Influence of enhanced efficiency fertilisation techniques on nitrous oxide emissions and productivity response from urea in a temperate Australian ryegrass pasture

Additional keywords: 3,4-dimethylpyrazolephosphate, fine particle spray, nitrification inhibitor, N-(n-butyl) thiophosphoric triamide, urease inhibitor.

High nitrogen (N) fertiliser inputs, commonly as surface-applied granular urea, are typical in pasture-based dairy systems in Australia, resulting in low N use efficiency (NUE). Application of N fertilisers, such as urea, can lead to emissions of the greenhouse gas nitrous oxide (N₂O). Fertiliser-induced N₂O emissions from crops and grasslands total 0.9 million tonnes of N per year globally (IFA 2001). In Australia, agriculture represents 76% of the national total N₂O emissions, with around 50% of this from mineral fertiliser application (DCCEE 2011). Enhanced efficiency fertilisers (EEFs), including urease inhibitors, nitrification inhibitors, fine particle sprays, and controlled-release fertilisers, have the potential to decrease N losses from agriculture by either altering the rate of N transformations or slowing the release of N from fertiliser granules (Chen et al. 2008).

Various EEFs have been developed to target particular N loss pathways (Chen et al. 2008). Urease inhibitors, such as N-(n-butyl) thiophosphoric triamide (NBTPT), slow urea hydrolysis and are designed to decrease ammonia (NH₃) volatilisation from surface-applied granular urea. Nitrification inhibitors, such as 3,4-dimethylpyrazole phosphate (DMPP) and dicyandiamide (DCD), slow nitrification and are designed to decrease N loss as nitrate (NO₃^–), N₂O, oxides of nitrogen (NO_x) and dinitrogen (N₂). Liquid or fine particle spray fertilisers are designed to decrease NH₃ volatilisation and increase plant utilisation through enhanced foliar uptake (Dawar et al. 2012). The greatest benefit of the EEFs is expected in spring and autumn (urease inhibitor) and late winter–spring (nitrification inhibitor) when climatic conditions favour NH₃ volatilisation and denitrification losses, respectively. However, when a loss pathway is decreased as a consequence of EEF use and the rate of applied N remains the same, there is a risk of increased loss from an alternate pathway unless the additional N is used by the plant.

The nitrification inhibitors DMPP and DCD have been reported to decrease N₂O emissions in laboratory studies on Australian soils (DMPP) (Chen et al. 2010) and in field studies in pasture systems (DCD) (Di et al. 2007; Kelly et al. 2008; Di et al. 2010; Gilsanz et al. 2016). A recent review found them effective in temperate grassland-based agriculture (Li et al. 2013a). Rowlings et al. (2016) studied the impact of DMPP on NUE in sub-tropical pastures, but no studies have investigated its impact on N₂O emissions and NUE in Australian temperate pasture systems to date. Although urease inhibitors are effective at decreasing NH₃ volatilisation by slowing the rate of urea hydrolysis and reducing the risk of elevated pH that drives NH₃ formation, their impact can be variable due to the influence of climatic conditions, soil type, and land use on NH₃ volatilisation. A review of the literature shows that across a range of crops and pastures, use of a urease inhibitor led to a 20% to 88% decrease in NH₃ volatilisation compared with urea (Watson et al. 1990; Rawluk et al. 2001; Turner et al. 2010; Suter et al. 2013). Due to the impacts on NH₃ volatilisation, it is expected that N₂O emissions may increase relative to urea because of greater N in the system, but some studies have reported decreases in N₂O with the urease inhibitor (Zaman et al. 2009; Singh et al. 2013; Ding et al. 2014). Use of fine particle sprays and suspensions have been shown to decrease NH₃ and N₂O loss and increase productivity relative to urea when applied with and without urease or nitrification inhibitors (Di and Cameron 2006; Dawar et al. 2011). To develop effective strategies to mitigate N₂O emissions and provide productivity benefits to facilitate adoption of these strategies in Australian pasture systems, greater knowledge of the impact of EEFs under field conditions is required.

This paper reports on a field experiment using EEFs in a ryegrass seed crop. Surface applications of granular urea (40 kg N ha^–1) with and without EEFs (urease inhibitor NBTPT, nitrification inhibitor DMPP) were made regularly over an eight month period (autumn to summer) in 2010. A fine particle spray containing NBTPT and gibberellic acid was also used during autumn when expected NH₃ loss is high. The impact of these amendments on soil mineral N, N₂O emissions, and biomass production and N utilisation was determined.

The experiment was conducted at a ryegrass (Lolium perenne L.) seed crop site at Murroon in south-western Victoria, Australia (38°26ʹ10.18ʹʹS, 143°47ʹ34.57ʹʹE). Details of the site characteristics are described in Suter et al. (2013). Briefly, the Chromosol soil (Isbell 1996) has a topsoil (0–10 cm) with a silty loam texture (22% clay, 38% silt, 40% sand, from 0–25 cm depth), a pH_CaCl2 of 4.6, 0.2% total N, 2.7% total C, a bulk density of 1.23 g cm^–3, and a cation exchange capacity (CEC) of 4.98 cmol (+) kg soil⁻¹. Initial mineral N content (0–10 cm) was 13.4 kg N ha^–1 as ammonium (NH₄⁺) and 0.84 kg N ha^–1 as nitrate (NO₃^–). The soil porosity at 0–10 cm depth was 0.54. The site was used for ryegrass seed production and fenced off from sheep that grazed the site intermittently before (>3 weeks) establishing the experiment. The experiment was a replicated split block design of five blocks, separated by a 0.5 m buffer zone, with each treatment randomly assigned within each block to account for site spatial heterogeneity. Each treatment plot was 1 m × 2 m. The experiment commenced 12 April 2010 and finished 23 December 2010.

Local rainfall measured on site in 2010 was 803 mm. Climatic variables were measured on site with a weather station (model WXT510; Vaisala, Helsinki, Finland) until 3 October 2010 and afterwards from the two closest Bureau of Meteorology stations (Colac, 38.23°S, 143.79°E and Cape Otway, 38.86°S, 143.51°E) due to issues with the on-site weather station. Soil moisture and temperature were measured using capacitance probes (EnviroPro^®) inserted vertically into the ground and logging at 10 cm intervals.

Fertiliser (40 kg N ha^–1) was applied six times from April to October 2010 on 12 April, 7 June, 16 July, 3 September, 27 September and 21 October. Treatments were as follows:

1. control (C) (no fertiliser)

2. granular urea (U) (46% urea-N)

3. granular urea with the urease inhibitor, NBTPT (GU) (Green Urea 14™ (45.8% urea-N with ‘Agrotain^®’ @ 5 L t^–1 urea))

4. granular urea with the nitrification inhibitor, DMPP (DMPP) (Urea with ENTEC (46% urea-N with 0.4% of DMPP per unit of ammonium-N))

5. urea applied as a fine particle spray containing NBTPT and gibberellic acid (FPA) (46% urea-N with ‘Agrotain’ @ 1 L t^–1 of urea and gibberellic acid (10 g ha^–1)). Treatment 5 was applied once on 12 April 2010 to target NH₃ loss, which was expected to be high in autumn.

Three composite soil samples (0–10 cm depth) were collected from each plot using a corer (2.5 cm internal diameter) at regular intervals (~fortnightly), immediately dried (40°C) and sieved (<2 mm). Subsamples (20 g of 105°C dried soil equivalent) were extracted with 2 M KCl (1 : 5 soil  solution), by shaking for 1 h, filtering though Whatman No. 42 filter papers and were analysed for NH₄⁺-N, and NO₃^–-N using a SAN++ segmented flow analyser (Skalar Analytical B.V. 2005).

Nitrous oxide (N₂O) emissions were measured using manual chambers (23 cm diameter × 25 cm high) similar to those reported in Saggar et al. (2004). Open-topped chambers were inserted into the ground (5 cm) for the entire course of the experiment and were capped for 1 h during sample collection times. Gas flux measurements were collected at regular intervals (every second day for one week after fertilisation, and then weekly) from the capped chambers commencing between 1000 and 1200 hours. Three samples were collected at 0, 30 and 60 min after the chamber was capped. Collected samples (20 mL, injected into a 12 mL evacuated Exetainer^®, (Labco Ltd, United Kingdom)) were analysed by gas chromatograph (Agilent 6890) using an electron capture detector (ECD) with a lower detection limit of 0.2 ± 0.02 μL L^–1. Temperature was logged within the closed chamber (Tinytag Transit 2 TG-4080, Gemini Data Loggers). Soil moisture (ML2x theta probes (Delta-T Devices Ltd, Cambridge, UK)) and temperature (small portable temperature probe) were recorded at each sampling time.

From 3 August to 30 August 2010 free water was observed on the soil surface and gas samples were not collected during this time due to problems accessing the area and collecting samples without disturbing the soil and chambers. We hypothesised that any denitrification would be largely N₂ during this time and N₂O emissions would be minimal (Ciarlo et al. 2008). Recent work by Friedl et al. (2016) found that the N₂/(N₂+N₂O) ratio increased with increasing soil moisture (to 100% water filled pore space, WFPS) in sub-tropical pastures, and Harris et al. (2013) hypothesised that this was the reason for low N₂O emissions from cropping sites during periods when soil WFPS exceeded 90%.

Soil N₂O flux (kg N ha^–1 h^–1) was calculated using the linear regression (LR) model recommended by Venterea et al. (2012) and extrapolated to a daily N₂O emission. To compare treatment effects, cumulative emission was calculated by integration of the area under the daily flux curve for the period of measurement.

Biomass was measured on samples (2 m length × 0.4 m width; 0.8 m² cut per plot) collected using a lawn mower to simulate grazing rotations typical of the area at 24–28 days after fertiliser (DAF) (autumn and spring) and 42–46 DAF (winter). Biomass samples were collected on 10 May (28 DAF), 7 June (55 DAF), 19 July (42 DAF), 3 September (46 DAF), 27 September (24 DAF), 21 October (24 DAF), and 23 December (63 DAF) in 2010. The harvest on 7 June 2010 was included to assess the longevity of the impact of the EEFs following a time of expected high ammonia loss. The harvest on 23 December 2010 was 63 DAF, as the pasture was grown to seed production stage, and included both the stem and grain. All other harvests were vegetative only. Biomass was removed from the entire experimental area when each harvest was collected. Pasture N content was determined by the Kjeldahl digestion method with colourimetric analysis using a Lachat 8500 Flow Injection Analyser after samples were dried at 70°C for 72 h (Rayment and Lyons 2011).

Statistical analysis of treatment effects at each harvest was performed using the Fisher l.s.d. analysis of variance (ANOVA, P < 0.05) with Minitab17. In addition, significance of soil properties (mineral N, temperature and moisture), sample time and treatment, and the interaction between these factors, on daily N₂O flux (log₁₀ transformed) and dry matter production was assessed using the program RStudio, Version 0.97.248.

Rainfall and temperature data (Fig. 1) showed an initially dry autumn followed by a wet, cool winter (August) and a warm, moist spring (October–November). Air temperature ranged from a low of 0°C to a maximum of 31°C. Soil moisture increased from 12 April 2010 through to late August when the site became saturated, before drying off towards the end of the year (Fig. 2). Rainfall during spring (Fig. 1) caused soil moisture contents to fluctuate between mid-September through to December. The measured volumetric soil moisture (ϴ_v) ranged from 10% (19% WFPS) to 53% (98% WFPS) at 10 cm depth, 13% to 58% at 20 cm depth and 13% to 56% at 30 cm depth during the study. The upper reported ϴ_v for each depth represents saturation.

Fig. 1.  Rainfall and temperature at the Murroon field site (38°26ʹ10.18ʹʹS, 143°47ʹ34.57ʹʹE) from 1 April 2010 to 31 December 2010.

Fig. 2.  Soil volumetric water content (10, 20, 30 cm depth) and temperature (10 cm depth) at the Murroon field site (38°26ʹ10.18ʹʹS, 143°47ʹ34.57ʹʹE) from 12 April 2010 to 23 December 2010. Note: Soil porosity (0–10 cm) is 0.54, a volumetric water content of 54% is saturation.

Soil temperatures ranged from 4.7 to 28°C at 10 cm depth, 6.7 to 24°C at 20 cm depth and 8 to 22°C at 30 cm depth (Fig. 2), following a similar pattern to the ambient temperature (Fig. 1). There was a strong diurnal pattern of soil temperature with minimum temperature recorded between 0630 and 0900 hours and maximum between 1700 and 1800 hours.

Soil ammonium (NH₄⁺) levels fluctuated in response to fertiliser additions (Fig. 3), plant uptake and mineralisation. In spring, increased NH₄⁺-N was observed in all treatments, including the control, indicating that the increased soil temperature and moist conditions (Fig. 2) stimulated mineralisation from the abundant organic matter pool (total N and C contents of 0.2 and 2.7% respectively) (Hu et al. 2014). There was no overall significant difference in NH₄⁺-N between treatments. Slightly greater NH₄⁺-N was measured in the DMPP treatment compared with U many times before 27 September 2010, most noticeably on 19 July 2010 (23.2 ± 10.0 kg N ha^–1 for DMPP and 4.1 ± 0.2 kg N ha^–1 for U) and 11 August 2010 (49.0 ± 20.6 kg N ha^–1 for DMPP and 8.2 ± 1.1 kg N ha^–1 for U) (P < 0.1) (Fig. 3a). This is expected as the inhibitor slows ammonification, and Fangueiro et al. (2009) found this inhibitory effect can occur for extended periods (e.g. 100 days for DMPP). In this study, the suppression of nitrification over the wetter months (July to September) when there is an increased risk of NO₃^– leaching and denitrification losses, indicates that the inhibitor could provide real benefits in decreasing gaseous emissions (N₂O and N₂) and increasing biomass production. This is reflected in the lower NO₃^–-N levels in the DMPP treatment during July and August (see below). Lower NH₄⁺-N in the DMPP treatment after fertilisation on 27 September 2010 compared with U and GU was not expected but the reason for this is not clear.

Fig. 3.  Soil ammonium (a) and nitrate (b) for C (X), U (◆), DMPP (◇), GU (⬤) and FPA () showing timing of each fertilisation event () and harvest (+) with standard error of the mean of five replicates.

Addition of NBTPT to urea (GU) increased the quantity of NH₄⁺-N in the soil compared with U in autumn (19 April), early winter (5 July) and for most of spring (8 October, 21 October and 19 November). This results from decreased NH₃ loss, leading to greater N remaining in the soil. A concurrent study of NH₃ loss at the site found that using NBTPT decreased N loss as NH₃ in autumn to 3.7 kg N from 12 kg N with urea (Suter et al. 2013). However, in July and September 2010 the higher rainfall and soil moisture (Figs 1 and 2) would lower the potential for NH₃ loss from urea so no additional N would be ‘saved’ in the urease inhibitor treatment, indicating no benefit from using the urease inhibitor at this time.

Applying FPA once on 12 April 2010 did not alter soil NH₄⁺-N levels compared with U, and the soil NH₄⁺-N returned to baseline levels by 24 May 2010 (5.9 ± 0.6 kg N ha^–1 compared with 5.3 ± 0.5 kg N ha^–1 for C). Samples collected on 27 September 2010 showed all treatments had similar background levels of NH₄⁺-N (1.9 ± 0.2 kg N ha^–1 for C; between 2.2 ± 0.4 and 3.4 ± 0.2 kg N ha^–1 for the fertiliser plots (Fig. 3a)).

Soil NO₃^–-N levels also fluctuated in response to applied fertiliser N (Fig. 3b) and were lower than NH₄⁺-N. This indicates that NO₃^–-N was either: (1) being removed by the plant material, with NO₃^–-N the favoured form of N for plant uptake (Li et al. 2013b); or (2) was denitrified. Denitrification losses are expected to occur mostly over winter under anaerobic soil conditions (Butterbach-Bahl et al. 2013). In the early stages of the study, conditions were dry so biomass production was low (see Biomass section below), and plant uptake of NO₃^–-N would also be low. At this time the DMPP and FPA treatments decreased the amount of NO₃^–-N produced relative to U and GU thereby retaining N in the soil for subsequent plant uptake. In the FPA treatment, NO₃^–-N levels decreased to baseline levels one month after application, indicating plant uptake or immobilisation. The urease inhibitor (GU) increased NO₃^–-N levels compared with U throughout the study due to decreased NH₃ loss, and this was significantly more than that observed in the C and DMPP treatments (P < 0.01). During spring when high biomass production occurred, NO₃^–-N did not appear to respond to fertiliser application due to plant uptake, which is supported by the lack of difference between the NH₄⁺-N levels in the treatments at this time (Fig. 3).

There was a significant relationship between daily N₂O flux, day of sample collection, soil temperature, soil moisture, and the interaction between temperature and soil moisture, and temperature and day of sample collection, and GU (F(16, 251) = 36.65, P < 0.005). Daily N₂O emissions were comparatively low throughout autumn and early winter (April to early August) and higher in spring (Fig. 4a), reflecting soil moisture and temperature conditions (Fig. 2). This trend (Figs 2 and 4) is expected based on our understanding of the drivers of N₂O emissions (Rafique et al. 2012; Butterbach-Bahl et al. 2013; Huang et al. 2013; Bell et al. 2015). Nitrate, the product of nitrification and the substrate for denitrification, had a significant (P < 0.001) influence on daily N₂O flux. The lower N₂O emissions in autumn, despite NO₃^– being higher in all treatments (Fig. 3b) compared with the rest of the year, is due to lower soil moisture (Fig. 2), which limits denitrification.

Fig. 4.  Daily (a) and cumulative (b) N₂O flux for C (X), U (◆), DMPP (◇), GU (⬤) and FPA (▼) showing timing of each fertilisation event () with standard error of the mean of five replicates.

Minimum and maximum daily N₂O flux for each treatment was –0.28 to 24.56 g N₂O-N ha^–1 day^–1 for C; –0.6 to 50.93 g N₂O-N ha^–1 day^–1 for U; 0.07 to 94.51 g N₂O-N ha^–1 day^–1 for GU; –1.89 to 23.56 g N₂O-N ha^–1 day^–1 for DMPP; –0.28 to 15.95 g N₂O-N ha^–1 day^–1 for FPA. The lowest N₂O emission recorded occurred on 16 April 2010, four days after the first fertiliser application when soil moisture was low. For all but the FPA treatment (which was only used in autumn) the maximum N₂O flux occurred during spring (September to November), seven or more days after fertiliser was applied. This was due to the time required for urea hydrolysis, nitrification and denitrification and the warm, moist conditions. Similar time delays in N₂O production have been observed when cattle urine is added to soil (Bell et al. 2015).

The cumulative N₂O data showed that over eight months the emission of N₂O from C and U were 0.6 and 1 kg N₂O-N ha^–1. This represents one tenth of the denitrification loss reported by Eckard et al. (2003) in temperate Australian pastures (6 and 13 kg N ha^–1 year^–1 (N₂ and N₂O combined) from control and urea (200 kg N ha^–1 year^–1) treatments). Although the N₂O : N₂ ratio can vary widely depending on soil type and environmental conditions (Saggar et al. 2013), our reported results provide a reasonable estimate of N₂O emissions from temperate Australian pastures when compared with those of Eckard et al. (2003). The cumulative N₂O data show that although the fertiliser treatments caused differences in N₂O emissions relative to the control during the earlier part of the experiment (12 April to 7 June 2010) (Table 1), the absolute difference was small because of the low level of emissions occurring at that time (Fig. 4b) compared with the remainder of the year. At that time there was no significant difference in N₂O emissions between fertiliser treatments (Table 1).

Table 1.  Average cumulative N₂O emissions ± standard error from: (1) 12 April to 7 June 2010; (2) 12 April to 23 December 2010; (3) 16 July to 23 December 2010; and (4) 27 September to 23 December 2010 calculated from integration of the daily emissions data (Fig. 4)
Within columns, different letters indicate a significant difference (P < 0.05); NA, not applicable (treatment only applied on 12 April 2010)

DMPP decreased fertiliser-induced N₂O emissions by 76% during the study relative to U by decreasing NO₃^–-N production, the substrate for denitrification (Fig. 3). The impact of DMPP on N₂O emissions was most noticeable during spring (September to November) when fertiliser-induced emissions with DMPP decreased by 95% (to 9 ± 63 g N₂O-N ha^–1) compared with U (240 ± 146 g N₂O-N ha^–1), with N₂O essentially reaching background (control) levels (419 ± 146 g N₂O-N ha^–1 for C, 431 ± 64 g N₂O-N ha^–1 for DMPP) (Table 1, Fig. 4a). This occurred despite low soil NO₃^– levels for all treatments during spring (Fig. 3b), indicating increased plant uptake due to the spring growing conditions. The calculated reductions in N₂O emissions with DMPP were similar to those reported elsewhere (Menéndez et al. 2009; Di and Cameron 2012; Misselbrook et al. 2014). The observed results indicate that DMPP is an effective N₂O mitigation tool for temperate Australian pasture systems.

Addition of NBTPT (GU) resulted in a 153% increase in fertiliser-induced N₂O emissions (1044 ± 265 g N₂O-N ha^–1) relative to U (412 ± 149 g N₂O-N ha^–1), with the greatest impact occurring from 16 July to 23 December 2010 (GU; 935 ± 250 g N₂O-N ha^–1, U; 314 ± 150 g N₂O-N ha^–1) (Table 1). Other studies have reported decreased N₂O emissions with urease inhibitors (Dawar et al. 2011; Sanz-Cobena et al. 2012; Singh et al. 2013). From previous work, urease inhibitors can stimulate or decrease denitrification depending on the inhibitor type and concentration used (Yeomans and Bremner 1986; Zhengping et al. 1991). Zhengping et al. (1991) observed no inhibitory effect on denitrification with NBTPT. If the urease inhibitor prevents NH₃ loss and the rate of N applied is the same as for urea, as is the case here (40 kg N ha^–1), then there is more NH₄⁺-N retained in the soil that can undergo nitrification and denitrification to produce N₂O. The greatest impact of elevated emissions from the urease inhibitor occurred after the 16 July 2010 application. This was unexpected because in July NH₃ loss was assumed to be low, based on winter climatic conditions, and in September the measured NH₃ loss was low (Suter et al. (2013). So, the applied N remaining in the soil for the U and GU treatments should have been similar, but for GU the urea would have been released more slowly making the substrate for nitrification and N₂O production available for longer compared with urea. The impact of GU on N₂O, therefore, depends on the temporal synchronisation of the substrates for N₂O production (NH₄⁺ for nitrification and NO₃^– for denitrification) and conditions conducive for denitrification. This explains the differences reported here and in the literature (Ding et al. 2014). Whereas the mineral N data (Fig. 3) shows similar levels of NH₄⁺-N for U and GU, there is a trend for greater NO₃^– with GU, supporting our hypothesis.

Average biomass production responded to N application but, excluding spring, the response was slightly lower than that reported for long-term N studies in the same region (McKenzie et al. 2003) (Table 2). There was a significant relationship between dry matter production, date of sample collection, treatment, and the interaction between sample date and soil NH₄⁺ levels and pasture dry matter (F(33, 116) = 45.35, P < 0.005). Total plant production increased and was significantly greater (P < 0.001) for the final harvest on 23 December 2010 (Table 2) because this contained both vegetative biomass and seed. Biomass production was significantly lower (P < 0.001) at the 7 June 2010 harvest as a result of the period between fertilisation and harvest (55 DAF) and the removal of part of the applied N in the first pasture harvest (10 May 2010, 28 DAF). Applying urea did not significantly increase biomass relative to the control at this harvest indicating insufficient N remaining in the soil (Fig. 3a, b). DMPP, GU and FPA significantly (P < 0.05) increased biomass production compared with C by an average of 81 ± 15 kg dry matter (DM) at this harvest. FPA significantly increased biomass relative to U at this harvest, which may be due to the presence of gibberillic acid (Biddiscombe et al. 1962). The greater response to the inhibitors in autumn, when conditions were drier, is similar to the results reported by Rowlings et al. (2016). The lack of a consistent productivity benefit with the nitrification inhibitor, despite the decrease in N₂O emissions, was due to little N being lost via denitrification (measured as N₂O), limited leaching in these texture contrast soils (Chromosols) (Isbell 1996), and the presence of sufficient N from fertilisation in all treatments.

Table 2.  Mean biomass production (kg DM ha^–1) for each biomass cut, agronomic efficiency of applied N (kg pasture increase per kg N applied), and apparent recovery of applied N (net kg N taken up per kg N applied) ± standard error
Significant differences (P < 0.05 (Fisher test)) between treatment for each harvest date and section are indicated by different letters. DAF, days after fertiliser application; DoG, days of growth

The agronomic efficiency (kg pasture increase per kg N applied) and apparent recovery of N (net kg N taken up per kg N applied) were significantly (P < 0.05) greater in spring (September to October) and December than in autumn and winter, reflecting the seasonal pattern of pasture production for the region as reported by Suter et al. (2013) (Table 2). The lowest response to N occurred in July (average 7.7 ± 0.3 kg DM per kg of N for fertiliser treatments) where pasture was cut at 42 DAF (Table 2) due to low mineral N available for pasture growth (Fig. 3). There was a trend for increased agronomic efficiency and recovery of applied N with DMPP and GU relative to U for the May, June, July and 3 September harvests (Table 2), as well as for FPA in May and June. At other harvests this was not observed. The lack of significant difference in the agronomic efficiency and apparent recovery of N with the EEFs corresponds with the response seen in other studies (O’Connor et al. 2012; Misselbrook et al. 2014; Bell et al. 2015). This is most likely a result of a combination of sufficient N being available for pasture growth across all treatments and other factors (water, temperature, environment) limiting pasture production. This, plus the impact of management, was concluded as the main reason for variable responses in productivity and agronomic efficiency trials using the inhibitors in a meta-analysis by Abalos et al. (2014). Previous research has found productivity benefits from the use of urease and nitrification inhibitors, with the greatest response from the urease inhibitor and greater efficiency observed where lower N rates were used (30 kg N ha^–1 compared with 60 kg N ha^–1) (Zaman et al. 2013). For example, Watson et al. (1990) found that NBTPT increased both yield and N efficiency by 11%. Dawar et al. (2011) concluded that using a fine particle spray instead of granular urea and addition of NBTPT, both increased biomass production (by 27% and 38%, respectively) and N response efficiency (by 9 and 13 kg DM kg N^–1, respectively) compared with granular urea. This is a much greater response than observed here, which may indicate the importance of season, with Dawar et al. (2011) applying the fertiliser in spring. The June harvest showed that GU, DMPP and FPA all gave a production benefit when the time since fertilisation was extended, by either providing greater mineral N due to decreased NH₃ loss (GU and FPA) or retaining N in a form less prone to loss (DMPP) (Fig. 3). This reflects the ability of the inhibitors to work for extended periods (Fangueiro et al. 2009). However, the results reported are for a single event and due to the importance of environment and management on the performance of the EEFs (Abalos et al. 2014) extrapolation to different seasons and conditions is problematic.

Urease inhibitors target NH₃ loss and greatest benefits from these products are expected under conditions where NH₃ loss is high, i.e. moist, warm, windy conditions (Suter et al. 2013). No benefit from these products is expected during high rainfall periods, particularly when rain falls soon after fertiliser application, as NH₃ loss is low. Conversely, for the nitrification inhibitors we would expect to see the greatest benefit at times when there is high leaching and denitrification losses, i.e. during the wetter months. Studies on corn (Zea mays L.) found variable yield responses with a nitrification inhibitor concluding that profitable yield benefits were only likely to occur under conditions where spring and summer rainfalls were greater (40%) than the long-term average (Kyveryga and Blackmer 2014). Another study reported that one inhibitor (urease) was effective in increasing yield but not the other (nitrification) in the studied cropping system (Kawakami et al. 2012). This indicates that the dominant loss pathways will differ between systems, years and seasons. A synopsis of pasture trials conducted on 132 paddocks in 37 farms across New Zealand found overall biomass benefits from the use of a nitrification inhibitor (DCD) (19% increase), with variation around this dependent upon the region (Carey et al. 2012). The authors of the New Zealand study point out that this benefit is greater than reported on previous individual small plot studies (Carey et al. 2012).

This study found that urea boosted pasture productivity in southern Australian rainfed pasture systems by between 0.7 and 18 kg DM (vegetative) per kg of N applied, with soil moisture and climate dictating the seasonal response to N. Overall, N₂O emissions from applied fertiliser were low, but greater when conditions were conducive for denitrification (spring). Use of the nitrification inhibitor, DMPP, was consistently effective as a tool for mitigating N₂O emissions from urea across all seasons (autumn, winter and spring) with a 76% reduction recorded over eight months. However, DMPP did not significantly boost pasture productivity, with any observed biomass response influenced by seasonal conditions. The urease inhibitor, NBTPT (GU), is not an effective tool for mitigating N₂O emissions from urea, particularly during periods of high emissions (spring), with 53% more fertiliser-induced N₂O emissions produced relative to granular urea over the eight months. The importance of climate conditions for ammonia loss caused inconsistent productivity benefits from the use of GU. The potential for FPA to mitigate N₂O and increase pasture production requires investigation during periods of high emissions. These results highlight the need to use EEFs at times when their targeted loss pathway is important. This can be achieved by identifying the temporal changes in loss pathways (volatilisation, leaching and denitrification) based on climatic conditions and soil background N and C levels. To gain maximum productivity benefits from the use of EEFs other factors that restrict pasture growth, such as soil moisture and temperature, need to be effectively managed and N inputs should be lowered also.