I'm not entirely sure if it's the physical aperture or a "reported" aperture. The iris in the lens is not automatically smaller when you zoom in but less light is transmitted through the lens. Now this design is the same for all macros, however as you zoom into 1:1 my Micro-Nikkor 105mm f/2.8 will report apertures from f/5.6 to f/56 or thereabouts, but the Vivtar Series 1 105mm f/2 reports f/2-f/22 regardless of zoom.
The camera responds accordingly with the correct shutter due to recording light but I think it effects the "3D Matrix Metering" algorithms somehow. This would be similar to an 70-200 with a Nikon 1.7x TC reporting f/5.6 as the max aperture, whereas with a Kenko Pro 300 1.7x TC reports f/2.8 despite the light actually being 2 stops lower.
What I'm saying is I don't think the image was "really" shot at f/45