Each plot displays the number of CpG sites for which adversity significantly predicted methylation, after controlling for covariates and correcting for multiple comparisons using (a) a Bonferroni threshold (p<1×10−7, n=38 sites) and (b) a False Discovery Rate (FDR) correction q < 0.05 (n=380 sites). The distribution of theoretical models chosen first by the LARS procedure for top CpG sites was significantly different than expected by chance, with exposure to adversity during sensitive periods, especially during very early childhood, more frequently predicting methylation.