV, coefficient of variation; LSD, least significant difference; h2, broad-sense heritability; *, **significant at p=0.05 and 0.01 levels, respectively.
aDisease severity scores of the maize association panel.
Frontiers in Genetics 06 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
TABLE 2 QTL detected by integrated composite interval mapping analysis for resistance to GLS and NCLB in the DH population evaluated in multiple locations.
Trait name QTL namea Chr Position (cM) Left markerb Right markerb LOD PVE (%) TPVE (%) Add Fav parent
GLS DS qGLS1_54 1 163 S1_283894617 S1_53456776 8.95 5.33 64.19 0.22 CML546
qGLS1_186 1 372 S1_190286762 S1_185978658 21.86 15.17 −0.28 CML511
qGLS1_185 1 383 S1_185978658 S1_143231392 11.34 9.01 0.21 CML546
qGLS2_30 2 208 S2_30710232 S2_32668550 3.79 2.24 0.11 CML546
qGLS3_151 3 92 S3_157562360 S3_150546157 5.31 3.37 0.13 CML546
qGLS5_07 5 189 S5_7548544 S5_1579511 3.62 2.25 −0.11 CML511
qGLS5_16 5 284 S5_15869219 S5_23093956 5.76 3.56 −0.14 CML511
qGLS7_158 7 105 S7_158889984 S7_158892468 4.49 2.58 −0.11 CML511
qGLS9_129 9 154 S9_135788881 S9_129671108 4.27 3.53 −0.13 CML511
qGLS10_50 10 217 S10_43765534 S10_54916081 3.19 6.21 0.22 CML546
NCLB DS qNCLB1_230 1 70 S1_229375633 S1_232878545 6.20 4.27 64.94 0.12 CML546
qNCLB1_303 1 429 S1_303106691 S1_304299395 3.13 2.08 0.08 CML546
qNCLB2_220 2 164 S2_223388206 S2_32056786 5.40 3.82 0.16 CML546
qNCLB3_02 3 163 S3_2734515 S3_1173815 4.95 3.57 0.11 CML546
qNCLB3_50 3 185 S3_65853211 S3_12761976 9.74 8.76 0.18 CML546
qNCLB4_200 4 272 S4_200040593 S4_201402668 3.50 4.58 −0.12 CML511
qNCLB5_83 5 80 S5_82971208 S5_160085856 3.60 4.56 0.13 CML546
qNCLB5_195 5 113 S5_194106967 S5_198705622 3.60 2.76 0.10 CML546
qNCLB6_137 6 204 S6_136207036 S6_137005821 3.21 2.17 −0.09 CML511
qNCLB6_153 6 335 S6_151834390 S6_153165363 3.95 3.58 −0.20 CML511
qNCLB7_120 7 96 S7_121214712 S7_47406965 7.51 5.58 −0.14 CML511
qNCLB7_174 7 257 S7_170932028 S7_174093748 3.43 2.44 −0.10 CML511
qNCLB8_171 8 276 S8_170418369 S8_171776990 7.85 5.77 0.14 CML546
GLS DS, gray leaf spot disease severity; NCLB DS; northern corn leaf blight disease severity; LOD, logarithm of odds; add, additive effect; PVE, phenotypic variance explained; fav parent,
parental genotype from where a favorable allele is contributing.
aQTL, name composed by the trait code followed by the chromosome number in which the QTL was mapped and a physical location of the QTL. QTL names are italicized.
bThe exact physical position of the marker can be inferred from the marker’s name, for example, S1_82702920: chromosome 1; 82,702,920 bp (Ref Gen_v3 of B73).
(Figure 3). There were weak positive and significant correlations explained 64.19% of total phenotypic variance (Table 2). All
between SD and PH/EH (Figure 3). 10 QTLs detected for GLS DS were also consistently detected for
GLS AUDPC values (Supplementary Table S3). For NCLB DS,
13 QTLs distributed in all chromosomes except for chromosomes
3.3 Construction of the genetic linkage map 9 and 10, individually explained 2.2%–8.7% which together
and QTL analyses contributed 64.94% of total phenotypic variation. AUDPC values
for NCLB revealed nine QTLs together explained 45% of total
The linkage map for the CML511×CML546 DH population was phenotypic variance (Supplementary Table S3). Three QTLs on
constructed with a total of 1,264 SNP markers. The genetic linkage chromosomes 3, 6, and 7 were consistent across NCLB DS and
map spanned a total map length of 3,344.9 cM with 2.65 cM as the AUDPC values. Among the agronomic traits, for AD, nine QTLs
average distance between two adjacent markers. The linkage map, as detected together explained 63% of the total phenotypic variance,
shown in Supplementary Figure S2, covered most of the maize and one QTL on chromosome 8 (qAD8_137) was a major effect QTL
genome. which explained 15.84% of phenotypic variance ((Supplementary
Several QTLs associated with resistance to GLS and NCLB with Table S4). For SD, six QTLs were detected which together explained
small additive effects were detected through inclusive composite 60% of the total phenotypic variance. There were three QTLs on
interval mapping. QTL analyses revealed 10 QTLs distributed on chromosomes 1, 4, and 8 that were consistent for both AD and SD.
chromosomes 1, 2, 3, 5, 7, 9, and 10 for GLS DS which individually For PH, four QTLs were identified which together explained 52% of
explained 2.6%–15.2% of phenotypic variance and together total phenotypic variance. One major effect QTL (qPH8_129)
Frontiers in Genetics 07 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
Association analyses revealed 11 and 18 SNPs significantly
associated with GLS DS and NCLB DS, respectively (Figure 3;
Table 3). For GLS-DS, the identified SNPs were distributed
across all chromosomes except for chromosomes 3, 6, and 7
(Figure 3). The Manhattan plot revealed that for GLS DS, the
highest peak was reported on chromosomes 4 and 8 (Figure 3A),
while for NCLBDS, the highest peak was reported on chromosome 8
(Figure 3C). Furthermore, we determined whether any of these
significant GLS or NCLB disease-associated SNPs identified in the
association analysis is co-located with the QTL for GLS or NCLB in
the biparental DH population. Two SNPs on chromosome 1 (S1_
192041854 and S1_253381765) were co-located with qGLS1_54
detected for both GLS DS and AUDPC values in the DH
population (Table 2; Table 3). Another SNP on chromosome 9
(S9_130213878) was found to be collocated within the qGLS9_129
and qG_AUDPC9_129 QTL region (Table 2; Table 3). Among the
16 SNPs identified for NCLB, marker S5_83980678 is found to be
within the region of NCLB DS QTL qNCLB5_83 (Table 2; Table 3).
To elucidate the molecular and physiological mechanisms
FIGURE 2 controlling GLS and NCLB DS, candidate genes were identified.
Pairwise Pearson correlation analysis for eight traits evaluated in On all chromosomes, a total of 24 candidate genes were discovered
three field trials in the biparental CML511×CML546 DH population. AD, (Table 3). Four candidate genes closely associated with the SNPs for
anthesis date; SD, silking date; GLS, gray leaf spot, NCLB, northern
corn leaf blight; AUDPC, area under disease progress curve; PH, GLS resistance were identified, namely, Zm00001eb077270,
plant height; and EH, ear height. The x marks indicate values that are Zm00001eb034870, Zm00001d017831, and Zm00001eb211960
not significant at p < 0.05. (Table 3). There were six candidate genes with defense response
annotations that were associated with SNPs for NCLB resistance
(Zm00001eb201110, Zm00001eb035640, Zm00001eb232660,
detected on chromosome 8 explained 22% of phenotypic variance. Zm00001eb285080, Zm00001eb144960, and Zm00001d001787)
For EH, six minor effect QTLs and one major effect (qEH8_128) (Table 3).
QTL were detected which together explained 56% of total Genomic prediction captures all variations from small to large
phenotypic variance. QTL mapping predominantly revealed that effects, which helps in improving complex traits. Prediction
the additive gene effect defined the gene action for resistance to GLS correlations obtained from cross-validations are commonly used
and NCLB. to know the effectiveness of genomic predictions for different traits.
In this study, we applied genomic predictions within the DH
population and association panel for disease traits and also for
3.4 GWAS analysis agronomic traits (Figure 4).
As expected, the mean prediction correlations for the DS traits
After a quality check of GBS SNP markers, 230,743 SNPs were were higher in the DH population (GLS DS = 0.75, NCLB DS = 0.68)
retained for the final association analyses (Supplementary Figure than those in the association panel (GLS DS = 0.63, NCLB = 0.49)
S3). The kinship matrix for these 239 lines was projected in the form (Figure 4). This was because of the highly relatedness among the DH
of a heatmap which shows the magnitude of the relationship lines compared to the lines in the association panel. In addition,
between the individuals (Supplementary Figure S4). This clearly relatively high-average prediction correlations were obtained for the
showed that there is no strong population structure in the other traits when validated in the DH population, namely, 0.73, 0.66,
association panel used here. PCAs revealed five subpopulations 0.71, 0.72, 0.75, and 0.71 for GLS-AUDPC, NCLB-AUDPC, AD, SD,
within the panel (Supplementary Figure S5). PH, and EH, respectively.
Association analyses for GLS DS and NCLB DS data were
performed with GLM, MLM, and FarmCPU models (Figure 3).
For both GLS and NCLB DS traits, for the GLMmodel, the observed 4 Discussion
p-values showed higher deviation from the expected p-values which
may cause high false positives. On the other hand, for the MLM GLS and NCLB are economically important foliar diseases of
model, though the observed p-values were closer to the expected maize. Understanding their genetic basis of resistance is valuable to
p-values, overfitting of the model is observed. For the FarmCPU designing an effective breeding strategy (Beyene and Prasanna,
model, the observed p-values were close to the expected p values and 2020). The DH population used in this study was artificially
were more effective in controlling the false positives (Figure 3). The inoculated with GLS; however, the locations evaluated were also
FarmCPU model is known to use both fixed and random effects a hotspot for NCLB, so we observed both GLS and NCLB disease
models iteratively which helps in avoiding overfitting of the model symptoms in the same population. For early scoring, symptoms for
by stepwise regression (Liu et al., 2016). Therefore, in this study, we both diseases were clearly distinguishable, which helped to score the
used the FarmCPU model in the association mapping. data with more accuracy. Scoring at a later stage of disease
Frontiers in Genetics 08 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
FIGURE 3
(A, C) Manhattan plots for the GWAS of GLS and NCLB disease severity in the maize association mapping panel. The dashed horizontal line of the
Manhattan plot depicts the significance threshold value of p < 8 × 10−5. The x-axis indicates the SNP location along the 10 chromosomes, with
chromosomes separated by different colors. Q–Q plots (B, D) of the estimated -log10(p) from association panel for GLS-DS and NCLB_DS traits. The
black line bisecting the plot in Q–Q plots represents the expected p-values with no associations present. The blue line represents the observed
p-values using the simplest model GLM(PCA + G) where the association between a phenotype and markers is directly detected. The pink line represents
the observed p-values using the MLM (PCA + K + G) model. The green line represents the observed p-values using the FarmCPU model. G, genotype
(fixed); PCA, three principal components (fixed); and K, kinship matrix (random).
development was tricky due to bigger blight merging with leaf spots, programs. Observed heritability estimates for GLS, NCLB, and other
so for the analyses, we used the third DS score for both GLS and agronomic traits in the DH population are consistent with earlier
NCLB. The association mapping panel was also evaluated in natural studies (Wisser et al., 2011; Benson et al., 2015).
hot spots for GLS and NCLB, and scoring was performed only once
at a grain-filling stage when clearly distinguished GLS and NCLB
symptoms were observed. Therefore, the collected DS data represent 4.1 Trait correlations
the real response of these lines to the respective diseases.
Nevertheless, both diseases appearing at the same growth stage There were moderate correlations in GLS and NCLB DS scores
and in the same experiment can lead to some confounding effects. for the biparental DH population between the field trials,
Most of the lines in both the DH population and the association indicating that trait expression was relatively consistent between
panel fall into the categories of resistant and moderately resistant, the evaluated locations (Supplementary Table S2). On the other
with a few in moderately susceptible but none in the completely hand, a significantly negative correlation was observed between DS
susceptible category (Figure 1). Overall, the phenotypic data in this data of GLS and NCLB, with the flowering time traits AD and SD in
study showed a normal distribution for NCLB and GLS DS scores in this study (Figure 2). This implies that lower values for AUDPC
both the biparental CML511×CML546 DH population and the (implying higher levels of disease resistance) corresponded with
association panel which supports the quantitative nature of longer AD or SD. Such negative correlations have also been
resistance in these diseases (Nyanapah et al., 2020). The parental reported in other studies (Asea et al., 2009; Wisser et al., 2011;
line CML511 exhibited a moderate level of resistance to GLS Kolkman et al., 2020). On the contrary, some studies reported a
congruent with the observations of Kibe et al. (2020a). We also positive correlation between GLS resistance and flowering time
observed significant genotypic and genotype × environment (Balint-Kurti et al., 2008; Zwonitzer et al., 2010; Benson et al., 2015;
interaction variance and moderate-to-high heritabilities in both Mammadov et al., 2015; Liu et al., 2016). This suggests the cautious
the DH population and the association panel, indicating good use of flowering time in the selection of lines for resistance to GLS
prospects for introgressing GLS and NCLB resistance in breeding and NCLB.
Frontiers in Genetics 09 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
TABLE 3 Chromosomal position and SNPs significantly associated with GLS disease severity (GLS_DS) and northern corn leaf blight disease severity (NCLB-DS)
detected by SNP-based GWAS in the association mapping panel.
SNPa Chr MLM MAFb H&Bc Effect Candidate Gene annotation
p-value p-value gene
GLS-DS
S4_170027069 4 8.68E-09 0.35 0.00 0.16 Zm00001eb189650 K13120—protein FAM32A (FAM32A)
S8_155438805 8 1.86E-08 0.47 0.00 −0.20 Zm00001eb360540 Cation efflux protein
S5_214099678 5 7.99E-08 0.29 0.01 0.18 Zm00001eb254100 Zinc finger FYVE domain-containing protein
S10_112359288 10 3.09E-06 0.13 0.13 −0.20 Zm00001eb421180 Copper transport protein atox1-related (abiotic stress)
S2_29666484 2 4.25E-06 0.38 0.13 −0.13 Zm00001eb077270 Wall-associated receptor kinase galacturonan-binding
domain (defense)
S1_253381765 1 4.51E-06 0.40 0.13 −0.13 Zm00001eb049550 Sel-1-like protein
S9_130213878 9 4.53E-06 0.10 0.13 −0.24 Zm00001eb393380 No associated annotations
S2_55483916 2 4.53E-06 0.26 0.13 0.13 ZM00001EB083120 No associated annotations
S1_192041854 1 1.98E-05 0.26 0.51 0.15 Zm00001eb034870 DNA damage–repair/toleration protein (plant defense)
S5_208091867 5 2.37E-05 0.21 0.55 0.12 Zm00001eb251710 Brevis radix domain/regulator of chromosome condensation
(plant defense)
S5_2923669 5 8.36E-05 0.24 1.00 0.13 Zm00001eb211960 NAD-dependent epimerase/dehydratase//cinnamoyl-COA
reductase-like
NCLB-DS
S8_157881780 8 4.41E-15 0.15 0.00 −0.57 GRMZM2G059590 Uncharacterized protein LOC103636219
S8_12990030 8 1.07E-09 0.16 0.00 −0.27 Zm00001d008560 No associated annotations
S4_212315234 4 1.11E-09 0.16 0.00 0.25 Zm00001eb201110 ATP-binding cassette transporter//ABC transporter (plant
defense)
S6_147054037 6 1.75E-07 0.08 0.01 0.25 Zm00001eb285130 TPR repeat-containing THIOREDOXIN TDX
S8_54094984 8 2.11E-07 0.49 0.01 −0.17 Zm00001eb341620 Thioesterase superfamily member-related
S1_194762510 1 6.65E-07 0.22 0.03 −0.26 Zm00001eb035640 AUX/IAA protein//B3 DNA-binding domain//auxin
response factor//DNA-binding pseudobarrel domain (plant
defense)
S1_280826386 1 1.48E-06 0.28 0.05 0.17 Zm00001d034003 Seed maturation family protein
S9_153843575 9 3.08E-06 0.45 0.09 0.16 Zm00001eb400490 Pre-mRNA-processing factor
S5_83980678 5 3.39E-06 0.14 0.09 0.22 Zm00001eb232660 Helicase superfamily/ATP-binding domain (plant defense)
S1_18630129 1 1.05E-05 0.45 0.24 −0.13 Zm00001eb006570 WD and tetratricopeptide repeats protein 1 (WDTC1)
S6_146813774 6 1.64E-05 0.06 0.34 −0.21 Zm00001eb285080 Protein kinase domain (plant defense)
S3_173387708 3 3.49E-05 0.09 0.67 −0.21 Zm00001eb144960 Lipoxygenase (plant defense)
S2_923555 2 4.77E-05 0.32 0.79 −0.16 Zm00001d001787 Cleavage and polyadenylation specificity factor subunit 5
(plant defense)
S7_155701108 7 4.77E-05 0.14 0.79 0.21 Zm00001d021552 Protein of unknown function
S4_233626821 4 5.87E-05 0.47 0.88 −0.15 Zm00001eb204230 Voltage- and ligand-gated potassium channel
S7_8047716 7 6.13E-05 0.24 0.88 −0.12 Zm00001d018877 Plastocyanin-like domain (Cu_bind_like)
aThe exact physical position of the SNP can be inferred from the marker’s name, for example, S5_51353429: chromosome 5; 51353429 bp (Ref Gen_v2 of B73).
bMinor allele frequency. Candidate gene names are italicized.
cFalse discovery rate calculated by using the Benjamini and Hochberg correction method.
In Kenya and Uganda, the main maize-growing area is the major concern which are probably due to the synergistic
frequently affected by GLS and NCLB pathogens (Borchardt interactions of both pathogens. However, a weak correlation was
et al., 1998; Nsibo et al., 2021). When both pathogens affect the observed between GLS DS and NCLB DS. One of the QTLs
maize at the same time, more pronounced necrotic symptoms are identified for GLS (qGLS2_30) was in proximity with the QTL
Frontiers in Genetics 10 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
FIGURE 4
Box-whisker plots for the accuracy of genomic predictions assessed by five-fold cross-validation within association and DH population. AD, days to
anthesis; SD, days to silking; PH, plant height; EH, ear height; GLS, gray leaf spot; AUDPC, area under the disease progress curve; NCLB, northern corn leaf
blight.
identified for NCLB (qNCLB2_220). In the comparison of SNPs 2001; Shi et al., 2007; Balint-Kurti et al., 2008; Berger et al., 2014; He
associated with GLS and NCLB DS in association mapping, it was et al., 2018; Lopez-Zuniga et al., 2019; Sun et al., 2021). The
observed that two SNPs (S1_194762510 and S1280826386) chromosome bin 1.06 also harbors resistance genes to several
associated with NCLB DS were collocated within GLS QTL other diseases like common rust, southern leaf blight (SLB), ear
(qGLS1_54) (Tables 2, 3). This suggests that there are some rot, and NCLB (Freymark et al., 1993; Wisser et al., 2006; Chung
common regions contributing to resistance for both diseases. On et al., 2010b; Zwonitzer et al., 2010; Poland et al., 2011; Jamann et al.,
the other hand, the observed weak correlation between the DS of the 2015). Chung et al. (2010b) demonstrated that the NCLB resistance
two diseases could be attributed to the different infection strategies QTL at bin 1.06 was important to protect the host against fungal
of the associated pathogens. Cercospora zeina is an apoplastic penetration of E. turcicum using an introgression line population.
necrotroph and a hemibiotroph, while E. turcicum is apoplastic The chromosomal region has also been associated with effects on
but then enters the vascular system of the leaf (Kotze et al., 2019). diverse traits such as grain yield and its components, anthesis silking
They also exploit different pathogenicity factors in causing disease interval, and root and shoot traits under both water stress and
symptoms (Swart et al., 2017; Human et al., 2020). optimal water environments (Ribaut et al., 1996; Tuberosa et al.,
2002; Landi et al., 2010). The concentrated mapping of QTL for
several traits, including multiple disease resistance in this
4.2 QTLs associated with GLS resistance chromosomal region, provides breeders and geneticists an
opportunity to dissect them further and find tightly linked
Most of the QTLs detected for GLS DS were also detected for flanking markers so that this region can be utilized to develop
GLS AUDPC (Table 2). This was well supported by the observed cultivars with multiple disease resistance.
strong correlation (r = 0.99) between GLS DS and GLS AUDPC The qGLS2_30 QTL identified in this study in the chromosomal
(Figure 2). A major QTL for GLS resistance (DS and AUDPC), bin 2.04 overlapped with QTL reported in earlier studies using
qGLS1_186, which explained 15.16% of the phenotypic variance, different mapping populations (Balint-Kurti et al., 2008; Lennon
overlapped with qGLS1_185 which also explained 9.01% of the et al., 2017). The QTL qGLS3_151 is placed between 150 and
phenotypic variance. Intriguingly, this major QTL has favorable 157 Mbp in the chromosomal bin 3.05, which has previously
alleles from the donor parent CML511. Kibe et al. (2020a) using the been identified as conditioning resistance to SLB and GLS
CML550×CML511 DH population also detected major effect QTL (Zwonitzer et al., 2010; Kump et al., 2011). The QTL qGLS7_158
qGLS1-155 which was located within the physical position of positioned between 157 and 159 Mbp in the chromosomal bin
154–157 Mbp which overlapped with the QTL detected in the 7.04 was also previously reported by Berger et al. (2014) for GLS
present study (qGLS1_185) spanning between 143 and resistance. Another QTL qGLS5_16 in the chromosomal bin 5.03 is
185 Mbp. Sun et al. (2021) also fine-mapped a major effect QTL also known to have several reported markers for GLS resistance in a
at the 187–189 Mb region, and the reported flanking markers would number of association mapping studies (Bubeck et al., 1993;
be useful to validate in tropical germplasm. Clements et al., 2000; Lehmensiek et al., 2001; Shi et al., 2007;
The chromosome bin 1.06 was described as a QTL hotspot for Zhang et al., 2012; Benson et al., 2015). Overall, many of the QTLs
GLS resistance as many studies reported earlier (Lehmensiek et al., detected in the present study overlapped between the biparental
Frontiers in Genetics 11 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
CML511 × CML546 DH population and the association panel as bins associated with the qualitative Ht genes (Galiano-Carneiro
well as earlier studies (Supplementary Table S5). The major effect and Miedaner, 2017).
QTL for both GLS on chromosome 1 is of immediate interest to be An association study on NCLB revealed a significant marker
used in resistance breeding. linked to a candidate gene Zm00001eb201110 on chromosome
Among the nine candidate genes identified for GLS resistance 4 which encodes for an ATP-binding cassette (ABC) transporter.
through association mapping, one on chromosome 2 Plant proteins with this function are known to be associated with
(Zm00001eb077270) encodes a putative receptor-like protein resistance to fungal and bacterial pathogens through the
kinase which are transmembrane signaling proteins that are able transmembrane transport of jasmonic acid or antimicrobial
to sense changes in the extracellular environment such as pathogen secondary metabolites (Zhang et al., 2020). Using GWAS,
invasion (Decreux et al., 2006; Qi et al., 2023). Another candidate many studies showed the association of ABC transporter genes
gene on chromosome 1 (Zm00001eb034870) encodes DNA with NCLB resistance (Poland et al., 2011; Ding et al., 2015).
damage–repair/toleration protein that harbors a leucine-rich Another candidate gene, Zm00001eb285080, on chromosome
repeat domain which serve as the first line of defense in response 6 encodes a protein kinase, a function known to be important
to pathogen-associated molecular patterns (Ng and Xavier, 2011). in regulating the response of plants to pathogen attack (Lehti-
The candidate gene Zm00001eb144960 on chromosome 3 encodes Shiu and Shiu, 2012). There is strong evidence that protein
lipoxygenases which are known to be associated with pest resistance, kinases play a pivotal role in resistance to NCLB (Poland
response to wounding, and plant defense mechanisms where it was et al., 2011; Ding et al., 2015; Kolkman et al., 2020).
reported to be involved in the early response to pathogen attack
(Peng et al., 1994). Overall, the detected candidate genes in the
association study have annotations inferring direct or indirect 4.4 QTLs associated with agronomic data
involvement in plant defense.
The major QTL for flowering time qAD8_137 was collocated
with QTLs qSD8_137 for SD, qPH8_129 for PH, and qEH8_128 for
4.3 QTL and SNPs associated with resistance EH ((Supplementary Table S4). These QTLs also explained the
to NCLB major effect of phenotypic variance of 15.8%, 21.4%, 22.39%, and
22.98% for AD, SD, PH, and EH, respectively. Several studies also
This study identified 13 QTLs for NCLB DS and nine QTLs recognized chromosomal bin 8.05 as a hotspot for flowering time
for NCLB AUDPC. Three of these QTLs were common for both QTL and genes (Balint-Kurti et al., 2008; Buckler et al., 2009; Van
the DS and AUDPC NCLB traits. The first example of this was Inghelandt et al., 2012). Interestingly, two qualitative resistance
QTLs qNCLB3_50 and qN_AUDPC3_50 that co-localized in the genes, Ht2 and Htn1, were also detected on the same
same position in bin 3.04. This significant QTL for NCLB chromosomal bin 8.05 (Galiano-Carneiro and Miedaner, 2017;
resistance has favorable alleles from the parent CML546 (the Hurni et al., 2015). The genetic mechanisms underlying flowering
more resistant parent). Previous research reports also identified time in this study were largely characterized by additive gene action.
bin 3.04 as a QTL hotspot conditioning resistance to multiple These results agree with the findings of Buckler et al. (2009) who
diseases including NCLB, SLB, and GLS (Lehmensiek et al., 2001; reported that variations in days to flowering are due to the joint
Wisser et al., 2006; Shi et al., 2007; Zwonitzer et al., 2010; Kump effect of many minor QTLs with additive effect.
et al., 2011; Liu et al., 2016; Lennon et al., 2017; Martins et al., Intriguingly, some of the QTLs associated with flowering time
2019). overlapped with the NCLB and GLS resistance QTL. For instance,
The QTL qNCLB5_83 was positioned in the chromosomal bin qAD1_60 for flowering time shared the same flanking markers as
5.04. According to Miedaner et al. (2020), up to eight QTLs have qGLS1_54 (Table 2; Supplementary Table S4), and the two SNPs
been localized in this bin showing the importance of this region (S1_192041854, S1_253381765) for GLS DS and two SNPs (S1_
for NCLB resistance. Interestingly, the SNP identified through 194762510, S1_2800826386) for NCLB DS detected through
association mapping (S5_83980678) is positioned within this association mapping are also positioned in this region.
QTL region (Table 2, 3). It is associated with a candidate gene Another QTL qGLS9_129 also had the same flanking markers
(Zm00001eb232660) that encodes a DNA helicase/ATP-binding as qAD9_130, and one SNP from association mapping (S9_
domain. This type of domain has a catalytic function in 130213878) was also detected in the same region for GLS DS.
unwinding of the double-stranded DNA that is instrumental The NCLB QTL qNCLB1_230 overlapped with qAD1_227 on the
in the repair of damaged DNA and DNA replication (Koonin, maize chromosomal bin 1.07 (Table 2; Supplementary Table S4).
1993). The QTLs qN_AUDPC2_188 and qPH2_176 overlapped on
Similarly, qNCLB6_153 for NCLB DS overlapped with qN_ chromosomal bin 2.06, sharing the flanking markers. This was
AUDPC6_153 for the AUDPC on chromosome 6 (bin 6.05). Up further supported by the positive correlation between PH and
to four QTLs were reported from different studies in the same NCLB AUDPC (Figure 2). On the other hand, Galiano-Carneiro
region (Miedaner et al., 2020). Another pair of QTL, qNCLB8_ et al. (2021) reported a negative correlation between PH and
171 and qN_AUDPC8_171, corresponded with a previously NCLB DS. There were no common QTL regions identified for PH
reported NCLB resistance QTL by Galiano-Carneiro et al. and EH that spanned the same chromosomal regions as GLS DS
(2021). Interestingly, none of the NCLB DS QTLs detected in and AUDPC. This is supported by the observed negative
this study were found in the same position with chromosomal correlation between the two traits (Figure 3).
Frontiers in Genetics 12 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
4.5 Genomic prediction of disease and Data availability statement
agronomic traits
The datasets presented in this study can be found in online
Compared to the association panel, the high-prediction repositories: data.cimmyt.org/dataset.xhtml?persistentId=hdl:
correlations in the DH population for GLS and NCLB could 11529/10548956 and zenodo.org/records/10046213.
be attributed to higher similarity or relatedness of individuals
between the training set and the prediction set (Figure 4; Lorenz
and Smith, 2015). GLS DS and GLS AUDPC exhibited slightly Author contributions
higher prediction accuracies compared to NCLB DS and other
agronomic traits. Agronomic traits, such as AD, SD, PH, and EH, DO: conceptualization, data curation, formal analysis, validation,
are characterized by more complex genetic networks, under the and writing–original draft. MD: conceptualization, funding acquisition,
control of numerous QTLs, and affected by the influence of the project administration, resources, supervision, and writing–review and
environment (Wallace et al., 2014). This presents a challenge in editing. DB: conceptualization, methodology, resources, supervision,
improving them through phenotypic selection (Du et al., 2021). and writing–review and editing. YB: funding acquisition, investigation,
Kibe et al. (2020a) reported low-to-moderate prediction project administration, supervision, and writing–review and editing.
correlations within populations and high values when different DN: conceptualization, formal analysis, methodology, and
related populations were combined and used in prediction. writing–review and editing. CJ: data curation, formal analysis,
Similarly, Galiano-Carneiro et al. (2021) reported prediction methodology, and writing–review and editing. SM: investigation and
accuracies of moderate levels (0.55) for prediction within writing–review and editing. MG: data curation, formal analysis,
families. Technow et al. (2013) reported prediction accuracies methodology, software, visualization, writing–original draft, and
of 0.58 and 0.55 when using a small population size of writing–review and editing.
75 individuals. It has been reported that there is no difference
among hybrids advanced through the genomic selection or
phenotypic selection in their response to NCLB and GLS, with Funding
the genomic selection being relatively cheaper than the
phenotypic selection (Beyene et al., 2019; 2021). Overall, the The authors declare financial support was received for the research,
use of genomic selection has potential to improve the resistance authorship, and/or publication of this article. This study was supported
to GLS and NCLB in breeding populations and could lead to the by the Kenya–South Africa joint science and technology research
development of multiple disease-resistant lines and hybrids. collaboration. The grants received from the National Research Fund,
Kenya, and the National Research Foundation (NRF), South Africa
(grant # 105806), toward this research are hereby acknowledged. This
5 Conclusion study was also supported by CIMMYT-Nairobi. CIMMYT received
support from the United States Agency for International Development,
GLS and NCLB are the major biotic stresses that hinder maize Foundation for Food andAgriculture Research (FFAR), and the Bill and
production in high-yielding maize-growing areas in East Africa, Melinda Gates Foundation (BMGF) under AG2MW (Accelerating
such as western Kenya. The use of genomic tools can provide Genetic Gains in Maize and Wheat for Improved Livelihoods,
useful information to fast track the development of disease- B&MGF Investment ID INV-003439) project, the CGIAR Research
resistant varieties. In this study, we aimed to identify and Program on MAIZE.
validate genomic regions associated with GLS and NCLB
resistance in biparental and association mapping populations
evaluated in multiple locations in western Kenya. We identified Acknowledgments
10 and 11 QTLs for GLS resistance and 18 and 16 QTLs for NCLB
resistance in the DH population and association mapping The authors thank the CIMMYT field and laboratory technicians
population, respectively. We detected a major QTL for GLS for phenotypic evaluations and sample preparation for genotyping.
resistance, qGLS1_186, which explained 15.2% phenotypic They are also grateful to the Buckler Lab at Cornell University,
variance and qNCLB3_50 for NCLB resistance, explaining United States, for genotyping the maize populations and providing
8.8% of the phenotypic variance. Several common QTL the marker information and Diversity Arrays Technology (DArT),
regions between linkage mapping and association mapping Canberra, Australia for genotyping DH population. Part of the data in
and between NCLB and GLS AUDPC traits were detected. A this manuscript was from the thesis work conducted by DO.
negative correlation between flowering time and severity of the
two diseases was reported. Several QTLs identified in the present
study were also co-localized with the QTL previously mapped for Conflict of interest
GLS and NCLB resistance. Our study highlights that the
combined use of linkage mapping and genomic selection is an Authors DO and CJ were employed by Crop Science Division
effective strategy for the improvement of resistance. Genomic Bayer East Africa Limited.
prediction sheds light on new ways to improve breeding for The remaining authors declare that the research was conducted
disease resistance with optimum allocation of resources and lays in the absence of any commercial or financial relationships that
the foundation for a new era of resistance breeding. could be construed as a potential conflict of interest.
Frontiers in Genetics 13 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
The authors declare that they were editorial board members of reviewers. Any product that may be evaluated in this article, or
Frontiers, at the time of submission. This had no impact on the peer claim that may be made by its manufacturer, is not guaranteed or
review process and the final decision. endorsed by the publisher.
Publisher’s note Supplementary material
All claims expressed in this article are solely those of the authors The Supplementary Material for this article can be found online
and do not necessarily represent those of their affiliated at: https://www.frontiersin.org/articles/10.3389/fgene.2023.1282673/
organizations, or those of the publisher, the editors, and the full#supplementary-material
References
Alvarado, G., López, M., Vargas, M., Pacheco, A., Rodríguez, F., Burgueño, J., et al. molecular avenues for combating abiotic stress tolerance in plants. Editor S. H. Wani
(2015). META-R (multi environment trail analysis with R for windows) version 5.0 - (Academic Press), 135–150. doi:10.1016/b978-0-12-813066-7.00009-7
CIMMYT research software dataverse - CIMMYT dataverse network. Available at:
Chen, G., Wang, X., Long, S., Jaqueth, J., Li, B., Yan, J., et al. (2015). An update on
https://data.cimmyt.org/dataset.xhtml?persistentId=hdl:11529/10201 (Accessed April
inflammation in the acute phase of intracerebral hemorrhage. Mol. Breed. 36 (1), 4–8.
15, 2021).
doi:10.1007/s12975-014-0384-4
Asea, G., Vivek, B. S., Bigirwa, G., Lipps, P. E., and Pratt, R. C. (2009). Validation of
Chung, C. L., Longfellow, J. M., Walsh, E. K., Kerdieh, Z., Van Esbroeck, G., Balint-
consensus quantitative trait loci associated with resistance to multiple foliar pathogens
Kurti, P., et al. (2010b). Resistance loci affecting distinct stages of fungal pathogenesis:
of maize. Phytopathology 99 (5), 540–547. doi:10.1094/phyto-99-5-0540
use of introgression lines for QTL mapping and characterization in the maize--
Ashburner, M., Ball, C. A., Blake, J. A., Botstein, D., Butler, H., Cherry, J. M., et al. Setosphaeria turcica pathosystem. BMC Plant Biol. 10, 103. doi:10.1186/1471-2229-
(2000). Gene ontology: tool for the unification of biology. The Gene Ontology 10-103
Consortium. Nat. Genet. 25 (1), 25–29. doi:10.1038/75556
Cimmyt, (2005). Laboratory protocols. CIMMYT applied molecular genetics
Balint-Kurti, P. J., Wisser, R. J., and Zwonitzer, J. C. (2008). Use of an advanced laboratory.
intercross line population for precise mapping of quantitative trait loci for gray leaf spot
Clements, M. J., Dudley, J. W., and White, D. G. (2000). Quantitative trait Loci
resistance in maize. Crop Sci. 48, 1696–1704. doi:10.2135/cropsci2007.12.0679
associated with resistance to gray leaf spot of corn. Phytopathology 90 (9), 1018–1025.
Bateman, A., Coin, L., Durbin, R., Finn, R. D., Hollich, V., Griffiths-Jones, S., et al. doi:10.1094/phyto.2000.90.9.1018
(2004). The Pfam protein families database. Nucleic Acids Res. 32, D138–D141.
Crossa, J., Pérez-Rodríguez, P., Cuevas, J., Montesinos-López, O., Jarquín, D., de los
Database issue). doi:10.1093/nar/gkh121
Campos, G., et al. (2017). Genomic selection in plant breeding: methods, models, and
Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: a perspectives. Trends Plant Sci. 22 (11), 961–975. doi:10.1016/j.tplants.2017.08.011
practical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. B Stat.
Crous, P. W., Groenewald, J. Z., Groenewald, M., Caldwell, P., Braun, U., and
Methodol. 57, 289–300. doi:10.1111/j.2517-6161.1995.tb02031.x
Harrington, T. C. (2006). Species of Cercospora associated with grey leaf spot of
Benson, J. M., Poland, J. A., Benson, B. M., Stromberg, E. L., and Nelson, R. J. (2015). maize. Stud. Mycol. 55, 189–197. doi:10.3114/sim.55.1.189
Resistance to gray leaf spot of maize: genetic architecture and mechanisms elucidated
Decreux, A., Thomas, A., Spies, B., Brasseur, R., Van Cutsem, P., and Messiaen, J.
through nested association mapping and near-isogenic line analysis. PLoS Genet. 11 (3),
(2006). In vitro characterization of the homogalacturonan-binding domain of the wall-
e1005045. doi:10.1371/journal.pgen.1005045
associated kinase WAK1 using site-directed mutagenesis. Phytochemistry 67 (11),
Berger, D. K., Carstens, M., Korsman, J. N., Middleton, F., Kloppers, F. J., Tongoona, 1068–1079. doi:10.1016/j.phytochem.2006.03.009
P., et al. (2014). Mapping QTL conferring resistance in maize to gray leaf spot disease
Dekkers, J. C. (2007). Prediction of response tomarker-assisted and genomic selection
caused by Cercospora zeina. BMC Genet. 15, 60. doi:10.1186/1471-2156-15-60
using selection index theory. J. Anim. Breed. Genet. 124 (6), 331–341. doi:10.1111/j.
Berger, D. K., Mokgobu, T., de Ridder, K., Christie, N., and Aveling Theresa, A. S. 1439-0388.2007.00701.x
(2020). Benefits of maize resistance breeding and chemical control against northern leaf
Ding, J., Ali, F., Chen, G., Li, H., Mahuku, G., Yang, N., et al. (2015). Genome-wide
blight in smallholder farms in South Africa. South Afr. J. Sci. 116 (11-12), 113–119.
association mapping reveals novel sources of resistance to northern corn leaf blight in
doi:10.17159/sajs.2020/8286
maize. BMC Plant Biol. 15, 206. doi:10.1186/s12870-015-0589-z
Beyene, Y., Gowda, M., Olsen, M., Robbins, K. R., Pérez-Rodríguez, P., Alvarado, G.,
Du, L., Zhang, H., Xin, W., Ma, K., Du, D., Yu, C., et al. (2021). Dissecting the genetic
et al. (2019). Empirical comparison of tropical maize hybrids selected through genomic
basis of flowering time and height related-traits using two doubled haploid populations
and phenotypic selections. Front. Plant Sci. 10, 1502. doi:10.3389/fpls.2019.01502
in maize. Plants (Basel) 10 (8), 1585. doi:10.3390/plants10081585
Beyene, Y., Gowda, M., Pérez-Rodríguez, P., Olsen, M., Robbins, K. R., Burgueño, J.,
Elshire, R. J., Glaubitz, J. C., Sun, Q., Poland, J. A., Kawamoto, K., Buckler, E. S., et al.
et al. (2021). Application of genomic selection at the early stage of breeding pipeline in
(2011). A robust, simple genotyping-by-sequencing (GBS) approach for high diversity
tropical maize. Front. Plant Sci. 12, 685488. doi:10.3389/fpls.2021.685488
species. PloS One 6, e19379. doi:10.1371/journal.pone.0019379
Beyene, Y., and Prasanna, B. M. (2020). Accelerating genetic gains in maize and
Freymark, P. J., Lee, M., Woodman, W. L., and Martinson, C. A. (1993). Quantitative
wheat: genetic gains in CIMMYT maize breeding program in Africa. Available at
and qualitative trait loci affecting host-plant response to Exserohilum turcicum in maize
https://excellenceinbreeding.org/sites/default/files/u1025/CIMMYT%20Maize%
(Zea mays L.). Theor. Appl. Genet. 87 (5), 537–544. doi:10.1007/bf00221876
20Genetic%20Gains%20in%20Africa%20%28AGG-EiB%20Workshop%3B%20Nov%
2018%202020%29.pdf (Accessed August 20, 2021). Galiano-Carneiro, A. L., Kessel, B., Presterl, T., and Miedaner, T. (2021).
Intercontinental trials reveal stable QTL for Northern corn leaf blight resistance in
Borchardt, D. S., Welz, H. G., and Geiger, H. H. (1998). Genetic structure of
Europe and in Brazil. Theor. Appl. Genet. 134 (1), 63–79. doi:10.1007/s00122-020-
Setosphaeria turcica populations in tropical and temperate climates. Phytopathology
03682-1
88 4, 322–329. doi:10.1094/phyto.1998.88.4.322
Galiano-Carneiro, A. L., and Miedaner, T. (2017). Genetics of resistance and
Bradbury, J. P., Zhang, Z., Kroon, D. E., Casstevens, T. M., Ramdoss, Y., and Buckler,
pathogenicity in the maize/Setosphaeria turcica pathosystem and implications for
E. S. (2007). TASSEL: software for association mapping of complex traits in diverse
breeding. Front. Plant Sci. 8, 1490. doi:10.3389/fpls.2017.01490
samples. Bioinformatics 23 (19), 2633–2635. doi:10.1093/bioinformatics/btm308
Glaubitz, J. C., Casstevens, T. M., Lu, F., Harriman, J., Elshire, R. J., Sun, Q., et al.
Bubeck, D. M., Goodman, M. M., Beavis, W. D., and Grant, D. (1993). Quantitative
(2014). TASSEL-GBS: a high-capacity genotyping by sequencing analysis pipeline.
trait loci controlling resistance to gray leaf spot in maize. Crop Sci. 33 (4), 838–847.
PLOS ONE 9 (2), e90346. doi:10.1371/journal.pone.0090346
doi:10.2135/cropsci1993.0011183x003300040041x
Goodstein, D. M., Shu, S., Howson, R., Neupane, R., Hayes, R. D., Fazo, J., et al. (2012).
Buckler, E. S., Holland, J. B., Bradbury, P. J., Acharya, C. B., Brown, P. J., Browne, C.,
Phytozome: a comparative platform for green plant genomics. Nucleic Acids Res. 40,
et al. (2009). The genetic architecture of maize flowering time. Science 325 (5941),
D1178–D1186. doi:10.1093/nar/gkr944
714–718. doi:10.1126/science.1174276
He, W., Yang, L., Leng, Y., Zhang, B., Yang, J., Li, L., et al. (2018). QTL mapping for
Challa, S., and Neelapu, N. R. R. (2018). “Chapter 9 - genome-wide association studies
resistance of maize to grey leaf spot. J. Phytopathology 166 (3), 167–176. doi:10.1111/
(GWAS) for abiotic stress tolerance in plants,” in Biochemical, physiological and
jph.12673
Frontiers in Genetics 14 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
Human, M. P., Berger, D. K., and Crampton, B. G. (2020). Time-course RNAseq Liu, L., Zhang, Y. D., Li, H. Y., Bi, Y. Q., Yu, L. J., Fan, X. M., et al. (2016). QTL
reveals Exserohilum turcicum effectors and pathogenicity determinants. Front. mapping for gray leaf spot resistance in a tropical maize population. Plant Dis. 100 (2),
Microbiol. 11, 360. doi:10.3389/fmicb.2020.00360 304–312. doi:10.1094/pdis-08-14-0825-re
Hurni, S., Scheuermann, D., Krattinger, S. G., Kessel, B., Wicker, T., Herren, G., et al. Lopez-Zuniga, L. O., Wolters, P., Davis, S., Weldekidan, T., Kolkman, J. M., Nelson,
(2015). The maize disease resistance gene Htn1 against northern corn leaf blight R., et al. (2019). Using maize chromosome segment substitution line populations for the
encodes a wall-associated receptor-like kinase. Proc. Natl. Acad. Sci. U. S. A. 112 (28), identification of loci associated with multiple disease resistance. G3 (Bethesda) 9 (1),
8780–8785. doi:10.1073/pnas.1502522112 189–201. doi:10.1534/g3.118.200866
Jaccoud, D., Peng, K., Feinstein, D., and Kilian, A. (2001). Diversity arrays: a solid Lorenz, A. J., and Smith, K. P. (2015). Adding genetically distant individuals to
state technology for sequence information independent genotyping. Nucleic Acids Res. training populations reduces genomic prediction accuracy in barley. Crop Sci. 55 (6),
29 (4), E25. doi:10.1093/nar/29.4.e25 2657–2667. doi:10.2135/cropsci2014.12.0827
Jamann, T. M., Balint-Kurti, P. J., and Holland, J. B. (2015). QTLmapping using high- Mammadov, J., Sun, X., Gao, Y., Ochsenfeld, C., Bakker, E., Ren, R., et al. (2015).
throughput sequencing. Methods Mol. Biol. 1284, 257–285. doi:10.1007/978-1-4939- Combining powers of linkage and association mapping for precise dissection of QTL
2444-8_13 controlling resistance to gray leaf spot disease in maize (Zea mays L.). BMC Genomics
16, 916. doi:10.1186/s12864-015-2171-3
Kanehisa, M., and Goto, S. (2000). KEGG: kyoto encyclopedia of genes and genomes.
Nucleic Acids Res. 28 (1), 27–30. doi:10.1093/nar/28.1.27 Martins, L. B., Rucker, E., Thomason, W., Wisser, R. J., Holland, J. B., and Balint-
Kurti, P. (2019). Validation and characterization of maize multiple disease resistance
Khan, S. U., Saeed, S., Khan, M. H. U., Fan, C., Ahmar, S., Arriagada, O., et al. (2021).
QTL. G3 (Bethesda) 9 (9), 2905–2912. doi:10.1534/g3.119.400195
Advances and challenges for QTL analysis and GWAS in the plant-breeding of high-
yielding: a focus on rapeseed. Biomolecules 11 (10), 1516. doi:10.3390/biom11101516 McDonald, B. A., and Linde, C. (2002). Pathogen population genetics, evolutionary
potential, and durable resistance. Annu. Rev. Phytopathol. 40, 349–379. doi:10.1146/
Kibe, M., Nair, S. K., Das, B., Bright, J. M., Makumbi, D., Kinyua, J., et al. (2020a).
annurev.phyto.40.120501.101443
Genetic dissection of resistance to gray leaf spot by combining genome-wide
association, linkage mapping, and genomic prediction in tropical maize germplasm. Meng, L., Li, H., Zhang, L., andWang, J. (2015). QTL IciMapping: integrated software
Front. Plant Sci. 11, 572027. doi:10.3389/fpls.2020.572027 for genetic linkage map construction and quantitative trait locus mapping in biparental
populations. Crop J. 3 (3), 269–283. doi:10.1016/j.cj.2015.01.001
Kibe, M., Nyaga, C., Nair, S. K., Beyene, Y., Das, B., Suresh, L. M., et al. (2020b).
Combination of linkage mapping, GWAS, and GP to dissect the genetic basis of Miedaner, T., Boeven, A. L. G., Gaikpa, D. S., Kistner, M. B., and Grote, C. P. (2020).
common rust resistance in tropical maize germplasm. Int. J. Mol. Sci. 21 (18), 6518. Genomics-assisted breeding for quantitative disease resistances in small-grain cereals
doi:10.3390/ijms21186518 and maize. Int. J. Mol. Sci. 21 (24), 9717. doi:10.3390/ijms21249717
Kilian, A., Wenzl, P., Huttner, E., Carling, J., Xia, L., Blois, H., et al. (2012). Diversity Munialo, S., Dahlin, S., Onyango, C., Kosura, W., Marstorp, H., and Oborn, I. (2019).
arrays technology: a generic genome profiling technology on open platforms. Methods Soil and management-related factors contributing to maize yield gaps in western Kenya.
Mol. Biol. 888, 67–89. doi:10.1007/978-1-61779-870-2_5 Food Energy Secur. 9. doi:10.1002/fes3.189
Kinyua, Z. M., Smith, J. J., Kibata, G. N., Simons, S. A., and Langat, B. C. (2010). Status Murithi, A., Olsen, M. S., Kwemoi, D. B., Veronica, O., Ertiro, B. T., Suresh, L. M.,
of grey leaf spot disease in Kenyan maize production ecosystems. Afr. Crop Sci. J. 18 (4), et al. (2021). Discovery and validation of a recessively inherited major-effect QTL
183–194. doi:10.4314/acsj.v18i4.68647 conferring resistance to maize lethal necrosis (MLN) disease. Front. Genet. 12, 767883.
doi:10.3389/fgene.2021.767883
Knapp, S. J., Stroup, W. W., and Ross, W. M. (1985). Exact confidence intervals for
heritability on a progeny mean basis. Crop Sci. 25 (1), 192–194. doi:10.2135/ Ng, A., and Xavier, R. J. (2011). Leucine-rich repeat (LRR) proteins: integrators of
cropsci1985.0011183x002500010046x pattern recognition and signaling in immunity. Autophagy 7 (9), 1082–1084. doi:10.
4161/auto.7.9.16464
Kolkman, J. M., Strable, J., Harline, K., Kroon, D. E., Wiesner-Hanks, T., Bradbury, P.
J., et al. (2020). Maize introgression library provides evidence for the involvement of Nsibo, D. L., Barnes, I., Kunene, N. T., and Berger, D. K. (2019). Influence of farming
liguleless1 in resistance to northern leaf blight. G3 (Bethesda) 10 (10), 3611–3622. practices on the population genetics of the maize pathogen Cercospora zeina in South
doi:10.1534/g3.120.401500 Africa. Fungal Genet. Biol. 125, 36–44. doi:10.1016/j.fgb.2019.01.005
Koonin, E. V. (1993). Escherichia coli dinG gene encodes a putative DNA helicase Nsibo, D. L., Barnes, I., Omondi, D. O., Dida, M. M., and Berger, D. K. (2021).
related to a group of eukaryotic helicases including Rad3 protein. Nucleic Acids Res. 21 Population genetic structure and migration patterns of the maize pathogenic fungus,
(6), 1497. doi:10.1093/nar/21.6.1497 Cercospora zeina in East and Southern Africa. Fungal Genet. Biol. 149, 103527. doi:10.
1016/j.fgb.2021.103527
Korsman, J., Meisel, B., Kloppers, F. J., Crampton, B. G., and Berger, D. K. (2012).
Quantitative phenotyping of grey leaf spot disease in maize using real-time PCR. Eur. Nyanapah, J. O., Ayiecho, P. O., Nyabundi, J. O., Otieno, W., and Ojiambo, P. S.
J. Plant Pathology 133 (2), 461–471. doi:10.1007/s10658-011-9920-1 (2020). Field characterization of partial resistance to gray leaf spot in elite maize
germplasm. Phytopathology 110 (10), 1668–1679. doi:10.1094/phyto-12-19-0446-r
Kotze, R. G., van der Merwe, C. F., Crampton, B. G., and Kritzinger, Q. (2019). A
histological assessment of the infection strategy of Exserohilum turcicum in maize. Plant Peng, Y. L., Shirano, Y., Ohta, H., Hibino, T., Tanaka, K., and Shibata, D. (1994). A
Pathol. 68 (3), 504–512. doi:10.1111/ppa.12961 novel lipoxygenase from rice. Primary structure and specific expression upon
incompatible infection with rice blast fungus. J. Biol. Chem. 269 (5), 3755–3761.
Kump, K. L., Bradbury, P. J., Wisser, R. J., Buckler, E. S., Belcher, A. R., Oropeza-
doi:10.1016/s0021-9258(17)41924-7
Rosas, M. A., et al. (2011). Genome-wide association study of quantitative resistance to
southern leaf blight in the maize nested association mapping population. Nat. Genet. 43 Poland, J. A., Bradbury, P. J., Buckler, E. S., and Nelson, R. J. (2011). Genome-wide
(2), 163–168. doi:10.1038/ng.747 nested association mapping of quantitative resistance to northern leaf blight in maize.
Proc. Natl. Acad. Sci. U. S. A. 108 (17), 6893–6898. doi:10.1073/pnas.1010894108
Landi, P., Giuliani, S., Salvi, S., Ferri, M., Tuberosa, R., and Sanguineti, M. C. (2010).
Characterization of root-yield-1.06, a major constitutive QTL for root and agronomic Prasanna, B. M., Cairns, J. E., Zaidi, P. H., Beyene, Y., Makumbi, D., Gowda, M., et al.
traits in maize across water regimes. J. Exp. Bot. 61 (13), 3553–3562. doi:10.1093/jxb/ (2021). Beat the stress: breeding for climate resilience in maize for the tropical rainfed
erq192 environments. Theor. Appl. Genet. 134 (6), 1729–1752. doi:10.1007/s00122-021-
03773-7
Lehmensiek, A., Esterhuizen, A. M., van Staden, D., Nelson, S. W., and Retief, A. E.
(2001). Genetic mapping of gray leaf spot (GLS) resistance genes in maize. Theor. Appl. Price, A. L., Patterson, N. J., Plenge, R. M., Weinblatt, M. E., Shadick, N. A., and Reich,
Genet. 103 (5), 797–803. doi:10.1007/s001220100599 D. (2006). Principal components analysis corrects for stratification in genome-wide
association studies. Nat. Genet. 38 (8), 904–909. doi:10.1038/ng1847
Lehti-Shiu, M. D., and Shiu, S. H. (2012). Diversity, classification and function of the
plant protein kinase superfamily. Philos. Trans. R. Soc. Lond B Biol. Sci. 367 (1602), Qi, H., Zhu, X., Shen, W., and Zhang, Z. (2023). A novel wall-associated kinase
2619–2639. doi:10.1098/rstb.2012.0003 TaWAK-5D600 positively participates in defense against sharp eyespot and fusarium
crown rot in wheat. Int. J. Mol. Sci. 24 (5), 5060. doi:10.3390/ijms24055060
Lennon, J. R., Krakowsky, M., Goodman, M., Flint-Garcia, S., and Balint-Kurti, P. J.
(2017). Identification of teosinte alleles for resistance to southern leaf blight in near Rashid, Z., Sofi, M., Harlapur, S. I., Kachapur, R. M., Dar, Z. A., Singh, P. K., et al.
isogenic maize lines. Crop Sci. 57 (4), 1973–1983. doi:10.2135/cropsci2016.12.0979 (2020). Genome-wide association studies in tropical maize germplasm reveal novel and
known genomic regions for resistance to Northern corn leaf blight. Sci. Rep. 10 (1),
Leonard, K. J., and Suggs, E. G. (1974). Setosphaeria prolata, the Ascigerous state of
21949. doi:10.1038/s41598-020-78928-5
Exserohilum prolatum. Mycologia 66 (2), 281–297. doi:10.1080/00275514.1974.
12019603 Ribaut, J. M., Hoisington, D. A., Deutsch, J. A., Jiang, C., and Gonzalez-de-Leon, D.
(1996). Identification of quantitative trait loci under drought conditions in tropical
Lipka, A. E., Tian, F., Wang, Q., Peiffer, J., Li, M., Bradbury, P. J., et al. (2012). GAPIT:
maize. 1. Flowering parameters and the anthesis-silking interval. Theor. Appl. Genet. 92
genome association and prediction integrated tool. Bioinformatics 28 (18), 2397–2399.
(7), 905–914. doi:10.1007/bf00221905
doi:10.1093/bioinformatics/bts444
Saito, B. C., Silva, L., Andrade, J., and Goodman, M. (2018). Adaptability and stability
Littell, R. C., Milliken, G., Stroup, W. W., Wolfinger, R., and Schabenberger, O.
of corn inbred lines regarding resistance to gray leaf spot and northern leaf blight. Crop
(2007). SAS for mixed models, 61. Cary, NC, USA: SAS Institute.
Breed. Appl. Biotec 18, 148–154. doi:10.1590/1984-70332018v18n2a21
Frontiers in Genetics 15 frontiersin.org
Omondi et al. 10.3389/fgene.2023.1282673
Sanchez, D. L., Liu, S., Ibrahim, R., Blanco, M., and Lübberstedt, T. (2018). Genome- Vivek, B., Kasango, J., Chisoro, S., and Magorokosho, C. (2007). Fieldbook: software
wide association studies of doubled haploid exotic introgression lines for root system for managing a maize breeding program: a cookbook for handling field experiments, data,
architecture traits in maize (Zea mays L.). Plant Sci. 268, 30–38. doi:10.1016/j.plantsci. stocks and pedigree information. Mexico, DF: CIMMYT.
2017.12.004
Wallace, J. G., Bradbury, P. J., Zhang, N., Gibon, Y., Stitt, M., and Buckler, E. S. (2014).
Sánchez-Sevilla, J. F., Horvath, A., Botella, M. A., Gaston, A., Folta, K., Kilian, A., et al. Association mapping across numerous traits reveals patterns of functional variation in
(2015). Diversity arrays technology (DArT) marker platforms for diversity analysis and maize. PLoS Genet. 10 (12), e1004845. doi:10.1371/journal.pgen.1004845
linkage mapping in a complex crop, the octoploid cultivated strawberry (Fragaria ×
Wang, N., Yuan, Y., Wang, H., Yu, D., Liu, Y., Zhang, A., et al. (2020). Applications of
ananassa). PLoS One 10 (12), e0144960. doi:10.1371/journal.pone.0144960
genotyping-by-sequencing (GBS) in maize genetics and breeding. Sci. Rep. 10, 16308.
Sansaloni, C., Franco, J., Santos, B., Percival-Alwyn, L., Singh, S., Petroli, C., et al. (2020). doi:10.1038/s41598-020-73321-8
Diversity analysis of 80,000 wheat accessions reveals consequences and opportunities of
Welz, G., and Geiger, H. (2000). Genes for resistance to northern corn leaf blight in
selection footprints. Nat. Commun. 11 (1), 4572. doi:10.1038/s41467-020-18404-w
diverse maize populations. Plant Breed. 119, 1–14. doi:10.1046/j.1439-0523.2000.
Shaner, G. E. (1977). The effect of nitrogen fertilization on the expression of slow- 00462.x
mildewing resistance in knox wheat. Phytopathology 77, 1051. doi:10.1094/phyto-67-
Welz, H. G. (1998). Genetics and epidemiology of the pathosystem Zea mays/
1051
setosphaeria turcica.
Shi, L.-y., Li, X.-h., Hao, Z.-f., Xie, C.-x., Ji, H.-l., Lü, X.-l., et al. (2007). Comparative
Wenzl, P., Carling, J., Kudrna, D., Jaccoud, D., Huttner, E., Kleinhofs, A., et al. (2004).
QTLmapping of resistance to gray leaf spot in maize based on bioinformatics.Agric. Sci.
Diversity arrays technology (DArT) for whole-genome profiling of barley. Proc. Natl.
China 6 (12), 1411–1419. doi:10.1016/S1671-2927(08)60002-4
Acad. Sci. U. S. A. 101 (26), 9915–9920. doi:10.1073/pnas.0401076101
Sserumaga, J. P., Makumbi, D., Assanga, S. O., Mageto, E. K., Njeri, S. G., Jumbo, B.
Wisser, R. J., Balint-Kurti, P. J., and Nelson, R. J. (2006). The genetic architecture of
M., et al. (2020). Identification and diversity of tropical maize inbred lines with
disease resistance in maize: a synthesis of published studies. Phytopathology 96 (2),
resistance to common rust (Puccinia sorghi Schwein). Crop Sci. 60 (6), 2971–2989.
120–129. doi:10.1094/phyto-96-0120
doi:10.1002/csc2.20345
Wisser, R. J., Kolkman, J. M., Patzoldt, M. E., Holland, J. B., Yu, J., Krakowsky, M.,
Sun, H., Zhai, L., Teng, F., Li, Z., and Zhang, Z. (2021). qRgls1. 06, a major QTL
et al. (2011). Multivariate analysis of maize disease resistances suggests a pleiotropic
conferring. resistance to gray leaf spot disease in maize. Crop J. 9, 342–350. doi:10.1016/
genetic basis and implicates a GST gene. Proc. Natl. Acad. Sci. U. S. A. 108 (18),
j.cj.2020.08.001
7339–7344. doi:10.1073/pnas.1011739108
Swart, V., Crampton, B. G., Ridenour, J. B., Bluhm, B. H., Olivier, N. A., Meyer, J. M.,
Xiao, Y., Liu, H., Wu, L., Warburton, M., and Yan, J. (2017). Genome-wide
et al. (2017). Complementation of CTB7 in the maize pathogen Cercospora zeina
association studies in maize: praise and stargaze. Mol. Plant 10 (3), 359–374. doi:10.
overcomes the lack of in vitro cercosporin production. Mol. Plant Microbe Interact. 30,
1016/j.molp.2016.12.008
710–724. doi:10.1094/MPMI-03-17-0054-R
Zhang, H., Jing, W., Zheng, J., Jin, Y., Wu, D., Cao, C., et al. (2020). The ATP-binding
Tang, Y., Liu, X., Wang, J., Li, M., Wang, Q., Tian, F., et al. (2016). Gapit version 2: an
cassette transporter OsPDR1 regulates plant growth and pathogen resistance by
enhanced integrated tool for genomic association and prediction. Plant Genome 9, 2.
affecting jasmonates biosynthesis in rice. Plant Sci. 298, 110582. doi:10.1016/j.
doi:10.3835/plantgenome2015.11.0120
plantsci.2020.110582
Technow, F., Bürger, A., and Melchinger, A. E. (2013). Genomic prediction of
Zhang, Y., Xu, L., Fan, X., Tan, J., Chen, W., and Xu, M. (2012). QTL mapping of
northern corn leaf blight resistance in maize with combined or separated training
resistance to gray leaf spot in maize. Theor. Appl. Genet. 125 (8), 1797–1808. doi:10.
sets for heterotic groups. G3 (Bethesda) 3 (2), 197–203. doi:10.1534/g3.112.004630
1007/s00122-012-1954-z
Tuberosa, R., Salvi, S., Sanguineti, M. C., Landi, P., Maccaferri, M., and Conti, S.
Zhao, Y., Gowda, M., Liu, W., Würschum, T., Maurer, H. P., Longin, F. H., et al.
(2002). Mapping QTLs regulating morpho-physiological traits and yield: case studies,
(2012). Accuracy of genomic selection in European maize elite breeding populations.
shortcomings and perspectives in drought-stressed maize. Ann. Bot. 89 (7), 941–963.
Theor. Appl. Genet. 124 (4), 769–776. doi:10.1007/s00122-011-1745-y
doi:10.1093/aob/mcf134
Zwonitzer, J. C., Coles, N. D., Krakowsky, M. D., Arellano, C., Holland, J. B.,
Van Inghelandt, D., Melchinger, A. E., Martinant, J. P., and Stich, B. (2012). Genome-
McMullen, M. D., et al. (2010). Mapping resistance quantitative trait loci for three
wide association mapping of flowering time and northern corn leaf blight (Setosphaeria
foliar diseases in a maize recombinant inbred line population-evidence for multiple
turcica) resistance in a vast commercial maize germplasm set. BMC Plant Biol. 12, 56.
disease resistance? Phytopathology 100 (1), 72–79. doi:10.1094/phyto-100-1-0072
doi:10.1186/1471-2229-12-56
Frontiers in Genetics 16 frontiersin.org