I've often found these gradient effects from the moon, light pollution etc. can be much worse if there is cloud about. Pat, are any of your subs affected by cloud? In which case reprocess without them, and/or use "median" rather than "average" stacking.
All is not lost with your data. There are a few ways you can process out even this much gradient. Have a look at
http://www.youtube.com/watch?v=yTEVMH_WE80