Ok, thanks Nate.
I have it implemented successfully now, with the box2d bodies positioned at each skeleton region and animated along with them. My original issue was simply that I was trying to verify one step at a time, and the bodies weren't properly rotated until the transformations in the update() method in the java example.
I do have another issue now, and that is with flipping the character horizontally. I've used
skeleton->flipX
to flip the animation. As I understand it though, there is no easy translation in box2d, as there are only transforms. Thus my bodies are messed up as soon as my character walks left.
I've read a couple suggested options, and that is to destroy and recreate the box2d bodies/shapes when changing direction, or maintain two separate body structures (one for right and one for left animations) and swap them in and out when necessary. I don't really like either and it is starting to seem I am fighting against the physics sim rather than working with it. For one, using static bodies for each region part was already less than ideal.
I think I am starting to settle on the same solution spudworks gave, and that is a much simpler dynamic body structure with minimal added fixtures/sensors for other body parts I want to interact with specifically (the head perhaps). If anyone had tips on using the more complex body structure, that would also be appreciated.
This is turning into more of a box2d specific discussion than Spine related, so I'll try to post only Spine related questions further in this thread should I encounter any.
Thanks for the help