Early primary root growth and its relation with drought-tolerance related traits measured in field conditions.

(A) Stress patterns identified by clustering of simulated fraction of transpirable soil water (FTSW) trend during the crop growth of 2000-2021 years in Bambey station. FTSW = 0.3 (horizontal red line) was considered as the onset of water stress. Each point is the average of a few daily FTSW based on the biological day which was used instead of daily FTSW to highlight the critical stages (panicle initiation and flowering, black lines). Thick lines and shaded area show the mean and two times standard error, respectively. Blue red and green line correspond to early-stress, late-stress, and no-stress respectively. (B) Root length in 122 pearl millet inbred lines derived from West and Central African landraces. Root length was measured at 6 days after germination in a paper-based hydroponic system installed in a growth chamber. Points represent mean +/- se. (C) Correlation between root length and the field measured traits using adjusted lsmeans across both years. The Pearson correlation coefficients are indicated for each pair of traits. (D) Linear regression between root length and stay-green. Red points represent the lsmean for the 6 inbred lines that were common between the two field trials.

Genetic dissection of primary root length in pearl millet and cellular analysis of root apical meristem and primary root growth of contrasted pearl millet lines in two experimental systems. (A) Manhattan plot of the GWAS by lfmm ridge method (Caye et al., 2019). The horizontal axes correspond to the map position of each of the 392,216 SNPs identified by GBS in a group of 122 inbred lines. The vertical axes indicates the -logĭ0 p value of the statistic. The dash line delimits the threshold for highly significant SNPs (p value < 10-4). Highlighted in red, significant SNP markers above the 0.05 FDR significance threshold. (B) BSA identification of significant regions associated with primary root length using bulks of contrasted F2 lines from a bi-parental cross. The plot shows the Euclidean Distance statistic profile (y axes) across the seven pearl millet chromosomes (x axes). The dash line indicates the 95% confidence interval threshold for the localisation of significant regions. In both plots, the shaded area delimits the extent of the six significant regions identified by BSA and therefore the overlap with significant SNPs identified by GWAS (A) and the correspondence with the BSA peaks found (B). (C) Confocal image of the root tip of the contrasted inbred lines for primary root growth (ICML- IS-1115 and SL2 with slow and fast growth respectively). (D) Estimation of cell production rate for each genotype according to Beemster et al (2002). (E) Maximum cell length reached in the root elongation zone for each genotype. *** p-value < 0.0001, ** p-value < 0.001, * p-value < 0.01, ns not significant.

Identification of candidate genes for primary root length in pearl millet. (A) Annotated genes in 100 kb region of chromosome 1 harbouring significant SNPs identified by GWAS and coincident with a BSA significant region. Blue asterisk shows the position of the SNP with the most significant association. Grey arrows represent the predicted genes according to the reference genome (Varshney et αl, 2017). The red arrow represents the new predicted gene, PgGRXC9, identified as a result of reannotation of the genomic interval using de novo pearl millet assembly based on contig sequences. (B) GWAS Manhattan plot zoomed in the 100 kb interval on chromosome 1. Dots represent the -log 10 (pvαl) of the lfmm ridge statistic (y axis) for the SNPs identified in the region. The dashed horizontal line shows the threshold for highly significant SNPs. Positions and physical distance between genes and markers are displayed in bp. (C) Primary root length phenotype associated to the allelic variants of the two significant SNPs in the region: chrl _231264526 (Nc/c=26, NGG= 74)). (D) Expression of PgGRXC9 for the two alleles from the RNAseq data. * p-value < 0.01. (E) Transversal section in the elongation zone of a primary root showing the expression profile of PgGRXC9 as indicated by RNAscope. Scalebar = 50 µm.

The Arabidopsis homolog of PgGRXC9 regulates root growth in Arabidopsis. (A) Phylogenetic tree of ROXYs protein sequences. This tree has been obtained from pairwise alignments of all whole protein sequence pairs, using Neighbor­Joining method and Jukes-Cantor distance matπx. GRXC1 (grey), a member of Class I GRX, is used as the tree root. Arabidopsis ROXY 18-21 are represented in green, and PgGRXC9 is represented in purple. (B) Contrasted primary root phenotype in NosO and roxyl9 in 17 days old seedlings grown in agar plates. (C) Primary root length of NosO (N=27) and roxyl9 (N=25). (D) Cell elongation rate. (E) Estimation of the cell production rate assuming a steady root grow th rate. * p-value < 0.01, ns: not significant.

Lapse between the first and the second significant rainfall.

Significant rainfall was considered as more than 10 mm rainfall per day. Data from the Bambey experimental station.

Covariation of root length, plant height and drought related traits including maximum quantum efficiency of photosystem II (FvFm) and performance index of photosynthesis (PI) measured at 32 days after sowing, and stay-green and plant survival measured at 42 days after sowing.

Mean values measured in 2018 (A) and 2020 (B) field trials were used for Principal Component Analysis (PCA). The contribution of each PCA axis (PC1 and PC2) is indicated on the graph.

GWAS Manhattan plot and p-values QQ plots obtained with five methods. (A) Analysis of variance or ANOVA, (B) Efficient Mixed Model Association or EMMA, (C) Mixed lineal model or MLM, (D) Latent Factor Mixed Model using an stochastic algorithm (MCMC), or LFMM_LEA, (E) Latent Factor Mixed Model using the ridge algorithm or LFMM_ridge.

Contrasted primary root length of lines ICML-IS 11155 and SL2.

(A) on paper roll 7 days after germination (NICML-IS 11155 = 32, NSL2 = 34) and (B) soil column using by X-ray tomography 10 days after germination (NICML-IS 11155 = 4, NSL2 = 5, P-value= 0.058).

Distribution phenotypes F2 and parents ICML-IS 11155 and SL2.

(A) Distribution of the phenotype parents vs F2 (histogram or boxplot). (B) Table of averages.

Cell length profile in the root meristem of (A) SL2 and (B) ICML-IS 11155.

Expression pattern of PgGRXC9 in the root tip as revealed by RNAscope.

Transversal sections in the root cap, elongation zone and mature zone of the root. Scalebars: 50 μM. Drawing created with BioRender.com

Protein alignment of PgGRXC9 and Arabidopsis ROXY members.

(A) Whole sequence alignment of PgGRXC9 and the 21 ROXY from Arabidopsis thaliana. The first 120 N-terminal amino acids of PgGRXC9 are not represented in this alignment. (B) C-terminus alignment of ROXY proteins. This alignment displays the L**LL (orange) and the ALWL (light blue) C-terminal motifs, important for binding to TGA and TPL/TPR. The green bar and purple triangle mark the ROXY18-21 Arabidopsis subclass and the PgGRXC9, respectively.

Drought-tolerance related traits in selected inbred lines grown in field conditions in 2018 and 2020.

After sowing, seeds were irrigated with 30 mm of water to allow germination and plants grew for 42 days without any further irrigation. Plant height, stay-green and survival were measured at 42 days after sowing while photosynthesis related traits (FvFm: maximum quantum efficiency of photosystem II; PI: performance index of photosynthesis) were measured at 32 days after sowing. Root length was measured 6 days after germination on a paper-based hydroponic system in growth chamber. Numbers indicate mean ± se.

GWAS p-values for the significant SNP markers identified through LFMM (ridge) using four other GWAS methods

Significant genomic regions identified by Bulk Segregant Analysis (BSA) for primary root length at the 95% confidence interval.

Common GWAS and BSA Marker trait associations for primary root length.

Top ten GO enriched terms in the differentially expressed genes between contrasting lines for root growth. Annotated: number of genes annotated with the corresponding GO term in the pearl millet genome; Significant: number of genes annotated in the differentially expressed genes with the corresponding GO term; Expected: number of corresponding GO term expected in the differentially expressed genes; Rank: rank of p-value in the classic Fisher test; p-value: p-value of the classic Fischer test; Weight01: weight01 in the Fischer test taking into account hierarchical links between GO terms.

Lines selected for field trials. Average root length represent the value observed in paper growth system in the original phenotyping performed for GWAS/