# Methods for analysing wildlife DNA methylation data

## TL;DR

This paper reviews methods for analyzing DNA methylation in wildlife, focusing on age estimation and health insights, using bottlenose dolphins as a case study.

## Contribution

The paper highlights limitations of epigenetic clocks in wildlife and suggests the need to separate age and health analyses.

## Key findings

- Epigenetic clocks can estimate age but are less effective for health indicators in wildlife.
- Age and health analyses are confounded, making accurate predictions challenging.
- Wildlife methylation data offer insights into population health and age structure.

## Abstract

The analysis of DNA methylation data for wildlife conservation is gaining momentum as the technology for quantifying the methylome becomes mainstream. The use of epigenetic information extracted from tissue samples can be used for estimating chronological age, individual traits and phenotypic variation. Methylation data present an exciting opportunity to study wildlife populations, with the potential to provide insights into age structure, vital rates and health. However, the statistical methodology for answering the emerging research questions has been developed and mostly applied in the human biomedical setting. We review the key methodologies commonly used in wildlife settings, and methods that have been used only in human studies so far that could improve our understanding of wildlife epigenomic changes. We show how the different methods relate to each other and how they link to research questions, illustrating each approach with data from a case study, a large dataset from wild bottlenose dolphins (Tursiops spp.) from the US southeast and Gulf coast. Estimating chronological age from models called epigenetic clocks and understanding the relationship between epigenetic indicators of health and exposure to stressors are both key goals in wildlife settings; however, we show that a single model cannot do both accurately. This is a fundamental limitation of clock-type models and might explain why some age-related health conditions have been found to be related to epigenetic age and others not. Decoupling the analysis of age and health is challenging because the two are confounded but is especially important in wildlife settings where age prediction is often the main analytical objective.