First of all, we need better engines. We need better engines and better probes to be able to investigate candidate systems up close. We won't send humans before we have thoroughly evaluated the environment and ensured it's, if not safe, habitable. It's OK if they take centuries to arrive, but it would be really great if they didn't. They could also be accelerated and decelerated at much more than 1G, something we would prefer not to do with a live payload.<p>Second, we need better engines, better robots and better manufacturing technology. Once the right planet is found, we'll send supplies and robots that can build the colony mostly from local resources. Hopefully, all that can withstand much greater accelerations than humans would.<p>Only then we can send humans, with human-rated engines (we can't do much more than 1G for any significant part of the trip) and a complete habitat that can sustain the population for the duration of the trip (as seen from the passengers - relativistics may apply, depending on our progress with the all-important engine thing)<p>So, before we have better engines (and get rid of the bureaucratic problems with nuclear reactors in space), it's really a waste of time to think about how large the crew should be. We can have that data if we build marginally better engines (yes, they are key) and colonize other planetary surfaces and asteroids. Eventually, some populations may opt to isolate themselves.<p>Also, don't forget we may end up learning a lot about genetics well before we can build a relativistic engine. It's perfectly reasonable to imagine we could reintroduce (or introduce, or remove) any genetic trait we want in the population at any time we need to.