Nitrous oxide emission from intensive grassland in Canterbury, New Zealand

Abstract

The work presented in this thesis was carried out in three phases: a study phase, an observation phase and an application phase. In the first phase, estimates of nitrous oxide emissions from New Zealand were made with two methods. The first method used available N₂O flux data from earlier studies and related them to ecosystems in New Zealand while the second method used regression equations to calculate the amount of N₂O -N lost from nitrogen inputs into these ecosystems. The mean emissions estimated using the two methods were 71,000 and 79,000 tonnes N₂O -N year⁻¹ respectively for New Zealand with uncertainties of 30 to 40 %.

Two field experiments were carried out during the observation phase. Experiment one (December 15, 1992 to December 22, 1993) focused on the quantification of nitrous oxide emissions and the factors driving the gas flux from urine-unaffected, urine-treated and ploughed young (<5 yrs.) intensive grassland. The treatments were simulated autumn and spring ploughing events and the simulation of single sheep urination events in summer, autumn, winter and spring using synthetic urine at a rate of 500 kg N ha⁻¹.

A soil cover technique was used to collect gas samples in the field. Daily nitrous oxide losses, averaged over a year for the urine-unaffected background plots, the spring and autumn ploughing events and the summer, autumn, winter and spring urine applications were 0.7, 1.5, 1.3, 1.2, 5.1, 2.7 and 3.7 g N₂O-N ha⁻¹ day⁻¹ respectively. Compared to the background emissions, the ploughing events resulted in doubling of the nitrous oxide emissions, with the urine-affected areas emitting approximately 1.5 to 7 times more nitrous oxide. Nitrogen lost as N₂O -N from the applied nitrogen in the urine patches amounted to 0.08, 0.37, 0.20 and 0.27% for the summer, autumn, winter and spring applications respectively with uncertainties (CV) of approximately 10 to 38%.

Dinitrogen fluxes were measured by the ¹⁵N method (the synthetic urine was labelled with ¹⁵N at an enrichment of 50 atm%) but ¹⁵N labelled gas fluxes were below the detection limit (approx. 200 g N₂-N ha⁻¹ day⁻¹) most of the time.

In addition to gas flux measurements, soil moisture content, soil temperature, soil inorganic nitrogen (NO₃⁻ and NH₄⁺), pH, conductivity and various micrometeorological observations were performed. It was observed that nitrous oxide emissions were mainly driven by a combination of soil water suction, soil temperature and soil inorganic nitrogen with 80 % of the total nitrous oxide from the four urine applications being lost when the soil water suction was below 150 cm (cm of water).

During a second field experiment (April 1994) the relative proportions of nitrous oxide emitted by nitrification and denitrification as well as dinitrogen from denitrification were identified by selectively inhibiting the nitrification pathway and the last reduction step of the denitrification sequence. This determination was carried out by the in-field incubation of soil cores in the presence of 0 Pa, 5 Pa and 10 kPa acetylene. Results were combined with nitrous oxide measurements made concomitantly using the soil cover technique. Observations were made from urine affected young and old ( > 20 yrs.) intensive grassland each at three soil moisture levels (<55 cm, 55 - 150 cm and > 150 cm suction). Significant differences were observed among the soil moisture treatments but not between the two pastures.

For each -soil moisture treatment, the fractions of the total nitrous oxide emitted by either nitrification or denitrification were regressed against the fraction of the total mineral-N present as either NH₄⁺-N or NO₃⁻-N. The resulting linear relationships were used to calculate nitrous oxide fluxes from both nitrification and denitrification for the data set of experiment one.

From the total nitrous oxide emitted by the four urine applications in experiment one, an average of 30 % (range 13 % (summer) to 57 % (spring)) was calculated to be a result of nitrification.

Similar relationships were developed for dinitrogen via denitrification. It was calculated that the N₂ flux during experiment one was on average 4.3 times higher (range 0.7 (winter) to 16.5 (summer)) than the nitrous oxide emissions. Comparisons with the ¹⁵N method showed that results calculated from the indirect acetylene method and the direct ¹⁵N method agreed well.

Measurements of nitrous oxide emissions at several times throughout the days of observation during experiment two provided information about the diurnal characteristic of nitrous oxide fluxes.

In the application phase of the work in this thesis, two mechanistic models were developed. From the findings of experiment two a model was developed for the diurnal characteristic of nitrous oxide fluxes based on the Arrhenius function. It was found that the diurnal characteristic of nitrous oxide emitted from urine affected grassland is consistent with temperature induced activity changes of microorganisms in the top 5 cm of soil.

Michaelis-Menten kinetics were applied in a second model using results from both experiments to calculate seasonal and annual nitrous oxide fluxes. The model is based on the determination of the Michaelis-Menten parameters, Ym and α and has soil water suction, soil temperature, soil inorganic nitrogen (NH₄⁺ and NO₃⁻) as well as rainfall as input parameters. In order to obtain Ym and α parameters the data set of experiment one (approximately 400 data points) was divided into two kinetic categories and further subdivided into temperature-suction classes. Unique parameters for each class were determined for each observation day during experiment one in order to calculate nitrous oxide fluxes from both nitrification and denitrification.

The total modelled nitrous oxide flux was compared with the measured nitrous oxide using several validation tests, including a paired-t test and estimation of the sum of squares. It was found that the estimations by the model were within the range of measured uncertainties of nitrous oxide fluxes from the urine affected intensive grassland.

It is concluded that more extensive data sets of nitrous oxide emissions and related factors are needed from different ecosystems and climatic regions. This would improve our understanding of nitrous oxide producing processes and would lead to the development of more accurate mechanistic models as well as reducing the uncertainties currently present in the estimates of the anthropogenic nitrous oxide flux from New Zealand.