The nitrogen balance of three long-term agroecosystems on a boreal soil in western Canada

Nitrogen (N) budgets can be used to quantify the flows of N in agroecosystems and to account for differences in losses and retention of N. The objective of our study was to develop 24-year N budgets for three diverse cropping systems on a boreal soil at Breton, Alberta, Canada: AER - an agroecological 8-year rotation, with N inputs from legumes [fababean (Vicia faba L.), red clover (Trifolium pratense L.), alfalfa (Medicago sativa L.)] and manure; CF - a continuous perennial grass-legume forage system, with N inputs from fertilizer (18 kg N ha(-1) yr(-1)) and white clover (Trifolium repens L.); and CG - a continuous annual grain system, with N fertilizer (90 kg N ha(-1) yr(-1)). We were able to compile detailed N budgets, demonstrate accumulation of soil N, and attribute differences in N flow and permanence to treatment effects. For AER and CG, net inputs almost exactly matched gains in soil N. The AER system had the highest N flow and the largest net N accumulation. Soil total N mass to 30 cm depth increased in all systems during 1980-2005, but increases were smaller in CG (0.59 Mg N ha(-1)) than in AER (1.90 Mg N ha(-1)) and CF (1.63 Mg N ha(-1)), showing the effect of legumes, perennial species, and manure in the latter systems. The proportion of total N inputs retained as soil N with organic N inputs in AER (44%) was about twice that with synthetic N fertilizer in CG (23%). The CF system had the lowest productivity and the least N loss to the environment (4 kg N ha(-1) yr(-1), compared to 28 for AER and 24 for CG). The proportion of N inputs lost to the environment was 16% for AER and 24% for CG. In CF, gains of soil N exceeded apparent net N inputs, perhaps because we under-estimated N inputs from clover. Estimate of legume N input was one of the larger sources of uncertainty. The study affirmed the value of N budgets in evaluating agroecosystem performance, and identified AER and CF as productive and sustainable systems due to their minimal reliance on external N inputs and small N losses to the environment. (C) 2008 Elsevier B.V. All rights reserved.