Fertiliser and seasonal urine effects on N₂O emissions from the urine-fertiliser interface of a grazed pasture

Abstract

Significant areas of ruminant-grazed pastures are simultaneously covered by excreted urine and fertiliser nitrogen (N). However, the effect of overlapping N inputs on nitrous oxide (N₂O) emission factors has not been studied. Three rates of ¹⁵N-labelled urea fertiliser were applied with either no urine, an autumn-urine or a spring-urine application. These treatments were applied to perennial ryegrass pasture (Lolium perenne L.) and N₂O fluxes were determined over 373 days using standard static closed chamber techniques. Cumulative N₂O-N fluxes ranged from 766 to 4332 g N₂O-N ha^–1 (0.36%–0.74% of total N applied) and were lowest in the absence of urine; however, no fertiliser rate effect occurred regardless of urine presence or season of application. Urine-elevated N₂O-N fluxes followed urine applications for up to 40 days, resulting in lower fertiliser contributions to the N₂O-N fluxes at these times. Total ¹⁵N recoveries as N₂O-N were ≤0.04% and did not differ with fertiliser rate.

Keywords:

• emission factor
• fertiliser
• nitrous oxide
• urine
• ¹⁵N

Introduction

Nitrous oxide (N₂O) is a potent greenhouse gas (GHG) and is projected to play a significant role in stratospheric ozone depletion (Ravishankara et al. 2009). The global atmospheric N₂O concentration has increased by 20% from 270 ppb in pre-industrial times (1750), to 324 ppb in 2011, with the rate of increase over the past three decades equal to 0.73 ± 0.03 ppb yr⁻¹ (Ciais et al. 2013). New Zealand is unique in that, due to its large pastoral agricultural sector, N₂O makes up a large proportion (17.2%) of its GHG emissions inventory (Ministry for the Environment 2012). In most other countries, N₂O emissions contribute only 5%–10% of total GHG emissions (UNFCCC 2003). Understanding and mitigating agricultural N₂O emissions from pastoral soils is therefore a key priority for New Zealand, if it is to achieve targets for reducing GHG emissions.

A significant proportion of the increase in the tropospheric N₂O concentration has been attributed to anthropogenic emissions associated with the expansion and intensification of agricultural practices, particularly animal production systems (Davidson 2009; Reay et al. 2012; Clough et al. 2013). It is estimated that approximately 40% of global N₂O emissions arise via microbial transformation (e.g. denitrification) of agricultural nitrogen (N) inputs (Denman et al. 2007; Davidson 2009; Schils et al. 2013).

Within grazed pasture systems, N inputs include the deposition of excreta (urine and dung) and the application of fertiliser N, often as urea. The area of pasture covered by ruminant excretal returns depends on grazing management. Under intensive rotational grazing conditions in New Zealand, with stocking rates of 2–3.5 cows per hectare, Moir et al. (2011) found that, on average, 5.8% of the grazed paddock area received urine during a single grazing event, while on an annual basis, the average urine patch coverage for the grazed paddock was 23.2% of the total area. Dennis et al. (2011) found similar results for Irish pastures when stocking rates were 2–2.9 cows per hectare, with 14.1%–20.7% of the soil surface wetted by urine on an annual basis. In grazed pastures, the rates of fertiliser application typically do not exceed 200 kg N ha^–1 per annum and these annual applications are often split into multiple applications in an effort to optimise fertiliser N supply with plant demand over the grazing season (NZFA 2013). These fertiliser applications are, more often than not, carried out immediately after a grazing event to maximise dry matter production in the interval between grazing.

It is well documented that N₂O emissions increase under ruminant urine patches due to the introduction of an organic carbon source, increased availability of mineral N substrate and increased soil moisture (de Klein et al. 2003; Clough et al. 2005, 2009; de Klein & Ledgard 2005; Di & Cameron 2008). This is because the amount of N deposited in urine patches far exceeds the pasture plants’ immediate demands (Haynes & Williams 1993) and so the excess N is readily available for numerous N₂O production mechanisms (Wrage et al. 2001). Increased N₂O emissions also occur following fertiliser N applications (de Klein et al. 2001; Hyde et al. 2006; Shcherbak et al. 2014).

Nitrous oxide emission factors (EF) are used to determine excreta and fertiliser contributions for a given country's N₂O inventory (IPCC 2006). They represent the ratio of N₂O-N emitted from a soil that receives an N input, minus the N₂O-N emitted from the soil not receiving N, divided by the amount of N applied. The IPCC guidelines stipulate default emission factors for EF₁ (N additions from mineral fertilisers) and EF_3PRP (excretal N inputs to pasture range or paddock) as 0.01 and 0.02 kg N₂O-N kg^–1 N input, respectively (IPCC 2006). However, many countries have also determined ‘country specific’ EF values. For example, in New Zealand the emission factors for assessing N₂O contributions from urine (EF_3PRP) and fertiliser N (EF₁) are both set at 0.01 kg N₂O-N kg^–1 N input.

To date, the IPCC guidelines have assumed a linear relationship between N₂O emissions and the rate of N input. However, it is becoming increasingly accepted that increasing the rate of fertiliser N inputs can lead to an exponential increase in the rate of N₂O emissions, particularly when N fertiliser application rates exceed plant requirements (McSwiney & Robertson 2005; Hoben et al. 2011; Signor et al. 2013; Masaka et al. 2014; Shcherbak et al. 2014). Thus when fertiliser N is applied to grazed pastures and covers urine patches (areas already high in N) the percentage of fertiliser N applied that is subsequently evolved as N₂O might be expected to be greater than when fertiliser is deposited on to non-urine-affected soils. The rationale for this is based on the fact that following urine deposition, the soil N concentration under the urine patch is high and in excess of the pasture plants’ immediate demand, and so a further input of N from fertiliser enhances the microbially available N in the soil. Furthermore, urine patches can enhance denitrifying conditions by promoting soil carbon release (Monaghan & Barraclough 1993; Lovell & Jarvis 1996; Clough et al. 2004). Considering that the entire pasture area of a dairy farm receives fertiliser, and that c. 20% of such a farm's pasture area may be exposed to ruminant urine, the consequences of an elevated EF₁ value in the urine patch area may be significant for N₂O inventory assessments. For example, assuming a 200 ha dairy farm applies 200 kg N ha^–1 per annum then, using an EF₁ value of 0.01 kg N₂O-N kg^–1 N input, 400 kg of N₂O-N will be emitted per annum. If, over the course of the fertiliser applications, 20% of the fertiliser falls on to urine patches where, assuming no seasonal effects, the combined effect of these N inputs is to elevate the fertiliser EF₁ value to an assumed rate of 0.02 kg N₂O-N kg^–1 N input, then the total fertiliser-derived N₂O emissions would increase by 20%.

Even though the application of fertiliser post grazing is routinely practised, there appear to have been no studies specifically designed to evaluate fertiliser N₂O emissions following fertiliser N deposition on to urine patches. We hypothesised that fertiliser N applied on to urine patches would exacerbate N₂O-N emissions from the fertiliser N applied. Thus the objectives of this research were to use ¹⁵N-labelled urea fertiliser to determine: (1) if the deposition of urea fertiliser on to a urine patch enhanced total and/or fertiliser-derived N₂O emissions (EF₁); (2) if such effects were seasonal; and (3) if the fertiliser rate affects seasonal effects.

Materials and methods

A lysimeter experiment was performed to measure pastoral N₂O emissions under different urea fertiliser rates and urine treatments as previously reported (Buckthought et al. 2015). In brief, 36 undisturbed soil monolith lysimeters, 50 cm diameter, 70 cm deep, were collected from the AgResearch No. 1 Dairy Farm in Ruakura, Hamilton, New Zealand (37.779°S, 175.315°E). The soil, a free-draining Horotiu silt loam (Typic Orthic Allophanic), was under ryegrass (Lolium perenne L.) and white clover (Trifolium repens) and had not been grazed or fertilised for at least 4 months prior to the lysimeter collection. The basic soil properties are presented in Table 1.

Table 1 Basic soil properties (0–7.5 cm depth).

Soil properties	Value
pH	5.9
Total C (g kg^–1)	55
Total N (g kg^–1)	6.7
Olsen P (mg kg^–1)	30
Sulphate sulphur (mg kg^–1)	6
Potassium (cmolc kg^–1)	0.65
Calcium (cmolc kg^–1)	8.1
Magnesium (cmolc kg^–1)	1.57
Sodium (cmolc kg^–1)	0.14
Cation exchange capacity (cmolc kg^–1)	23
Total base saturation (%)	45

The lysimeters were collected on 14 December 2010 and transported to a field trench facility where they were installed in the ground at the same level as the surrounding pasture surface. The lysimeters immediately received 25 kg N ha^–1 as urea followed by 10 mm of water to stimulate pasture growth and to break drought conditions. Prior to the beginning of the experiment, 800 mm of water (>1 pore volume) was applied to the lysimeters over a 10 day period to leach any antecedent nitrate (NO₃–) from the soil. The experiment began on 21 February 2011 and continued until 28 February 2012 (373 days).

Experiment design and treatments

Nine treatments, each with four replicates, were arranged in a randomised complete block design. Three treatments consisted of urea fertiliser, applied in solid form at rates of 0, 200 and 400 kg N ha^–1 yr^–1 evenly split into applications of 0, 25 and 50 kg N ha^–1, respectively, but with nil urine applied. These are subsequently referred to as F0U0, F2U0 and F4U0 (Table 2). A further three treatments had the same fertiliser treatments, but they also received a single bovine urine-N treatment applied in the autumn on day 73, at a rate of 800 kg N ha^–1 (applied c. 15 min prior to the fertiliser) and these treatments are subsequently referred to as F0UA, F2UA and F4UA (Table 2). The final three treatments also received the same fertiliser and urine treatments, except the urine was applied in spring, on day 193, and these treatments are subsequently referred to as F0US, F2US and F4US (Table 2).

Table 2 Details of urine and fertiliser treatments applied to lysimeters.

Code	Treatment	Annual ^15N fertiliser rate (kg N ha^–1 yr^–1)	Urine (kg N ha^–1 yr^–1)
F0U0	Control	0	0
F2U0	^15N Urea 200	200	0
F4U0	^15N Urea 400	400	0
F0UA	Urine (autumn)	0	800
F2UA	^15N Urea 200 + Urine (autumn)	200	800
F4UA	^15N Urea 400 + Urine (autumn)	400	800
F0US	Urine (spring)	0	800
F2US	^15N Urea 200 + Urine (spring)	200	800
F4US	^15N Urea 400 + Urine (spring)	400	800

In order to determine the urea fertiliser's contribution to the N₂O emissions, it was enriched with ¹⁵N (5 atom%). The 200 kg N ha^–1 yr^–1 fertiliser represented standard fertiliser practice for dairy farms in the Waikato region, while the 400 kg N ha^–1 yr^–1 rate represented a more intensive system. Fertiliser applications were promptly irrigated in with 10 mm of deionised water to reduce NH₃ volatilisation losses (Black et al. 1987).

Fresh dairy cow urine was collected from the AgResearch Tokanui dairy farm, 1 day prior to each urine application, and standardised to a concentration of 8.0 g N L^–1. Urine (1.8 L) was applied on either 4 May 2011 (day 73, for autumn treatments) or 31 August 2011 (day 192, for spring treatments). The urine concentration and volume were selected based on previously reported pastoral grazing studies (Petersen et al. 1956; Davies et al. 1962; Richards & Wolton 1976; Safley et al. 1984; Haynes & Williams 1993). The urine was irrigated in with 10 mm of water to reduce NH₃ volatilisation. When water was applied to irrigate in fertiliser and/or urine, the same rate of irrigation (10 mm) was applied to all lysimeters, including controls. Timings of treatment applications are shown in Fig. 1.

Figure 1 Timing of lysimeter treatment applications.

Nitrous oxide sample collection and analysis

The lysimeter casings were affixed with stainless steel annular water-filled channels which facilitated sealing of the headspace chambers. The headspace chambers (internal volume 17.6 L) were equipped with a three-way stopcock and a sealable vent (ensuring equilibrated air pressure). The chambers were insulated with Styrofoam to avoid temperature fluctuations during gas measurements.

Using a 50 mL syringe equipped with a three-way stopcock, headspace gas was gently flushed twice through the syringe. A 10 mL sample was then collected and injected into a previously evacuated (<0.01 atm) 6 mL glass vial (Exetainer^®; Labco, High Wycombe, UK). Sampling occurred 0, 30 and 60 min after positioning headspace chambers, with ambient air samples also collected to determine background N₂O concentrations. Nitrous oxide sampling was undertaken twice a week for the first 2 weeks following any fertiliser and/or urine treatment application, and once a week thereafter. A total of 59 N₂O measurements were collected over the 373 day experiment.

Gas samples were also collected after 180 min for N₂O-¹⁵N analysis by taking a 20 mL sample and injecting this into a previously evacuated 12 mL vial (Exetainer^®; Labco, High Wycombe, UK). Headspace gas samples for N₂O-¹⁵N were collected less frequently, once a week for the first 2 weeks following fertiliser and/or urine application, and once every 2 weeks thereafter. Samples for ¹⁵N₂O analysis were collected from all the ¹⁵N fertiliser treatments, but from only two replicates of the control (F0U0) treatment.

Analysis of the headspace gas was performed using a gas chromatograph (GC) (model 8610; SRI Instruments, Torrance, CA, USA) interfaced to a liquid autosampler (model 222XL; Gilson, Middleton, WI, USA). The autosampler had been modified for gas analysis by substituting a purpose-built (PDZ-Europa, Crewe, UK) double-concentric injection needle, allowing the entire gas sample to be flushed swiftly into the GC. The GC configuration included two 0.3 cm OD stainless steel columns packed with Haysep Q connected in series, oxygen-free dry N carrier gas (40 mL min^–1) and a ⁶³Ni electron capture detector at 320 °C. Reference gases (0.363–35.2 µL L^–1, BOC, Auckland, New Zealand) were used to prepare standard curves. The detection limit for N₂O was 0.01 µL L^–1 greater than the ambient concentration. Analysis of gas samples for ¹⁵N enrichment was performed using a continuous flow isotope ratio mass spectrometer (Sercon 20/20; Sercon, Cheshire, UK). The samples were analysed for N₂O concentration and ¹⁵N enrichment. The N₂O-¹⁵N recovered as a percentage of the total ¹⁵N fertiliser applied was calculated according to Cabrera & Kissel (1989). Soil N₂O fluxes were calculated using the method of Hutchinson & Mosier (1981). Cumulative N₂O fluxes were determined by integrating fluxes over time using a trapezoidal method. The contribution of the fertiliser ¹⁵N-derived N₂O to the total flux was determined by taking the ratio of ‘moles of ¹⁵N enriched N₂O evolved’ to the ‘total moles of N₂O evolved’ (enriched plus unenriched) and expressing this as a percentage.

Leachate and pasture N dynamics were also measured using this experimental setup, the results of which are published in Buckthought et al. (2015).

Throughout the experiment, climate data including rainfall were collected daily from the Ruakura weather station (NIWA 2007), located 0.1 km from the lysimeter site.

Statistical analysis

Statistical analyses were performed using the ANOVA directive of GenStat (15th edition). The least significant difference (LSD) was used to determine statistical variation between the sample means at the 5% significance level. The standard error of the difference (SED) was calculated by dividing the LSD (5%) by the t value (at the 5% significance level). Non-normally distributed data were log-transformed (natural log) prior to statistical analysis. Raw N₂O-N flux data that was ≤0 was adjusted to a value of 0.001 g N₂O-N ha^–1 d^–1 (half the minimum recorded flux) prior to log transformation. Where transformation was required, all conclusions were drawn from the analysis on the transformed scale; however, to maintain standard units, the means and error presented are the back-transformed data.

Results

Climate data

Total rainfall over the study period was 1307 mm (Fig. 2), which is higher than the annual average (1200 mm). This rainfall did not include the 80 mm (8 × 10 mm) simulated rainfall applied after the split fertiliser applications. There was uncharacteristically high rainfall in December 2011 (220 mm) and early January 2012 (130 mm) that resulted in summer soil drainage. The daily air temperature ranged from 1.0 to 23.5 °C and averaged 12.4 °C. Daily soil temperatures at 10 cm depth averaged 13.9 °C, ranging from 4.5 to 22 °C. There were 16 frost days during the study. Winter drainage from soil commenced on day 35 (late March 2011) and continued until day 250 (late October 2011), however most of the drainage occurred between days 69–145 (May–July 2011) (Fig. 2).

Figure 2 Daily and cumulative rainfall and cumulative drainage at experimental site over the duration of study. Note this does not include the 80 mm (8 × 10 mm) simulated rainfall applied after the split fertiliser applications.

Nitrous oxide emissions

Daily N₂O-N fluxes

In the absence of fertiliser and urine, the F0U0 treatment had N₂O-N fluxes ranging from 0–6 g N₂O-N ha^–1 d^–1, except on day 95 where the flux was 20 g N₂O-N ha^–1 d^–1 (Fig. 3A). In the non-urine treatments that received fertiliser (F2U0 and F4U0) the N₂O-N fluxes ranged from 3–41 g N₂O-N ha^–1 d^–1 (Fig. 3A). Following autumn-urine application, N₂O-N fluxes increased to peak at 310, 210 and 143 g N₂O-N ha^–1 d^–1 in treatments F0UA, F2UA and F4UA, respectively (Fig. 3B). Within 26 days of autumn-urine being applied, the N₂O-N fluxes equalled those of the non-urine treatments (Fig. 3B). After spring-urine application, N₂O-N fluxes peaked twice. The first peak occurred 13 days after urine was applied, with mean fluxes of 81, 88 and 55 g N₂O-N ha^–1 d^–1 in the F0US, F2US and F4US treatments, respectively (Fig. 3C). The second peak occurred 33 days after urine application with mean fluxes of 141, 104 and 83 g N₂O-N ha^–1 d^–1 for the F0US, F2US and F4US treatments, respectively (Fig. 3C). After 40 days, the spring-urine-induced N₂O-N fluxes had returned to control levels (Fig. 3C).

Figure 3 Mean temporal N₂O emissions, daily rainfall and soil moisture from A, nil urine; B, autumn urine; C, spring urine treatments. F0, F2 and F4 = fertiliser rates of 0, 200 and 400 kg N ha^–1; and U0, UA and US = nil, autumn- and spring-urine applications, respectively. Bold arrows indicate urine + ¹⁵N fertiliser application, other arrows indicate split fertiliser applications. Error bars = LSR (5%), error is only shown if LSR is >5, n = 4. Note different y axis scales.

Following fertiliser application events there were short periods of elevated N₂O-N fluxes, most readily seen in the absence of urine (Fig. 3C). On days 3, 79, 81, 101, 103, 107, 123, 205, 225 and 264, the N₂O-N fluxes were, on average, greater (P < 0.05) under the 400 kg N ha–¹ fertiliser rate than the 200 kg N ha–¹ rate. Urine application (spring or autumn) resulted in higher (P < 0.001) daily N₂O-N fluxes, when comparing the given seasonal urine treatment with those treatments not receiving urine.

Fertiliser ¹⁵N enrichment and recovery as N₂O-N

The ¹⁵N enrichment of the N₂O-N evolved ranged from 0.38–4.13 atom% ¹⁵N. A fertiliser rate effect occurred with N₂O-¹⁵N enrichment higher under the 400 kg N ha^–1 rate on every day of measurement (P < 0.05 to P < 0.001) except day 39, when averaged across urine treatments. Between days 74 and 107, an interaction between urine and fertiliser treatments resulted in lower N₂O-¹⁵N enrichment in autumn-urine treatments and higher N₂O-¹⁵N enrichment in non- or spring-urine treatments where 400 kg N ha^–1 of fertiliser was applied. From days 193 to 225, a urine treatment effect also caused N₂O-¹⁵N enrichment to be lower (P < 0.05 to P < 0.001) under spring-urine treatments, when averaged across fertiliser treatments. Cumulative recoveries of fertiliser ¹⁵N as N₂O-¹⁵N were ≤0.04% (Fig. 4). Higher ¹⁵N recovery occurred under the autumn-urine treatments (0.04%; P < 0.05) than in the spring-urine treatments (0.03%). Cumulative recoveries of fertiliser ¹⁵N as N₂O-¹⁵N were not affected by fertiliser rate or fertiliser by urine treatment interactions.

Figure 4 Mean cumulative recovery of N₂O-¹⁵N from ¹⁵N affected treatments. F0, F2 and F4 denote fertiliser rates of 0, 200 and 400 kg N ha^–1; U0, UA and US denote nil, autumn- and spring-urine applications, respectively. Bold arrows indicate urine + ¹⁵N fertiliser application, other arrows indicate split ¹⁵N fertiliser applications. Error bar = LSD (5%), n = 4.

The urea-¹⁵N fertiliser contribution to the daily N₂O-N emission fluxes ranged from 0.7%–75.2% with urine-N and/or soil-N contributions making up the difference (24.8%–99.3%). The fertiliser contribution to the N₂O-N emissions was greater (P < 0.05) under the 400 kg N ha^–1 treatment (Fig. 5). On days 74 to 107, following autumn-urine application (day 73), there was a lower (P < 0.01) contribution from fertiliser-¹⁵N to the N₂O fluxes evolved (Fig. 5). An interaction (P < 0.05) between the fertiliser and urine treatments occurred on days 81 and 87, following autumn-urine application, with non-autumn urine treatments having N₂O fluxes with a higher fertiliser-¹⁵N contribution at 400 kg N ha^–1 yr^–1 (Fig. 5). There was also a urine treatment effect following spring-urine application (day 193), from days 193 to 225, with a lower (P < 0.05) fertiliser-¹⁵N contribution to the N₂O flux evolved following the spring-urine application. No interaction between urine and fertiliser treatments was observed following the spring-urine application.

Figure 5 Mean fertiliser contribution to the N₂O flux from ¹⁵N affected treatments, expressed as a percentage of the total N₂O flux. F0, F2 and F4 denote fertiliser rates of 0, 200 and 400 kg N ha^–1; U0, UA and US denote nil, autumn- and spring-urine applications, respectively. Bold arrows indicate urine + ¹⁵N fertiliser application, other arrows indicate split ¹⁵N fertiliser applications.

Cumulative N₂O fluxes and emission factors

Cumulative N₂O-N fluxes over the 373 day period ranged from 766–4332 g N₂O-N ha^–1 (Table 3). The lowest cumulative fluxes (P < 0.05) occurred in the absence of urine where there was no fertiliser treatment effect (Table 3; Fig. 6). The highest cumulative N₂O-N fluxes occurred in the autumn-urine treatment, where the F2UA and F4UA treatments had higher (P < 0.05) N₂O-N fluxes than the spring-urine treatments (Table 3; Fig. 6). Fertiliser rate did not affect the magnitude of these cumulative N₂O-N fluxes under autumn-urine applications (Table 3). The cumulative N₂O-N flux from the F0UA treatment did not differ from any of the spring-urine treatments. All spring-urine treatments had higher cumulative N₂O-N fluxes compared with the non-urine treatments (P < 0.05), but there was no fertiliser rate effect (Table 3; Fig. 6). The rate of increase in cumulative N₂O-N emissions was most rapid after urine application events (Fig. 6).

Table 3 Cumulative N₂O-N fluxes as a percentage of total N applied and as an emission factor for total N applied.

Treatment	Description	Cumulative N2O-N flux (g N2O-N ha^–1)	Cumulative N2O-N flux (% total N applied)	Emission factor (%) for total N applied
F0U0	0N urea, nil urine	766a	ND^1	ND^1
F2U0	200N urea, nil urine	1470a	0.74a	0.35ab
F4U0	400N urea, nil urine	1732a	0.43b	0.24b
F0UA	0N urea, autumn urine	5292bc	0.66ab	0.57a
F2UA	200N urea, autumn urine	6104b	0.61ab	0.53ab
F4UA	400N urea, autumn urine	6436b	0.54ab	0.47ab
F0US	0N urea, spring urine	4144c	0.52ab	0.42ab
F2US	200N urea, spring urine	3785c	0.38b	0.30ab
F4US	400N urea, spring urine	4332c	0.36b	0.30ab
LSD (5%)^1		1657	0.20	0.19

Different letters indicate a significant difference between treatments at the 5% (P < 0.05) level.

¹Not determined since zero N applied.

Figure 6 Mean cumulative N₂O emissions from all treatments. F0, F2 and F4 = fertiliser rates of 0, 200 and 400 kg N ha^–1; and U0, UA and US = nil, autumn- and spring-urine applications, respectively. Bold arrows indicate urine + N fertiliser application, other arrows indicate split fertiliser applications. Error = LSD (5%), n = 4.

Cumulative N₂O fluxes as a percentage of the total N applied (fertiliser plus urine) were lowest (P < 0.05) in the spring-urine treatments receiving fertiliser and the F4U0 treatment, and highest (P < 0.05) in the F2U0 treatment (Table 3), and ranged from 0.36%–0.74% of the total N applied (Table 3). Calculating emission factors using the combined total N inputs showed the highest emission factor occurred in the autumn-urine treatment receiving no fertiliser, with further fertiliser inputs having no effect on the emission factor (Table 3). The combined emission factor in the spring-urine treatments did not differ from the autumn-urine treatments and was also unaffected by fertiliser inputs (Table 3). In the absence of urine but with 400 kg of N applied (F4U0) the emission factor was significantly lower (Table 3) than in the autumn-urine treatment without N (F0UA). The recovery of ¹⁵N as N₂O demonstrates that the emission factor values, based solely on fertiliser N applied (EF₁), were ≤0.04 on average, as noted above.

Discussion

Increased N₂O fluxes following urine application in both the spring and autumn were typical of those previously observed following urine application to pasture soils. For example, Di et al. (2007) measured an N₂O peak of c. 300 g N₂O-N ha^–1 day^–1 under autumn-applied urine on a Horotiu soil, which was very similar to that measured under autumn urine in this study. Previous studies have, however, reported much higher N₂O emission peaks of 500–2000 g N₂O-N ha^–1 day^–1 from under autumn-applied urine (Di & Cameron 2003, 2006; Clough et al. 2009) and up to 700 g N₂O-N ha^–1 day^–1 from spring-applied urine (Di & Cameron 2003). Reasons for higher daily fluxes from some of these previous studies may include higher urine-N application rates of up to 1000 kg N ha^–1 and the application of irrigation (Di & Cameron 2003, 2006). Urine application enhances pasture soil N₂O emissions because it provides a substantial quantity of N substrate, over and above the plants’ immediate demand. This surplus N becomes available for the microbial processes, such as nitrification, nitrifier-denitrification and denitrification, which generate N₂O (Wrage et al. 2001). In addition the application of ruminant urine also releases soil carbon (Lambie et al. 2012) which may enhance heterotrophic denitrification and N₂O formation.

Nitrous oxide fluxes from urine patches may also vary with the season in which urine deposition occurs. If ruminant urine is deposited in the wetter, colder months of the year (autumn/winter), higher N₂O emissions may occur relative to the warmer and drier months of the year (spring/summer) (Luo et al. 1999; de Klein et al. 2004; Di & Cameron 2008). This is because in the autumn/winter months there is reduced pasture growth and a decreased demand for N, consequently, there is a greater availability of NO₃– substrate for denitrifying bacteria, along with higher soil water contents generating more anaerobic sites for denitrification to occur. These factors may explain the higher daily N₂O-N fluxes observed under the autumn-urine treatments. In the current study, the cumulative N₂O fluxes observed under the autumn-urine treatment in the absence of fertiliser were not statistically different from those measured in the spring-urine treatment (Table 2), although there was a trend for the autumn-urine treatment to have higher cumulative fluxes, possibly due to a decrease in plant demand for N since soil moisture contents during the periods of elevated N₂O fluxes were very similar (Fig. 3). Greater replication may have shown a statistical difference.

The periods of increased N₂O flux observed following the spring and autumn urine applications in the current study, at 40 and 26 days respectively, were also characteristic of the enhanced N₂O flux periods previously observed following ruminant urine application to pasture (e.g. Clough et al. 2009; Ledgard et al. 2014) and are a function of substrate supply and resulting microbial activity. The longer period of elevated daily fluxes in the autumn-urine treatment may have been due to a reduced pasture demand for N at this time.

Nitrous oxide emission factors for urine patches generally range from 0.1%–4% of the urine applied (de Klein et al. 2001), but may reach up to 14% on poorly drained clay soils (Lovell & Jarvis 1996; de Klein et al. 2001). Thus, in the current study, urine-induced N₂O emission factors in the absence of fertiliser (EF_3PRP) were at the lower range of these values (Table 2). This may have been a consequence of the higher than average, well-distributed annual rainfall contributing to soil moisture conditions that favour complete denitrification of N₂O to N₂. Alternatively, the higher than average rainfall may have caused nitrate to be moved deeper into the soil profile, thus reducing the potential for N₂O emission at the soil surface due to further complete denitrification occurring.

Increasing fertiliser rates did not increase cumulative N₂O emissions greater than background levels, probably due to the split fertiliser application regime which saw only 25 and 50 kg N ha^–1 fertiliser, respectively, applied at any one time, thus supporting the practice of split fertiliser applications as a practical and effective N₂O loss mitigation strategy (Di & Cameron 2002).

The total fertiliser ¹⁵N recovery as N₂O-¹⁵N was comparatively low (≤0.04%) when compared with other fertiliser studies where ¹⁵N recovery of urea as N₂O has been observed to range from c. <1%–3.5% (e.g. Di & Cameron 2008; Phillips et al. 2013). Differences in ¹⁵N recovery occur due to studies varying in duration, season, management and fertiliser. In this current study, the fate of urea fertiliser was followed, for the first time, over a full year and over eight split applications of urea-¹⁵N. Given that fertiliser applications were applied to optimise pasture production, and only applied when pasture was actively growing, this may have contributed to reduced N₂O emissions from the fertiliser pool. Alternatively, the above average rainfall may have promoted fertiliser N to be lost as N₂ as noted above. It is also possible that higher frequency of sampling for N₂O fluxes may have resulted in a greater recovery of ¹⁵N as N₂O.

The initial objective of this study was to determine if the application of urea fertiliser over a urine patch enhanced the total and/or fertiliser-derived N₂O emissions. The results demonstrate that application of fertiliser, regardless of rate, did not increase cumulative N₂O-N losses from autumn- or spring-urine treatments (Table 3). Expressing the cumulative N₂O-N flux as a percentage of the total N applied (fertiliser plus urine) also showed no increases as a consequence of fertiliser application or rate. Similarly, the combined emission factor (EF_1+PRP), based on total N inputs did not increase as a result of fertiliser application, regardless of rate. In fact there was a tendency, although not statistically significant, for the combined emission factor to decline as fertiliser rate increased (Table 2). The fact these various measures (relating N₂O emissions to N inputs) demonstrated no significant additional N₂O emissions as a result of increasing N inputs, is at odds with the original hypothesis that N₂O emissions would increase following fertiliser addition to urine patches. This suggests that the fertiliser N applied, in addition to urine, was used more efficiently or that its loss was not as N₂O.

Since the annual fertiliser N inputs (200 or 400 kg N ha^–1 yr^–1) were applied as eight evenly split fertiliser dressings, the rates of N application did not exceed the pasture's ability to utilise the fertiliser N (Di & Cameron 2002). Conversely, bovine urinary-N excreta deposition rates are known to exceed pasture N requirements by several-fold, contributing to gaseous and leaching losses of N (Haynes & Williams 1993; Cameron et al. 2013). The current study shows that, within the fertiliser-excreta interface, urine application elevated N₂O fluxes for 40 and 26 days following spring and autumn urine applications, respectively, with the bulk of this additional N₂O coming from urine-N, since fertiliser contributions to the total N₂O flux were diminished for c. 33 days regardless of season that urine was applied (Fig. 5). Thus, under the management conditions imposed, whereby 200 kg of fertiliser N ha^–1 was typical of dairy farm systems, N₂O emissions from fertiliser-¹⁵N were influenced by urinary-N derived N₂O emissions for only four of the eight split-fertiliser applications, at most. Thus while total N rates increased with fertiliser a large percentage of this fertiliser N was not influenced by urine-N and likely to be used more efficiently by the pasture. This potentially explains the trend for the combined emission factor (EF_1+PRP) to decline as fertiliser inputs increased under a urine patch. Thus on a farm system basis, the effect of a urine patch on annual fertiliser-derived N₂O-N losses was negligible.

Conclusions

It can be concluded that, under the management conditions imposed in the current study, the overlap of fertiliser at the fertiliser-urine patch interface does not lead to an increase in the N₂O emission factors for either fertiliser or urine excreted on to pasture, EF₁ and EF_PRP, respectively, and that there were no significant seasonal or fertiliser rate effects as a result of fertiliser-urine overlaps.

Acknowledgements

This research was funded by the Pastoral 21 programme, a collaborative venture between DairyNZ, Fonterra, Dairy Companies Association of New Zealand, Beef + Lamb NZ and the Ministry of Business, Innovation and Employment. The statistical analyses were carried out by Dr John Waller, formerly of AgResearch Ltd. Nitrous oxide analysis and ¹⁵N analysis was carried out by Manjula Premeratne and Roger Cresswell, respectively, of Lincoln University.