the p-values were computed on the non-transformed fold changes.

the log fold change is there only to show you in an easier-to-interpret manner what the fold changes were

in a log2 transform -2 and 2 are at the same amount away from 0 whereas the original values would be 4x and 0.25x

Secondly, where can I find the error between biological replicates that was found in the RNA-seq experiment? I did not analyze the raw data myself; I had a vendor send me the DGE data. I can see DGE levels for individual samples/replicates, but I can't find already calculated error statistics between replicates.

This depends on how the data was processed. Most tools will provide some sort of standard error for the effect size, which is what I assume you're asking for. Ask the vendor for the full results/relevant info.