2.3. Statistical analysis
Before addressing the effects of cropping system treatments during our study period, we evaluated the effects of land use legacy (i.e., blocks 1–3 in corn and blocks 4 and 5 in grass-alfalfa) with PERMANOVA. Separating samples by past land use accounted for $7% of the variability in the 2008 data (P = 0.04), but only 1.1% of variability in 2013 (P = 0.64), indicating these effects were relatively small at the onset of our study and had disappeared statistically 5 years later.