Figure 1. The centromere map of 251 rice mini‐core germplasm accessions (A) Schematic representation of the construction of the rice centromere map. The pink region on the chromosome indicates the CentO‐enriched region (CoER), and the blue region on the chromosome indicates the peri‐CoER. (B–D) Tukey's multiple comparisons test of the CoER length (B) and the number of CoER elements with CentO satellite (C) and long terminal repeats (LTRs) (D) across different subpopulations in the complete CoER dataset of the centromere map. Significance is represented in alphabetical notation (P < 0.05). (E) Landscape of CoER length (i), CentO repeat copy (ii) and LTR number (iii) in the complete CoERs across different chromosomes. Ob, Og, Or, Osi, Osj, As, Af and All refer to Oryza barthii, O. glaberrima, O. rufipogon, O. sativa indica, O. sativa japonica, African rice, Asian rice, and the 251 accessions, respectively.

Figure 2. High diversity analysis of CentO repeats in the rice centromere map (A) CentO repeat copy across all complete CentO‐enriched regions (CoERs), with red bar highlighting the 155 and 165 bp CentO satellite repeats. (B) Heatmap showing the identity of CentO consensus sequences on each CoER between centromeres. With light red (lowest) to dark red (highest), based on the average similarity ratio (%) of the consensus sequences on each chromosome for each complete CoER. (C) Maximum likelihood phylogenetic tree of CentO consensus repeats from 2,188 complete CoERs. Tracks from outer to inner circles indicate: subpopulation distribution, centromere distribution of different chromosomes. (D) Comparison of CentO satellite repeats from two clades. (i) Two bar charts showing the distribution of different lengths of CentO repeats in each subpopulation within Clade I and Clade II. The x‐axis shows the CentO monomer length, and the y‐axis shows the ratio of CentO repeats with that subpopulation. (ii) Two pie charts showing the ratio of different lengths of CentO repeats within Clade I and Clade II. The CentO repeat lengths category is shown in the legend. Excluded were the CentO repeat lengths that are <10% of the total. (E) Frequency of CentO sequence variant sites across the consensus repeats from the CoERs. The color‐coded DNA sequence is shown on left (dark green, A; red, T; orange, C; light green, G; ‐, gap).

Figure 3. Characterization of long terminal repeats (LTRs) in the rice centromere map (A, B) The linear regression curves show large correlation between the number of LTRs (y‐axis) and the distance from the centromere (x‐axis) (except for some accessions on chromosomes 9 and 11, P‐values are < 2.2e‐16 by Spearman correlation analysis) (A), and the number of LTRs within CentO‐enriched regions (CoERs) (y‐axis) and the CoER length (x‐axis) (B). The linear regression curves for each subpopulation are shown as R‐values and P‐values in the legend. (C) Number variation of LTR elements in complete CoERs dataset across different subpopulations and chromosomes. The x, y, and z axes represent the mean, SD, and coefficient of variation (CV) of LTR numbers, respectively. (D) The genome‐wide distribution of LTRs in complete CoERs across different subpopulations. The x‐axis represents the distance from the CoER origin to the flanking regions (200 kb) on either side. The y‐axis represents the LTR density, and was calculated by dividing the number of LTRs in the CoER length. The subplot that shows the LTR distribution patterns in the 5 Mb flanking regions of CoER (above the main plot). (E) The LTR distribution patterns with a representative example of the CoERs on chromosome 1. Upper panel, density plot of LTR numbers in CoER start sites and genomic regions on chromosome 1. Lower panel, heatmap of LTR number distribution in each CoER start site and genomic regions on chromosome 1. The red dashed line indicates the CoERs. The window size is 200 kb. The heatmap shows the LTR density between different genome regions on chromosome 1 for each CoER, ranging from blue (lowest) to red (highest). The color‐coded representation indicates different subpopulations of accessions. (F, G) The percentage of LTRs with different insertion times (F) and different LTR families (G) in different genomic regions across complete CoERs. The regions are defined as the core centromere region (CoER) (i), the 1 Mb flanking region of the CoER on either side (peri‐CoER) (ii), the rest of the genome (non‐(peri)CoER) (iii), and genome‐wide (iv). Ob, Og, Or, Osi, and Osj refer to Oryza barthii, O. glaberrima, O. rufipogon, O. sativa indica, and O. sativa japonica, respectively.

Figure 4. Identification of a centromeric gene OsMAB based on structural variation – expression quantitative trait loci (SV‐eQTL) analysis (A) Local Manhattan plot of OsMAB expression level and SV. The gray region in the Manhattan plot indicates centromere region, and the centromere position information was obtained from the CENH3 (centromere‐specific histone H3‐like protein) chromatin immunoprecipitation ‐ sequencing of the telomere‐to‐telomere (T2T) Nipponbare genome. The black dashed line represents P‐value threshold (P‐value = 2.48e‐5). (B, C) Fragments per kilobase of transcript per million mapped reads (FPKM) (B) of OsMAB and tiller numbers (C) of accessions with (Hap.2) or without (Hap.1) and the deletion in the promoter of OsMAB. **P < 0.01, ***P < 0.001 by Wilcox test (B, C). (D) Morphology of NIP and osmab mutant. Bar, 10 cm. (E) Sequencing comparison of mutation sites in the wild type and the osmab mutant. (F) Gene structure and variation sites within coding sequences (CDs) of the OsMAB. (G) Tiller numbers of NIP compared with osmab mutant (n = 12). ***P < 0.001 by Student's t‐test. (H) Subcellular localization of OsMAB (bar, 5 μm).