Adding OpenMP reduction clause to loop inside a function

I have a function that contains an OpenMP-parallelized for loop, which calls a callback at each iteration, similar to this:

template<class Callback>
void iterate(const Callback& callback, bool parallel) {
    #pragma omp parallel for if(parallel)
    for(int i = 0; i < 1000; ++i) {
        callback(i);
    }
}

The caller can control whether the loop is parallelized using the parallel argument. Is there a way to allow the caller to specify variables on which reduction should be done, like this:

int counter = 0;
#pragma omp ??? reduction(+ : counter)
iterate([&counter](int i) {
    counter++;
}, true);

One way seems to be to put #pragma omp parallel outside the function at the call site, and #pragma omp for inside the function. But this restricts how the function can be used, and seems to be non-standard.

Is there way to do this using nested parallel constructs, the loop contruct with its bind clause, or possibly another openmp feature?