Agriculture, Ecosystems and Environment 136 (2010) 310–317
Contents lists available at ScienceDirect
Agriculture, Ecosystems and Environment journal homepage: www.elsevier.com/locate/agee
A preliminary study to model the effects of a nitrification inhibitor on nitrous oxide emissions from urine-amended pasture Donna L. Giltrap a,*, Jagrati Singh a,b, Surinder Saggar a, Mohammad Zaman c a
Landcare Research Ltd., Private Bag 11052, Palmerston North, New Zealand Institute of Natural Resources, Massey University, Palmerston North, New Zealand c Summit-Quinphos NZ Ltd., New Zealand b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 7 May 2009 Received in revised form 19 August 2009 Accepted 20 August 2009 Available online 9 September 2009
New Zealand’s grazed pastures receive large quantities of nitrogen (N) inputs from animal excreta and chemical fertilisers. While N promotes pasture growth, surplus N can cause environmental problems by leaching into waterways or by nitrifying and denitrifying to form the greenhouse gas, nitrous oxide (N2O). Various approaches have been attempted to mitigate the economic and environmental impacts of N losses. One such approach is the use of nitrification inhibitors (NIs). The value of these inhibitors in mitigating N losses in grazed pasture partly depends on their rate of biodegradation and persistence in soils. A simplified model of nitrification inhibition was used with the process-based NZ-DNDC model to investigate the effect of NI dicyandiamide (DCD) on transformations of N to nitrate (NO3) and subsequent reduction to N2O in a grazed pasture system receiving cow urine at application rate of 600 kg N/ha. The modelled N2O emissions with and without DCD application were comparable to the field measurements on Tokomaru silt loam soil, assuming that the effect of the DCD was to decrease the nitrification rate by about 70%. An attempt was also made to simulate the effect of the biological degradation of DCD by exponentially decreasing the inhibitor effectiveness with time. However, this did not improve the fit of the modelled N2O emissions over the 50-day measurement period. Further refinements including the effects of soil type, and changes in NI concentration throughout the soil profile over time and its subsequent effect on N transformations will be developed as more experimental data become available. ß 2009 Elsevier B.V. All rights reserved.
Keywords: Dicyandiamide (DCD) Grazed pastures Greenhouse gas mitigation Measured emissions NZ-DNDC Urine-N
1. Introduction New Zealand’s grazed pasture systems receive N inputs from animal excreta, application of chemical fertiliser and biological N fixation of atmospheric dinitrogen (N2) by legumes. Nitrogen is essential for plant growth, but surplus N is likely to be lost during N transformation processes (nitrification/denitrification) to the atmosphere as N2O or into waterways as NO3. In New Zealand the agricultural sector accounts for 48% of total greenhouse gas emissions, about a third of which is due to direct and indirect N2O emissions from soils (MfE, 2009). New Zealand N2O emissions have increased by 23% between 1990 and 2007 (MfE 2009), mainly due to the sharp increase in the use of fertiliser-N and associated excretal-N inputs. Therefore several approaches that have the potential for reducing N2O emissions
* Corresponding author. Tel.: +64 6 353 4820; fax: +64 6 353 4801. E-mail address:
[email protected] (D.L. Giltrap). 0167-8809/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.agee.2009.08.007
from agricultural soils thereby substantially reducing New Zealand’s greenhouse gas inventory are being investigated. One proposed method for reducing N losses is the use of nitrification inhibitors (NIs) that inhibit nitrification, a microbial process in which ammonium (NH4+) is converted to NO3 (Abbasi and Adams, 2000; Cookson and Cornforth, 2002). This reduces the amount of soil NO3 available for reduction to N2O via denitrification. In addition, N in the form of NH4+ is less susceptible to leaching than NO3 and therefore remains available for plant uptake for longer. Current commonly used NIs are dicyandiamide (DCD), nitrapyrin and 3,4-dimethylpyrazole phosphate (DMPP). Of these, DCD is the most widely used NI in New Zealand as it is cheaper, less volatile and relatively soluble in water, so the majority of research has focused on DCD. At present the use of DCD on grazed pastures is not widespread in New Zealand (3.5% of the effective dairying area was estimated to be using DCD, MfE, 2009), but the use of NIs is seen as a potential mitigation option for the future. The effectiveness of an NI varies with soil type, pH and temperature. A laboratory-scale study investigating the effectiveness of the NI DCD in reducing N2O emissions, as well as its rate of
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
decay under controlled conditions in three New Zealand soils, contrasting in texture, mineralogy and organic carbon levels, was reported in Singh et al. (2008). These results showed the halflife of DCD and the amount by which DCD addition reduced the total N2O emissions from a urine patch over a 50-day period varied with soil type. In this field study, we investigated the effect of cow urine (600 kg N/ha) with DCD (7 kg/ha) applied to Tokomaru silt loam soil (Argillic-fragic Perch-gley Pallic) on N2O emissions where climate effects must also be taken into account. We then modelled the effect of the DCD on N2O emissions using the process-based NZ-DNDC model. The processes that produce N2O within agricultural soils are complex, with many interactions. Therefore, it is very difficult to develop an empirical model with sufficient flexibility to cover the full range of management practices, and soil and climate conditions found in practice and also account for the impacts of mitigation technologies. Farquharson and Baldock (2008) reviewed the processes and relationships that produce N2O and the strategies that have been used to model these emissions. Frolking et al. (1998) compared the N2O emissions predicted by four process-based models (DayCENT, DNDC, ExpertN and the NASA-Ames version of the CASA model). All four models generated similar results for the general cycling of nitrogen through agroecosystems, but the models varied in the N trace gas fluxes predicted. NZ-DNDC is a modified version of the process-based model DNDC (denitrification–decomposition), described by Li et al. (1992) that can simultaneously model agricultural trace gas emissions, NO3 leaching and crop yield. DNDC consists of five submodels that model soil thermal-hydraulic flows, plant growth, decomposition, fermentation and denitrification. Nitrous oxide is emitted both as part of the nitrification and denitrification reactions. Nitrification processes occur during aerobic soil conditions and the N2O emissions from nitrification are modelled as a function of soil temperature and NH4+ concentration. Denitrification reactions occur under anaerobic conditions and the model partitions N-substrates between the aerobic and anaerobic fractions based on the soil water-filled pore space (WFPS). DNDC models the relative growth and death rates of denitrifying bacteria and the N2O and N2 produced during denitrification are determined by the relative reaction kinetics of each stage of the denitrification sequence (NO3 ! NO2 ! N2O ! N2). The model can be run at field or regional scale and has been applied internationally to a wide range of systems in a number of countries including Australia (Kiese et al., 2005), Belgium (Beheydt et al., 2007), China (Li et al., 2004; Xu-Ri et al., 2006), Germany (Butterbach-Bahl et al., 2004; Neufeldt et al., 2006), India (Babu et al., 2006), New Zealand (Giltrap et al., 2008; Saggar et al., 2004, 2007a,b), UK (Brown et al., 2002), USA (Farahbakhshazad et al., 2008; Li et al., 1996). NZ-DNDC is a modified version of DNDC version 8.6K. The model modifications for New Zealand grazed pasture conditions are described in Saggar et al. (2004, 2007a). The changes made to NZ-DNDC include accounting for seasonal changes in pasture growth, N input from animals, and water balance/soil moisture status (Saggar et al., 2007a). The model was validated against field measurements from two dairy pastures with contrasting soils (Saggar et al., 2004), and a sheep pasture (Saggar et al., 2007b). A series of sensitivity tests was conducted using NZ-DNDC to demonstrate the model’s sensitivity to changes in climate, soil organic carbon, fertiliser management, and grazing regimes with respect to pasture production and N2O emission predictions (Saggar et al., 2007a). Some additional changes were made to NZ-DNDC for this study. The following update (Changsheng Li, personal communication)
311
was incorporated to account for change in soil pH arising from transformations between soil NH3 and NH4+:
DpH ¼
0:5 DNH3 V WFPS
(1)
where DNH3 is the change in NH3-N, V is the pore volume and pH is constrained to be between 3 and 11. In addition, we removed the clause that automatically set the anaerobic volume fraction to 0.9 when the soil was above field capacity. The aim of this study was to compare the performance of DCD in field conditions with the N2O emissions reductions found under controlled conditions, and to compare the results with a simple model of nitrification inhibitor effects within the process-based NZ-DNDC model as the first stage towards developing a more complete model of nitrification inhibitor effects. 2. Materials and methods 2.1. Experimental design The experiment was conducted on a Massey University dairy farm located near Palmerston North (408210 S, 1758390 E) New Zealand. The site was a dairy-grazed pasture comprising predominantly perennial ryegrass (Lolium perenne) and white clover (Trifolium repens) on a Tokomaru silt loam soil. This formed part of a Tech NZ-funded collaborative project (involving Landcare Research, HortResearch, AgResearch, Massey University and Summit-Quinphos NZ Ltd.) to evaluate the effect of inhibitors in mitigating N losses in grazed pastures. In this paper we examine three treatments that were replicated three times: control (water); urine; and urine + DCD. Urine was applied in May 2005 at 600 kg N/ha and DCD at 7 kg/ha in a completely randomised block design of 1 m 1.5 m plots separated by a 0.5 m buffer zone. The area was fenced off 2 months before the start of the experiment to avoid excretal-N deposition from grazing animals. A pre-conditioning harvest (20 mm height) was taken before applying the treatments. Further details about the experimental methods and results are available in Zaman et al. (2009) 2.2. N2O measurements Nitrous oxide emissions were measured for 50 days using the closed-chamber technique (Saggar et al., 2004). One chamber per plot was inserted 100 mm into the soil 1 day before measurements were made. On each sampling day, three gas samples were taken from each chamber after closing the chambers at times t0, t30, t60 (i.e., 0, 30 and 60 min, respectively). The gas samples collected were transferred to evacuated vials and then analysed using a Shimadzu GC-17A gas chromatograph equipped with a 63NiElectron capture detector (Hedley et al., 2006). Basal N2O emissions were measured 1 day before the application of treatments. Measurements of N2O emissions were taken daily for the first week to capture short-term changes in N2O fluxes. Subsequent measurements were made on alternate days for the next 2 weeks, and then twice a week in the 4th week. For the remaining period of 2 weeks, measurements were taken once a week. The N2O flux was estimated from the measurements, and cumulative fluxes were calculated as described by Saggar et al. (2004, 2007b). 2.3. Statistical analysis The total N2O emissions over the 50-day period were calculated using the trapezium method to integrate between the measured days. An ANOVA analysis using the SAS (version 8) statistical
312
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
software package was used to calculate the least significant difference (at a 5% threshold) between the three treatments.
The ‘‘goodness of fit’’ of the models to the measured results was assessed by calculating the root mean square error (RMSE). The RMSE is defined as:
2.4. Modelling
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P 2 i ðP i Oi Þ RMSE ¼ n
The input data required by NZ-DNDC are climate data (daily rainfall, maximum and minimum air temperature and solar radiation), soil properties (soil texture, organic carbon, pH, bulk density and WFPS at field capacity and wilting point) and farmmanagement data (timing and amount of fertiliser application, grazing and irrigation). Saggar et al. (2007a) found it necessary to add a ‘‘water retention layer’’ for grazed pastures to simulate the effect of soil compaction from treading on reducing water flow. However in this study there was no grazing, so no water retention layer was used. The application of urine at 600 kg N/ha was simulated in NZDNDC as a urea fertiliser application. The additional 7.5 mm water content of the urine was added to the daily rainfall. NZ-DNDC has a simple method for simulating nitrification inhibitors applied with fertiliser using two parameters: duration and effectiveness (Neff). The duration is the number of days after application that the nitrification reaction will be inhibited. Neff is a number from 0 to 1 that represents the fraction by which the nitrification rate is reduced. The inhibited rate of nitrification is given by: N i ¼ ð1 N eff ÞN 0
(4)
where Oi is the ith observed value, Pi is the model prediction for the ith observation, and n is the total number of observations. RMSE has the same units as the measured value. Table 1 shows the initial inputs used in the model simulations. The measured values for the initial soil NO3-N and NH4+-N concentrations in the control treatments were used. The initial WFPS used in the model were selected to minimise the RMSE between the measured and modelled soil WFPS (in the top 5 cm). 3. Results Fig. 1(a)–(c) shows the measured soil NH4+-N, NO3-N and N2O emissions for the three treatments: control (only water added), urine-only (applied at a rate of 600 kg N/ha), urine + DCD (urine
(2)
where Ni is the inhibited nitrification rate and N0 is the nitrification rate in the absence of an inhibitor. A similar approach has also been used by Del Grosso et al. (2009), where the impact of NIs was modelled with DAYCENT by reducing the calculated nitrification rates by 50% for 2 months after the inhibitor was applied. Multiple simulations were run using different values of Neff which were then compared to the experimental results. In reality the effectiveness of an inhibitor depends upon its concentration and the soil properties, while the inhibitor is susceptible to leaching and experiences biological decay at a rate that is temperature-dependent. As the DCD degrades exponentially, we also looked at the effect of using a Neff that exponentially decayed with time. For the exponential decay model the Neff for day t + 1 was defined as: ln 2 N tþ1 ¼ N t exp t 1=2
(3)
where t1/2 is the half-life of DCD in days. The average daily temperature during the field trial was 10.5 8C, which corresponds to t1/2 = 70 based on the relationship between DCD half-life and soil temperature developed by Kelliher et al. (2008). A more advanced model would change the decay rate in response to changes in daily temperature and model the concentration gradient of DCD throughout the soil profile.
Table 1 Model input values used in simulations. Parameter
Value
Bulk density Clay content Initial soil NH4+-N Initial soil NO3-N Initial WFPS pH Soil organic carbon at surface Soil texture Soil WFPS at field capacity Soil WFPS at wilting point
1.1 g/cm3 23% 9.1 mg N/kg soil 11.2 mg N/kg soil 41% 5.62 0.039 kg C/kg soil Silt loam 62% 28%
Fig. 1. Measured soil (a) NH4+-N concentration (0–100 mm), (b) NO3-N concentration (0–100 mm) and (c) N2O emissions for the control, urine-only, and urine + DCD treatments. Urine was added at a rate of 600 kg N/ha, DCD was added at 7 kg/ha. For the control treatment only water was added. Error bars represent the standard deviation of the three replicates.
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
313
Fig. 2. Measured and modelled soil (a) WFPS (0–50 mm), (b) NH4+-N concentration (0–100 mm), (c) NO3-N concentration (0–100 mm) and (d) N2O emissions for the control (water-only) treatment. Error bars represent the standard deviation of the three replicates.
applied at a rate of 600 kg N/ha plus DCD added at 7 kg/ha). The effect of DCD addition with urine is to slow the conversion of NH4+ to NO3 resulting in the urine + DCD treatment having higher NH4+-N concentrations, lower NO3-N concentration and lower N2O emissions relative to the urine-only treatment. Fig. 2(a)–(d) shows the measured soil WFPS, NH4+-N and NO3N concentrations and N2O emissions from the control (water-only) treatment compared to the modelled results. There was reasonable agreement between the modelled and predicted values with RMSEs of 4.4 mg NH4+-N/kg soil, 2.5 mg NO3-N/kg soil and 15 g N2O-N/ha/d for soil NH4+-N, NO3-N and N2O-N, respectively. Initially the modelled urine-only treatment produced a poor agreement with measured results for soil NO3, NH4+ and N2O emissions. In this study the majority of NH3 emissions from a urine patch during autumn application occurred on the first day following application and after 5 days the emissions were back to (or only marginally above) the levels of the control for summer and spring applications (Zaman et al., 2009). However, the model simulated only 38% of the NH3 emissions occurring within the first 5 days, suggesting that the rate of NH3 volatilisation was too slow. A urine treatment differs from a urea fertiliser application (with additional water) in several ways that might affect the NH3 volatilisation rate. First, the urine was applied at a rate similar to that of a urine patch (600 kg N/ha), which is higher than usual rate for urea fertiliser. Urine is applied as a liquid whereas urea fertiliser is applied as a solid, therefore there could be differences in the volatilisation rates and transport within the soil. Finally, urine is not purely composed of urea and contains other chemicals such as uric and hippuric acid that could affect the rates of transformation. The coefficient of the rate of transfer of NH3 from the liquid to gas
form was adjusted to minimise the RMSE of modelled soil NH4+. The measured and modelled soil NH4+ levels using the original and modified model are presented in Fig. 3. Fig. 4(a)–(c) shows the measured and modelled soil NH4+-N and NO3-N concentrations and N2O emissions, respectively. The RMSE values were still reasonable (50 mg NH4+-N/kg soil, 50 mg NO3-N/ kg soil and 37 g N2O-N/ha/d). Overall simulated N2O emissions were close to the measured emission except that the model appeared to under-predict the N2O emission peak on day 12. To improve the model capability to accurately estimate soil NH4+-N
Fig. 3. Measured soil NH4+-N concentration (0–100 mm) following a urine application of 600 kg N/ha with model prediction before and after modifying the NH3 volatilisation rate.
314
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
Fig. 5. Measured and modelled N2O emissions for the urine plus DCD treatment. For the model simulations the effectiveness of DCD as a NI (Neff) was maintained as a constant from 0 (no nitrification inhibition) to 1 (complete nitrification inhibition). The urine was applied at a rate of 600 kg N/ha and DCD applied at 7 kg/ha. Error bars represent the standard deviation of the three replicates.
1.62 kg N2O-N/ha for the control, urine-only and urine plus DCD treatments, respectively. The least significant difference (at the 5% level) was 1.01 kg N2O-N/ha, so the addition of DCD with urine resulted in a statistically significant reduction in N2O emissions over the 50-day period relative to the urine-only treatment. The corresponding NZ-DNDC predictions of total N2O emissions over this time period were 0.48 and 3.0 kg N2O-N/ha for the control and urine-only treatments. In the urine + DCD treatment, the total N2O measured over the 50-day period was 1.6 0.6 kg N/ha, representing a decrease in N2O emissions of about 50% compared to the urine-only treatment. 3.1. Constant Neff model
Fig. 4. Measured and modelled soil (a) NH4+-N concentration (0–100 mm), (b) NO3-N concentration (0–100 mm) and (c) N2O emissions for the urine-only treatment. Urine applied at a rate of 600 kg N/ha. Error bars represent the standard deviation of the three replicates.
and NO3-N levels and emissions of NH3 further data on Ntransformations under urine patches are needed. Table 2 shows the results of ANOVA analysis using the SAS (version 8) statistical software package. The total N2O emissions measured over the 50-day experiments were 0.81, 3.37 and
Fig. 5 shows the measured N2O emissions compared with the NZ-DNDC predictions, using a range of constant values for Neff for the urine + DCD treatment. Note that even with no nitrification occurring (Neff = 1), there were still some N2O emissions from denitrification of existing soil NO3. Table 3 shows the total N2O emissions predicted as Neff varied. The measured N2O emission (both as an absolute figure and relative to the urine-only measurement) lies between the values for Neff = 0.6 and 0.8. This is also where the minimum model RMSE occurred. This suggests that DCD in Tokomaru silt loam soil slowed nitrification by 60–80%. Note that Neff represents the reduction in nitrification rate rather than the reduction in N2O emissions. In this study, the reduction in N2O emissions was less than the reduction in nitrification rate. This is because denitrification is a significant source of N2O emissions and NO3 already in the soil at the time of DCD application is available for denitrification as can be seen from the model simulation with Neff = 1.
Table 3 The effect of Neff on NZ-DNDC predictions of N2O emissions using constant Neff over the 50-day period. Neff
Total 50-day N2O emission (kg N2O-N/ha)
Total 50-day N2O emission as % of urine-only treatment
Model RMSE (kg N2O-N/ha/d)
0 0.2 0.4 0.6 0.8 1.0 Measured
3.08 2.72 2.28 1.73 1.01 0.06 1.6 0.6
100 88 74 56 33 2 50 20
0.042 0.034 0.027 0.023 0.026 0.039
Table 2 Mean N2O emitted over a 50-day period from plots receiving various treatments. Treatment
N2O emitted over 50-day period (kg N/ha)
Control (water-only) Urine (600 kg N/ha) Urine (600 kg N/ha) + DCD (7 kg/ha) Least significant difference (5%)
0.81 3.37 1.62 1.01
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
315
Fig. 6. Measured and modelled N2O emissions for the urine plus DCD treatment. For the model simulations the effectiveness of DCD as a NI (Neff) decayed exponentially with a half-life of 70 days with the initial Neff varying from 0 (no nitrification inhibition) to 1 (complete nitrification inhibition). The urine was applied at a rate of 600 kg N/ha and DCD applied at 7 kg/ha. Error bars represent the standard deviation of the three replicates.
3.2. Exponentially decaying Neff model Fig. 6 shows the measured N2O emissions compared with the NZ-DNDC predictions, using an exponentially decaying Neff (with a range of starting values for Neff) for the urine + DCD treatment. After 35 days the N2O emissions are very similar for all initial Neff values. Table 4 shows the total N2O emissions predicted using an exponentially decaying Neff as the initial value varied. The measured N2O emission (both as an absolute figure and relative to the urine-only measurement), allowing for uncertainties, agrees with the modelled values using an Neff 0.8. The minimum model RMSE with respect to N2O emissions occurred using an initial Neff = 0.8. 3.3. Comparing the constant and exponentially decaying Neff models Fig. 7(a)–(c) shows the measured soil NH4+-N and NO3-N concentrations and N2O emissions for the urine + DCD treatment compared to the best fit model predictions using a constant Neff of 0.7 and an exponentially decaying Neff initially set to 0.8. Both models simulated NO3-N concentrations and N2O emissions that were in reasonable agreement with the measured values, while the models under-predicted the soil NH4+-N during the first 14 days. Both models behaved similarly for the first 12–16 days, after which the exponentially decaying Neff model tended to predict higher N2O emissions and soil NO3 levels and lower soil NH4+ concentrations. This was expected as the exponentially decaying Neff value dropped from 0.8 to below 0.7 after 13.5 days. The RMSEs
Table 4 The effect of initial Neff on NZ-DNDC predictions of N2O emissions using an exponentially decaying Neff over the 50-day period. Neff
Total 50-day N2O emission (kg N2O-N/ha)
Total 50-day N2O emission as % of urine-only treatment
Model RMSE (kg N2O-N/ha/d)
0 0.2 0.4 0.6 0.8 1.0 Measured
3.08 2.87 2.64 2.38 2.08 1.74 1.6 0.6
100 93 86 77 68 57 50 20
0.042 0.036 0.031 0.027 0.026 0.028
Fig. 7. Measured and modelled soil (a) NH4+-N concentration (0–100 mm), (b) NO3-N concentration (0–100 mm) and (c) N2O emissions for the urine + DCD treatment. Urine applied at a rate of 600 kg N/ha and DCD applied at 7 kg/ha. The constant Neff model assumes that DCD reduces nitrification by 70% for the duration of the experiment (i.e., Neff = 0.7), while in the exponential Neff model the Neff value decays exponentially from an initial value of 0.8 with a half-life of 70 days. Error bars represent the standard deviation of the three replicates.
for the modelled soil NH4+-N were 67 and 65 mg NH4+-N/kg soil for the constant and exponentially decaying Neff models, respectively. For the soil NO3-N concentrations and N2O emissions the constant Neff model performed slightly better with RMSEs of 15 mg NO3-N/kg soil and 23 g N2O-N/ha/d, respectively compared to the exponentially decaying Neff model with corresponding RMSEs of 17 mg NO3-N/kg soil and 26 g N2O-N/ha/d. The total modelled N2O emission over the 50-day period was 1.4 kg N/ha for the constant Neff model and 2.1 kg N/ha for the exponentially decaying Neff model compared to the experimental measurement of 1.6 0.6 kg N/ha.
316
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
4. Discussion The use of NIs is being considered as a potential mitigation method to help reduce New Zealand’s N2O emissions (Ministry of Agriculture and Forestry, 2006). However, N2O emissions reductions from using NIs need to be clearly demonstrated and quantified if they are to be counted towards international agreements such as the Kyoto Protocol. From the laboratory experiments of Singh et al. (2008) we saw that not only did N2O emissions vary with soil type, but also the emission reductions obtained using DCD as an NI varied. This variation could in part be explained by the difference in DCD decomposition rates in the different soils. Other factors controlling the effectiveness of DCD in different soil types (e.g., temperature, soil pH buffering capacity and nitrification rates) need to be better understood to enable a better estimate of the N2O emissions reductions possible on a national scale and a better indication of the regions where the use of NIs can be of maximum benefit. In this study two approaches were used to model the effectiveness of the nitrification inhibitor (Neff). In the constant Neff model, Neff was assumed to remain constant for the entire duration. In the exponential Neff model, the biological decay of the applied DCD was simulated by using a Neff value that exponentially decreased over time. The rate of decay of DCD has been found to depend upon both the type of soil (Singh et al., 2008) and temperature (Guiraud et al., 1989; Di and Cameron, 2004). However, using an exponentially decaying Neff did not improve the model predictions substantially compared to the constant Neff simulations (in fact, the RMSE for the exponential Neff model was higher than the constant Neff model for soil NO3 and N2O emissions). An implicit assumption of the exponential Neff model was that the effectiveness of the NI is directly proportional to its concentration in the soil. However, logically this cannot be the case over all possible inhibitor concentrations as it is not possible for the nitrification rate to be reduced by more than 100%. The laboratory study by Singh et al. (2008) found no significant difference in N2O emissions over a 58-day period between urine and DCD treatments with DCD applied at 10 mg/soil and 20 mg/kg soil. In addition, the half-life used in the Neff model was calculated using Eq. (4), which does not take soil type into account. Singh et al. (2008) found the half-life of DCD (applied at 10 mg/kg soil) in Tokomaru soil at 25 8C to be 10.0 1.2 days, while the half-life calculated using Eq. (4) would be 20.6 days, which suggests that Eq. (4) over-predicts the half-life of DCD in Tokomaru soil. Further studies are needed to quantify the relationships between soil type, inhibitor concentration and inhibitor effectiveness. This is the first time this model has been used to assess the changes in N2O emissions with DCD in a grazed pasture system. The addition of DCD to urine-N in the field plots resulted in significant decreases in N2O emissions, with the magnitude of the decreases being in the 50 20% range (depending on season of application) as reported with the application of DCD in lysimeter studies in New Zealand (e.g., Di and Cameron, 2002, 2003; Di et al., 2007; Zaman and Blennerhassett, 2010). Using DCD with urine in the Tokomaru soil resulted in a reduction in N2O emissions of 50 20% relative to using urine-only. This is somewhat lower than the N2O emission reduction of 85–90% found by Singh et al. (2008) using 10 mg DCD/kg urine-treated soil under controlled conditions. However, there were a number of differences between the controlled laboratory experiment and this field study that could potentially account for the different emission reductions. First, the temperature and soil moisture content were kept constant in the laboratory experiment, while the field trial was exposed to variations in temperature and rainfall. In the laboratory, soil moisture was maintained at 80% of field capacity, which would
have restricted denitrification. In the field, 89.5 mm rain fell over the 50-day period and could have produced regions within the soil where the soil moisture level was temporarily >62% WFPS, which would have increased denitrification. The temperature used in the controlled experiment was 25 8C, significantly higher than the average temperature of 10.5 8C during the field trial. The higher temperature of the laboratory experiment would have enhanced not only the rate of DCD degradation, but also the rate of denitrification. In the laboratory, there was no uptake of N from plants as there would have been in the field. Also, in the field, the urine and DCD were applied to the soil surface, whereas they were mixed with the whole soil in the laboratory experiment. Finally, the laboratory experiment used only the top 0–0.1 m of the soil, whereas in the field total N2O emissions from all soil depths were measured. While laboratory experiments are useful for gaining an understanding of the processes involved in N2O emissions, they are not always accurate predictors of emissions under field conditions. The NZ-DNDC model was able to simulate the effect of DCD on N2O emissions by assuming DCD reduced the nitrification rate by a constant factor, Neff, for a fixed period of time. Simulating the biological degradation of DCD in the soil using an exponentially decreasing Neff soil did not improve the model performance compared to using a constant Neff over the 50-day period. However, if the model had been run for longer time period it would have been necessary to account for the finite lifetime of DCD in soil. As NZ-DNDC is a process-based model, in the longer term it would be desirable to refine the model to add the relationships between soil type, DCD degradation, DCD concentration and Neff as these relationships are quantified. Although the models produced reasonable estimates of N2O emissions over the 50-day period, soil NH4+ and NO3 levels following urine application tended to be underestimated, so further work is needed on the model’s handling of N-transformations following urine application. 5. Conclusion The use of DCD added to urine applied during winter 2005 to a dairy pasture field plots on Tokomaru silt loam reduced N2O emissions by 50 20% relative to the urine-only treatment. The NZDNDC model was able to simulate the effect of DCD on N2O emissions in Tokomaru soil for a 50-day period under field conditions by assuming a constant fractional reduction (Neff) in the nitrification rate. The best agreement with the experimental results was found when Neff was 0.7. An attempt to model the exponential decay effect of DCD in the soil did not improve the model performance. When the NI effects are measured over a longer period it would be necessary to account for the degradation of DCD, but such a model would need to consider the interactions between soil type, temperature, DCD concentration and inhibitor efficacy. This study has highlighted the potential of the NZ-DNDC model to assess the emissions reductions from the use of NIs and points to several important aspects that need to be addressed in future research. Which processes regulate the efficacy of NIs at site/point application scales, and what conclusions can be drawn about their likely short-, medium- and long-term effectiveness and required application rates/frequencies? What are the relationships between soil properties, temperature and DCD degradation and efficacy, and how can these be incorporated into the NZ-DNDC model? Studies addressing these issues are currently underway. Acknowledgements Funding for this research was provided by the New Zealand Foundation for Research Science and Technology. Thanks to Dr. Des Ross for his comments on the initial draft and Anne Austin for
D.L. Giltrap et al. / Agriculture, Ecosystems and Environment 136 (2010) 310–317
internal editing. We would also like to thank the creator of DNDC, Dr. Changsheng Li, for his ongoing advice and support. References Abbasi, M.K., Adams, W.A., 2000. Estimation of simultaneous nitrification and denitrification in grassland soil associated with urea-N using 15N and nitrification inhibitor. Biology and Fertility of Soils 31, 38–44. Babu, Y.J., Li, C., Frolking, S., Nayak, D.R., Adhya, T.K., 2006. Field validation of DNDC model for methane and nitrous oxide emissions from rice-based production systems of India. Nutrient Cycling in Agroecosystem 74, 157–174. Beheydt, D., Boeckx, P., Sleutel, S., Li, C., van Cleemput, O., 2007. Validation of DNDC for 22 long-term N2O field emission measurements. Atmospheric Environment 41, 6196–6211. Brown, L., Syed, B., Jarvis, S.C., Sneath, R.W., Phillips, V.R., Goulding, K.W., Li, C., 2002. Development and application of a mechanistic model to estimate emission of nitrous oxide from UK agriculture. Atmospheric Environment 36 (6), 917–928. Butterbach-Bahl, K., Kesik, M., Miehle, P., Papen, H., Li, C., 2004. Quantifying the regional source strength of N-trace gases across agricultural and forest ecosystems with process based models. Plant Soil 260 (1–2), 311–329. Cookson, W.R., Cornforth, I.S., 2002. Dicyandiamide slows nitrification in dairy cattle urine patches: effects on soil solution composition, soil pH and pasture yield. Soil Biology and Biochemistry 34, 1461–1465. Del Grosso, S.J., Ojima, D.S., Parton, W.J., Stehfest, E., Heistemann, M., DeAngelo, B., Rose, S., 2009. Global scale DAYCENT model analysis of greenhouse gas emissions and mitigation strategies from cropped soils. Global and Planetary Change 67, 44–50. Di, H.J., Cameron, K.C., 2002. The use of a nitrification inhibitor, dicyandiamide (DCD), to decrease nitrate leaching and nitrous oxide emissions in a simulated grazed and irrigated grassland. Soil Use Management 18, 395–403. Di, H.J., Cameron, K.C., 2003. Mitigation of nitrous oxide emissions in spray irrigated grazed grassland by treating the soil with dicyandiamide, a nitrification inhibitor. Soil Use Management 19, 284–290. Di, H.J., Cameron, K.C., 2004. Effects of temperature and application rate of a nitrification inhibitor, dicyandiamide (DCD), on nitrification rate and microbial biomass in a grazed pasture soil. Australian Journal of Soil Research 42, 927–932. Di, H.J., Cameron, K.C., Sherlock, R.R., 2007. Comparison of the effectiveness of a nitrification inhibitor, dicyandiamide, in reducing nitrous oxide emissions in four different soils under different climatic and management conditions. Soil Use Management 23, 1–9. Farahbakhshazad, N., Dinnes, D.L., Li, C., Jaynes, D.B., Salas, W., 2008. Modeling biogeochemical impacts of alternative management practices for a row-crop field in Iowa. Agriculture Ecosystem and Environment 123, 30–48. Farquharson, R., Baldock, J., 2008. Concepts in modelling N2O emissions from land use. Plant Soil 309, 147–167. Frolking, S.E., Mosier, A.R., Ojima, D.S., Li, C., Parton, W.J., Potter, C.S., Priesack, E., Stenger, R., Haberbosch, C., Do¨rsch, P., Flessa, H., Smith, K.A., 1998. Comparison of N2O emissions from soils at three temperate agricultural sites: simulations of year-round measurements by four models. Nutrient Cycling in Agroecosystem 52, 77–105. Giltrap, D.L., Saggar, S., Li, C., Wilde, H., 2008. Using the NZ-DNDC model to estimate agricultural N2O emissions in the Manawatu-Wanganui region. Plant Soil 309, 191–209.
317
Guiraud, G., Marol, C., Thibaud, M.C., 1989. Mineralization of nitrogen in the presence of a nitrification inhibitor. Soil Biology and Biochemistry 21, 29–34. Hedley, C.B., Saggar, S., Tate, K.R., 2006. Procedure for fast simultaneous analysis of the greenhouse gases: methane, carbon dioxide, and nitrous oxide in air samples. Communication is Soil Science and Plant Analysis 37, 1501–1510. Kelliher, F.M., Clough, T.J., Clark, H., Rys, G., Sedcole, J.R., 2008. The temperature dependence of dicyandiamide (DCD) degradation in soils: a data synthesis. Soil Biology and Biochemistry 40, 1878–1882. Kiese, R., Li, C., Hilbert, D.W., Papen, H., ButterbachBahl, K., 2005. Regional application of PnET-N-DNDC for estimating the N2O source strength of tropical rainforests in the Wet Tropics of Australia. Global Change Biology 11 (1), 128–144. Li, C., Frolking, S., Frolking, T.A., 1992. A model of nitrous oxide evolution from soil driven by rainfall events: 1. Model structure and sensitivity. Journal of Geophysical Research 97, 9759–9776. Li, C., Narayanan, V., Harriss, R.C., 1996. Model estimates of nitrous oxide emissions from agricultural lands in the United States. Global Biogeochemical Cycles 2, 297–306. Li, C., Mosier, A., Wassman, R., Cai, Z., Zheng, X., Huang, Y., Tsuruta, H., Boonjawat, J., Lantin, R., 2004. Modeling greenhouse gas emissions from rice-based production systems: Sensitivity and upscaling. Global Biogeochemical Cycles 18, GB1043 1010.1029/2003GB002045. Ministry of Agriculture and Forestry, 2006. Sustainable Land Management and Climate Change: Options for a Plan of Action. Ministry of Agriculture and Forestry, Wellington, New Zealand. Ministry for the Environment, 2009. New Zealand’s Greenhouse Gas Inventory 1990–2007. Ministry for the Environment, Wellington, New Zealand. Neufeldt, H., Schafer, M., Angenendt, E., Li, C., Kaltschmitt, M., Zeddies, J., 2006. Disaggragated greenhouse gas emission inventories from agriculture via a coupled economic-ecosystem model. Agriculture, Ecosystems and Environment 112, 233–240. Saggar, S., Andrew, R.M., Tate, K.R., Hedley, C.B., Rodda, N.J., Townsend, J.A., 2004. Modelling nitrous oxide emissions from dairy-grazed pastures. Nutrient Cycling in Agroecosystem 68, 243–255. Saggar, S., Giltrap, D.L., Li, C., Tate, K.R., 2007a. Modelling nitrous oxide emissions from grazed grasslands in New Zealand. Agriculture Ecosystem and Environment 119, 205–216. Saggar, S., Hedley, C.B., Giltrap, D.L., Lambie, S.M., 2007b. Measured and modelled estimates of nitrous oxide emission and methane consumption from a sheepgrazed pasture. Agriculture Ecosystem and Environment 122, 357–365. Singh, J., Saggar, S., Giltrap, D., Bolan, N., 2008. Decomposition of dicyandiamide (DCD) in three contrasting soils and its effect on nitrous oxide emission, soil respiratory activity and microbial biomass—an incubation study. Australian Journal of Soil Research 46, 517–525. Xu-Ri, H., Niu, H.-S., Li, C.-S., Wang, Y.-S., Wang, M.-X., 2006. Uncertainties in upscaling N2O flux from field to 18 18 scale: A case study for Inner Mongolian grasslands in China. Soil Biology and Biogeochemistry 38, 633–643. Zaman, M., Blennerhassett, J.D., 2010. Effects of the different rates of urease and nitrification inhibitors on gaseous emissions of ammonia and nitrous oxide, nitrate leaching and pasture production from urine patches in an intensive grazed pasture system. Agriculture Ecosystems and Environment 136, 236–246. Zaman, M., Saggar, S., Blennerhassett, J.D., Singh, J., 2009. Effect of urease and nitrification inhibitors on N transformation, gaseous emissions of ammonia and nitrous oxide, pasture yield and N uptake in grazed pasture system. Soil Biology and Biochemistry 41, 1270–1280.