The effects of irrigation on carbon balance in an irrigated grazed pasture system in New Zealand

Abstract In New Zealand, increasing areas of dryland farming are being converted to irrigated farming, and there are conflicting findings whether this will lead to gains or losses of soil organic carbon (SOC). In this study, we used 2 years of eddy-covariance data from an irrigated, grazed dairy pasture in Canterbury, New Zealand and compared observed gas exchange fluxes to those from the process-based CenW model. CenW simulations agreed very well with observations for evapotranspiration, gross primary production, and ecosystem respiration rates with Nash-Sutcliffe model efficiencies ~0.8–0.9 for daily and weekly mean values. Model efficiency for net ecosystem productivity was ~0.7–0.8 for daily and weekly averages. The mean simulated net ecosystem carbon balance (NECB) over the 2 year period was 0.6 t C ha−1 y−1 which was comparable to the measured NECB of 1.0 ± 0.3 t C ha−1 y−1. The model also highlighted some processes where current understanding is incomplete. We found that gross primary production (GPP) was substantially reduced after grazing. While some GPP reduction was expected due to reduced leaf area, measured rates were only about half (reduced by a further ~30 kg C ha−1 d−1) of that expected due to reduced leaf area alone. This probably resulted from the combined effect of trampling damage by cattle, a minor contribution of foliage coverage by dung pats, and some additional leaf physiological effect when previously shaded parts of the foliage were suddenly exposed to full sun to which they were not fully adapted. GPP then gradually recovered over the subsequent 2 weeks. CenW was then used to compare SOC levels after 50 years of irrigated dairy farming with those under a non-irrigated system. We found that SOC responses to irrigation interacted strongly with imposed grazing patterns. When we used the same grazing regime under irrigated and non-irrigated conditions, the application of irrigation resulted in a small increase of 4 ± 3 t C ha−1 (mean ± standard error) over the range of weather conditions typical for this site. However, with this grazing management, the non-irrigated system had very low productivity (