Andy McGuire, who earned an MS with me some years ago, doing research on cover crops, has just posted a provacative essay titled "Don't Mimic Nature on the Farm, Improve It." He contrasts a well-known agroecologist saying agriculture should "mimic nature" with statements from natural-ecosystem ecologists (and my book) denying the perfection of natural ecosystems and the "balance of nature" hypothesis. He concludes that:
"We can, with ingenuity, wisdom, and a good dose of humility, purposefully assemble systems that outperform natural ecosystems in providing both products and ecosystem services."
I agree, but with some reservations. If the overall organization of natural ecosystems isn't necessarily perfect -- see this discussion -- then it should be possible for us to improve on it, at least by agricultural criteria.
Let's consider a specific example: a pasture grazed by dairy cattle. We want to maximize milk production, subject to various constraints that include long-term sustainability and minimizing pollution. Our reference natural ecosystem is the prairie, grazed by bison, that once occupied the same land. The natural sex ratio of bison is 50:50, like other mammals, but we can get more milk from the same land with a female-biased sex ratio. Similarly, the ratio of nitrogen-fixing legumes to grasses in a natural prairie depends on their relative survival and reproduction, not on how much nitrogen the ecosystem needs for maximum productivitiy. Given the economic and environmental costs of nitrogen fertilizer, we might want to increase the abundance of legumes in our pasture, relative to the natural prairie.
But how? What combination of McGuire's "ingenuity, wisdom, and... humility" will lead to increased legume abundance with the fewest negative side effects? And are we even sure increasing legume abundance is a good idea?
Planting additional legume seed each year might cause enough soil disturbance to increase erosion. Low doses of a grass-specific herbicide would increase costs and perhaps pollution. Ingenuity might suggest introducing a mild pathogen that would slow grass growth without killing it. Humility, though, would identify some of the risks with that approach (for example, the pathogen might evolve greater virulence) and the possiblity of additional, unrecognized risks.
I would suggest trying several approaches in limited experiments (not including the pathogen option!), then doing longer-term and larger-scale tests of those that seem more promising. These tests may find problems that weren't apparent in short-term, small-scale experiments. (Similarly, we need more long-term monitoring of transgenic crops once they're in widespread commercial use. For example, an herbicide-resistant weed mutant is much more likely to arise on millions of acres than on one acre.)
But natural-ecosystem ecologists could play an important role also. For example, it might help us to know what combination of factors limits legume abundance in the natural prairie. If the preferences of grazing animals are key, can we enhance legume survival through grazing management? I don't mean to suggest that agronomists would never think of this without information from natural ecosystems, but comparisons among systems can often reveal patterns that aren't obvious in a single system.
Also, legumes and grasses have different soil-resource requirements. In particular, phosphorus fertilization can favor legumes more than grasses, although the long-term availability of phosphorus fertilizers is a concern. Differences in resource requirements among species have sometimes been proposed to explain the high species diversity of some natural ecosystems. On the other hand, a recent paper in Nature showed that differences in resource requirements among tropical tree species aren't enough to prevent loss of diversity during the seedling stage, when fungicide sprays reduce the diversity-enhancing effects of more-abundant species suffering more losses to pathogens.