I note from the comments to the article:
Tilman et al. (Reports, 8 December 2006, p. 1598) argued that low-input high-diversity grasslands can provide a substantial proportion of global energy needs. We contend that their conclusions are not substantiated by their experimental protocol. The authors understated the management inputs required to establish prairies, extrapolated globally from site-specific results, and presented potentially misleading energy accounting.
Tilman et al. suggest that LIHD plantings could provide a sustainable source of harvestable biomass for fuel production, but they reported sample yields from an experiment in which nearly all the biomass was burned in situ, not harvested. Although several plant nutrients are lost from burned vegetation as gases or particulates, most cations are returned to the soil (2). With mechanical harvest, all nutrients are removed. Although legumes can replace nitrogen, nutrient replacement will be an important requirement for many marginal, and especially acidic, soils if yields are to be sustained. Limestone additions would be required to maintain symbiotic N2 fixation on soils with poor pH buffering capacity. Liming represents a major energy input (3, 4).
More seriously, the experimental approach of Tilman et al. is a form of double accounting with respect to carbon. The authors estimated harvestable biomass from small samples taken in late summer, then burned the remaining biomass on the plots the following spring . Combustion of this sort is incomplete, so some, if not most, of the soil C sequestration they measured is almost certainly due to charcoal additions that would not have occurred with harvest for biofuel production. Burning also has multiple, and often unpredictable, effects on prairie plant ecology. In general, burning reduces the presence of woody species in mixed stands, as the authors observed (1), but also helps control other undesirable species and may increase root biomass, tillering, soil temperature, and nitrification (2). With the exception of the decline in woody species, these benefits would not accrue with mechanical harvest of herbaceous perennials.
Now, Tilman et al put up a robust defense to the criticism, but the debate indicates that Tilman's conclusions are still a ways from entering the canon.
My concern is that the careful management required to maintain his rosy picture is quite likely to go out the window when oil supplies are down to 40 mbpd. As I said in other posts, anytime we bring another biome under human exploitation it ends up degraded. I have no reason to expect our behaviour will be any more exemplary in this case, so I will retain my hard-line skepticism until we have operational proof that my concerns are unfounded.