univariate test for comparison


currently i'm testing a tree-based algorithm based on permutation tests (doesn't really matter for the question) for a paper. For every variable (100 in my simulation) I get a p-value based on the null hypothesis of independence between the variable and the response Y.
Now I should compare my p-values with the p-values from a (classical) univariate test.

Which test should I take? (response and predictor variables are continuous)
And means "univariate" that I should calculate a single test for every variable to get a p-value?