A field effect transistor's drain and source are theoretically interchangeable, however in practise, MOSFETs always have a parasitic body diode, which is always connected cathode to source in a P channel device or cathode to drain for an N-channel device.
In this circuit, the drain is acting as the source and the source as the drain. When the power is connected up in the correct polarity, the MOSFET turns on, bypassing the internal body diode, dropping a much lower voltage, than the usual 0.6V to 1V we'd normally expect from a silicon diode. When the polarity is reversed, the MOSFET turns off and the diode is also reverse biased, so blocks the current. If the MOSFET were connected up the other way, it would not block the current, when the polarity is reversed.