So it's a simulationwise limitation? I can live with it without a problem. One can fly RNAV with setting the heading bug manual or use RNAV as reference system. Just trying to understand the backgrounds.

Why does the autopilot not react then as it does with NAV or GPS?

Because each navigation unit (including the RNAV) actually carries its own autopilot. In order for one unit to take over flying, the others must relinquish control. Some units support this, while others don't. If the currently selected GPS does not play nice and separate from the autopilot cleanly (or support this feature at all), then I cannot step in with my own avionics.

With the KLN 90B installed I have that rocker switch and via the MD41 I can selecet GPS or NAV. The MD41 decides which signal is shown by the MC60 (or any other NAV/OBS indicator) and then used by the autopilot. To show the RNAV signals as a third option there's that rocker switch and the MC60 does show the RNAV signals correctly. Why does the autopilot not react then as it does with NAV or GPS?

'NOTE: The autopilot in this aircraft is capable of receiving navigation input from the RNAV511,
but will only do so when the no-GPS avionics configuration is selected from the tablet interface.
When operating without a GPS, the navigation source selector button and integrated
annunciators will read “NAV/RNAV”, instead of “NAV/GPS”. '