Frequently the control circuit lacks the power to energize a relay coil directly. It cannot source (or sink) much current at the output.
A transistor needs only a tiny current to turn it on. And it can sink or source sufficient current to energize a relay coil.
We tend to use NPN type because its method of operation is a little more intuitive. Also NPN's were easier to manufacture and had better performance specs in the earlier years of solid state technology.
Generally you substitute PNP's by flipping a circuit head over heels, and reversing diode directions, while keeping positive supply at top. Then you drop in PNP's where there are NPN's.
PNP type operation will be more certain if the bias sees low impedance to the positive supply. You turn it on by extracting current through the bias terminal. Putting the load in the emitter leg of a PNP may interfere with it being sufficiently biased. (In PNP nomenclature emitter means emitter of current holes.)