OVERVIEW OF STRATEGIES FOR REDUCING NITROGEN LOSSES

The best way of achieving the dual and generally conflicting goals of increased productivity and reducing N losses to the environment is to ensure that ‘more for less’ is achieved, i.e. more milk per animal or per unit of dry matter (DM) intake, or more DM per unit of N input. Reactive N losses to the environment can be further reduced through the adoption of practices that minimize N loss risk, or capture N before it enters waterways. Table 1 provides an overview of some available options.

Table 1. Summary of the key ‘efficiency’ or ‘reduced nitrogen loss risk’ measures for grazed pastoral dairy systems

THE POTENTIAL OF INTEGRATED MEASURES FOR REDUCING NITROGEN LOSSES: CASE STUDY EXAMPLES FROM GRAZED NEW ZEALAND DAIRY FARMS

In 2001 the New Zealand dairy industry initiated a study in which regionally representative catchments that were predominantly used for dairy farming were monitored for water quality and flow, as well as changes in farm management practices that were likely to have an impact on contaminant losses (including N) to water. Some of the key trends in stream water quality are documented in Wilcock et al. (Reference Wilcock, Monaghan, Quinn, Srinivasan, Houlbrooke, Duncan, Wright-Stow and Scarsbrook2013) and show how, as expected, measured N losses have increased in catchments where land use has changed from lower intensity sheep–beef farming to dairy farming. Farm scale modelling based on detailed farm surveys of farm inputs, managements and saleable outputs has also shown how N leaching losses have increased as dairy farms have intensified by, in most cases, importing more purchased feed and N fertilizer (Table 2). This link between dairy farm intensification and increased loss of N to water has also been observed in catchments in South-West Victoria, Australia by Smith et al. (Reference Smith, Western and Hannah2013), who also found some evidence to suggest that this link may have become decoupled during the second half of the 20-year monitoring period due to various changes in farm management. However, Smith et al. (Reference Smith, Western and Hannah2013) could not specify which changes may have had the greatest effect.

Table 2. Attributes of average farms in the catchment case study used for modelling the potential benefits of implementing a range of integrated mitigation measures on grazed New Zealand dairy farm systems

* Inchbonnie catchment study commenced in 2004.

† Some farms within the Toenepi and Waiokura catchments discharge effluent to waterways after treatment via a 2-pond (anaerobic then aerobic) system.

‡ Replacement stock mostly grazed off-farm for Waikakahi and Bog Burn catchments.

§ 1 kg MS (milksolids)=c. 12 litres milk.

# Uncovered pads typically used for over-wintering herds.

†† Fertilization rates for non-effluent/effluent blocks.

This farm and catchment monitoring has clearly demonstrated the challenges for reducing reactive N losses against a backdrop of farm intensification and land use change from lower intensity sheep farming to dairy. Although dairy systems have become more N efficient, the rate of productivity gains has typically been greater than the rate of efficiency gains. Some important metrics that describe this intensification for the farms in the present study are (Table 2):

• • Purchased feed sourced via imported supplements or N fertilizer has increased by c. 5 and 4% per annum on average, respectively.

• • Milk production per hectare has increased by c. 2% per annum (ranging from 1 to 4% per annum).

• • Estimated N leaching and N₂O losses per hectare have increased by c. 2% per annum on average; when expressed on a per unit product basis, these losses remained either constant (N leaching) or decreased by only c. 1% (N₂O emissions) per annum on average.

Of note is the improved fertilizer management that has occurred on effluent blocks, where both N and P fertilizer inputs have generally reduced as farmers and fertilizer representatives have considered the returns of these nutrients in applied effluent.

To evaluate the potential for offsetting the increases in N loss between 2001 and 2009, farms were remodelled to identify some of the most effective mitigation options, applied singly or in combination, and explore whether any of the measures resulted in pollution swapping. The scenarios were derived based on the most recent survey (2009; Table 1) of typical farm practices and the suite of mitigation measures potentially available and relevant, as discussed above (but excluding edge-of-field attenuation options). The scenarios considered were:

1. 1. Eliminating applications of fertilizer N to some of the catchment farms during winter.

2. 2. Optimal reuse of farm dairy effluent to minimize direct losses during application to land and ensure that N fertilizer inputs account for N returns in the applied effluent.

3. 3. Improving animal performance through better feeding and using cows with higher genetic merit. This was assumed to allow a slight reduction in animal replacement rates (to 18%).

4. 4. Substituting half of the N-fertilized grass with purchased supplement that contained a higher energy-to-protein ratio.

5. 5. Applying the nitrification inhibitor DCD to pasture areas.

6. 6. Restricting the duration of pasture grazing to 10 h/day during autumn and the first 2 months of winter for Toenepi, Waiokura and Inchbonnie catchment farms, where cows are mostly grazing on pasture during this period. For the Bog Burn and Waikakahi farms that currently use a forage brassica crop to over-winter cows, it was instead assumed that cows would be housed in a herd shelter for 10 weeks over winter.

7. 7. A scenario where all of the above options are assumed to be implemented.

The mitigation modelling results for farms in 2009 are shown in Figs 1 and 2 and have been normalized to losses modelled for the case study farms based on farm survey information gathered for 2001. This modelling analysis indicates that for all catchments the N leaching losses (without mitigations) have increased by 3–30% since 2001, with the highest increases observed in the Toenepi and Inchbonnie catchments. Most of the individual mitigation options would have relatively small effects on offsetting these increases in N leaching. For example, eliminating N fertilizer applications during winter was relevant to only two of the five catchment farms, and was predicted to reduce losses by c. 10% for these farms relative to the 2009 base farm losses, while improved effluent management was estimated to reduce losses by only 2–4% relative to 2009 base farm losses. However, neither option could offset the increase in leaching losses observed since 2001. Using fewer cows with higher genetic merit could offset the increase in losses since 2001 in two of the five catchments only. Replacing N-fertilized grass with a higher-energy supplement and applying the nitrification inhibitor DCD were slightly more effective measures, reducing N leaching losses by on average 15 and 16% relative to the 2009 base farm losses on average, respectively, and could in most cases offset any increases observed since 2001. It should be noted that the current modelling assumes that supplements were produced outside of the catchments and hence does not account for any off-farm effects on N losses. The single most effective measure for reducing N leaching losses from the farms was to limit the duration of pasture grazing during autumn and winter months. Averaged across the catchments, this was estimated to reduce N leaching losses by c. 20% relative to the 2009 base farm losses.

Fig. 1. Modelled nitrogen leaching losses per hectare for the catchment farms in 2009 assuming a range of individual mitigation practices have been implemented. Results have been normalized to losses estimated for each catchment Base farm in 2001. FDE, farm dairy effluent; SR, stocking rate; BW, cow breeding worth; DCD, dicyandiamide.

Fig. 2. Modelled nitrous oxide emissions per hectare for the catchment farms assuming a range of individual mitigation measures have been implemented. Results have been normalized to emissions estimated for each catchment Base farm during 2001. FDE, farm dairy effluent; SR, stocking rate; BW, cow breeding worth; DCD, dicyandiamide.

The analysis also suggested that intensification resulted in an increase in N₂O emissions (without mitigations) of 0–25% since 2001, with the highest increases observed in the Toenepi and Waikakahi catchments. As for N leaching, the implications of each mitigation measure on N₂O emissions were variable. Only three of the mitigation practices would consistently deliver reductions in per-hectare N₂O emissions relative to the 2009 base farm emissions: improving animal performance combined with a reduction in stocking rate (5% reduction on average, ranging from 3 to 7%), introducing a higher-energy supplement to the cow diet (15% reduction on average, ranging from 8 to 21%) and applying DCD to pasture (15% reduction on average, ranging from 6 to 23%). The use of a higher-energy supplement or DCD could in most cases also offset any increases in emissions since 2001. Limiting the duration of pasture grazing during autumn and winter months was estimated to lead to greater emissions of N₂O from the North Island Toenepi and Waiokura farms compared with the 2009 base farms, whereas for the South Island catchments this option resulted in reductions of 2–26% in N₂O emissions. These estimated reductions in the South Island catchments were also large enough to offset any increases observed since 2001. The increased emissions in the North Island catchments were due to the modelled increases in losses from the stored manures and effluents that were accumulated in the off-paddock herd shelter. This is an example of how some mitigation options can potentially result in pollution swapping; in the case of the Waiokura and Toenepi farms, the 17 and 21% reductions in estimated N leaching losses per hectare due to off-paddock grazing resulted in an estimated increase in total N₂O emissions per hectare of 19 and 29%, respectively, even though these estimates account for a reduction in indirect N₂O emissions due to reduced N leaching. In contrast, the restricted pasture grazing strategy was estimated to reduce both N leaching and N₂O emissions, relative to the 2009 base scenario for the three South Island model farms. The contrasting modelled responses in N₂O emissions between the North and South Island catchment farms are assumed to be attributable to the climatic factors that are influencing the N losses modelled from the stored manures (relatively large increases in emissions modelled for the warmer North Island farms) or excreta deposited to pasture and winter crops (greater benefit observed for restricting grazing duration for the cooler, wetter South Island farms).

Figures 1 and 2 indicate that mitigation practices will need to be collectively implemented to achieve reductions in N losses large enough to offset any future increases due to on-going intensification on the catchment farms. If the options above are categorized into ‘efficiency’ (1–3), ‘mitigation’ (4 and 5) and ‘system changes’ (6), and progressively implemented and modelled in that order, the magnitude of (technically) achievable reductions in N loss becomes more apparent. The system change was implemented last because it was assumed that the restricted grazing system (measure 6) would be the least favoured N leaching mitigation measure due to the cost and management complexity introduced by this measure and the potential pollution swapping that could occur in some areas. Model estimates of leaching, N₂O emissions, NH₃ volatilization, denitrification and total N losses were included to more fully consider any implications of pollution-swapping that may potentially arise from these sets of collective measures (Fig. 3). Loss estimates have been normalized to 2001 for each catchment farm to remove the effects of skewing that would otherwise be apparent if actual losses per hectare were plotted on individual axes.

Fig. 3. Model estimates of leaching, nitrous oxide emissions, ammonia volatilization, denitrification and total N losses from the catchment farms for scenarios where efficiency (□; scenarios 1–3), mitigation (○; scenarios 4 and 5) or system change (Δ; scenario 6) measures are assumed to be collectively and progressively implemented. Base farm losses for 2001 are shown as solid black lines (⧫). Lowest losses are at the centre and increase with distance outwards.

Although there is variability between the catchment farms, this modelling of the introduction of collective measures for an average farm suggests that ‘efficiency’ measures (scenarios 1–3) could reduce N leaching losses from the 2009 Base farms by 7–21% (mean of 13%). A co-benefit of introducing these would be a reduction in N₂O emissions of c. 7%. Although these N leaching reductions are significant, they are only estimated to reduce losses, on average, to levels calculated for the 2001 Base farms (Fig. 3). Implementation of both ‘efficiency’ and ‘mitigation’ measures (scenarios 1–5) would reduce total N leaching and N₂O emissions from the 2009 Base farms by c. 30%, or by c. 22% if compared against the 2001 Base farms. Finally, implementation of all the strategies, including ‘system change’ (scenario 6), is predicted to reduce N leaching and N₂O emissions from the 2009 Base farms by 44% (range 33–61%) and 22% (range 11–32%) on average, respectively; or by 34 and 12% relative to the 2001 Base farms, respectively. At a technical level, these potential reductions are substantial, although they do require progressively greater management expertise and, in some cases, may incur significant cost. However, the ‘efficiency’ measures (scenarios1–3) are likely to result in no cost or even a cost saving, as they represent increases in total farm efficiency (i.e. more milk per cow or per unit of DM). The ‘mitigation’ measures, in particular the restricted grazing option, are likely to incur cost and may require incentives (e.g. carbon trading) or regulations (e.g. nutrient loss capping to protect water quality) to support their implementation. Fortunately, in most cases the combined strategies do not result in pollution swapping or greater total N losses to the environment (Fig. 3). Exceptions to this general finding are the greater denitrification losses modelled for the Toenepi, Waiokura and Waikakahi farm scenarios. As observed for N₂O losses, denitrification losses increased on the Toenepi and Waiokura farms when all practices were collectively implemented due to the effects of storing effluent captured on the off-paddock confinement facility. Reasons for the greater denitrification losses modelled for the Waikakahi farm are unclear, although they are very small in absolute terms at only 1 or 2 kg N/ha/year.

IMPLICATIONS FOR CATCHMENT FARMS

These modelling assessments suggest that integrating a range of strategic and tactical management and mitigation options can reduce N losses to the environment, while maintaining milk productivity. It is apparent from this analysis, however, that there is no ‘one size fits all’ approach to the challenge of reducing N losses from these dairy systems and that the ability to reduce losses will vary between individual farms depending on their existing management practices and level of farm inputs. While most of the ‘efficiency’ options evaluated are in principle equally relevant to all the catchment farms, some of the mitigation practices are inappropriate due to the considerable variability evident in soil and climate factors for the catchment farms. The most obvious example of this is the low effectiveness of DCD in the Inchbonnie catchment, where the extremely high rainfall (4·5 m per annum) was the cause of the low predicted reductions in N leaching and N₂O emissions of only 2 and 6%, respectively. In contrast, in the drier Bog Burn catchment estimates of N leaching and N₂O emissions were reduced by 26 and 17% due to DCD application, respectively. Another example is the use of a restricted grazing strategy during autumn and winter to reduce urinary N returns to pasture ahead of winter rainfall and drainage. Although this is modelled to deliver N leaching reductions of 17–21% for the warmer North Island catchments of Toenepi and Waiokura, the unintended consequences of implementing this measure are the modelled increases in N loss via NH₃ volatilization and N₂O emissions from the herd shelter and stored effluents (Fig. 3). This is the only scenario where an individual measure is predicted to actually result in an increase in total N losses, in this case by 12 and 1% for the Toenepi and Waiokura catchment farms, respectively. Although increases in NH₃ volatilization are also modelled for the restricted autumn–winter grazing management scenario when applied to the cooler South Island catchment farms, these increases are relatively small, and total N losses are still 6–10% lower than modelled for the equivalent Base farms.

The cost associated with implementing each of the management options evaluated above is obviously an important consideration that will govern likely rates of adoption by farmers. Although these costs are subject to the price volatility often attached to farm input costs and product returns, it is reasonable to assume that the efficiency measures evaluated in the present paper will incur little or no cost and in fact are likely to deliver a net return where superior animal performance has been assumed. The net costs associated with the mitigation practices (DCD and lowered dietary N content) will fluctuate depending on input prices and milk returns. Recent evidence presented by Gillingham et al. (Reference Gillingham, Ledgard, Saggar, Cameron, Di, De Klein, ASPIN, Currie and Christensen2012) would suggest that the small to modest increases in pasture production in response to DCD application are likely to at least partly off-set the costs of applying this product in many situations. The economic consequence of modifying dietary N content is, however, subject to a number of price variables, the most important of which are the cost of the low-N feed, such as maize or cereal silage/grain, and the price of milk (Jensen et al. Reference Jensen, Clark and Macdonald2005). The measure that is likely to incur greatest cost is the restricted grazing strategy, which requires significant capital investment and some degrees of on-going management cost associated with the required changes in animal and effluent management practices. An unintended consequence of this strategy could be that farmers might use the housing or stand-off facility to increase stock numbers by bringing in more supplements to maximize the return on investment. As a result, reactive N cycling in the system could be intensified and total N losses potentially higher, although losses per unit of product are likely to be lower. Given these cost implications, and the pollution swapping potential discussed above, the restricted grazing strategy is likely to have more limited applicability than the other options considered here.

FUTURE OPTIONS AND RESEARCH NEEDS

The catchment case study analysis suggests that improved management and mitigation practices can partially offset increased N losses associated with recent farm intensification. However, on-going intensification will require future options to minimize or reduce reactive N losses to the environment. These future options could include genetic solutions such as continuous improvement of the genetic merit of dairy cows (Beukes et al. Reference Beukes, Gregorini and Romera2011) or the development of pasture species with low N concentration to minimize N excretion in predominantly grazed systems (Parsons et al. Reference Parsons, Rasmussen, Liu, Xue, Ball and Shaw2013). Other options include those that accelerate the reduction of N₂O to environmentally benign N₂ during denitrification. This reduction process is the only known sink of N₂O in pastoral systems and is catalysed by the microbial enzyme nitrous oxide reductase (N₂OR) encoded by the nosZ gene (Thomson et al. Reference Thomson, Giannopoulos, Pretty, Baggs and Richardson2012). Manipulating the soil physicochemical and/or environmental factors that influence nosZ gene expression and/or N₂OR activity could provide scope for accelerating N₂O reduction in urine patches, thereby reducing N₂O emissions from grazing livestock systems. Such manipulations or technologies could also have a spillover benefit in that they could be used in areas where the risk of NO₃ ⁻ leaching poses an environmental threat. Using the technologies in combination with technologies or practices that enhance total denitrification (e.g. riparian areas or wetlands) would result in reduced nitrate pollution to waterways without increasing the risk of N₂O emissions from these areas. Future options could also focus on further exploiting the spatial and temporal variability in soil, environmental and climatic drivers, to further increase NUE (‘more for less’) or the effectiveness of mitigation options such as nitrification inhibitors (‘reduced N loss risk’). For example, Shepherd et al. (Reference Shepherd, Phillips, Snow, Currie and Christensen2011) showed that 40–50% of urine deposited in late summer/early autumn (i.e. well before the start of the drainage season) could be lost through leaching. Therefore, targeting N mitigations early could substantially reduce the risk of N leaching, particularly in summer-dry areas where pasture growth rates, and thus N uptake, are limited. Similarly, the spatial and temporal variability in soil moisture, a key driver of N losses (van der Weerden et al. Reference Van der Weerden, Kelliher and de Klein2012), could be exploited by using real-time soil water monitoring to improve tactical decision making for fertilizer, manure and/or grazing management to reduce N leaching and N₂O emissions.

Finally, evaluation of the whole dairy system (i.e. dairy farm and associated land used for feed production) is important when assessing the effectiveness of management interventions that tighten the N cycle. This scale of analysis adds considerable complexity and places much reliance on modelling tools that attempt to simulate system responses relative to a change in land use management. Although the knowledge of biological systems and their response to farm management interventions remains surrounded by uncertainty, the challenging and pressing issues at hand require use of the modelling tools available and acknowledge the uncertainties inherent in model outputs. Recognizing these uncertainties can then guide future research needs and modelling tool development.

The generous support and contribution by the many farmers and land managers working within the catchments is gratefully acknowledged. We also thank Alison Rutherford, Chris Smith, David Houlbrooke, Alec McGowan and Bog Longhurst for their assistance in undertaking the farm surveys. This work was supported by a range of funding parties: the Pastoral21 (Environment) programme (contract C10X0603), funded by the NZ Ministry for Science & Innovation (MSI), DairyNZ, Fonterra and Beef & Lamb NZ; the Sustainable Farming Fund, administered by the NZ Ministry for Primary Industries; and NZ MBIE (contract C10X1006).