Estimating age and investigating epigenetic changes related to health across multiple bottlenose dolphin populations

Epigenetic age estimation has the potential to influence conservation approaches for wildlife. DNA extracted from a small skin sample can be analyzed for methylation changes, which can be related to chronological age through an epigenetic clock to obtain a DNA-based age estimate (DNAm age). For a wide range of cetacean species, skin samples can be obtained remotely using dart biopsy. We tested multiple modeling approaches (elastic net regression, random forest regression, and a hybrid of elastic net and random forest classification) using DNA methylation data from 426 skin samples from observed-age wild bottlenose dolphins (Tursiops spp.), and 50 managed care dolphins, to develop an epigenetic clock. The hybrid model gave highly accurate age predictions for calf (<2 yrs., n = 28), subadult (2–15 yrs., n = 273) and adult (15–25 yrs., n = 110) age classes with median absolute errors (MAE) of 0.149, 1.46 and 3.25 yrs., respectively. In the oldest individuals (>25 yrs., n = 65), age predictions were less accurate, with MAE of 5.90 yrs. The overall MAE across all ages (n = 476) was 1.91 yrs. DNAm age has been linked to health in humans, with higher age residuals (i.e., higher DNAm age as compared to actual age) being associated with increased risk of morbidity and mortality. We compared dolphin age residuals with previously reported health scores that are predictive of one year mortality risk. Our findings suggest that analysis of DNA methylation patterns is a viable approach, not only for estimating age of dolphins, but potentially also to assess individual and population health.