(I don't know what the Amazonian frog eats; my point is that biotopes are complex and second- and third-order effects need to be monitored. That's why smaller interventions - e.g. making humans poisonous to mosquitoes, pushing them to the evolutionary path of avoiding humans - would be better than whole eradication.)