How many species would we need? I don't think that question has an answer, because of the somewhat vague definition of what a species is. Culex pipiens, restuans, and quinquifasciatus are very similar mosquito species that readily hybridize to form completely viable offspring; would you need *all three of them*? You might; these are important disease vectors, both human (Saint Louis and Japanese Encephalitis) and animal (dog heartworm), and its quite possible that some genetic populations don't spread certain diseases nearly as well.
And why would you need that? It turns out that pathogens and disease vectors might play an important part in maintaining ecosystem diversity. Hantavirus is common in rodents for example. Differences in hantavirus strains might prevent one population of rodents from taking over the range of another. In effect by co-evolving with a pathogen, a population can use it as a natural defense. This diversity in turn contributes to the resilience of the overall population to environmental changes.
Suppose meadow vole populations A and B live next to each other, but invade the other's territory. There is an environmental change that wipes out one of them, say B. Then A is free to spread into B's territory, and overall the population of voles looks pretty much the same. But if B had previously overrun A's territory, then the voles would have been wiped out.
The operation of the biosphere is immensely complex. The more you know about it, the less plausible things like terraforming seem. I think it may be possible to create a self-sustaining generation ship, particularly if the enclosure is very large and energy is essentially limitless. But I think such an environment would be a dead end. I don't think it would be possible to bootstrap anything like the Earth's biosphere on another planet, at least not for millions of years.