Background DNA methylation in the form of 5-methylcytosine (5mC) is the most abundant base modification in animals. However, 5mC levels vary widely across taxa. Whilst vertebrate genomes are hypermethylated, in most invertebrates, 5mC concentrates on constantly and highly transcribed genes (gene body methylation GbM) and, in some species, on transposable elements (TEs), a pattern known as ‘mosaic’. Yet, the role and developmental dynamics of 5mC and how these explain interspecific differences in DNA methylation patterns remain poorly understood, especially in Spiralia, a large clade of invertebrates comprising nearly half of the animal phyla. Results Here, we generate base-resolution methylomes for three species with distinct genomic features and phylogenetic positions in Annelida, a major spiralian phylum. All possible 5mC patterns occur in annelids, from typical invertebrate intermediate levels in a mosaic distribution to hypermethylation and methylation loss. GbM is common to annelids with 5mC, and methylation differences across species are explained by taxon-specific transcriptional dynamics or the presence of intronic TEs. Notably, the link between GbM and transcription decays during development, alongside a gradual and global, age-dependent demethylation in adult stages. Additionally, reducing 5mC levels with cytidine analogues during early development impairs normal embryogenesis and reactivates TEs in the annelid Owenia fusiformis. Conclusions Our study indicates that global epigenetic erosion during development and ageing is an ancestral feature of bilateral animals. However, the tight link between transcription and gene body methylation is likely more important in early embryonic stages, and 5mC-mediated TE silencing probably emerged convergently across animal lineages. Overall design: Methylomes of three annelid species (O. fusiformis, C. teleta and D. gyrociliatus) at different developmental timepoints to investigate the 5mC dynamics during embryogenesis and the life cycle of this animals using WGBS, EM-seq and Nanopore sequencing. To investigate the impact of DNA methylation during development, we combined cytidine analogs, RNA-seq and EM-seq.