Welcome, fellow Dr. Tatiana fans!
Olivia Judson's latest column includes a good summary of work in my lab on cooperation between soybean plants and the rhizobium bacteria that (typically) provide them with nitrogen. As she points out, "cheating" is less likely to evolve in symbiont populations if they are transmitted in eggs or seeds, relative to symbionts that are acquired from the environment. In the former, if the host dies before reproducing, the symbiont dies, too. Symbionts without brains (bacteria, say) can't anticipate the effects of their actions; it's just that those whose genetically programmed behavior increases host survival become more common over generations.
Similarly, low symbiont diversity within an individual host may favor symbiont investment in costly activities that benefit the host. If each host has many different symbionts, on the other hand, then helping the host indirectly benefits competing symbionts sharing that host.
Rhizobium bacteria reach new host plants through soil, not via seeds, and they can do so even if the host dies without reproducing. Furthermore, each individual plant has multiple strains of rhizobia, which should undermine cooperation. Why then, do most rhizobia use their limited energy supply to fix nitrogen, giving most of it to the host plant? Why not use that energy for their own reproduction, instead?