Plant materials and genotyping
A RIL population consisting of 204 lines was developed from a cross between two Chinese maize inbred lines DH4866 (resistant) and T877 (susceptible). The resistant line DH4866 was derived from a cross between the two Chinese elite inbred lines 7922 and Ye478. The susceptible line T877 was developed from a cross between an American hybrid 78599 and a Chinese elite inbred line E28 [28]. This population was generated, identified, maintained, and provided by Jiangsu Yanjiang Institute of Agricultural Sciences, Nantong, China, and was not deposited elsewhere.
The RIL population and two parental inbred lines were genotyped with an Affymetrix microarray CGMB56K SNP Array, which contains 56,000 maize SNPs, made by China Golden Marker (Beijing) Biotech Co., Beijing, China. After quality control, 9,780 high-quality polymorphic SNPs were used to construct linkage map. The genetic linkage map contained 1,868 bin markers and spanned a total length of 3,081.8 cM, with a mean interval length of 1.65 cM. Detailed information on construction of the bin map has been reported previously [19].
Field trials and phenotyping
The RIL population and its parents were planted at three locations of China in 2019: Sanya, Hainan Province (SY, 108°E, 18°N); Nantong, Jiangsu Province (NT, 120°E, 31°N) and Xinxiang, Henan Province (XX, 113°E, 35°N). Each genotype was grown in single rows with 3.0 m in length and spaced 0.6 m between rows with a planting density of 65,000plants/hm2, following a randomized complete block design with two replications per location. The field management followed common agricultural practice at the three locations.
Artificial inoculation with an aggressive isolate of F. graminearum (strain F0609) was conducted. The fungal inoculum was prepared according to Sun et al. (2018) [29]. Briefly, the strain was cultured on potato dextrose agar (PDA) medium for about two weeks, followed by culturing several agar plugs with fully grown mycelium in sterilized mung bean soup at 200 rpm, 28 °C for 2 ~ 3 d. The spore suspension was filtered, counted and adjusted to a concentration of 1 × 106 spores/ml supplemented with 0.001% surfactant Tween-20. For field inoculation, approximately 10 d after silk emergency, the primary ear of each plant per plot was pierced at the base, middle and top of a husk. 1–2 kernels were stabbed by a 10 cm needle, but not reaching the cob, and then three wounds were injected with 200 ul of spore suspension. At the physiological maturity stage, inoculated ears were harvested, and the disease severity of each ear was scored using a rating scale from 1 to 7 (1 = 0%, 2 = 1–3%, 3 = 4–10%, 4 = 11% to 25%, 5 = 26–50%, 6 = 51–75%, and 7 = 76–100% kernels showing visible disease symptoms) (Figure S1) [30]. The average of two replicates in each location was used for subsequent analysis.
The kernel assay was conducted according to the protocol for F. verticilloiedes described by Gao et al. (2007) [31]. Three replicates were included for each line, and at least three independent experiments were conducted with consistent results obtained.
Phenotypic data analysis
The phenotypic statistical analysis was performed using R version 3.6.0 software (https://www.r-project.org/). The correlation analysis for each location was conducted by the chart.Correlation function of the PerformanceAnalytics package [32]. Analysis of variance of disease severity was performed using the lmer function of the lme4 package [33] based on the following model: yij = µ + gi + ej + geij + εij, where yij is the trait measured, µ is the grand mean for all locations, gi is the genotypic effect for the ith genotype, ej is the locational effect for the jth location, geij is the genotype × location interaction effect, and εij is the residual error. The broad-sense heritability (H2) was estimated following Knapp et al. (1985) [34]: H2 (%) = σ2g/(σ2g + σ2ge/n + σ2e/nr), where σ2g is the genotypic variance, σ2ge is the interaction of the genotype with the location, σ2e is the error variance, n is the number of locations, and r is the number of replications. The best linear unbiased prediction (BLUP) values of each line was estimated to less environmental error.
QTL mapping
QTL mapping was performed using composite interval mapping (CIM) [35] in Windows QTL Cartographer 2.5 [36]. Model 6 of the Zmapqtl module was selected to detect QTLs and their effects with scanning every 0.5 cM at a window size of 10 cM. Forward-backward stepwise regression with five controlling markers was used to control the background from flanking markers. After 1000 permutations, the threshold logarithm of the odds (LOD) value was determined at a significance level of P < 0.05. The confidence interval of QTL positions was estimated with the 1.5-LOD support interval method. Multiple interval mapping (MIM) in Windows QTL Cartographer 2.5 was performed to estimate the interactions of identified QTLs with the Bayesian Information Criteria as the criteria [37].