While we do not typically consider the composite effects between wood members and the diaphragm for strength based load sharing, with the single exception of SPDWS section 3.1.1.1, I generally consider it a rational approach to use the 2.5ft area for consideration of point loads. The best way to rationalize this is to apply the entire load to a single rafter, and compare the local deflection to the adjacent member. If these members are spaced at 16" o.c., or even 24" o.c., you'll see the that the radius of curvature of the sheathing/diaphragm material overtop the members starts to grow, and as such, you'd start to expect load sharing to alleviate the deflected shape, otherwise, our assumption regarding the performance related to the stiffness of the structural elements would be invalid.
Basically, if you cause one member to deflect, but the adjacent one does not deflect similarly, there must be essentially zero stiffness associated with the members tying them together (diaphragm), otherwise, the load would redistribute.