Nitrous Oxide Emissions from the Application of Fertilizers: Source Partitioning

Emissions of N₂0 from soils is the result of complex interactions among agronomic practices, N fertilization regime (i.e. source, rate, timing, and place of N fertilizer application), soil physical and chemical properties, and weather conditions. The results of a given field study are therefore determined by site-specific conditions during the experiment, and their extrapolation in time and space is difficult. Meta-analysis is a statistical tool that has been widely used to analyze and integrate larger sets of research findings (Valkama et al., 2009; de Graaff et al. 2006; Van Groenigen et al. 2006; Hungate et al. 2010; Chivenge et al. 2011).

Current contact info:
Dr. Johan Six, Professor in Sustainable Agroecosystems
Department of Environmental System Science
ETH-Zurich (Swiss Federal Institute of Technology Zurich)
TAN F4, Tannenstrasse 1
8092 Zurich, Switzerland
Email:Jsix@ethz.ch
Phone: +41 44 63 2 84 83

Dr. Charlotte Decock, Postdoctoral Scholar
Department of Environmental System Science
ETH-Zurich
TAN F4, Tannenstrasse 1
8092 Zurich, Switzerland
Email: charlotte.decock@usys.ethz.ch



Justification

Emissions of N₂0 from soils is the result of complex interactions among agronomic practices, N fertilization regime (i.e. source, rate, timing, and place of N fertilizer application), soil physical and chemical properties, and weather conditions. The results of a given field study are therefore determined by site-specific conditions during the experiment, and their extrapolation in time and space is difficult. Meta-analysis is a statistical tool that has been widely used to analyze and integrate larger sets of research findings (Valkama et al., 2009; de Graaff et al. 2006; Van Groenigen et al. 2006; Hungate et al. 2010; Chivenge et al. 2011).

Meta-analysis is a set of statistical procedures aimed at synthesizing data reported from multiple studies to calculate summary effects and to assess the dispersion of effects (Borenstein et al., 2009). As with the application of general linear models as in Stehfest and Bouwman (2006), such methods can isolate effects of single variables by accounting for the effects of the many others that influence the observed emission in any given study. Addressing these multiple interactions requires a systematic database of as many observations as possible.


Objectives

To perform a meta-analysis on N₂0 emissions from fertilized agricultural soils under soybean-corn or continuous corn cropping systems in the U.S., with a focus on a subset of Corn Belt results reported in the scientific literature.


Methodology

Meta-analysis requires the following steps:

1- Determination of selection criteria for including results of a given study

1. a. Peer-reviewed; or reviewed and considered acceptable by at least two other research scientists who have published papers on N20 emission.
2. b. Duration sufficient to capture the growing-season emissions and preferably the full year.
3. c. Adequate methodology as per GRACENET.
4. d. Two groups can be formed:
1. Studies with more than one rate of N applied, from which a fertilizer-induced emission can be calculated from difference in emission divided by difference in rate. Units for analysis would include emissions per unit of N applied, and emissions per unit of yield response to N.
2. Studies in which N fertilizer was applied and reported. These would include the foregoing, along with many more studies of tillage, crop rotation, etc. Units of analysis would include emissions per unit area, and emissions per unit of crop yield.

2- Develop an EndNote database with details of all above studies.

3- Develop a spreadsheet database containing values for the following list of dependent and independent variables. Note that for Group i, variables are repeated for control N rate and control dependent variables to allow calculation per unit of applied N.

Dependent variables

1. Mean sum of measured emissions (kg N20-N/ha) over the whole year OR over the growing season.
2. Standard deviation of sum of measured emissions (kg N20-N/ha)
3. Crop yield (kg/ha)
4. Standard deviation of crop yield (kg/ha)

Independent variables

1. N source (class variable)
2. N rate (kg N/ha)
3. N timing (class variable- fall, spring, sidedress, split)
4. N placement (class variable- surface, incorporated, band, broadcast, etc.)
5. Total above-ground plant N uptake (where data are available)
6. Study# (linked to EndNote database)
7. Year (of growing season)
8. State/province
9. Site name
10. Latitude (degrees to 4 decimal places)
11. Longitude (degrees to 4 decimal places)
12. Crop type/rotation (species and product, e.g. corn for grain)
13. Tillage practice
14. Previous crop
15. Mean annual precipitation
16. Mean annual temperature
17. Aridity Index based on Holdridge
18. Soil classification down to series level
19. Soil texture class
20. Soil % sand
21. Soil % clay
22. Soil pH
23. Soil organic matter/soil C
24. Mean air temperature during measurement period.
25. Sum of precipitation during measurement period.
26. Sum of precipitation for 30 days following fertilizer application.
27. Emissions measurement start date
28. Emissions measurement end date

4- Statistical analysis of the dataset