Evaluating the effects of dicyandiamide (DCD) on nitrogen cycling and dry matter production in a 3-year trial on a dairy pasture in South Otago, New Zealand

Abstract

The nitrification inhibitor dicyandiamide (DCD) has been reported to significantly reduce nitrate () leaching and nitrous oxide (N₂O) emissions under urine patches in New Zealand trial conditions. However, results on the efficacy of DCD for increasing pasture production have been inconclusive. As part of a national series of nitrous oxide mitigation research (NOMR) trials, a 3-year study was conducted at a dairy farm in South Otago to examine the effect of DCD on pasture production and N leaching under mowing and/or (simulated) grazing conditions and on N₂O losses from dung and urine patches. Over the three trial years, DCD significantly inhibited nitrification in urine-amended soil, reducing peak concentrations in the soil by >50% and the N₂O emission factor from urine by 33%–72%. Nitrate concentrations in drainage under simulated grazing plots with or without DCD applications were low and consequently no effect of DCD was detected. The N₂O emissions and N leaching results suggested that high rates of denitrification occurred in the imperfectly draining soil, particularly in years 2 and 3 of the study. DCD significantly increased spring pasture dry matter (DM) yields from urine patches by 9%–21% in all years under mowing, but no significant annual effect on DM response was detected under either mowing or grazing. The observed increases in DM response in the urine patches under mowing were not detectable under the variable pasture responses under true grazing conditions.

Keywords:

• dicyandiamide
• DCD
• grazing
• nitrous oxide emissions
• nitrate leaching
• pasture production
• South Otago

Disclaimer

DCD has been withdrawn from use on farms in New Zealand and its use is now restricted to research undertaken on small plot trials or lysimeters. Protocols to ensure that DCD does not enter the food chain are supported by soil and plant testing for DCD residues.

Introduction

The pastoral dairy industry is New Zealand's largest agricultural export earner (Statistics NZ 2012) and its ongoing intensification and expansion is raising concerns about the environmental impacts and nitrogen (N) losses to water and air. In New Zealand's pastoral systems, urine deposited by grazing animals has been identified as the largest contributor to environmental N losses from dairy farms (Ledgard 2001). A single dairy cow urination can deposit a large amount of N equating to localised deposition of about 700–1000 kg N ha⁻¹ directly on to pasture (Haynes & Williams 1993). Much of this urine-N is quickly hydrolysed to ammonium () and transformed to nitrate () by soil bacteria known as nitrifiers. The amount of within the urine patch is well in excess of plant uptake and therefore prone to loss from the soil system via leaching, ammonia (NH₃) volatilisation, and dinitrogen (N₂) or nitrous oxide (N₂O) emissions. In New Zealand, leaching and N₂O emissions are the loss pathways that are of greatest concern (e.g. Ledgard 2001; Di & Cameron 2002). The potential for pollution to occur increases as the dairy industry expands and intensifies and animal numbers increase; therefore the fate of the excess produced in urine patches is subject to major research and regulatory scrutiny in New Zealand (e.g. Ledgard et al. 2009; de Klein & Monaghan 2011).

In recent years, the nitrification inhibitor dicyandiamide (DCD) has received much attention as a technology to reduce soil concentrations in urine patches deposited on to pasture in New Zealand (e.g. Di & Cameron 2002; Smith et al. 2008; Monaghan et al. 2009; de Klein et al. 2011). DCD temporarily inhibits the first step of the nitrification process (Di et al. 2010), thereby inhibiting the production of and thus reducing the risk of subsequent N losses. New Zealand research has shown that DCD can reduce N losses via leaching by 16%–83% (studies summarised in Cameron et al. 2014) and N₂O emissions by 0%–86% (studies summarised in de Klein et al. 2011). The retention of N in the form in soil prolongs the availability of N for pasture plant uptake and could thus potentially increase pasture production. In addition, an increase in pasture production is an economic incentive to use DCD as a farm management tool. The results to date on the effect of DCD on pasture production are varied, with >20% production increases reported in a trial by Moir et al. (2007) and nil effects in others (Monaghan et al. 2009; MacDonald et al. 2010). Following application to pasture, DCD is gradually degraded or leached from soil, and its persistence in soil is influenced by factors such as temperature, moisture, organic matter and pH (Amberger 1989; Kelliher et al. 2008; Menneer et al. 2008; Kim et al. 2012). The efficacy of DCD to reduce N losses may therefore vary according to local soil and climate factors; the timing of application in relation to these factors may have a large influence on the subsequent performance of DCD as an effective farm management tool.

To measure the efficacy of DCD for reducing N losses and increasing pasture production in dairy systems under a range of environmental conditions, a coordinated series of nitrous oxide mitigation research (NOMR) trials was conducted over 3 years, in the four key dairying regions in New Zealand. The results of trials in Waikato, Manawatu and Canterbury are presented in associated papers (Ledgard et al. 2014; Kim et al. 2014 and Cameron et al. 2014, respectively). This paper reports the results from the NOMR trial in South Otago.

The main objectives of the trial were to measure the effect of DCD applications to a South Otago poorly drained soil on:

• N₂O emissions from real cattle urine patches (gas chamber trial, 2009–11).

• leaching from simulated grazing plots (drainage trial, 2010–11).

• Pasture DM response, soil mineral N and soil DCD decay rates from synthetic cow urine applied to small plots (mowing trial, 2009–11).

• Pasture dry matter response under grazing conditions (grazing trial, 2009–11).

Materials and methods

Site description and climate data

The South Otago NOMR trial was conducted at the Telford Polytechnic Dairy Farm, from April 2009 to March 2012 (Gillingham et al. 2012). The trial site was located on Tokomairiro deep silt loam (Fragic Perch-gley Pallic soil, NZ Soil Classification, or Fragiochrept, USDA Soil Taxonomy [Hewitt 1998]), an imperfectly drained soil with moderate fertility and high water availability. The trial paddock (total area 2.17 ha) was cultivated and sown to a perennial ryegrass/white clover pasture in February 2009. General soil fertility and soil physical parameters of the trial paddock in each year are presented in Tables 1 and 2. Soil physical parameters were measured as described by Drewry et al. (1999). Rainfall and soil temperature (10 cm depth) data were obtained from an NIWA meteorological site located about 2 km from the trial site.

Table 1 Mean soil fertility measurements (0–7.5 cm depth) at the Telford NOMR 2009–11 trial site; n = 6.

Trial year	Olsen P (µg mL^−1)	K (MAF QT)	S(SO4) (ppm)	Ca (MAF QT)	Mg (MAF QT)	Na (MAF QT)	Organic C (%)	Organic matter (%)	Total Kjeldahl N (%)	pH
1	26	4	10	13	17	7	4.0	7	0.31	6.2
2	18	4	9	10	14	6	3.5	5.9	0.27	6.2
3	20	3	5	10	17	6	4.0	6.9	nd	5.9

nd, not determined.

Table 2 Mean soil physical measurements (0–5 cm depth) at the Telford NOMR 2009–11 trial site; samples taken within the mowing trial areas in trial years 1, 2 and 3; n = 6.

Trial year	Bulk density (g cm^−3)	Microporosity (vol% pores <300 µm)	Macroporosity (vol% pores >30 µm)	Total porosity (vol%)
1	0.94	45.5	18.9	64.4
2	0.96	47.1	16.6	63.6
3	0.90	59.6	6.4	66.0

The NOMR trial consisted of four component studies: an N₂O gas chamber trial, a drainage trial, a mowing trial and a grazing trial. All components examined the effect of DCD on various aspects of N cycling in a paddock within an operating farm system. The treatments, measurements and methods for each trial are described in sections below and the treatment rates and dates are summarised in Table 3.

Table 3 Overview of experimental treatments and measurements at the Telford NOMR 2009–11 trial site. In square brackets are the N application rates (kg N ha⁻¹) and dates. DCD was applied at 10 kg DCD ha⁻¹ in all trials.

Trial	2009–10 (yr 1)	2010–11 (yr 2)	2011–12 (yr 3)
N2O	Control Urine [547, 22 May] Urine+1DCD [22 May] Dung [1001, 22 May] Dung+1DCD [22 May]	Control Control+2DCD [17 May; 17 Aug] Urine [616, 17 May] Urine+2DCD [17 May; 17 Aug]	Control Urine [710, 20 Apr] Urine+2DCD [20 Apr; 20 Jul] Urine+5DCD [20 Apr; 23 May; 27 Jun; 20 Jul; 23 Aug]
N leaching		Simulated ‘grazing’^a +3DCD [15 Mar; 19 May; 16 Sep]	Simulated ‘grazing’^a +3DCD [17 Mar; 21 Apr; 20 Jul]
Mowing	Control^b Urine^b [700, 22 May] Urine+1DCD^b [22 May]	Control^b Control+2DCD^b [19 May; 9 Sep] Urine^b [700, 19 May] Urine+2DCD^b [19 May; 9 Sep] Urine-edge Urine-edge+2DCD Urea [25, 19 May; 37, 9 Sep] Urea+2DCD [19 May; 9 Sep]	Control Urine^b [700, 13 Apr] Urine+1DCD^b [14 Apr] Urine+2DCD^b (autumn/winter) [14 Apr; 21 Jul] Urine+3DCD [14 Apr; 21 Jun; 26 Aug] Urine+5DCD^b [14 Apr; 20 May; 21 Jun; 21 Jul; 26 Aug; 22 Sep] Urea [25, 20 Aug; 25, 22 Sep]Urea+2DCD (spring) [26 Aug; 22 Sep]
Grazing	Control +2DCD [21 May; 14 Sep]	Control +2DCD [18 May; 13 Sep]	Control +2DCD (autumn/winter) [11 Apr; 20 Jul] +5DCD [13 Mar; 11 Apr; 20 Jul; 21 Sep; 27 Oct] +2DCD (spring) [21 Sep; 27 Oct]

^a Cows were excluded from the N leaching plots, but each time the grazing trial was grazed, two real cow urine patches were applied at a rate of 700 kg N ha⁻¹ per patch.

^b Treatments sampled for soil analyses.

Gas chamber trial to measure DCD effect on N₂O emissions

Nitrous oxide measurements were made using a standardised chamber methodology (de Klein et al. 2003) in a small trial area that was fenced off at least 4 weeks prior to treatment application. A different area was used in each trial year. Within this area, stainless steel rings (0.25 m diameter in year 1 and 0.5 m diameter in years 2 and 3) that were fitted with a 2 cm tall, U-shaped ‘water trough’ flange were inserted 10 cm into the soil and left to ‘settle’ for at least 1 week. Urine and/or dung and DCD treatments were then applied to the rings in a randomised block design (n = 6) (Table 3).

Fresh urine or dung was collected from a local dairy farm 1–2 days prior to application, chilled at 4 °C and the total N concentration analysed. The urine was evenly applied within the stainless steel rings at a rate of 10 L m⁻² (equivalent to 700 kg N ha⁻¹), while 0.9 kg fresh cow dung was applied to a 20 cm diameter circle within the rings. DCD was dissolved in water (100 g DCD 8 L⁻¹) and sprayed on to the plots after the urine or dung applications at a rate equivalent to 10 kg DCD ha⁻¹. An equivalent amount of water was sprayed on to the control plots.

Gas sampling started on the day of treatment application in autumn (April or May) and continued twice weekly for 6 weeks, thereafter weekly for another 6 weeks and then fortnightly. Extra sampling was conducted after significant rainfall events (>10 mm) throughout the entire measurement period. Measurements continued for 7–8 months until the N₂O emissions from all treatments had receded to background levels.

At each sampling occasion, the water trough flanges were carefully filled with water (to avoid spillage within the chamber internal area) and a chamber cover with two open ports was placed into the trough. After placement of the cover, the ports were closed with Suba-Seals to provide an airtight cover. Headspace samples were taken at 0, 20 and 40 min (t0, t20 and t40) for the first 6 weeks after treatment application. For the remainder of each trial, headspace samples were taken at t0 and t30.

Nitrate leaching trial to determine DCD effect on loss in drainage

The leaching trial consisted of 20 plots (2 × 10 m) which were each hydrologically isolated by burying plastic polythene sheeting vertically into the ground to 1 m depth around the three up-slope boundaries of each plot. In the centre of each plot, a narrow trench was dug to 70 cm depth along the 10 m length of each plot, using a chain-trencher. A slotted drainage pipe was placed at the bottom of this trench, which was then backfilled with c. 40 cm of pea gravel and c. 30 cm of topsoil. The drainage pipes were positioned on a clay pan to capture subsurface drainage water from within each plot, with minimal loss below the pan. At the down-slope boundary of each plot, the drainage water collected through the slotted pipe exited the plots and was captured using tipping buckets and a siphon device as described by Monaghan et al. (2009). Subsamples were frozen and later analysed for and concentrations by flow injection analysis.

Stock was excluded from the trial area, but each plot received two simulated urine patches by applying real cow urine with watering cans within two 50 cm diameter rings (i.e. 2% of each plot area received urine at any one time). The urine was applied at the same rate as for the gas chamber trial (i.e. 10 L m⁻², at a rate of 700 kg N ha⁻¹ for each patch). The timing of the urine application coincided with the grazing rotation in the grazing trial plots (see below). Prior to urine application, all drainage plots were harvested using a push mower (within ±1 week of the grazing event) down to a residual of c. 1400 kg DM ha⁻¹. Plots were paired to give 10 blocks of two plots with two treatments randomly applied within blocks; control (no DCD) or +3DCD.

DCD was applied three times throughout the season, directly following the urine applications in early autumn, late autumn and early spring, at 10 kg DCD ha⁻¹ per application to the appropriate plots using a high-pressure handheld sprayer (Table 3).

Fortnightly soil sampling (0–10 and 10–20 cm depth) occurred in the 10 +3DCD plots until soil DCD concentrations were below detection limit (0.003 ppm) for at least 1 month. Soil DCD content was determined as for the mowing trial (see below).

Mowing trial to determine the DCD effect on DM production, soil mineral N and soil DCD

A mowing trial was established in autumn each year. For each trial year, a different area of the paddock was fenced off and used to measure annual pasture production, and soil , and DCD concentrations for 5–8 months, until all treatments had receded to background levels. The trial consisted of 51 plots (1.25 × 2 m²) in year 1 and 136 plots (0.5 × 5 m²) in years 2 and 3. The plots were arranged into 17 blocks with three (year 1) or eight (years 2 and 3) randomly applied treatments in each block with a buffer strip (approx. 0.5 m width) between each plot. Synthetic urine (700 kg N ha⁻¹ equivalent, urine composition according to de Klein et al. 2003) was applied to plots using watering cans. Fertiliser N (urea) was applied by hand to plots; fertiliser N application rates were determined according to Telford farm practice, which varied slightly between trial years. On the same day or the day following urine or urea application, DCD was applied (10 kg DCD ha⁻¹ equivalent) to appropriate plots using a high-pressure hand sprayer. Re-applications of DCD (10 kg DCD ha⁻¹ equivalent) were specific to each trial year. In the first year, DCD applications were applied during the autumn/winter period only. In following years, treatments were expanded to include DCD applications during the autumn, winter and spring periods (Table 3). In year 2, a treatment was included to assess the effect of DCD on pasture responses at the outer edges of the urine-treated area (urine-edge effect). The urine-edge effect plots were located directly adjacent to the urine plots (i.e. without a buffer strip between the plots) and DM production determined as per the protocol described below.

Pasture DM production was determined at regular intervals by mowing one or two strips in each plot. In the urine-edge effect plots, these strips were located at the boundary line with the urine-treated plots. The collected herbage was weighed and a subsample dried to determine percentage DM. In year 1, the mowing dates were based on the grazing rotations of the grazing trial and, as a result, pasture covers were very high at the time of mowing during spring and early summer. In subsequent years, the plots were therefore harvested as pasture covers reached (or as close as practical to) a target cover of 2800 kg DM ha⁻¹.

Soil samples (0–10 and 10–20 cm depths) were taken from six randomly selected blocks of selected treatments (see Table 3). Sampling occurred at regular intervals and lasted for 7–10 months until average soil DCD and soil mineral N reached background levels (i.e. not significantly different from control plots, P > 0.05). Samples were extracted in the laboratory using 2M potassium chloride (KCl) and sent to Ruakura (Hamilton, AgResearch Ltd) for soil and analyses by flow injection analysis. Separate subsamples were extracted with distilled water and analysed for DCD by high performance liquid chromatography (HPLC, Shimadzu HIC-6A) with adjustment for incomplete recovery as described by Welten et al. (2012).

Grazing trial to determine DCD effect on pasture production

The grazing trial was established to measure pasture production in response to DCD application under grazing conditions. The trial was conducted in the same area of the paddock for all years and, in years 1 and 2, consisted of 70 plots (100 m² per plot) with two treatments (control and +DCD; n = 35). DCD was applied twice per year (2 × 10 kg DCD ha⁻¹), once in autumn and once in spring (Table 3). The treatments were randomly applied within blocks. In year 3, all plots were divided into two, to establish 140 plots (50 m² per plot) and enable an additional two treatments to be included: DCD applied five times per year and DCD applied twice in spring (Table 3). For each DCD application, dissolved DCD (10 kg DCD in 800 L) was applied at 10 kg DCD ha⁻¹ using high-pressure knapsack sprayers. The area was grazed as closely as possible to the farm rotation practice (i.e. grazed at a target herbage mass of c. 2800 kg DM ha⁻¹ down to a residual target of c. 1450 kg DM ha⁻¹).

To determine pasture yield (kg DM ha⁻¹), pasture cover was measured at pre- and post-grazing for each grazing event using a Farmworks rising plate meter (Farmworks 2008), by taking c. 80 readings per plot. Pasture cover (kg DM ha⁻¹) was estimated from the plate meter ‘clicks’ using the standard DairyNZ all season's equation (Farmworks 2008).

Results

Climate data

The cumulative rainfall total in years 1 and 2 were similar to the long-term average of 730 mm (NIWA Meteorological Site No. 692706). In year 3, total rainfall marginally exceeded this long-term average (Fig. 1). In year 1 there was a substantial amount of rain prior to and immediately following the cultivation of the paddock and subsequent trial establishment (May 2009). This frequent rainfall pattern continued for much of the duration of the trial, particularly in the initial 12 weeks in year 1. In year 3 there was approximately 100 mm more rainfall prior to the start of the trial in autumn compared with previous trial years.

Figure 1 Rainfall data for all trial years at the Telford trial site. Data sourced from the NIWA meteorological site approximately 2 km from the trial site.

Soil temperatures at 10 cm depth were typically lowest during July and August each year (3-year average 4.6 °C) and highest during December (3-year average 14.6 °C). It should be noted that January and February recorded the highest average soil temperatures in all trial years (3-year average 14.9 °C), however most gas and soil sampling had ceased by this time (Fig. 2). Due to the earlier start date in trial year 3 (April, rather than May as in trial years 1 and 2), the average soil temperature was higher at the start of trial year 3 (6.6, 7.7 and 10.1 °C in years 1, 2 and 3, respectively).

Figure 2 Daily average soil temperature (0–10 cm depth) for the Telford trial site for all trial years. Data sourced from the NIWA meteorological station approximately 2 km from the trial site.

Soil , and DCD concentrations

The application of DCD inhibited nitrification in soil, thus reducing soil concentrations (0–20 cm depth) in urine patches for 8–16 weeks in all years compared with urine-only treatments (Fig. 3). In the urine-only treatment, soil concentrations peaked 3–5 weeks following treatment application (61.6, 69.3 and 87.3 kg NO₃-N ha⁻¹ in years 1, 2 and 3, respectively). DCD application reduced these peaks by >50% (24.4, 28.6 and 32.1, 27.4, 18.4 kg NO₃-N ha⁻¹ in years 1, 2 and 3, with the latter reporting values for one, two and five DCD applications, respectively). Soil concentrations also took longer to decline to background levels by a few weeks in the DCD treatments (Fig. 3). An spike in September for all treatments in year 3 cannot be readily explained.

Figure 3 Soil nitrate () concentrations (kg N ha⁻¹) in soil (0–20 cm depth) for trial years 1, 2 and 3 at Telford. Vertical bars represent the SEM; n = 10.

The application of urine substantially increased concentrations for 8–12 weeks. When DCD was applied, soil concentrations were higher than in the urine-only treatment and persisted for longer; from about 3 up to 12 weeks following application (Fig. 4). Increasing the number of DCD applications from one to five in year 3 resulted in little difference in soil concentrations.

Figure 4 Soil ammonium () concentrations (kg N ha⁻¹) in soil (0–20 cm depth) in trial years 1, 2 and 3 at Telford. Vertical bars represent the SEM; n = 10.

Soil DCD concentrations (0–10 cm depth) in the mowing trial ranged from 3–7 kg DCD ha⁻¹ immediately after application and reduced to non-detectable levels after 8–12 weeks. The applied DCD was never fully recovered in any of the trial years (Fig. 5). The highest DCD concentrations were measured in the first 2 weeks post-application in year 1 and following the third application in the +5DCD treatment in year 3 (7–8 kg DCD ha⁻¹ at 0–10 cm depth, Fig. 5). Apart from these two occasions, soil DCD concentrations rarely exceeded 5 kg DCD ha⁻¹ during any of the trial years. DCD concentrations in the 10–20 cm soil layer rarely exceeded 2 kg DCD ha⁻¹ across all years and treatments.

Figure 5 DCD concentrations (kg DCD ha⁻¹) in soil (0–10 cm depth) in trial years 1, 2 and 3 at Telford. Vertical bars represent the SEM; n = 10.

Nitrous oxide emissions

The application of DCD to urine patches was consistently effective at reducing peak daily N₂O emissions by 14%, 70% and 42% or 50% in years 1, 2 and 3 (two or five DCD applications, respectively, in year 3, Fig. 6). In all years, daily N₂O emissions peaked within the first 8 weeks after treatment application and returned to background levels within 3–4 months. In year 2, a second N₂O emission ‘envelope’ occurred approximately 12 weeks after urine application, which made a large contribution to the emissions total that year. In year 3, the peak N₂O emissions were at least threefold higher than in the previous years.

Figure 6 Daily N₂O emissions (g N₂O-N ha⁻¹ d⁻¹) from dairy cow urine and dung treated with or without DCD at the Telford trial site. Note the differences in the y-axis scale between the trial years. Vertical bars are the SEM; n = 6.

Total N₂O losses from urine patches varied between years (10, 23 and 41 kg N ha⁻¹ in years 1, 2 and 3, respectively) and were significantly higher in year 3 compared with other years (P < 0.01). In year 1, the N₂O emission factors (expressed as N₂O-N emitted as a percentage of N applied) from autumn-applied urine patches were significantly reduced (P < 0.01) by 72% following a DCD application in autumn (Table 4). In years 2 and 3, the N₂O emission factors from autumn-applied urine patches were significantly reduced by 54% (P < 0.01) and 33% (P < 0.05), respectively, following two DCD applications, one in autumn and one in winter. Increasing the number of DCD applications from two to five in year 3 resulted in a further reduction in N₂O emissions, with DCD reducing the emission factor by 50% (P < 0.001).

Table 4 N₂O emissions and the N₂O emission factor for the N₂O trial treatments at the Telford NOMR trial site for trial years 1 (2009–10), 2 (2010–11) and 3 (2011–12). Numbers in parenthesis are SEM; n = 6.

Trial yr/Treatment	Total N2O loss (kg N ha^−1)	Emission factor (% of N applied)	DCD reduction (%)
1
Control	4.1 (1.9)
Dung	5.8 (1.3)	0.17
Dung+DCD	4.6 (2.0)	0.05	71^NS
Urine	10.3 (1.8)	1.15
Urine+DCD	5.8 (1.2)	0.32	72*
2
Control	0.3 (0.1)
Control+2DCD	0.4 (0.2)
Urine	22.9 (4.4)	3.68
Urine+2DCD	10.9 (5.9)	1.70	54**
3
Control	1.4 (0.4)
Urine	40.8 (2.5)	5.5
Urine+2DCD	27.9 (5.9)	3.7	33*
Urine+5DCD	21.1 (3.3)	2.8	50**

NS, not significant.

* P < 0.05.

** P < 0.01.

In year 1, total N₂O emissions from dung treatments were low and similar to control treatments (5.8 and 4.1 kg N ha⁻¹ for dung and control treatments, respectively) and applying DCD did not significantly reduce those total emissions (Table 4).

Nitrate losses in drainage

Cumulative leaching losses in trial years 2 and 3 from the drainage trial site ranged from 9–16 kg N ha⁻¹ across treatments and trial years (Table 5). The application of DCD reduced the load in drainage by 8% and 11% in years 2 and 3, respectively, but these reductions were not statistically significant. There were no noticeable effects of DCD application on concentrations at individual drainage events, except on one occasion in year 2 of the trial, where the concentration of the DCD plots was significantly lower than that of the control plots (data not shown).

Table 5 Total annual pasture yield (t DM ha⁻¹) and nitrate leachate load in drainage (kg NO₃-N ha⁻¹) from the Telford NOMR leaching trial, years 2 (2010–11) and 3 (2011–12). Numbers in parenthesis are the SEM; n = 10.

Trial yr	Treatment	Total pasture yield (t DM ha^−1)	Pasture response (%)	Annual N-load in drainage (kg N ha^−1)	DCD reduction (%)
2	Control	9.2 (0.03)		16.3 (6.6)
	+3DCD	9.5 (0.03)	3.3^NS	15.0 (4.8)	7.8^NS
3	Control	13.0 (0.2)		9.8 (1.7)
	+3DCD	13.2 (0.2)	1.2^NS	8.8 (0.9)	10.9^NS

NS, not significant.

Pasture dry matter response

In the mowing trial, DCD significantly increased spring pasture dry matter (DM) in the urine-treated plots (Table 6). In year 1, spring and annual DM from the urine+DCD plots were 9% and 5% higher (P < 0.05), respectively, than from the urine-only plots. In year 2, spring and annual DM responses of the urine plots were increased by 12% and 5% (P < 0.05), respectively, due to DCD applications. In the urine-edge effect plots, DCD also significantly increased spring pasture DM (by 9%, P < 0.05), but not the annual DM production. In year 3, DCD applications increased spring and annual DM in the urine plots by 13%–21% (P < 0.05) and by 6%–11%, respectively, depending on the DCD treatment (one, two, three or five applications; P < 0.05 for three and five DCD applications only). In general, the DM responses increased with the number of DCD applications (Table 6). There were no significant pasture responses during summer. In the urea-treated plots, DCD application did not significantly affect DM response, with the urea and urea+DCD plots yielding the same amount in both year 2 and 3 (Table 6). Similarly, in year 2, DCD application did not increase DM yield of the control treatment.

Table 6 Total and seasonal pasture yield (t DM ha⁻¹) from the Telford NOMR small plot mowing trial, years 1 (2009–10), 2 (2010–11) and 3 (2011–12). Numbers in parenthesis are the SEM; n = 17.

Trial yr/Treatment	Annual pasture yield (t DM ha^−1)	DCD response (%)	^aSpring pasture yield (t DM ha^-1)	DCD response (%)	^bSummer pasture yield (t DM ha^−1)	DCD response (%)
1 (May 2009–Apr 2010)
Control	9.6		4.8		3.1
Urine	11.4		6.4		3.2
Urine+1DCD	11.9	5*	7.0	9*	3.2	0
2 (May 2010–Feb 2011)
Control	10.4 (0.1)		2.3 (0.0)		4.7 (0.1)
Control+2DCD	10.0 (0.1)	−4^NS	2.0 (0.0)	−15^NS	4.7 (0.1)	0
Urine	13.4 (0.1)		4.6 (0.1)		5.0 (0.1)
Urine+2DCD	14.0 (0.1)	5*	5.1 (0.1)	12*	5.1 (0.1)	0
Urine-edge	11.3 (0.1)		3.0 (0.1)		4.8 (0.1)
Urine-edge+2DCD	11.6 (0.1)	3^NS	3.3 (0.1)	9*	4.9 (0.1)	0
Urea	11.0 (0.1)		3.6 (0.0)		4.2 (0.1)
Urea+2DCD	10.9 (0.1)	−1^NS	3.4 (0.0)	−6^NS	4.2 (0.1)	0
3 (Apr 2011–Apr 2012)
Control	8.4 (0.1)		2.8 (0.0)		2.6 (0.1)
Urine	10.9 (0.1)		3.5 (0.0)		2.1 (0.1)
Urine+1DCD	11.3 (0.1)	7^NS	4.0 (0.0)	13*	2.1 (0.1)	0
Urine+2DCD (autumn/winter)	11.1 (0.1)	6^NS	4.0 (0.0)	13*	1.9 (0.1)	−10^NS
Urine+3DCD	11.9 (0.1)	11*	4.1 (0.0)	16*	2.2 (0.1)	5^NS
Urine+5DCD	12.0 (0.1)	11*	4.3 (0.1)	21*	2.2 (0.1)	5^NS
Urea	9.3 (0.1)		3.9 (0.0)		2.1 (0.1)
Urea+2DCD (spring)	9.3 (0.1)	0	3.9 (0.0)	0	2.2 (0.1)	5^NS

NS, not significant.

^a Spring: 1 September–30 November.

^b Summer: 1 December–28 February.

* P < 0.05.

In the grazed plots, DCD application did not significantly increase pasture yields on an individual, seasonal or annual basis (Table 7). The total pasture yields were difficult to compare between years due to the variation in the measurement periods and subsequent grazing rotations (Table 7); however, there is a consistent lack of DCD treatment effect across all years.

Table 7 Annual and seasonal pasture yield (t DM ha⁻¹) from the Telford NOMR Grazing trial, years 1 (2009–10), 2 (2010–11) and 3 (2011–12). Numbers in parenthesis are the SEM; n = 35.

Trial yr/Treatment	Total pasture yield (t DM ha^−1)	Pasture response (%)	Spring cumulative pasture yield^b (t DM ha^−1)	Summer cumulative pasture yield^c (t DM ha^−1)	Number of grazing episodes^d
1 (Sep 2009–Apr 2010)					8
Control	10.4^a (0.2)		5.4^a (0.03)	4.0 (0.05)
+2DCD	10.5^a (0.2)	1.0^NS	5.5^a (0.03)	4.1 (0.04)
2 (Sep 2010–Feb 2011)					6
Control	5.9 (0.03)		2.9 (0.03)	3.0 (0.02)
+2DCD	6.0 (0.03)	1.3^NS	2.9 (0.03)	3.0 (0.03)
3 (Apr 2011–Apr 2012)					10
Control	9.1 (0.03)		3.6 (0.03)	2.4 (0.02)
+2DCD (autumn/winter)	9.0 (0.03)	−1.3^NS	3.6 (0.03)	2.4(0.02)
+5DCD	9.6 (0.03)	5.4^NS	3.8 (0.04)	2.5 (0.03)
+2DCD (spring)	9.0 (0.02)	−1.4^NS	3.4 (0.02)	2.5 (0.02)

NS, Not significant.

^a Yield includes an independent measurement in October 2009 made within the trial paddock to replace lost data.

^b Spring: 1 September–30 November.

^c Summer: 1 December–28 February.

^d Average grazing rotation over the measurement period was 30, 30 and 37 days in trial years 1, 2 and 3, respectively.

Discussion

Soil , and DCD concentrations

The application of DCD significantly inhibited soil nitrification rates across all treatments in all years, as evidenced from the soil mineral N measurements. In general, concentrations in the DCD treatments were maintained at higher levels for 4–7 weeks longer than in the no-DCD treatments. This coincided with a >50% reduction in peak concentration following urine+DCD application over the same period. However, thereafter, soil levels in the DCD plots were elevated above control levels for up to 4 weeks longer than in the urine-only plots, which indicated that DCD effectiveness slowly decreased over time. These observations are consistent with previous findings (e.g. Smith et al. 2008; Di & Cameron 2012; Monaghan et al. 2013) and with the soil DCD measurements in our trial. Soil DCD concentrations declined to zero within 8–12 weeks after application, except in year 1 where DCD persisted at detectable concentrations for up to 18 weeks. The cooler soil temperatures in year 1 (Fig. 2) probably resulted in a slower degradation of DCD (Kelliher et al. 2008) compared with the other years. Immediately following application, DCD recovery in soil samples was not always achieved. The interception of DCD through adhesion to grass surfaces has been suggested as a cause for the low recovery rate (Gillingham et al. 2012). This is supported by the fact that DCD recovery commonly improved a few weeks following application (Fig. 5), suggesting that DCD may have been washed from grass surfaces over time. Incomplete DCD recovery could also have been due to DCD adsorption on soil organic matter (Welten et al. 2012). However, this effect would probably have been small because Welten et al. (2012) measured DCD recovery efficiencies ranging from 65% in a silty peat soil to 94% in the Tokomairiro soil of our study. In years 2 and 3, DCD recovery rarely exceeded 5 kg DCD ha⁻¹ in soil, even immediately following (re-)application(s). Antecedent weather and soil conditions may have modified application rate and subsequent efficacy of DCD to inhibit nitrification. Gillingham et al. (2012) indicated that rainfall and high soil moisture immediately prior to DCD application may have increased DCD leaching losses, although rainfall and soil moisture were high at the start of year 1 also. Field observations suggest that high wind may also affect application by dispersing DCD spray on to unintended areas. Soil and N₂O emissions results over all trial years suggest however that there was sufficient DCD present to inhibit nitrification for at least 8 weeks, and in some cases up to 12–15 weeks.

Nitrous oxide emissions

The inhibition of nitrification rates due to DCD application significantly reduced N₂O emissions from urine patches in all treatments and all years. Reductions in total N₂O losses ranged from 33%–72% (56% average over 3 years), which are within the range of other New Zealand studies (de Klein et al. 2011; Cameron et al. 2014; Kim et al. 2014; Ledgard et al. 2014). The majority of N₂O emissions occurred in the 6–8 weeks following treatment applications, and in year 3, DCD effectiveness in reducing N₂O emissions was higher with five compared with two applications. However, at c. NZ$100 ha⁻¹ per application, the costs of five DCD applications will most likely outweigh the additional benefit of reduced N₂O emissions. In year 1, total N₂O emissions from the dung treatment were low but DCD application still resulted in a similar proportional reduction in emissions as for the urine treatment. Total N₂O emissions from urine patches showed a large inter-annual variation, with emissions from patches applied in years 2 and 3 being two to four times higher than from urine applied in year 1. The N₂O emission factors for urine and dung (calculated as the amount of N₂O-N emitted as a percentage of the total urine-N applied; de Klein et al. [2011]) in year 1 were similar to the New Zealand Greenhouse Gas Inventory values of 1% and 0.25%, respectively (Ministry for the Environment 2013). In years 2 and 3, the emission factors for urine were four to six times higher than the New Zealand value and in the very top range of (or even higher than) values observed in all previous New Zealand studies (Kelliher et al. 2014). This inter-annual variation could be due to differences in weather conditions (warmer and wetter, particularly in year 3) or due to differences in soil biological and/or physical conditions in the trial areas. Although the trial areas for the three years were all located within the same paddock, soil micro-porosity was significantly higher in the year 3 trial area (Table 2). Additional measurements also suggested that the earthworm population was higher in the trial area in year 3 (data not shown), which could also have affected the N₂O emissions (e.g. Lubbers et al. 2013).

Nitrate leaching

DCD applications did not significantly affect nitrate leaching losses in drainage water. The estimated annual N leaching losses at our study site (9–16 kg N ha⁻¹) were low compared with N leaching losses measured in other grass-clover based grazed systems (Ledgard et al. 2009) but very similar to those measured in a 4-year study in Southland (7–15 kg N ha⁻¹; Monaghan et al. 2009). However, in their study, Monaghan et al. (2009) did find that DCD significantly reduced N leaching losses by 21%–56%. The low N leaching losses at our study site are also supported by relatively low peak soil concentrations of <75 kg N ha⁻¹ at 0–20 cm depth in the mowing trial. In comparison, maximum soil concentrations at the Tokanui NOMR site (Ledgard et al. 2014) were c. 150 kg N ha⁻¹. This may suggest that denitrification rates were higher in our imperfectly draining soil compared with the free-draining allophanic soil at Tokanui, thus minimising the risk of N leaching. The relatively high N₂O emission rates at our site, especially in years 2 and 3, would support this conclusion.

Pasture dry matter responses

In the mowing trial, DCD application significantly (P < 0.05) increased spring pasture responses from urine by 9%–12% across all years. Annual increases in DM response from DCD in urine treatments ranged from 3%–11% but these were not statistically significant. These DCD-induced DM responses are similar to those found following urine application in other mowing trails (e.g. Menneer et al. 2008; Sprosen et al. 2009; Cameron et al. 2014; Kim et al. 2014; Ledgard et al. 2014). In the grazing trial, DCD application did not increase pasture yield at individual grazing events, nor cumulative seasonal or annual DM yields. These results are in contrast to other South Island grazing trials (Moir et al. 2007; Carey et al. 2012), but similar to those obtained in some or all trial years of the other NOMR sites (Cameron et al. 2014; Kim et al. 2014; Ledgard et al. 2014). Similarly, in their 4-year grazing study in Southland, Monaghan et al. (2009) did not find any significant effect of DCD application on pasture responses. These authors suggested that the DM response appeared to be consistent with what could be expected based on the amount of additional N available due to reduced gaseous and leaching losses, and from decomposition of DCD itself. The incomplete DCD recovery that was observed in our study could have influenced the DCD effect on pasture responses in the grazing trial. However, the mowing trial results show that despite this incomplete DCD recovery, DM responses following urine application were significantly higher in the DCD plots. In addition, the soil mineral N and N₂O emission results indicated that sufficient DCD was present to inhibit nitrification for up to 3–4 months. It is therefore unlikely that the lack of pasture response in the grazing trial was due to reduced DCD availability in the soil. DM responses in a grazing trial are by definition much more variable than in mowing trials and small increases in DM production may be difficult to detect with current pasture measurement techniques.

Conclusion

DCD application(s) to a South Otago imperfectly drained soil inhibited nitrification rates and significantly reduced the N₂O emission factor from urine patches by 33%–72%. N₂O emissions from the site were relatively high, especially in years 2 and 3, which suggested that soil and climatic conditions were favourable for denitrification. As a result, N leaching losses were relatively low and no significant effect of DCD could be detected, despite soil concentrations in urine patches being significantly reduced following DCD application. DCD applications significantly increased spring and annual pasture responses in urine patches in the mowing trial, but no significant pasture responses were measured under grazing. Increasing the number of DCD applications during the season increased the effectiveness of DCD on increased pasture response and reduced N₂O emissions from urine patches.

Acknowledgements

The authors wish to thank Telford farm staff and the Invermay technical team for their assistance in the field and lab. These trials were conducted under the nitrous oxide mitigation research (NOMR) programme, jointly funded by the Ministry for Primary Industries (MPI; formerly the Ministry of Agriculture and Forestry), DairyNZ, Fonterra Co-operative Group Ltd, Ravensdown Fertiliser Co-operative Ltd, Ballance Agri-Nutrients Ltd and the Pastoral Greenhouse Gas Research Consortium (PGgRc).