Scott2016, pp. 264–265
for loops do not allow arbitrary modification of the counter variable. It also requires that iteration bounds are computed only once. That means that compiler can generate a code to pre-computes the number of iteration and then executes iteration check faster. (Decrement–compare to 0–jump can be performed with a single instruction on many processors.)
On the other hand, C's for loop is very generic and it is required to execute the boundary check on every iteration. C compiler needs to do more analysis to perform the same optimization that is free in Fortran. That might be another reason why Fortran used to generate faster code than C.