The "normally-closed" dual of a "normally-open" enhancement-mode MOSFET is, as you noted, a depletion-mode MOSFET. But depletion-mode MOSFETs are pretty uncommon (I've never needed one) and I don't know of any that can run at a continuous 16 A.
Though depletion-mode MOSFETs do have their uses, enhancement mode does fine for load switching. Usually it's as simple as inverting the logic in firmware. Output 0 for on and 1 for off.
If this isn't possible for you, your real problem is that enhancement-mode MOSFET logic is inherently inverting. (Specifically, in this configuration it's a common-source amplifier, which is an inverting amplifier.) Turning on an nMOS with a logic HIGH pulls the output LOW. Turning on a pMOS with a logic LOW pulls the output HIGH.
So if you can't change the firmware logic, just add another inverting stage, as such:

simulate this circuit – Schematic created using CircuitLab
Something along those lines should do the trick. Closing the switch pulls M1's gate low, which lets R1 pull M2 high. Opening the switch lets R3 pull M1's gate high, which shorts M2's gate to ground.
Make sure you choose transistors that can withstand a \$V_{GS}\$ of more than 24 V1, and your load transistor will also need to withstand a \$V_{\text{DS,off}}\$ of more than 48 V. You won't need a beefy load transistor on the first stage, just pick something that can withstand the voltages you need. I'd also recommend a flyback diode on any inductive load being switched by a MOSFET, otherwise you'll see damaging voltage spikes.
1 The IRF1010EZ you specified in your design can only handle a \$V_{GS}\$ of ±20 V, which is typical in my experience. You will have to limit the gate-source voltage, e.g. with an extra resistor or zener diode.