Hah, I was really confused for a second when I thought Muon's national lab map signature was the third map for the thread
Just as a reminder, this survey counts people in the county they attend religious services, not the county where they live. That explains a bit of the randomness in the first chart, especially for states with smaller counties where plenty of people cross the county line to go to church.
I kind of wish the Census Bureau asked about religious affiliation just for the numbers, maps, and trends but it's illegal:
http://www.census.gov/prod/www/religion.htm