An “insightful analysis” for the longly titled Which species? What kind of diversity? Which ecosystem function? Some problems in studies of relations between biodiversity and ecosystem function, Bengtsson, J. 1998. Applied Soil Ecology 10: 191-199.
Bell ringers: The sentences which most excited me were
1. “Diversity of functional groups, diversity within functional groups vs. total diversity”
(p.196) Despite the author’s claim that diversity is not a mechanistic driver of ecosystem function, it seems clear that we will identify any real mechanisms linking ecosystem function to gritty biology through the persistence of statistical correlations between units of stuff in ecosystems and the outcomes of those ecosystems. Divvying up diversity into inter- and intra-functional groups measures seems like a powerful step in finding the most suggestive statistical correlations.
2. “It is difficult to predict which species will be of importance in the future.”
(p. 197) I gather there is a good literature on this “natural insurance capital” theory, which is an exciting idea. Clearly there is a Gleason:Clementisan evolutionary component to the question of whether any given diversity of groups/species is best suited to the likely perturbations of their locale. It seems to contradict the author’s claim that “there is no mechanistic relationship between diversity and ecosystem function”. Perhaps not in the immediate term, but given the convincing argument for a consideration of time-dynamic processes, all ecosystem functions may be dependent on a future-proof “smart diversity”.
Mechanism and correlation: The author is clearly not a disciple of R.H. Peters and his “Critique of Ecology”, which advocates an abandon of mechanistic “narrative descriptions” (which Peters claims can’t predict outcomes or definitively answer questions). Rather, the author suggests correlation is a lesser kind of knowledge and that mechanism is the goal of real beef-eating scientists. I agree with him, but wonder if he’s forgotten that we get there through data, and if we pre-judge our data based on the existing canon of identified mechanisms, we may miss out on new candidates. This is especially important in an emerging field, where there may not be consensus around relevant mechanism. A bunch of possible ecosystem functions are listed, and there is an implication that those functions plus some other stuff that we also know are a good approximation of what ecosystems do. My intuitive response is that ecosystems are awfully complicated and our understanding of how they work is yet basic. I fully agree with the author that we’ve been way over-focused on divvying them up into units of species, but I’m skeptical that we now know how to best aggregate them.
Experiments and data aggregation: The kinds of experiments the author advocates for testing mechanism are awfully compelling (and perhaps I should more carefully read the ecotron paper now). They would be tough though. Time-dynamic-analysis, controlling for biomass, in real ecosystems when possible, is a high bar. Perhaps rather than insisting on defining “functional groups in consistent ways” a priori, we should be working on measuring our data at the least-aggregated level, and providing it in standardized formats into open repositories which would allow us to take on such cool-but-daunting studies in the “big science” format increasingly popular with the bioinformatics/molecular genetics crowd.