Nitrogen inputs and outputs for New Zealand from 1990 to 2010 at national and regional scales

Abstract

Reactive nitrogen (N) is increasingly added to the New Zealand environment because of increased sales of N fertilizer and increased human population. The Greenhouse Gas Inventory now reports in detail on changes for N losses from grazing animals from 1990 to 2010. Using animal numbers, we made assessments of N inputs and outputs for the 16 regions of New Zealand for 1990, 2001 and 2010 to assess temporal trends. Fertilizer sales have increased from 46 Gg N in 1990 to 329 Gg N in 2010, which leads to reduced biological N fixation by pastures. The import of oil-palm kernel has increased from zero to about 28 Gg N in 2010. Total N inputs are estimated to have increased from 689 Gg to 951 Gg N. The outputs of produce, leachate, gasses and sediment have increased from 771 to 866 Gg N; outputs to rivers may increase further if increases in outputs lag behind increases in inputs. Many of the inputs and outputs are well constrained because animal numbers have been used rather than land area, but uncertainties do exist for specific land-use classes. For example, the area of lifestyle blocks is approaching 800,000 ha and there is uncertainty regarding N inputs and outputs in this land use. There are also uncertainties in the amount of N fixation, the N loss by leaching in any one year, the amounts and fate of dissolved organic N, and the N content of eroded sediment. These uncertainties need to be resolved so that the amount of N stored in soils can be assessed. It seems likely that the N concentration of soils under dairying is increasing relative to the carbon concentration (i.e. soil C/N ratios are declining) but there is conflicting evidence as to whether the total N (and C) in these soils is increasing or decreasing.

Keywords:

• ammonia
• clover
• denitrification
• erosion
• nitrate
• rivers
• sediment

Introduction

Much of the New Zealand economy depends on agriculture, in particular pastoral agriculture, and the growth of pasture largely depends on water and nitrogen (N). Nitrogen is added to the soil mainly through biological N fixation by legumes, and more recently from urea fertilizer. This has increased the amount of reactive N in the environment which can cascade through the environment causing damage to waters, forests, biodiversity, air quality and human health (Parfitt et al. 2006). The Nanjing Declaration on Management of Reactive Nitrogen (Initrogen 2004), confirmed and amplified in Delhi (Iniforum 2010), called for regional assessments of N deficiencies and excesses and mitigation options, based on expert judgement of scientific knowledge and uncertainties. Therefore, it is important to continue to assess trends in inputs and outputs of N for New Zealand.

We reported on the N cycle for New Zealand, and this included annual estimates of inputs and outputs of N for each regional council (Parfitt et al. 2006, 2008). Recently new information has been made available; there have been annual assessments of New Zealand's Greenhouse Gas Inventory that reported on grazing animals and the feed intake (MFE 2011a); the fertilizer sold each year has been documented by FertResearch; the Overseer® Nutrient Budget model has been revised (AgResearch 2009); long-term average erosion can be assessed using the NZeem® model (Dymond et al. 2010); the National River Water Quality Network (NRWQN) has operated for more than 20 years (Davies-Colley et al. 2011); and we have measured ammonia deposition at several sites.

Here we estimate the annual flows of N into and out of plant and soil ecosystems, including livestock, and also flows involving waste treatment, cities and towns for 1990, 2001 and 2010. The boundaries are the oceans, the atmosphere above plant canopies, and the aquifers. We do not include the internal cycling of N from soil to plant, and plant to soil, which occurs several times within a given year, with the exception of N excreted by livestock on grazed pastures. Losses go to the groundwater, to the oceans, to the atmosphere, as exported produce, and by burial in the landscape. We discuss those N stocks and flows that are reasonably well understood, and also identify those that have large uncertainties, to identify areas for further research.

Methods

We used similar methods to those reported previously for the regions of New Zealand for 2001 (Parfitt et al. 2006, 2008), but with more recent spatial, census and statistics data, and with some changes to methodology (Table 1). The land-use areas generally have been used to estimate N inputs and outputs for woody vegetation, and the numbers of beef cattle, dairy cows, sheep and deer have been used to estimate N inputs and outputs for pasture. The only exception is that denitrification to nitrogen gas under pasture has been estimated using the area of pasture. The most recent Ministry of Agriculture and Forestry (MAF) data have been used for dry-matter intake and numbers of beef cattle, dairy cows, sheep and deer in 1990, 2001 and 2010 (Fiona Thomson, MAF, pers. comm.). The total number of dairy cows has increased from 3.4 million in 1990 to 6.0 million in 2010, while the number of sheep has decreased from 58 million in 1990 to 33 million in 2010 (Table 2). Average gross energy intake (MFE 2011a), together with the N concentration in the feed (Ausseill et al. 2010), has been used in calculations; the equivalent stock units were 92 million in 1990, 100 million in 2001, and 96 million in 2010. The historical human population was 3.44 million in 1990, 3.74 million in 2001, and 4.37 million in 2010 (Statistics NZ 2011).

Table 1  Changes in methodology and outputs for major processes compared with Parfitt et al. (2006) data for 2001. Figures in parentheses in Gg N.

	Method used by Parfitt et al. 2006	Method used in this paper	Change in output
Ammonia deposition	50% lost off-shore (106)	75% lost off-shore (44)	Decrease
Ammonia volatilization	kg/SU (212)	MFE (175)	Decrease
Leaching nitrate-N pasture	Soil solution (224)	Overseer (148)	Decrease
Leaching DON pasture	10% of nitrate-N (22)	6 kg/ha (49)	Increase
Denitrification (N2) soil	kg/ha (116)	Changed kg/ha (70)	Decrease
Erosion	C loss and C/N (190)	NZeem® (138)	Decrease

Table 2  Animal numbers in New Zealand.

	1990 (million)	2001 (million)	2010 (million)
Dairy	3.4	5.1	6.0
Sheep	58	40	33
Beef	45	45	39
Deer	1.0	1.6	1.2
Total stock units	92	100	96

Land area

The area of pasture, tussock, crop, horticulture, indigenous forest, scrub and exotic forest was obtained from the Land Resource Inventory, topographic maps, the Land Cover Data Bases (LCDB) for 1995 and 2004, and the LUCAS land use maps for 1990 and 2008 (Dymond et al. 2012), and from census information from MAF for 2001/02 and 2006/07 (Table 3). The major changes in land use between 1990 and 2010 have been the increase in the area of dairy farms, the retirement of tussock land from agriculture and conversion of farm land to non-agricultural lifestyle blocks (totalling 873,000 ha by 2010) (Andrew and Dymond, Landcare Research, pers. comm.); these last two changes account for more than 1 million ha in total. The area of planted forest was 1.48 million ha of forest in 1990. About 590,000 net ha of new forest has been established since 1990, and 37,000 ha of pine forest has been converted to dairy pasture in Waikato (Hill & Borman 2011), bringing the total planted forest area to 2.07 million ha (P. Newsome, Landcare Research, pers. comm.; MFE 2011b). The area used as milking platforms has increased from 1.5 million ha in 1990 to 1.6 million ha in 2010, with the area of dairy support at 0.65 million ha. Thus the area under dairying has increased, while the area under sheep has decreased. Overall, the area of grazed pasture has reduced from about 12.5 million ha in 1990 to about 11.2 million ha in 2010 but the area of dairy pasture has increased (Table 3). The area of land under sheep, beef cattle, dairy cows and deer for each regional council was obtained from Ausseill et al. (2010).

Table 3  Land use in New Zealand.

	1990 (million ha)	2001 (million ha)	2010 (million ha)
Dairy milking and support	1.5	1.9	2.2
Drystock pasture (including woody weeds)	9.5	8.7	8.4
Tussock grazed (estimate)	1.5	1.0	0.6
Tussock ungrazed*	1.2	1.65	2.0
Exotic forest*	1.5	2.1	2.1
Native forest*	8.2	8.1	8.1
Scrub*	0.9	0.8	0.8
Crop	0.34*	0.335* (0.42)^a	0.33* (0.37)^a
Horticulture	0.08*	0.08* (0.11)^a	0.08* (0.13)^a
Other*	2.1	2.1	2.1
Total NZ	26.7	26.7	26.7
Grazing total	12.5	11.6	11.2
Lifestyle (within drystock pasture)	0.1	0.5	0.8
*LCDB combined with LUCAS LUM woody vegetation changes.
^aAgricultural Production Census.

Biological N fixation

We used our previous equations to calculate biological N fixation by pastures, but now have adjusted the stock units using average gross energy intake for 1990, 2001 and 2008 (extrapolated to 2010) from MFE (2011a) calculations. We have allowed for a decrease in biological N fixation by pastures arising from the increased use of N fertilizer (Parfitt et al. 2006). Since most of the N fertilizer is applied to dairy farms we have applied the correction to the area of dairy farms in each region. The reduction in biological N fixation by dairy pastures was 2% for 1990, 12% for 2001 and for 2010 was 23%. We have also estimated a decrease in N fixation arising from the infestation of pastures with the clover root weevil (Sitona lepidus Gyllenhal) since 1996. Our estimates range from a 5% reduction in North Island to no correction in South Island in 2001, to a reduction of 10% throughout in 2010. This is evaluated in the discussion section.

The average N fixation by legumes in exotic forests is taken to be 40 kg N/ha/y for 4 years, fixation by broom (Cytisus scoparius L) and gorse (Ulex europaeus L) is assumed to average 30 kg N/ha/y (Parfitt et al. 2006). The N fixation in indigenous forest is taken to be 1.5 kg N/ha/y.

Wet and dry deposition

We used our previous estimates of N in rainfall; these were in the range 1–5 kg N/ha/y (Parfitt et al. 2006). The general input of N by wet deposition is taken to be 1.5 kg N/ha/y. The exceptions were 3 kg N/ha/y for Southland and Tasman, and 5 kg N/ha/y for West Coast, where the rainfall can be greater than 2 m.

We have measured ammonia-N dry deposition at 14 sites in Waikato and Manawatū using Ogawa passive collectors (Roadman et al. 2003) placed 2 m above pastures. The filters were changed every 3 months over a 2-year period. Extraction of ammonia from the filters was performed according to Roadman et al. (2003) followed by FIA analysis; ammonia-N deposition was calculated based on equations for diffusion into the filters (Roadman et al. 2003). The dry deposition was greatest adjacent to dairy farms (5–10 kg N/ha/y) and least at 900 m altitude near native vegetation (1 kg N/ha/y).

It has been suggested that dry deposition of ammonia on to tree canopies explains the high nitrate-N concentrations found in springs and head-water streams in the Ruahine Range (McGroddy et al. 2008). We also collected samples of spring waters and head-water streams at 16 sites in the upper Turitea (Palmerston North) water catchment in December 2006 when the soils were near field capacity; the nitrate-N ranged from 0.1 to 1.9 mg N/L (mean 0.6, median 0.3 mg N/L). Consistent with these data, we assume that ammonia-N is lost from urea and from urine spots (see below) but 25% is redeposited on to adjacent forest and trees by dry deposition.

Fertilizer and imports

The N and phosphorus (P) fertilizer consumed each year was obtained from Fert Research (Fig. 1). Imports of N in grain and in animal feed were estimated from imported tonnages from Statistics NZ. The import of oil-palm kernel, which has 2% N, has increased as shown in Fig. 2.

Figure 1  Total nitrogen and phosphorus fertilizer consumed in New Zealand in Gg (1000 tonnes) N and P per year. N = •, P = ○.

Figure 2  National imports of palm kernel expeller (Gg N) assuming 2% N.

Produce

The N exported has been calculated from the N concentration and tonnes of exported produce (I. Parminter, MAF, pers. comm.). The exports containing N are meat, casein, milk powder, cheese, wool and hides, and to a lesser extent butter, wood products, grain and fruit. The various products have been apportioned to the regions based on the number of cows and sheep, and on the area of forests, crops and horticultural land. The N consumed by the New Zealand population has been apportioned to the regions by population.

Loss in waters

Here, we have used the Overseer® model (version 5.4.9) to estimate the N leaching from soil profiles under dairy, sheep and beef land for two slope classes (above and below and equal to 15°) for each region. For drystock we used regional average stock units and fertilizer rates, and assumed, on average, pastures were ‘developed’ on the flat and rolling land, and ‘developing’ in hill country. For dairy farms we used data for each region from NZ Dairy Statistics (Dairy NZ 2010). We assumed that the dairy pastures in 1990 were ‘developed’ and by 2010 they were ‘highly developed’. For 2001, we modelled losses for both ‘developed’ and ‘highly developed’ pastures and took the mean value. We converted the N leaching loss to an average loss per animal, and then calculated a loss for the region. For grazed tussock we assumed the loss was 5 kg N/ha/y. We have measured leaching losses of dissolved organic N (DON) under pasture in the range 4–8 kg N/ha/y for undeveloped and highly developed pasture (Parfitt et al. 2009) and assume DON losses are 6 kg N/ha/y for pasture.

We also assume nitrate-N leached from soil is 40 kg/ha/y under cropping, 60 kg/ha/y for horticulture including vegetables (Parfitt et al. 2006) and 5 kg/ha/y for vineyards (S. Green, Plant & Food Research, pers. comm.).

Leaching of N in low-fertility ecosystems occurs mainly as DON (McGroddy et al. 2008; Parfitt et al. 2009). Nitrate-N is also leached from land with low inputs of N and is usually in the range 1–4 kg/ha/y (Parfitt et al. 1997; McGroddy et al. 2008; Parfitt & Ross 2011). We have assumed total leaching losses of N from all forests are 2 kg/ha/y. Losses from point sources (sewage, dairy factories, abattoirs) have been taken from Elliott et al. (2005) for 2001, and assumed to be 33% greater in 1990 and 33% less in 2010 as a result of improved waste treatment.

The leaching from soils is then split into leaching losses to the oceans and leaching losses by attenuation between the soil and the ocean. For the regions, the 2001 losses of soluble N to the oceans are taken from Parfitt et al. (2006), except that losses from Taranaki and Wellington are constrained by inputs to lower values. The sites with increased or decreased losses from 1998 to 2007 for each region are given by Ballantine et al. (2010), and it is assumed that this gives the trends for the years 2001–10. The trends range from a decreased N loss of 5% for Wellington, to an increased loss of 10% for Waikato. For 1990–2001, the loss of N in rivers is assumed to increase by 5–15% according to where N fertilizer was used. The leaching losses by attenuation are obtained by subtracting the losses to the ocean from the leaching from soils.

For 17 major rivers (see Appendix), data for total N and nitrate-N were obtained from the NRWQN together with daily mean flows, to determine whether there was a trend in N export between 1990, 2001 and 2010. The data were grouped into nominal dates of 1990 (from 1989 to 1996), 2001 (from 1997 to 2003) and 2010 (from 2004 to 2011). Rating curves were generated by using locally weighted regression (log flow vs log nitrate-N). Specifically, this was done using the LOESS function in R on data from each group. The rating curve was then used to predict nitrate-N for every day in the daily flow record. Log-bias correction (Ferguson 1987; Hicks et al. 2000) was applied. The LOESS predictions do not allow extrapolation, and the daily flow record contains values outside the monthly samplings, so missing values were estimated using the nearest total N and nitrate-N concentration on an adjacent day, working first backward and then forward if gaps remained. Noting that bias and variability in changes over time can be driven by changes in water volume associated with multi-year climate cycles, with an impact on 7-year means, comparisons of calculated export were made by extrapolating the rating curve on to the flow record for each 7-year period, and for the total 21-year record. The data were then plotted on a log-log plot of 2001 and 2010 vs. 1990. The resulting comparisons were analysed using log-log regression in JMP8 software, to assess the magnitude and significance of changes in N export. The data were centred by subtracting the mean of the 1990 log-transformed data, and the intercepts of regression lines could then be interpreted as the increase in N export expressed as a percentage.

Denitrification

Losses as N₂O for New Zealand are reported by MFE (2011a) in CO₂ equivalents as 8163 Gg in 1990 and 10,038 Gg in 2009. Losses of N₂O from agricultural soils are reported as 25 Gg in 1990, 32 Gg in 2002 and 31 Gg in 2009 (MFE 2011a). These have been converted to N and apportioned to regions by areas of farmland and stock units.

We assumed that loss of N as N₂ gas was 2 kg N/ha/y for tussock land, 4 kg N/ha/y for drystock and 5–24 kg N/ha/y for dairy land, increasing with N fertilizer and with poor drainage (Ruz-Jerez et al. 1994; Barton et al. 1999; Ledgard et al. 1999; Stevenson et al. 2011). We obtained the areas of poorly drained soils under pasture for each region from the Land Resource Inventory (LRI) and assumed that the proportional areas were the same for dairy land. We used 5 kg N/ha for 1990, 10 kg N/ha for well-drained soils and 20 kg N/ha for poorly drained soils in 2001, and 12 kg N/ha for well-drained soils and 24 kg N/ha for poorly drained soils in 2010. We used our previous estimates for the other annual loss of N₂ gas: 13 kg N/ha for cropped land, 1 kg N/ha for forest and 10 kg N/ha for the small areas of wetlands and organic soils. We used similar methods to those reported previously for denitrification from dairy effluent ponds and town sewage (Parfitt et al. 2006).

Total national emissions of NOx gases in 1990 were 99 Gg NOx mainly from fossil fuel combustion; in 2009 they were 152 Gg (MFE 2011a). We assume (as with ammonia) that 25% gets redeposited on New Zealand ecosystems. The input of NOx was apportioned to regions based on the population of the regions.

Ammonia

Ammonia-N volatilization from N fertilizer is assumed to be 10% (see review in MFE 2011a). Ammonia-N volatilization also arises from the N excreted on to pasture (Sherlock & Goh 1984; Jarvis & Ledgard 2002) and we have used the value of 10% of the N excreted (MFE 2011a), but since the ammonia-N arises from urine rather than from dung, we excluded the proportion of N in dung (assumed to be 33%) (Luo & Kelliher 2010). The N excreted by dairy cows, beef cattle, sheep and deer in 1990, 2001 and 2009 are given in MFE (2011a). We made no adjustment for allophane or soil pH since pH values are not available for regions.

Erosion

Erosional losses from soils deliver sediment to the ocean, lakes and reservoirs, while some sediment is buried in the landscape. The NZeem® model of Dymond et al. (2010) has been used to estimate losses of sediment for each region. The total N losses to the oceans (sediment and dissolved N) used in Parfitt et al. (2006) were based on the work of Elliott et al. (2005) and have been used here. Therefore percentage N in the sediment varied and N concentrations used ranged from 0.2% N in Gisborne where there is gully erosion (Gomez et al. 2003), to 2% N in more fertile regions. Values were: West Coast 0.3%; Hawke's Bay, Otago, Canterbury 0.4%; Southland, Manawatū 1.2% (Parfitt & Hill 2004); Northland 1.4%; Waikato, Taranaki, BOP, Auckland 2%.

We assumed that burial of sediment N on floodplains and hill slopes is approximately the same as sediment N losses to the sea (i.e. a sediment delivery ratio of 50%), except for the West Coast of South Island, where, because the steep Alps are adjacent to the ocean, there are few burial zones. The buried sewage sludge N is added to buried sediment N.

We assumed no change in the long-term average except for planting of exotic forest that reduced erosion by 2010 by about 2%.

Results

Inputs

The major change in the inputs of N between 1990 and 2010 has been the increased use of N fertilizer. This has increased from 46 Gg N in 1990 to 329 Gg N in 2010 (Fig. 1).

The increased use of N fertilizer has reduced the N fixation by clovers, particularly on dairy farms. As the clover weevil has also reduced N fixation, the estimated N fixation has reduced from 495 to 420 Gg N (Tables 4–6). This equates to an average reduction from 19 to 16 kg N/ha for New Zealand on an area basis; the N fixation by pastures, however, is on average about 40 kg N /ha.

Table 4  Annual inputs of N for New Zealand by region for 1990.

	Area (sq km)	Pasture legume (Gg N)	Other BNF (Gg N)	Atmosphere (Gg N)	Fertilizer (Gg N)	Imports (Gg N)	Total (Gg N)
Northland	12,696	26	3	3	3	1	36
Auckland	5104	13	1	1	1	9	26
Waikato	24,500	67	4	7	11	3	93
Bay of Plenty	12,271	13	4	3	3	2	24
Gisborne	8361	20	1	2	1	0	25
Hawke's Bay	14,172	42	2	4	2	1	51
Taranaki	7256	27	1	3	3	1	34
Manawatū-Wanganui	22,211	76	2	7	4	2	92
Wellington	8125	19	2	2	1	4	27
Tasman	9654	5	2	3	1	0	12
Nelson	424	0	0	0	0	0	1
Marlborough	10,494	9	1	2	0	0	13
West Coast	23,363	4	4	12	1	0	21
Canterbury	45,039	71	4	10	9	4	98
Otago	31,922	51	2	7	3	2	65
Southland	31,775	51	3	12	4	1	70
Total	267,367	495	37	79	46	33	689

Table 5  Annual inputs of N for New Zealand by region for 2001 (Gg N).

	Pasture legume	Other BNF	Atmosphere	Fertilizer	Imports	Total
Northland	25	3	4	14	1	47
Auckland	11	1	1	6	11	30
Waikato	76	5	10	61	4	156
Bay of Plenty	15	5	3	14	2	39
Gisborne	17	2	2	3	0	25
Hawke's Bay	37	3	4	10	1	55
Taranaki	24	1	3	15	1	44
Manawatū-Wanganui	71	3	7	25	2	108
Wellington	19	2	2	6	4	33
Tasman	6	3	3	4	0	16
Nelson	0	0	0	0	0	1
Marlborough	8	2	2	2	0	14
West Coast	6	4	12	7	0	29
Canterbury	79	4	11	48	5	147
Otago	51	2	7	14	2	76
Southland	54	3	13	20	1	91
Total	498	41	86	248	38	911
% of 1990 values	101	113	109	544	115	132

Table 6  Annual inputs of N for New Zealand by region for 2010 (Gg N).

	Pasture legume	Other BNF	Atmosphere	Fertilizer	Imports	Total
Northland	22	3	4	17	3	49
Auckland	9	1	1	6	14	31
Waikato	71	4	10	75	14	174
Bay of Plenty	14	5	3	17	5	43
Gisborne	15	2	2	6	0	25
Hawke's Bay	28	3	4	16	2	52
Taranaki	22	1	3	28	5	58
Manawatū-Wanganui	56	3	7	20	4	90
Wellington	14	2	2	7	5	30
Tasman	5	3	3	5	1	17
Nelson	0	0	0	0	0	1
Marlborough	5	2	2	2	1	11
West Coast	6	4	12	13	1	37
Canterbury	67	4	13	77	10	171
Otago	41	2	7	16	3	70
Southland	47	3	13	26	4	94
Total	420	41	88	329	73	951
% of 1990 values	85	112	111	723	221	138

The other major change in inputs has been from imported feed that has increased from 33 to 73 Gg N. Most of this is made up of imports of palm kernel cake that have increased by about 28 Gg N. Total inputs are estimated to have increased from 689 Gg N in 1990 to 911 Gg N in 2001 to 951 Gg N in 2010 (Fig. 3).

Figure 3  Total inputs (A) and outputs (B) of N for New Zealand for 1990, 2001 and 2010 (Gg).

The total inputs in kg N/ha for New Zealand are 26 kg/ha in 1990 and 36 kg/ha in 2010. The largest inputs on an area basis in 2010 are for Taranaki (80 kg/ha) and Waikato (71 kg/ha) and the least are for Marlborough (10 kg/ha) (Fig. 4).

Figure 4  Regional inputs and outputs of N for 1990 and 2010 for each region in kg/ha. The regions are in the same order as in Table 4. BNF = biological N fixation.

Outputs

There has been little change in the export of N in produce or the N stored in trees between 1990 and 2010. The erosion data are a long-term average, so will not capture decadal trends. While the export of dairy products has increased, the export of meat products has decreased.

Total leaching of N from soil profiles increased from 249 to 271 Gg N (Tables 7–9). The leaching ranges from 2 kg N/ha in forests to 56 kg N/ha under dairying in West Coast. Leaching of DON from grazed pasture amounts to 49 Gg N, and to 21 Gg N from forests in 2010. Total leaching from crops was 15 Gg N and from horticulture is 6 Gg N in 2010.

Table 7  Annual outputs of N for New Zealand by region for 1990 (Gg N).

	Produce	Leaching to ocean	Leaching to attenuation	Denitrification in soil	Denitrification in waters	Ammonia volatilization	Sediment Total	Trees, Total	Total
Northland	9	4	9	5	1	5	21	1	55
Auckland	5	2	4	5	4	3	3	1	27
Waikato	26	16	16	10	2	15	16	2	102
Bay of Plenty	6	4	5	4	1	3	5	1	29
Gisborne	5	2	7	4	0	4	11	1	33
Hawke's Bay	11	7	11	6	1	8	5	1	49
Taranaki	12	7	6	3	1	6	7	0	42
Manawatū-Wanganui	23	9	20	10	1	15	17	1	96
Wellington	6	3	5	4	2	4	7	1	31
Tasman	2	2	3	2	0	1	3	1	14
Nelson	0	0	0	0	0	0	0	0	1
Marlborough	3	1	4	3	0	2	2	1	16
West Coast	2	4	17	3	0	1	16	0	43
Canterbury	22	9	26	14	2	13	9	2	98
Otago	16	3	19	9	1	9	8	1	66
Southland	16	14	11	8	1	10	10	1	70
Total	163	88	161	90	17	97	140	15	771

Table 8  Annual outputs of N for New Zealand by region for 2001 (Gg N).

	Produce	Leaching to ocean	Leaching to attenuation	Denitrification in soil	Denitrification in waters	Ammonia volatilization	Sediment Total	Trees, fires	Total
Northland	9	5	10	7	1	7	21	1	61
Auckland	4	2	4	6	4	3	3	1	28
Waikato	34	18	24	16	3	25	16	2	138
Bay of Plenty	8	4	7	5	1	5	5	1	36
Gisborne	4	2	7	3	0	4	10	1	31
Hawke's Bay	9	7	10	7	1	8	4	1	48
Taranaki	11	8	5	5	1	7	7	0	44
Manawatū-Wanganui	19	11	18	11	1	16	16	1	93
Wellington	5	3	5	4	2	4	7	1	32
Tasman	2	2	3	2	0	1	3	1	16
Nelson	0	0	0	0	0	0	0	0	1
Marlborough	2	1	3	3	0	2	2	1	15
West Coast	2	5	17	3	0	2	16	0	47
Canterbury	24	9	29	19	3	19	9	2	113
Otago	14	3	18	11	1	10	8	1	67
Southland	16	15	10	10	1	12	10	1	75
Total	165	97	170	112	20	124	138	16	844
% of 1990 values	101	110	106	125	120	128	98	110	109

Table 9  Annual outputs of N for New Zealand by region for 2010 (Gg N).

	Produce	Leaching to ocean	Leaching to attenuation	Denitrification in soil	Denitrification in waters	Ammonia volatilization	Sediment Total	Trees, fires	Total
Northland	9	5	10	7	1	7	21	1	60
Auckland	3	2	3	7	5	3	4	1	28
Waikato	35	20	24	18	4	26	16	2	145
Bay of Plenty	8	5	6	6	1	5	5	1	38
Gisborne	4	2	6	3	0	3	10	1	29
Hawke's Bay	8	7	7	6	1	7	4	1	42
Taranaki	12	8	6	6	1	9	7	0	49
Manawatū-Wanganui	18	11	15	11	1	14	16	1	88
Wellington	5	2	5	4	2	4	7	1	30
Tasman	2	2	4	2	0	2	3	1	16
Nelson	0	0	0	0	0	0	0	0	1
Marlborough	2	1	2	4	0	1	2	1	13
West Coast	4	8	17	4	1	3	16	0	52
Canterbury	27	10	32	20	3	23	9	3	127
Otago	14	3	17	11	1	10	8	1	66
Southland	19	16	11	11	1	14	10	1	82
Total	167	104	167	119	22	132	138	16	866
% of 1990 values	103	117	104	132	134	136	99	110	112

The leached N moves to rivers and to aquifers. The soluble N exported to the oceans by rivers increased from 88 Gg N in 1990, to 97 Gg N in 2001 and to 104 Gg N in 2010 (Tables 7–9). The N leaching increased in regions where N fertilizer is used on dairy farms, and where there was an increase in dairy production, but decreased where the number of drystock have decreased (Gisborne and Hawke's Bay). More than 20 years’ worth of NRWQN data are now available to test whether trends in regional N leaching estimates are consistent with trends in 17 major rivers. The NRWQN data include nitrate-N and total N, noting that total N includes particulates but is important to consider in large river systems because it includes organic-N produced by algal uptake of soluble N.

The export of nitrate-N for 17 major NRWQN rivers generally did not increase between 1990 and 2001 but then increased significantly (P ≥ 0.002) by up to 10% from 2001 to 2010 (Fig. 5A). Most rivers showed increases in nitrate-N, the exceptions being the Whanganui, the Manawatū, the Tukituki and the Haast. The largest increases were for the Ōreti (176%), Buller (161%), Waimakariri (147%) and the Wairau (145%).

The export of total N for 17 major NRWQN rivers also did not increase between 1990 and 2001 but then increased significantly by 15% from 2001 to 2010 (Fig. 5B). When the data for nitrate-N were subtracted from those of total N to give N in sediment together with ammonium-N and DON, they showed that most of the increase arose from the increase in nitrate-N; the major exceptions were the increase in sediment N for the Manawatū (3.4 kg N/ha for 2001 to 4.7 kg/ha for 2010), the Hurunui (1.1 kg N/ha for 1990 to 3.0 kg/ha for 2010) and the Buller (2.7 kg N/ha for 1990 to 3.2 kg/ha for 2001).

Loss of N by erosion decreased slightly as a result of afforestation. The loss of sediment N in rivers was 75 Gg N (that includes West Coast sediment), and the loss of N by burial was 65 Gg N.

Regionally, soluble N lost to oceans ranged from 1 kg N/ha overall in Otago in 1990 to 12 kg N/ha in Taranaki in 2010, and generally increased with time for each region (Tables 7–9). This magnitude of export is consistent with NRWQN river data where the export of nitrate-N ranged from 1 kg N/ha overall for the Clutha catchment in 1990 to 8 kg N/ha for the Ōreti catchment (Southland) in 2010. The NRWQN data, summarized above and in Fig. 5, show an increase mainly between 2001 and 2010, and little or no increase from 1990 to 2001, possibly indicating a lag between changes in inputs and outputs that is not fully captured by our N budgeting methodology. The regional pattern of changes in river nitrate-N and total N export appears consistent with the positioning of the catchments in geographical regions in Tables 7–9.

Figure 5  N losses in catchments of major NRWQN rivers for 2001 (⋄) and 2010 (♦) plotted against 1990 data for nitrate-N (A) and total N (B). Also shown are the trend lines and the 1:1 line (broken line). The difference between 2010 and 1990 is significant (P ≥ 0.002). Our preferred method for estimating N losses is to use river flows from the entire period. The alternative is to use river flows from each ~7 year period only. This alternative method often yields results that are biased by climate patterns rather than N biogeochemistry.

Denitrification in waters and wetlands increased from 17 to 22 Gg N, and other denitrification increased from 90 to 119 Gg N. This mainly arises from the increase in dairy and human effluent.

Ammonia-N volatilization increased from 97 to 132 Gg N, again mainly from the increase in dairy production. Volatilization from N fertilizer increased from 3 Gg N to 31 Gg N and volatilization from urine spots increased from 93 to 99 Gg N. The volatilization from pasture increased from an average of 8 kg N/ha in 1990 to 13 kg N/ha in 2010. The dry deposition on forest and trees adjacent to pasture increased from 2.0 to 3.1 kg N/ha. It is assumed that the difference between ammonia volatilization and dry deposition is lost from New Zealand in the atmosphere to the oceans.

Total N outputs were estimated to be 771 Gg N (29 kg N/ha) in 1990, 844 Gg N in 2001 and 866 Gg N (32 kg N/ha) in 2010 (Tables 7–9) (Fig. 3). The largest outputs in 2010 were for Taranaki (68 kg N/ha) and the least were for Marlborough (13 kg N/ha) (Fig. 4).

Unaccounted-for N

The difference between inputs and outputs for New Zealand changed from a net loss of 82 Gg N in 1990 (3 kg N/ha on average), to a gain of 67 Gg N in 2001, to a gain of 85 Gg N in 2010. The difference for each region varies from a loss of 15 kg N/ha for Northland in 1990, to a gain of 12 kg N/ha for Waikato and Taranaki in 2010. The loss for Northland arises entirely from erosion, and the gains from input of N fertilizer that is then stored in soils.

If erosion, which arises from steeper land, is excluded from the calculation of differences, then on average the land was storing N. The difference between inputs and outputs gave a net gain of 58 Gg N in 1990 (2 kg N/ha on average), a gain of 204 Gg N in 2001 and a gain of 224 Gg N in 2010 (8 kg N/ha on average). Using these data, the regions losing N in 1990 were West Coast and Taranaki, and in 2010 only Marlborough had a net loss.

Discussion

Inputs

The major change in the N budget from 1990 to 2010 has been the increased use of N fertilizer, mainly on dairy farms where the use was more than100 kg N/ha in 2010, but also on sheep and beef farms with lower slope classes where the average use reached about 10 kg N/ha. The largest change occurred between 1990 and 2005, with a lesser change since 2005, when sales of N fertilizer appear to have reached a plateau. Sales were affected by low economic returns on farms from 2006 to 2008, the recession in 2008, higher prices of fertilizer, use of Overseer® (which promoted more careful use of fertilizer) and the drought in 2008/09. The increased use of oil-palm kernel on dairy farms has led to N inputs of 28 Gg N on these farms in 2010. This contributes to reactive N in the environment and if imports of feed continue to increase, the addition of reactive N to the New Zealand environment could be considerable, as shown in our paper on N scenarios (Parfitt et al. 2008). The increased use of N, together with the increase of N in dairy feed and excreta, has led to increased N leaching, denitrification and volatilization. To some extent the losses have been reduced from sheep farms because of the lower numbers of animals, and therefore the amount of reactive N in the environment has not increased as rapidly as it would have done if sheep numbers had remained constant.

The regions most affected are those with large numbers of dairy cows. The biggest changes have occurred in Canterbury and Otago where new dairy farms have been established, which use N fertilizer and irrigation. The milk solid production on these farms averages 1190 kg/ha in Canterbury and 1060 kg/ha in Otago (Dairy NZ 2010), leading to high average N losses estimated with the Overseer® model. A large part of the pasture in these regions, however, is under sheep and the N losses have tended to reduce from sheep farms between 1990 and 2010.

The infestation of pastures with the clover root weevil (Sitona lepidus Gyllenhal) has probably reduced N fixation but the amount of reduction is unknown. For 2010 we set the reduction at 10% for all pastures, but pasture production overall does not appear to have been affected in the past 20 years, because livestock production has been maintained. Possibly the use of N fertilizer has compensated for the effect of the weevil, or possibly the pasture ecosystem has adapted to the weevil. Nevertheless, this is a source of uncertainty in the N inputs.

Outputs

Leaching losses of N are estimated to have increased from 249 to 268 to 271 Gg N between 1990 and 2010, and these data arise from the output from Overseer®. Losses from point sources are estimated to be 8 Gg N in 1990 and 4 Gg N in 2010, and total leaching from forests and woodlots is about 42 Gg N from 1990 to 2010.

Leaching of DON from forests amounts to 21 Gg N, and leaching of DON from grazed pasture has been set at 6 kg N/ha/y and amounts to 49 Gg N in total. For pasture this estimate is based on a limited data set (Ghani et al. 2007, 2010; Parfitt et al. 2009; van Kessel et al. 2009) and it assumed not to be accounted for in Overseer®. There is considerable uncertainty in this number; the total N lost to the oceans, however, is constrained by the data from NIWA (Elliott et al. 2005). Ghani et al. (2011) reported that a large part of DON from pasture is bioavailable and can be converted to ammonium-N. Some DON, however, is found in spring waters and rivers (Parfitt et al. 2009); the concentration of DON in the Manawatū River is about 50% of the dissolved inorganic N (DIN) (Parfitt, unpublished data).

The largest source of dissolved N is from diffuse sources, mainly urine spots in pastures. The leaching loss of DIN from sheep pastures has reduced from 58 Gg N to 33 Gg N because sheep numbers have reduced between 1990 and 2010, but the loss of DIN has increased from 35 Gg N to 83 Gg N on dairy farms, whereas under beef it has increased from 33 to 38 Gg N. Concentrations of DIN may have reduced in streams near sheep farms, particularly in hill country, but have increased in streams and rivers near dairy farms which are usually in the lowlands. Ballantine et al. (2010) reported that nitrate-N had a meaningful decrease at 39 out of 545 sites in New Zealand between 1998 and 2007, but had a meaningful increase at 49 out of 545 sites. The largest decreases were for Taranaki (3/12 sites), Wellington (11/59) and Manawatū-Wanganui (2/13), and the largest increases were for Waikato (26/118), Bay of Plenty (BOP) (3/16) and Southland (Environment Southland 2010).

Our analysis of 17 rivers showed the largest increases in loss of nitrate-N are for the Ōreti, Buller, Waimakariri and the Wairau rivers. The increase has generally occurred from 1997 to 2011, and this is consistent with the analysis of Ballantine & Davies-Colley (2010). Our data for leaching to rivers showed increases from 88 to 97 to 104 Gg N for 1990, 2001 and 2010, respectively. Similarly, the data plotted in Fig. 5 is plotted as a logarithm so that the difference from 1990 can be interpreted as the increase in N export, yielding estimates of 9±2% and 8±2% relative to 1990 for nitrate-N and total N, respectively. An alternative method of extrapolating river flow data (not shown) yielded values of 10±3% and 4±2%, respectively; therefore we suggest the increase in N export from 1990 to 2010 is best estimated as being in the range of 4–10%.

The larger increase in N export to rivers in years surrounding 2010, compared with 2001 (Fig. 5), also suggests that there may be a lag between the increase in nitrate-N leached in soils (which is assumed to immediately match changes in inputs in Tables 7–9) and the increase in the nitrate-N in rivers. The mean residence time of waters can range from a few months (Motueka) to more than 60 years (pumice) (Stewart et al. 2007), and the lag will depend on the soils, geology and rainfall (Morgenstern et al. 2010), as well as the time-course of soil N saturation (Schipper et al. 2004). This suggests that the loads of nitrate-N in some rivers may continue to increase for many years after the land-use has intensified. Our analysis suggests that in coming decades New Zealand rivers could approach the same ratio of inputs to outputs observed in rivers draining to the North Atlantic (Parfitt et al. 2006). Evidence for decreases in N outputs for some catchments and increases in others emphasizes the importance of monitoring of sufficient quality and frequency to observe the impacts of economic shifts and policy changes that occur over periods of 5–7 years.

The loss of N to aquifers and streams was obtained from the difference between the total N leached and the amount of N lost to the oceans. The difference is assumed to be N lost by leaching to aquifers and lost during transport to major rivers. These losses include biological uptake (e.g. in streams and rivers) and denitrification, and they have not been apportioned beyond a loss to aquifers and streams.

Denitrification in soil has increased from 90 to 119 Gg N, and denitrification in waters and wetlands from 17 to 22 Gg N. The increase with time mainly arises from the increase in dairy production, and the increase in the human population. Estimates for average denitrification rates on several soils across New Zealand appear to be reasonably well defined (Ruz-Jerez et al. 1994; Barton et al. 1999; Ledgard et al. 1999; Stevenson et al. 2011), but denitrification is spatially and temporally extremely variable.

Ammonia-N volatilization has increased from 97 in 1990 to 132 Gg N in 2010, again mainly from the increase in dairy production. The numbers are given by the greenhouse gas inventory data but there are uncertainties in the estimates (MFE 2011a). The ammonia can be redeposited in rain, by dry deposition on tree canopies or can be lost to the atmosphere. The average dry deposition on trees downwind of pasture has increased between 1990 and 2010. Although the changes may seem small, deposition rates as low as 5 kg N/ha can alter the cycling of carbon in forests on nutrient-poor soils (Janssens et al. 2010) that are common in New Zealand (McGroddy et al. 2008) and can lead to losses of nitrate-N from forest streams.

Losses by erosion have decreased slightly as a result of afforestation. These losses are for a long-term average, and it is beyond the scope of this paper to report on the influence of weather in 1990, 2001 and 2010. We recognize that erosion losses, by burial and by transport to the oceans, occur in major events such as Cyclone Bola in 1988 and the Manawatū floods in 2004.

In 1990, the land in all regions was losing N, as estimated by subtracting outputs from inputs. This loss arises from erosion and from mineralization of soil N. If erosion was excluded (because it is episodic), and assuming most changes in soils occurred under pasture, then pasture-land in the regions gained small amounts of N (2–8 kg N/ha), except for Taranaki, Marlborough and West Coast, where there were net losses of N. In 2010, there was increased storage of N in soil, ranging from 8 kg N/ha/y in Otago to 41 kg N/ha/y in Bay of Plenty. The N that is stored probably arises mainly from the N fertilizer that has been applied on dairy land. Since erosion occurs mainly on non-dairy land, the actual storage in dairy land may be quite large, particularly on land that has been converted from drystock farming. Overseer® indicates that storage ranges from 10 to 30 kg N/ha/y depending on whether the pasture is highly developed or developed. This is consistent with data from resampling some of the 500 soils data set, which showed that soil total N (0–10 cm) has increased by about 20 kg N/ha/y on dairy farms (n = 93) and 25 kg N/ha/y on sheep and beef farms (n = 61) between 2002 and 2009 with a concurrent decrease in soil C/N ratio for both land uses (our unpublished data). This is not consistent, however, with the data from Schipper at al. (2010) who reported that 29 dairy soils (0–30 cm) lost 57 kg N/ha/y over 27 years, but this was for an older subset of dairy farms where N fertilizer generally was not added. The Schipper et al. (2010) study took place from about 1985 to 2008, highlighting the possible difference between the long-term and short-term response of soils to land-use intensification.

Unresolved gains and losses

There is considerable uncertainty in some of the inputs and outputs, whereas others are well constrained by the data. The Greenhouse Gas Inventory data (MFE 2011a) has allowed us to improve on the estimates of N inputs and outputs given in Parfitt et al. (2006, 2008). In particular, the feed intake by grazing animals is better defined, leading to a better estimate of N fixation by pastures. The loss of ammonia-N from pastures is also better defined using the inventory data. The estimate for 2001 is now 124 Gg N (Parfitt et al. 2008) compared with 212 Gg N (Parfitt et al. 2006) and 241 Gg N (Parfitt et al. 2008). The dry deposition of ammonia is also better defined.

The unresolved gains and losses include:

•	N fixation—the N fixation by pastures averages 40 kg N/ha consistent with the values reviewed in Heggie & Savage (2009). Uncertainties arise, however, in the estimation of N fixation by pasture. Not only is there a large year-to-year variation because of different weather conditions (Ledgard et al. 1999), but there are also no data on the effect of the clover root weevil on N fixation.
•	DON—leaching losses in pastures are calculated with Overseer® using average regional inputs; the losses of DON from pasture have a major effect on the budgets. If these losses were 100 Gg N in 2001 and 2010 instead of 49 Gg N, then outputs would be closer to inputs. It is assumed these losses are not included in Overseer®; this is an area requiring further research.
•	Losses in rivers—the losses of total N, DIN and DON in rivers vary with weather conditions (Ballantine et al. 2010). Losses are also changing as a result of mitigation measures on farms. These estimates are beyond the scope of this paper and we note our analysis represents a preliminary assessment of only 17 of the 77 NRWQN sites.
•	Sediment N—the loss of N by erosion would be better constrained if the N content of the sediment were available. The National Institute of Water and Atmospheric Research (NIWA) has begun to measure the N in sediment monthly in major rivers (NRWQN river samples), but samples from major floods, when most erosion occurs, will be required (R Davies-Colley, NIWA, pers. comm.).
•	Immobilization and mineralization—the 500 soil data set indicate that there have been significant decreases in soil C/N, suggesting N is becoming stored in pasture soils. There is uncertainty about the amount that can be stored because some soils appear to have a limit where the C/N ratio is about 9, whereas for other soils the limit is greater than 9 (Schipper et al. 2004).
•	Loss of farm land to lifestyle—the area of lifestyle blocks is estimated at 800,000 ha and much of this is under pasture. There is uncertainty as to how this land is being used.

Conclusions

This study provides an additional analysis of N inputs and outputs for New Zealand covering the period from 1990 to 2010. The construction of national N budgets should be an important component of N management as outlined under the Nanjing and Delhi Declarations. For New Zealand, most of the large inputs are from pastoral farming and these can be estimated from the Overseer® model. These inputs also constrain many of the outputs—products, leaching and volatilization. This exercise, however, indicates that there is large uncertainty in parts of the N cycle under pastures, including N fixation by pastures, loss of DON, the extent of mitigation and immobilization, and the influence of geology and weather on leaching and erosion. These are areas that require further research. Further refinement of these N budgets and scenarios should enable New Zealanders to make informed decisions regarding the trade-off between economic growth of the agricultural sector and the impacts of reactive N on the environment, and the degree of regulation that is required.

Acknowledgements

We are grateful to Graham Bryers and Kathy Walter (NIWA) for the NRWQN data and the daily mean flow data, and to many colleagues for assistance with data and many helpful discussions.

Appendix

National River Water Quality Network (NRWQN) sites

Site	Location	Catchment (ha)
CH2	Hurunui at SH1 bridge	2525
CH4	Waimakariri above old HW bridge	3076
DN4	Clutha at Balclutha	20582
DN5	Mataura at Seaward Downs	5139
DN8	Ōreti at Riverton HW bridge	2150
DN9	Waiau at Tuatapere	8098
GY1	Buller at Te Kuha	6309
GY4	Haast at Roaring Billy	1027
HM4	Waikato at Rangiriri	12429
HV2	Tukituki at Red bridge	2438
HV5	Moōhaka at Raupunga	2371
NN4	Wairau at Tuamarina	3494
RO5	Rangitāiki at Te Teko	2818
TK6	Waitaki at SH1 bridge	11909
WA4	Whanganui at Paetawa	6567
WA9	Manawatū at Ōpiki bridge	4846
WN3	Raumahanga at Waihenga	2368