This is certainly a curious one.
If we look at a couple of equations for % Gradient.
Climb gradient = 100% x ( (Thrust – Drag) / Weight)………Equation 1.
Climb Gradient = Approximately 100% x ( ROC / TAS)……Equation 2.
Strictly speaking it should be (ROC/Ground Speed) in the equation 2, but for small angles in still air the Ground Speed is approximately equal to TAS.
Both of the above equations are equally valid, so it must be possible to use both equations to explain the reason for the reduction in gradient shown by FE Hoppies post.
Looking at equation 2, if the TAS increases due to increased OAT, while the ROC remains constant, then the % gradient will indeed decrease.
Looking at equation 1, the increasing OAT will not change the weight, so if the flat rating keeps the thrust constant, the % gradient will decrease only if the drag increases.
One possible explanation might be that the increasing OAT causes the KIAS value of V2 to decrease. V2 is less than Vmd, so this decrease will increase the drag. This increased drag at constant thrust would cause a direct reduction is the % gradient in equation 1. It would also reduce the ROC, which would in turn cause a reduction in % gradient in equation 2.
I’m not stating that the above explanation is a fact. I am simply musing over the matter.