From Robert Walker’s well-considered Science 2.0 article explaining why terraforming Mars is a far more fraught operation than merely building working Biospheres, which themselves aren’t easy assignments:

“Our only attempt at making a closed Earth-like ecosystem so far on Earth, in Biosphere 2, failed. There, it was because of an interaction of a chemical reaction with the concrete in the building, which indirectly removed oxygen from the habitat. Nobody predicted this and it was only detected after the experiment was over. The idea itself doesn’t seem to be fundamentally flawed, it was just a mistake of detail.

In the future perhaps we will try a Biosphere 3 or 4, and eventually get it right. When we build self-enclosed settlements in space such as the Stanford Torus, they will surely go wrong too from time to time in the early stages. But again, you can purge poisonous gases from the atmosphere, and replenish its oxygen. In the worst case, you can evacuate the colonists from the space settlement, vent all the atmosphere, sterilize the soil, and start again.

It is a similar situation with Mars, there are many interactions that could go wrong, and we are sure to make a few mistakes to start with. The difference is, if you make a mistake when you terraform a planet, it is likely that you can’t ‘turn back the clock’ and undo your mistakes.

With Mars, we can’t try again with a Mars 2, Mars 3, Mars 4 etc. until we get it right.”