Field Measurement of Denitrification: II. Mass Balance and Sampling Uncertainty1

Denitrification determined from the total production of volatile denitrification products was compared with denitrification determined from the difference of fertilizer addition, plant uptake, residual soil N, and leaching. Field plots on Yolo loam (Typic Xerorthents) were instrumented with soil atmosphere samplers, soil solution samplers, tensiometers, and thermocouples. Plots were maintained at two different water contents, at three different carbon levels (cropped, uncropped, and manure addition), and at two different temperatures. Fertilizer was applied at the rate of 300 kg N ha⁻¹ as KNO₃ enriched with ¹⁵N. The flux of gases at the soil surface was measured from the accumulation of N₂O and ¹⁵N₂ beneath air-tight covers placed over the soil surface for 1 or 2 hours per day. Fertilizer N in plant, soil, and soil solution samples was determined from total N and ¹⁵N analyses. Soil ¹⁵NO₃⁻ vs. time and depth and soil-water flux at several samplers were used to determine leaching of fertilizer N. Denitrification measured directly was generally 0 to 30 kg N ha⁻¹ less than that determined by difference for the wetter treatments. For the drier treatments, denitrification measured directly was 12 to 65 kg N ha⁻¹ less than that determined by difference due to the inability to measure very small gas fluxes of ¹⁵N₂ over relatively long time periods and inability to continuously monitor gas fluxes. Uncertainties (95% confidence intervals) associated with the difference method were primarily due to the large sampling variability in leaching (±20–110% of applied N), although variability in soil residual NO₃⁻ (±5–30% of applied N) and residual organic N (±0–14% of applied N) measurements were also large. The uncertainty in the leaching component was also increased due to systematic error associated with overestimating the solute flux because of anion exclusion or immobile water. Although uncertainty in the direct method due to sampling variability (±0–30% of applied N) was less than that of the difference method, systematic errors may result in an underestimation of denitrification for some situations.