Family-wise error rate in T test and ANOVA

I'm using one psychological questionnaire that consists of 3 sub scales. As part of the data analysis, I want to carry out 2 Independent sample T tests and 1 one-way ANOVA for each sub scale. I need to do the ANOVA because the independent variable in this case (age) has 4 levels. So my question is:

When controlling for family-wise error (e.g. with Bonferroni's correction) should I calculate the adjusted p-values only based on the 2 T tests or also add the ANOVA (that is, 3 tests in total)? I cannot decide because the independent variable in the ANOVA test is not dichotomous and I'm not sure if the p-values are comparable to the ones from the T tests.

Thank you! Any help will be much appreciated!