The detailed ChIP protocol is in the Supplemental Material. For sequencing, DNA fragments were made into libraries using the NEBNext Ultra II FS kit (New England Biolabs). Libraries were pooled and sequenced on a NextSeq 500 platform (Illumina) with paired-end 150-nt runs at high output (developing seeds) or single-end 75-nt runs at high output (mature embryos). Adaptor sequences were removed using Trimmomatic (Bolger et al. 2014 (link)); trimmed reads were then aligned on the Arabidopsis TAIR10 genome using Bowtie2 (Langmead and Salzberg 2012 (link)). Duplicated reads are removed using the Picard Tool suite, and coverage was calculated and mapped to DMRs using Bedtools. All distributions were compared with the first WT sample distribution using the Kolmogorov–Smirnov test with R.