2.2. Global Methylation Analysis via ELISA

We measured global methylation levels in 20 randomly selected individuals per temperature (26 °C, 28 °C and 31 °C) via methylation-specific ELISA using the MethylFlashTM Global DNA Methylation (5-mC) ELISA Easy Kit (Epigentek), following the manufacturer’s instructions. DNA samples were diluted to 10 ng/uL for ELISA. Reactions were run in a Chromate 4300 machine at the Iowa State University proteomics facility. All plates included a positive control standard curve from the kit plus a turtle-specific standard curve of eight standards obtained by serially diluting (1:1) a sample of pooled DNA from all individuals.

The normality of the absorbance values was tested using QQ plots in RRPP [50], and results indicated that no data transformation was necessary (Supplementary Figure S1). We converted the absorbance values to methylation percentage, following the equation [51]:

where 5 mC% = percentage of 5-methylcytosines, OD = optimal density, NC = negative control, Slope = standard curve slope, S = input DNA in ng. For the statistical analysis of these values, we evaluated first if the standard curves of positive control (provided with the kit) and our turtle standard curves were linear using a generalized linear model (GLM). Then, we performed an ANCOVA to compare slopes between these two types of standard curves. Next, we tested for differences in global methylation using ANOVA. Tests were applied first to the calculated methylated percentages that represent the total 5-mC fraction in the sample accounting for the kit’s specificity in detecting DNA methylation, given that it is calculated proportionally to the OD intensity measured [51]. Second, ANOVA was applied to the absorbance values, following a traditional ELISA data analysis [52]. As the interaction between temperature and sex was not significant for the analysis of 5 mC% in the full factorial ANOVA (p > 0.05), we performed a reduced ANOVA that excluded the interaction term. Additionally, because the temperature and sex terms were not significant in the reduced model, we then tested for differences combining samples by sex (26 °C male + 28 °C male and 31 °C female + 28 °C female), and by temperature (31 °C, 26 °C and 28 °C). On the other hand, because the sex and temperature interaction was significant for the absorbance values (p < 0.05) we did a pairwise comparison of all temperature by sex combinations.