From my understanding of that abstract, it's the _pre-agricultural_ erosion rates are lower than assumed, so the tolerance value for agricultural land has been set too high.

From the conclusion:

> We find that pre-agricultural erosion rates in the midwestern United States are on the order of 0.0001–0.1 mm yr–1. Soil loss tolerance values of 1 mm yr–1 are one to four orders of magnitude higher than pre-agricultural erosion rates. Similarly, the agricultural erosion rates are 10–1000 times greater than the preagricultural erosion rates. Our results indicate that tolerable soil erosion, as currently defined, will lead to the depletion of midwestern soils.

It seems like this rather indicates that the USDA erosion tolerance is too high to meaningfully slow the rate of erosion. 1mm is not a useful target given that the average pre-agricultural rate of erosion is 0.04mm--this paper says that the pre-agricultural erosion rate was much lower than expected and this means modern practices are still causing more damage than previously understood and tolerance should be lowered by an order of magnitude.

Yes, lower than USDA limit. Except it turns out USDA limit is 25x higher than pre-agriculture erosion rates (which I guess no one had actually measured before).

Aren't those the "pre-agricultural erosion" rates that are lower than the current rates?