A more reliable design for biodiversity study?

Naeem and Li 1 present the results of a microcosm study in which species diversity of organisms within trophic groups was varied. They conclude that the existence of multiple species within these groups enhanced the “reliability” of these systems, that is, the increased likelihood of a consistent level of performance over a given unit of time. But there are problems with their study. For the least diverse communities, one predator species was randomly selected from a selection of two, one autotroph from a selection of three, one consumer of bacteria from a selection of five, and one omnivore from a selection of six. Meanwhile, with the most diverse communities, both predator species and all three autotroph species were used, three consumer bacteria were chosen from the five and three omnivores from the six.