Brian.G,
Don't mean to beat a dead horse on this topic, but I would disagree with Mr. Wolkner's explanation. While applying a given normal force over a smaller contact area would increase the interface unit pressure, it would have no real effect on axial separation of the mating faces, nor on the ability of that interface to transfer shear forces.
The same is true for serrations. The included ramp angle of the serration surfaces does indeed produce a greater surface normal force for a given axial bolt preload. But that surface normal force is also more quickly relieved when the rod loads reverse.
The only realistic approach to reducing fretting at the rod beam/cap joint is to design the joint with enough elasticity to prevent the fastener preload from being relieved under all operating conditions.