The Performance Trends software has a validated 1968 stock 340 example engine so I ran that and compared it to your data.... It matches up nicely at 2500 then it makes no sense. Cam timing is not going to give you that sort of a curve... personally I would start by seriously questioning the dyno and test itself unless you can find something seriously wrong with the engine (like it has lost a lobe on the cam or something like that)....
Thanks for the graph. Shows just how bad things are. I sit here staring at it trying to think what could be so wrong. The engine sounds good and strong during pulls. In fact I asked about the dyno possibly as the cause. That did not get me anywhere.