Inferences from epigenetic information in an ecological context: a case study of early-life environmental effects on DNA methylation in zebra finches

DNA methylation (DNAm) plays a key role in mediating phenotypic responses to environmental variation. Various approaches exist to link DNAm variation to phenotypes, ranging from single nucleotide resolution to the development of composite indexes. We here discuss the conceptual differences between these approaches through a case study using whole genome, longitudinal DNAm data from adult zebra finches raised in experimentally created small or large broods. Specifically, we (i) identified CpG sites where DNAm was affected by brood size, and (ii) developed two DNAm-based indexes of the early-life environment (DMSi: based on differentially methylated sites with respect to rearing brood size; SMLmi: supervised machine learning-derived index optimized to predict brood size). We also compared results obtained by either merging or analyzing longitudinal DNAm samples separately and discussed how the permanence or transience of DNAm changes shapes responses to environmental variation through life. Our findings confirm that early-life environment leaves lasting DNAm signatures detectible in adulthood, and this effect is stronger later in life. Importantly, both indexes predicted early-life growth rates, demonstrating that DNAm-based indexes can be leveraged to retroactively quantify aspects of early-life conditions, providing a powerful novel tool in the study of wild populations.