That output leads to some thoughts...

1. 4.5 million voxels is a lot, you are likely oversampling
the data (which should not affect FDR correction).

2. It might be helpful to include your actual commands,
even for 3dhistog.

3. Unless the data is quite high resolution, it does not
look like you provided that gray matter mask to the
3dhistog command. What was your command?

4. The fraction of voxels at high correlation values is not
so large. Maybe the correlation is not so high. Does it
actually look like you should have significance based on
the thresholded data view in afni?

5. The reason there is no correlation of 1.0 is presumably
because your seed is an ROI or radius average. Again,
your commands would help clarify the applied steps.

- rick