One problem with really giving a 'correct' answer to your question is that the techniques you mention aren't necessarily distinct. You could have a subtractive synthesizer that uses wavetable oscillators; An additive synth might also use wavetables for efficicency; an FM synth with many operators can in theory do everything an Additive synth can do, and so on.
There's also the issue that what you find in real-world products isn't necessarily a representation of what's possible in theory. This is particularly true of additive synthesis - it's a very powerful technique that can in theory produce any sound, but it can take a lot of CPU power, so the 'additive synthesizers' you can buy tend to be quite limited.
The bottom line is that any of these synthesis techniques can produce incredibly complex sounds - but what sounds you can actually produce on a particular bit of equipment depends on the specific facilities available on that equipment. Many of the most important facilities for creating complex sounds are the facilities available for modulation (such as assigning LFOs to various parameters), and the effects processing available. Many incredibly deep and complex synth patches are very thin and weedy with all the effects turned off.
An experienced synthesist could often produce a sound using one of these techniques, and then produce a passable imitation of it using one or more of the other techniques. In many cases, it's knowledge that gives you power!
If you're just starting to learn synthesis, subtractive is often popular for being quite intuitive and giving you controllable results without too much effort.
