S1 Fig.

Field trial experimental design. The design follows a completely randomized block layout with four replicates. Each replicate consists of 10 sub-blocks, each containing 16 plots. Each plot (detail shown in the top right) is planted with three rows of 10 plants of the same genotype, with a spacing of 0.9 m between rows and 0.3 m between plants. The red dots indicate locations where soil samples were collected at 4 different depths (0–140 cm). Leaf samples, soil analyses, and root traits were also studied (illustrations shown at the bottom right). The blue icons represent irrigation pumps. S2 Fig. Correlation plot for soil ion content in 2021 (A) and 2022 (B) field sites. Heatmap representing Pearson’s correlation coefficients between soil ion concentrations measured at four different depths (0–140 cm). Color gradients indicate the Pearson’s correlation coefficient. Non-significant correlations at a p-value threshold of 0.05 are indicated with a cross. S3 Fig. Boxplot representing variation in ion content in 2021 and 2022 in the PMIGAP panel. Ion content is represented as mg/kg. p-values from the Wilcoxon test are represented. S4 Fig. Correlation plot for leaf ion content of the PMIGAP panel measured during the experimental field study. Heatmap representing Pearson’s correlation coefficients between BLUEs of all accessions of the panel observed in 2021 (A) and 2022 (B). Color gradients indicate the Pearson’s correlation coefficient. Non-significant correlations at a p-value threshold of 0.05 are indicated with a cross. S5 Fig. Correlation plot for ion content, root (A) and agro-morphological (B) traits in 2021 and 2022. Heatmap representing Pearson’s correlation coefficients between ion content, root and agro-morphological traits. Color gradients indicate the Pearson’s correlation coefficient. Significant correlations at a p-value threshold of 0.05 are indicated in bold. Roots traits are number of metaxylem vessels (MX_Number), mean area of metaxylem vessels (Meansize_MX), sclerenchyma pixel sum (SCL_Area), total area of the root section (RootArea, µm²), total area of metaxylem vessels (Totalsize_MX, µm²), total area of the stele (SteleArea, µm²), Ratio between stele area and root area (SR_Ratio), ratio between SCL and root area (SCL_Ratio). Agromorphological traits are: number of days after sowing when 50% of plants in the plot show flowering (DTF), 1000 grains weight (PMG, g), total grain weight at harvest (GW, g), number of tillers on three plants measured at maturity (Tiller_number), shoot dry biomass of plants phenotyped for root traits (SDW, g) at 49 DAS in 2021 or 42 DAS in 2022, shoot dry biomass of three plants harvested at maturity (SDW_Maturity, g), plant height from soil to flag leaf at maturity (HSDF, cm). S6 Fig. Histogram representing variation in ion content in 2021 (A) and 2022 (B) in the PMIGAP panel. Ion content is represented as mg/kg. Data used for the GWAS analysis follow a normal distribution for all the variables studied. S7 Fig. Manhattan plot and Quantile-Quantile (QQ) plots of Cadmium. GWAS result using the Fisher combining method with LFMM, EMMA, and BLINK. The red line indicates the significance threshold for the respective methods. QQ Plot indicated that the GWAS models fitted well to the data, with observed p-values distributed uniformly and showing inflation at higher values. S1 Table. Passport data of the different pearl millet lines used in the study. S2 Table. List of all SNPs identified by GWAS. Information includes the name of the ion (Ion), SNP identifier (SNP), chromosome number (Chrom), SNP position on the chromosome (POS), most significant p-value (pvalue_max), method with the most significant p-value (significant_method), names of all methods that detected the SNP (methods), and whether the SNP was detected in 2021, 2022, using the combined Fisher method, or all (Year). S3 Table. List of marker trait association (MTA) retained. Information includes the name of the ion (Ion), SNP identifier (SNP), chromosome number (Chrom), SNP position on the chromosome (POS), most significant p-value (pvalue_max), method with the most significant p-value (significant_method), names of all methods that detected the SNP (methods), and whether the SNP was detected in 2021, 2022, using the combined Fisher method, or all (Year). S4 Table. List of QTL obtained based on linkage disequilibrium. Information includes the name of the quantitative trait loci (QTL_name), chromosome number (chrom), number of significant MTA in the QTL (nbr Sig_snp), region of the QTL (start and end of the QTL in 50 kb around the most extreme or significant MTA; QTL_Pos), variable name used for GWAS of variables without (content: cont) or with (residual: res) flowering time influence, whether detected QTL is identified by GWAS of the ion or residuals, or both (Type), the ion name (Ion). S5 Table. List of marker trait association (MTA) retained based on criteria selection and identified by GWAS on residuals of the linear regression between ion content and flowering time. Information includes the name of the ion (Ion), SNP identifier (SNP), chromosome number (Chrom), SNP position on the chromosome (POS), most significant p-value (pvalue_max), method with the most significant p-value (significant_method), names of all methods that detected the SNP (methods), and whether the SNP was detected in GWAS of the ion or residuals, or both (Type). S6 Table. List of marker trait association (MTA) retained based on criteria selection and identified by GWAS on flowering time. Information includes the name of the trait (Day_flowering), SNP identifier (SNP), chromosome number (Chrom), SNP position on the chromosome (POS), most significant p-value (pvalue_max), method with the most significant p-value (significant_method), names of all methods that detected the SNP (methods). S7 Table. Gene Expression level in leaves and roots. Information includes gene name (Gene) and his level expression in leaves (Leaf), in roots (crown, lateral and primary).
(ZIP)