Bottom line: It appears to have no negative effect and may have a positive one.
I used the scenario builder to test 6 different sets of hill berries 5 times each. After testing the hill gathering rate, I went back to SB and removed the hills, i.e., set everything to elevation 1, and reran the tests for flat ground.
My data actually surprised me. Not only was there not a decrease in overall gathering efficiency, statistically you can say at the 60% confidence level that with the same placement, hill berries show a very minor efficiency improvement.
Representative pics follow. Click on the thumbnails for larger pics.
This is the unmodified random map with the addition of 6 peons and one granary. Here is the screenshot immediately after collecting all the berries with the hills still intact. And here is the same configuration once all the hills are made low. I may do a little more investigation, or not, depending on what interest this thread receives. Keep your stick on the ice.
Anyway, in all my tests, there was only one case where the flat berry collection beat the average time of the hill berries, and since it was the first data point I collected, I suspect it may have been a mistake on my part on getting started.