New Zealand dairy farm system solutions that balance reductions in nitrogen leaching with profitability – a case study

ABSTRACT

This study tested prescribed management practices to reduce nitrogen (N) leaching by 20% while maintaining or improving profitability relative to an existing farm management baseline (Control). N leaching and profitability were estimated for a South Canterbury case study dairy farm using Overseer® Nutrient Budgets and FARMAX Dairy. Three practices were used: (1) reducing N in cows’ diets through low-N feed (fodder beet), (2) recapturing N from soils through catch crops (oats) and (3) diluting urinary N by including plantain in cows’ diet. While most treatments reduced N leaching, significant management inputs were required to achieve a 20% reduction from the Control. Plantain was identified as the key forage for reducing N leaching from the milking platform. Fodder beet and oats had little impact, due to the small area cropped on the milking platform and low dietary substitution. However, they increased profitability relative to the Control. Only one scenario, employing all three forages with biannual direct drilling of plantain, achieved the target, reducing N leaching by 27% and increasing profitability by 2% compared with the Control. The implications of this modelling study for real-life application are that a combination of measures will be needed to achieve large environmental and economic targets.

KEYWORDS:

• Nitrogen
• nitrate leaching
• plantain
• fodder beet
• catch crop
• oats
• modelling
• dairy farm system
• mixed pasture
• profitability

Introduction

Humanity faces major environmental, social and economic challenges in demands for food, water, nutrients and energy (Janzen et al. 2011). In part, these challenges are driven by growing human populations and their need for greater food and water security. Intensification of production is one method used to address increasing food demand.

Intensification of farming animals can involve increases in inputs, such as water, feed, agrichemicals, and stocking rate per hectare of land, to increase production (Eurostat 2017; Ministry for the Environment and Statistics New Zealand 2018). In New Zealand, major intensification of animal production has occurred over the last 50 years, characterised by greater inputs per unit area (Parliamentary Commissioner for the Environment (PCE) 2004; MacLeod and Moller 2006), and the conversion of sheep and beef farms to dairy farms in Canterbury and Southland (Ministry for the Environment and Statistics New Zealand 2018). This widespread intensification has resulted in environmental degradation, including the pollution of freshwater bodies by nitrogen (N) leaching from farms (Parliamentary Commissioner for the Environment (PCE) 2004; Ministry for the Environment and Statistics New Zealand 2018).

Some approaches for reducing N leaching can increase operating costs and reduce revenue or operating profit (Doole and Romera 2015; Muller and Neal 2019). However, strategies involving alternative plant species appear to show more favourable financial outcomes while reducing N leaching (Beukes et al. 2017, 2018). Between 2013 and 2019, the Forages for Reduced Nitrate Leaching (FRNL) programme examined a range of potential management practices to reduce N leaching, by at least 20%, from dairy, arable and mixed livestock farm systems. The approach was to incorporate forages with particular characteristics into farm systems (DairyNZ 2018). The FRNL mitigations investigated targeted N leaching from dairy cow urine patches by (1) reducing N in cows’ diets through low-N feed (e.g. fodder beet), (2) recapturing N from soils through catch crops (e.g. oats) and (3) diluting urinary N by including plantain in cows’ diet (Beukes et al. 2017, 2018).

Some FRNL animal and field trials and modelling showed that when plantain made up at least 30% of a cow’s diet, the N load of urine patches (i.e. amount of N per unit of urinated area), the main source of N loss to water from agricultural land (Di and Cameron 2002), was significantly reduced (Judson and Edwards 2016; Box et al. 2017; Dodd et al. 2019a). Based on FRNL research outcomes, incorporating plantain, fodder beet and oat catch crops into the model farm system was expected to reduce the farm’s estimated N leaching. It was less clear if these mitigation options could maintain the farm’s milksolids (MS) production and profit.

To test this, modelling was deemed necessary as some measurements could not be made at the scale needed and account for considerable year-to-year variation on farms due to weather conditions, responsive management, and market prices.

The objective of this study was to model management options using combinations of FRNL solutions aimed at reducing N leaching by 20% from current farming methods (Control scenario) while maintaining profitability. Based on the results of the FRNL programme and preliminary farm financial estimates, it was hypothesised that by integrating fodder beet, oats and plantain into the existing farm system, it would be possible to reduce N leaching by 20% and increase profitability compared with the Control.

Methods

This study modelled the application of FRNL practices to an existing south Canterbury dairy farm. Scenarios were analysed using FARMAX Dairy (Version 7.1.2.41) for physical and financial performance. Outputs from Farmax were then analysed using OverseerScience (Overseer® Version 6.5.2) and post-model processing to determine N leaching mitigation potential. Overseer reports N leaching (kg N ha⁻¹), however the model is calibrated for nitrate-N (Science Advisory Panel 2021). Hence, the Overseer’s estimates of N leaching (kg N ha⁻¹) as reported in this paper should be interpreted as nitrate-N leaching (kg NO₃-N ha⁻¹). All scenarios were assumed to be at steady state, i.e. no transition period was factored into the analyses. Treatments using fodder beet, oats and plantain were evaluated.

Physical and financial data from the study farm were used as a base to evaluate FRNL management strategies. Management strategies were considered successful if three main criteria were met: (1) management solutions were practical, (2) farm profitability was maintained and (3) a 20% reduction in N leaching relative to the Control treatment was achieved. In Overseer®, N leaching is defined as the movement of N below 60 cm soil depth i.e. below the pasture root zone (Watkins and Selbie 2015).

Farm system description

A hypothetical Baseline scenario (the Control, NC₀ in Table 1) was a simplified representation of the 2017/2018 observed dairy season for a farm in south Canterbury. The observed system was simplified to allow representation in the models and to generalise the results (especially the financials) for the Canterbury region as follows.

Table 1. Names of the model scenarios and descriptions of the crop and plantain treatments applied to the Control.

Crop treatment	Plantain treatment
	No plantain	No maintenance	Maintenance of new pastures in 4^th year	Maintenance of new pastures in 4^th and 7^th year	Maintenance of new pastures every 2^nd year
NC	NC0	NC1	NC2	NC3	NC4
FB	FB0	FB1	FB2	FB3	FB4
BC	BC0	BC1	BC2	BC3	BC4

Notes: “Maintenance” means that pastures are direct-drilled with 4 kg plantain seed ha⁻¹ in spring to maintain/increase the proportion of plantain in the pasture sward (see Table 2). NC = no crops; FB = fodder beet; BC = both fodder beet and oats crops. NC₀ is considered the Control (no plantain, fodder beet or oat catch crop). The total milking platform area was 312.8 ha with a 12.6 ha fodder crop block rotating through the milking platform area.

The farm’s dominant soil type was Claremont moderately deep (40–80 cm) poorly drained silty-loam soils with high profile available water (90 mm between 0 and 60 cm; https://smap.landcareresearch.co.nz/). The average annual rainfall was 509 mm; the potential evapotranspiration was 772 mm; and the average ambient air temperature was 11.1 °C. The system modelled was a 95% irrigated, 312.8 ha milking platform. Standard perennial ryegrass/white clover (PR/WC) pastures were fertilised at 282 kg N ha⁻¹ yr ⁻¹ on the 184.5 ha non-effluent block and 275 kg N ha⁻¹ yr⁻¹ on the 128.3 ha effluent block, with 10% of the farm undergoing pasture renewal each year. Pasture was irrigated using a combination of travelling, spray line and pivot irrigators (12%, 4% and 79% of the milking platform, respectively) from September to May, based on the soil water budget (plant available water with 70% trigger point). Where plantain (PL) was included in renewed pastures as a treatment, the same irrigation schedule was used as for PR/WC pasture – it was assumed that there was no difference in ME or growth between PR/WC and PR/WC + PL pastures (Pembleton et al. 2015; Minnée et al. 2020). The system was stocked at 3.7 Kiwi crossbred cows ha⁻¹ (1142 milking cows at peak), producing 1600 kg MS ha^{−1 ­­}yr⁻¹ (445 kg MS cow⁻¹ yr⁻¹). This was a spring calving system with Planned Start of Calving on 3 August and Planned Start of Mating on 28 October. Imported supplementary feeds included wheat grain, protein supplement and pasture silage, amounting to approximately 330 kg dry matter (DM) eaten cow⁻¹ yr⁻¹ (Table A1). Other supplemental feed was homegrown: 19 t DM pasture silage harvested, 315 t DM fodder beet crop (25 t DM ha⁻¹yr⁻¹, grazed in situ by lactating and dry cows) and 126 t DM oat silage (10 t DM ha⁻¹yr⁻¹). Fodder beet was sown in October and grazed by the end of May on 12.6 ha (4%) of the milking platform. The crop was irrigated from November to March based on soil moisture sensor readings, using probes/tapes to achieve 70% plant available water. Oat crops were sown by early June, as paddocks became available after fodder beet grazing. Oats were harvested in stages in September and October.

All wintering of non-lactating cows and rearing of young stock was done on a hypothetical support block. The stock consisted of 1235 non-lactating cows from June to mid-August, and year-round 373 R1s (rising one-year olds, weaned replacement heifers from four months of age in their first season of grazing) and 270 R2s (rising two-year olds, i.e. replacement heifers in their second season of grazing, before their first calf).

Financial inputs (costs, milk price, etc.) were based on Marlborough-Canterbury region averages for the 2017–2018 season (or as close as possible) (Askin and Askin 2016; DairyNZ and Livestock Improvement Corporation 2019). The 2017 milksolids price of $6.50/kg MS was used. No premiums or added value were included (DairyBase DairyNZ 2019). See Tables A2, A4 and A5 in the Appendix for further financial data used in the Farmax scenarios.

Treatments

This section describes the key parameters that were considered when creating the model scenarios for each treatment. These key parameters were vital for ensuring the scenarios for all treatments could be directly compared with the Control. Refer to Table 1 for treatments.

Maintaining milksolids (MS) production was a key strategy that ensured all scenarios were directly comparable and that the effects of the three forages and their management could be distinguished. Maintaining parameters such as dry-off dates, culling decisions, dietary ME and pasture management were essential. Changes in these parameters would have affected MS production (Waghorn et al. 2007) and potentially N leaching (Di and Cameron 2002; Clark et al. 2019). Crop and pasture production affected the quantity and quality of feed available and influenced if supplemental feed was necessary to maintain MS production. Milksolids production was maintained through isoenergetic calculations, i.e. ensuring that dietary ME was maintained between all treatments, regardless of the diet composition (Appendix, Table A1). Future modelling work in this area may explore the consequences of varying these critical management practices.

The treatments applied in the modelled scenarios were based on forages and practices identified in the FRNL programme:

1. Reducing the amount of N consumed by the cow in autumn through low-N feed (fodder beet, FB treatment). In the Overseer modelling, fodder beet was assumed to be grazed for two hours per day during late lactation (autumn) to reduce the total amount of N ingested.

2. Recycling N via oat catch crops sown immediately following grazing of fodder beet on the milking platform (both crops, BC treatment). The oats were ensiled and fed to non-lactating cows in the paddock in the following season. This area was resown in perennial pastures as part of the regrassing programme in spring.

3. Diluting N excreted in urine by feeding pastures and pasture silage containing plantain, especially during mid-summer/autumn (plantain treatments 1 to 4; Table 1).

The three practices were used to develop the crop and plantain treatments applied to the modelled farm (Table 1). In no-crop treatments (NC), imported feeds were used in place of fodder beet and oat crops grown on the milking platform in FB and BC treatments. The imported feeds used in NC had similar feed characteristics and popularity in Canterbury to fodder beet and oat silage (DairyNZ 2017; Dalley et al. 2017; Edwards et al. 2017). For example, barley grain was most similar to fodder beet based on its low N and high ME contents, and pasture silage was most similar to oat silage in terms of DM, ME and N contents (Dalley et al. 2017). Treatment diets for lactating dairy cows are represented in Figure 1.

Figure 1. Treatment diets presented as feed offered to lactating dairy cows in Farmax. N.B. feed offered is not equivalent to feed eaten as utilisation is not taken into account here. NC = no crops; FB = fodder beet; BC = both fodder beet and oats crops.

Crop treatments

Except where plantain treatments were applied, pasture swards were PR/WC mixes. Two crop treatments, Fodder Beet only (FB) and fodder beet followed by oats (Both Crops; BC), were introduced as part of the regrassing scheme on the milking platform. For the Control (No Crops; NC), 10% of the milking platform area was regrassed/cropped each spring in all treatments. Where FB and BC treatments were applied, 6% of the farm was regrassed with permanent (new) pasture and 4% of the farm was sown with fodder beet in October, grazed in situ by lactating cows in April and May. For FB treatments, grazed fodder beet was followed by a fallow period then new pasture. For BC treatments, grazed fodder beet was followed by oats sown by June as paddocks became available. The oat crop was harvested for silage in stages from September to October (due to the varying sowing dates) and followed by permanent pasture.

Plantain treatments

Four treatments were applied to introduce plantain to the milking platform in the Control and the two crop treatments. Treatment 1 involved incorporating plantain into the pasture base by sowing 4 kg plantain seed ha⁻¹ with 14 kg perennial ryegrass seed ha⁻¹ and 4 kg white clover seed ha⁻¹ during the regrassing programme in spring. Ten percent of the milking platform was regrassed each year. As plantain does not persist well in ryegrass/white clover + plantain (RG/WC + PL) swards (Dodd et al. 2019b), and it was impractical to grow pure plantain across the entire farm (Mangwe et al. 2019), plantain treatments 2–4 involved increasing the frequency of direct drilling 4 kg plantain seed ha⁻¹ into existing PR/WC + PL mixtures in spring, to recover the proportion of plantain in the sward (Table 1).

The plantain treatments were as follows:

1. Newly established swards included plantain (PR/WC + PL), but no maintenance of the plantain content occurred – 0% of the farm area was redrilled with plantain in later years.

2. Swards were direct-drilled with plantain in their fourth year after establishment. This is 10% of the farm area each spring.

3. Swards were direct-drilled with plantain in their fourth and seventh years after establishment, which is 20% of the farm area each spring.

4. Swards were direct-drilled with plantain every second year, starting two years after establishment. This is 40% of the farm area each spring.

Plantain persistence assumptions

The persistence curve, representing the proportion of pasture DM as plantain in PR/WC + PL swards, was a major assumption for this case study. The persistence curve assumed that the plantain content of the sward (DM basis) was 50% in the first year, 40% in the second year, 20% in the third year and 10% in the following years (Table 2). Direct-drilling plantain, beginning in October, was assumed to return plantain populations to those of first-year pastures, i.e. 50% sward DM as plantain. A sensitivity analysis of plantain persistence was performed to determine how variations in persistence could affect treatment responses. This used values where plantain establishment was not as successful and plantain content in the pasture decreased more rapidly than the original persistence values (Bryant et al. 2019; Dodd et al. 2019b) (Table 2).

Table 2. Persistence values for plantain in perennial ryegrass/white clover + plantain mixed pastures showing the decline in plantain presence over time.

Age of pasture	Proportion of pasture dry matter as plantain
	Original	Sensitivity analysis
1^st year pasture and newly maintained pasture	50%	40%
2^nd year pasture	40%	20%
3^rd year pasture	20%	<15%*
4+ year old pasture	<15%*	<15%*

Notes: The persistence values were used for all scenarios and were based on personal communication with D. Chapman, 24th June, 2019 and M. Dodd, 1st July, 2019. The sensitivity analysis of the persistence values was based on recent data from different regions in New Zealand, mainly from the Tararua area (Dodd et al. 2019b), showing that the establishment and persistence of plantain in PR/WC + PL swards may be less than the initial curve assumed for the original scenarios. *Indicates pastures that have <15% plantain (effectively 0%) and therefore will have negligible impact on nitrogen leaching.

For this case study, the cost of maintenance via direct drilling was estimated to be $200 ha⁻¹_, which is $20 per effective ha of the whole farm when 10% of the farm is drilled each year, as is the case in the NC₂, FB₂ and BC₂ scenarios (see Appendix, Table A2). The cost of direct drilling alone was $100 ha⁻¹ (Askin and Askin 2016).

Nitrogen leaching estimation

For the Control (NC­₀), FB₀ and BC₀ treatments (no plantain), N leaching values were calculated using Overseer. The following section applies only to scenarios where plantain treatments were applied. This is because at the time of modelling, plantain was not represented in Overseer.

Post-model processing for plantain treatments

The use of the OverseerScience model meant that the N loading rate in the urine patch could be changed from the default 750 kg N ha⁻¹ yr⁻¹. A lower loading rate was used as a proxy for the urine dilution effect of the presence of plantain. Overseer captures the risk of N leaching with the urine N load parameter and does not take account of area covered by urine patches.

A linear relationship between ingested %PL and urinary N concentrations (Equation 1(1) $\mathrm{R}\mathrm{e}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{r}\mathrm{y}\mathrm{N}\mathrm{l}\mathrm{o}\mathrm{a}\mathrm{d}=\text{-}0.6205\left(\text{\%}\mathrm{P}\mathrm{L}\mathrm{o}\mathrm{f}\mathrm{d}\mathrm{i}\mathrm{e}\mathrm{t}\mathrm{D}\mathrm{M}\right)+1$(1) ) was assumed for lactating dairy cows, based on data from Box et al. (2017) and Minnée et al. (2020). This relationship had an R² value of 0.83. To estimate the reduction in urinary N load, the proportion of dietary DM as plantain (%PL, between 0 and 1) was first calculated. (1) $\mathrm{R}\mathrm{e}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{i}\mathrm{n}\mathrm{u}\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{r}\mathrm{y}\mathrm{N}\mathrm{l}\mathrm{o}\mathrm{a}\mathrm{d}=\text{-}0.6205\left(\text{\%}\mathrm{P}\mathrm{L}\mathrm{o}\mathrm{f}\mathrm{d}\mathrm{i}\mathrm{e}\mathrm{t}\mathrm{D}\mathrm{M}\right)+1$(1)

For diets containing <15%PL, it was assumed that there was no effect (0%) on urinary N load and therefore no impact on N leaching.

The proportion of the diet consisting of plantain was estimated. Feeds containing plantain were: PR/WC + PL pasture and PR/WC + PL pasture silage harvested from the milking platform. Feed was classed into one of four categories:

1. 1^st year pasture,

2. 2^nd year pasture,

3. 3^rd year pasture,

4. Supplements, other than pasture silage and 4+ year old pasture.

For the purpose of calculating the %PL of the total diet calculation, supplements and 4+ year old pasture were combined as one category because pastures four years or older had <15% plantain cover (Table 2) and, other than homegrown pasture silage, none of the supplements contained plantain. This category had no effect on urinary N loading.

Consumption of PR/WC + PL pasture and silage were assumed to have the same impact on urinary N load, i.e. plantain did not have to be ‘fresh’ to affect urinary N (Judson and Edwards 2016). To estimate the %PL of the total diet DM, the proportion of the diet from each feed category was multiplied by its proportion of plantain. An example is provided in the Appendix, Table A3.

N leaching results were predicted by OverseerScience, keeping the farm’s soil type and climate the same across all scenarios. Urinary N load in the urine patch (an OverseerScience input) was estimated for each scenario where plantain treatments were applied. Equation (1) was used to estimate the relative urinary N load for each scenario as compared to the Control (no plantain, NC₀). The annual N leaching output from Overseer for each feed category was then used to estimate farm-scale N leaching under the different plantain treatments (Appendix, Table A3).

Results

The support block has been excluded from the results as it was the same across treatments and no treatments were applied to it. Therefore, treatments did not affect the overall N leaching or financial performance of the support block.

Effectiveness of crop and plantain treatments for reducing nitrogen leaching

This section compares all crop and plantain treatments to the Control, NC₀. Whole farm N leaching for NC₀ was 26 kg N ha⁻¹ yr⁻¹.

Crop treatments

Across all ha of the milking platform, neither crop treatment i.e. addition of the fodder beet crop (FB₀) or fodder beet and oat crops (BC₀) affected N leaching compared with the Control (Table 3).

Table 3. Milking platform nitrogen (N) leaching and operating profit for all treatments.

		N leached		Operating profit		Cost of mitigation
Treatment	Description of plantain treatment	kg N ha^−1 yr^−1	Change compared to Control	$ ha^−1	Change compared to Control	$ ha^−1 kg^−1 N mitigated compared to Control
NC0 (Control)	No plantain	26		2342
NC1	No maintenance	24	-8%	2328	-1%	7
NC2	Maintenance in 4th year	22	-15%	2306	-2%	9
NC3	Maintenance in 4th and 7th year	20	-23%	2283	-3%	10
NC4	Maintenance every 2nd year	18	-31%	2239	-5%	13
FB0	No plantain	26	0%	2432	4%
FB1	No maintenance	24	-8%	2418	3%	-45
FB2	Maintenance in 4th year	23	-12%	2396	2%	-16
FB3	Maintenance in 4th and 7th year	21	-19%	2374	1%	-6
FB4	Maintenance every 2nd year	19	-27%	2328	-1%	2
BC0	No plantain	26	0%	2501	6%
BC1	No maintenance	24	-8%	2487	6%	-81
BC2	Maintenance in 4th year	22	-15%	2467	5%	-35
BC3	Maintenance in 4th and 7th year	21	-19%	2443	4%	-19
BC4	Maintenance every 2nd year	19	-27%	2385	2%	-6

Notes: The reductions are reported relative to the Control. Bolded lines indicate treatments that achieved the targeted 20% reduction in N leaching. N = nitrogen; NC = no crops; FB = fodder beet; BC = both fodder beet and oats crops. NC₀ is considered the Control (no plantain, fodder beet or oat catch crop). Maintenance means direct drilling of plantain into existing pasture to maintain/increase the proportion of plantain in the sward.

Leaching from the cropping paddocks was 56 kg N ha⁻¹ yr⁻¹ for FB₀ and 47 kg N ha⁻¹ yr⁻¹ for BC₀. Therefore, addition of a winter catch crop of oats after autumn grazing of fodder beet reduced N leaching by 16% from the crop paddocks. At the milking platform level, the effects of cropping and utilising a catch crop were negligible due to the small proportion of total area being in crop.

Plantain treatments

When plantain was introduced via PR/WC + PL pastures (with no subsequent direct drilling of plantain content i.e. Treatment 1), N leaching was reduced from the Control by 8% for NC₁ and 7% for FB₁ and BC₁. More frequent direct-drilling of plantain to maintain a sufficient plantain content in pasture resulted in higher plantain content of the diet and therefore greater reductions in N leaching (Table 3).

As the frequency of direct-drilling of plantain increased, the size of the difference between the FB and BC treatments remained the same, except for treatment FB₂ which had less effect on N leaching than BC₂ (Table 3 and Figure 2). This is the only treatment that showed any effect due to the presence of a catch crop, which could simply be due to rounding. With more frequent direct-drilling of plantain, N leaching from the NC treatments decreased more than for plantain treatments applied to FB and BC (Table 3), due to the relatively greater area of pasture containing plantain.

Figure 2. Changes in farm operating profit and nitrogen (N) leaching relative to the Control (NC₀ = no plantain, fodder beet or oats catch crop; white triangle). The dotted line marks the targeted N leaching reduction of 20%. NC = no crops; FB = fodder beet; BC = both fodder beet and oats crops.

To achieve the initial objective of a 20% reduction in N leaching, it was necessary to direct-drill plantain in existing PR/WC + PL pastures every second year, regardless of crop treatment (treatments NC₄, FB₄ and BC₄) or, in the case of NC₃ in years 4 and 7 after sowing (Table 3). The frequency of direct drilling for NC₄, FB₄ and BC₄ required 40% of the milking platform pasture area to be direct-drilled and 10% to be renewed each year. The plantain treatments with less frequent direct drilling (NC₁-NC₂, FB₁-FB₃ and BC₁-BC₃) reduced N leaching but did not achieve the 20% reduction target.

Plantain persistence

The sensitivity analysis, which assumed a faster decline in plantain persistence, resulted in a 19% reduction in N leaching, at most (Table 4). This was for both NC₄ and BC₄. In both cases, plantain was direct-drilled every second year. The assumptions made for plantain persistence had an important effect its ability to reduce its N leaching output in the of the modelled farm system.

Table 4. Milking platform nitrogen (N) leaching results for the plantain persistence sensitivity analysis.

Treatment	kg N ha^−1 yr^−1	Change compared to Control
NC0	26
NC1	25	-4%
NC2	24	-8%
NC3	23	-12%
NC4	21	-19%
FB0	26	0%
FB1	25	-4%
FB2	24	-8%
FB3	23	-12%
FB4	22	-15%
BC0	26	0%
BC1	25	-4%
BC2	24	-8%
BC3	23	-12%
BC4	21	-19%

Notes: The reduction in N leaching is reported relative to the Control. No treatments achieved the targeted 20% reduction in N. NC = no crops; FB = fodder beet; BC = both fodder beet and oats crops. NC₀ is considered the Control (no plantain, fodder beet or oats catch crop).

Cost of mitigation: nitrogen leaching vs. operating profit

The dairy business operated year-round regardless of where the animals were located so splitting costs, such as animal health, between the milking platform and support block was difficult. In addition, no changes were made to the support block that altered the financial predictions. The financial results are therefore reported for the whole farm system with differences ascribed to the changes on the milking platform. Refer to Table 3 and Figure 2 for the sections below.

When the FB₀ crop treatment was applied, operating profit was increased by 4%. When BC₀ was implemented, operating profit increased by 6%. Both are reported relative to the Control and had minimal effect on N leaching.

When plantain was included in No Crop treatments (NC₁-₄), operating profit declined by up to 5%. Where plantain treatments FB₁-₃ were applied to FB₀, operating profit declined, however these treatments were still more profitable when compared with the Control. Treatment FB₄ was 1% less profitable than the Control. The plantain treatments applied to Both Crops (BC₁-₄) reduced operating profit relative to BC₀, but these treatments were still more profitable than the Control (and reduced N leaching). Increasing the frequency of direct-drilling plantain reduced N leaching but also reduced operating profit due to the cost of the seed and the direct-drilling.

In terms of the cost of N mitigation, plantain-only treatments (NC_1-4) reduced profit by $7 to 13 ha⁻¹ kg⁻¹ N mitigated (change in farm profit ÷ change in N leached, from Table 3). As more N leaching was mitigated, the cost of reducing N leaching increased (total and per kg N leaching reduced). For FB_1-4 the cost of N leaching mitigation also increased with greater N leaching mitigation (from an increased profit of $45 to a reduction of -$2 ha⁻¹ kg⁻¹ N mitigated). Where both crops were modelled with plantain, mitigating N leaching resulted in increased profit, but again the increase in profit was smaller with more effective mitigation: BC_1-4 resulted in an increase in profit of $81 to $6 ha⁻¹ kg⁻¹ N mitigated (Table 3).

Four treatments (NC₃, NC₄, FB₄ and BC₄) achieved the initial objective of reducing N leaching by at least 20%. In NC₃, maintenance occurred in years 4 and 7 after sowing. For the remaining treatments, plantain was direct-drilled every second year – the most frequent maintenance schedule. The annual cost for NC₃ was $59 ha⁻¹ NC₄ was $103 ha⁻¹ and $14 ha⁻¹ for FB₄, compared with the Control (NC₀). The relative benefit for BC₄ was $43 ha⁻¹, i.e. this treatment increased operating profit. Of all treatments explored, BC₄ was the only treatment that reduced N leaching by more than 20% and increased operating profit above that of the Control.

Irrigation and drainage

Irrigation and drainage data as estimated by Overseer are presented in Table 5. There was no difference in irrigation rates applied to pasture – on average, all scenarios received 354 mm yr⁻¹. The only difference in irrigation was due to presence of fodder beet crop in FB and BC treatments. Fodder beet crops received an average of 15 mm yr⁻¹. Oat crops were not irrigated. Annual rainfall was 509 mm. The presence of the oat crop (BC treatments) resulted in 6 mm less drainage from the crop block than for FB treatments.

Table 5. Whole farm and fodder beet/oat crop blocks average irrigation and drainage.

Irrigation (mm season^−1)						Drainage (mm yr^−1)
Treatment	Spring	Summer	Autumn	Winter	Annual	Annual
Pasture
NC	84	205	65	0	354	137
FB	84	205	65	0	354	137
BC	84	205	65	0	354	137
Fodder beet/oat crops
NC
FB	3	11	2	0	15	102
BC	3	11	2	0	15	96

Note that in Overseer, fodder crops (fodder beet and oat crops) are rotated through the pasture blocks annually.

Discussion

Feasibility of plantain, fodder beet and catch crops as mitigation options

Previous studies (Beukes et al. 2017, 2018; Pinxterhuis and Edwards 2018) explored the effects of multi-species pastures, fodder beet and catch crops on N leaching at farm scale, via modelling. However, the effect of plantain on N leaching has not been fully examined. Plantain has only relatively recently been shown to be a significant component of a multi-species pastures approach to decreasing N leaching due to its dilution effect on the urinary N concentration of grazing animals (Box et al. 2017; Minnée et al. 2019). This mechanism had not been reflected fully in the farm systems models used. This study focusses on the effect of urine dilution on N leaching through the incorporation of plantain in the diet.

We demonstrated it was possible to reduce N leaching by >20% when plantain was included in the sward at establishment, but only when direct-drilled into the existing mixed swards every second year, irrespective of the inclusion of fodder beet or oats. In three cases, this resulted in a reduction of operating profit: by $53 ha⁻¹ (NC₃), $103 ha⁻¹ (NC₄) and $14 ha⁻¹ (FB₄), and in the case of BC₄ an increase of $43 ha⁻¹ compared with the Control NC₀.

Nitrogen leaching from crop treatments

Dalley et al. (2020) and Waghorn et al. (2019) both reported reductions in urinary N excretion when dairy cows were fed diets containing 23–60% fodder beet in late lactation. Therefore, we expected that pasture supplemented with fodder beet in late lactation (autumn in New Zealand) would reduce overall N leaching, unless leaching on the hectares where the fodder beet was grown was increased to such an extent that it negated the dietary benefits. The latter was indeed the case, and N leaching from the total milking platform area was the same between the Control (NC­₀) and the two crop treatments without plantain (FB₀ and BC₀). Any benefits for N leaching of the catch crop treatment were not noticeable for the total milking platform area, given that the crop area on the milking platform was small (only 4% of the effective area).

Including an oat catch crop immediately following autumn-grazed fodder beet resulted in a 19% reduction in N leaching at the crop block level, in line with soil mineral N observations from Malcolm et al. (2020) and modelling by Teixeira et al. (2016). The effectiveness of oat catch crops in reducing N leaching was dependent on the length of time it was in the ground and the date of sowing – later sowing dates reduce the opportunity for N uptake before the final harvest (Malcolm et al. 2016). In addition, N is leached when drainage occurs (Di and Cameron 2002; Selbie et al. 2015), so drainage following grazed fodder beet crops may happen before a catch crop can be established so less N will be recovered.

Nitrogen leaching from plantain treatments

The plantain treatments explored in this study resulted in far greater reductions in N leaching than the crop treatments. While strategies need to be tailored for different farms (see below), it is likely that of the three forages investigated, plantain will be the easiest to implement and most effective, although the most costly. Plantain reduced farm-scale N leaching the most, achieving up to a 31% reduction in NC₄ compared with the Control, NC₀ (Table 3).

Overall, the Control + plantain treatments (NC₁-NC₄) leached less than their respective crop + plantain treatments (FB₁-FB₄ and BC₁-BC₄) when plantain treatments were applied. The greater impact of plantain on reducing N leaching compared with the crop treatments was due to: (i) the area of the milking platform affected by plantain (96%) was far greater than the area cropped (4%), and (ii) the greater length of time plantain contributed to dietary DM on the milking platform (10 out of 12 months) compared with fodder beet (3 out of 12 months), impacting on the N load of urine patches (Judson and Edwards 2016; Box et al. 2017; Dodd et al. 2019a; Minnée et al. 2020). The more plantain there was at the milking platform scale, the greater the reduction in predicted N leaching.

Research at the time of this study suggested that plantain in PR/WC + PL pasture silage had the same effect as grazed PR/WC + PL pasture on the concentration of urinary N (Judson and Edwards 2016). This was viewed as an advantage as PR/WC + PL silage could be deferred for feeding in February-May – the time of the year where plantain could be most advantageous for reducing N leaching (Shepherd et al. 2011; Selbie et al. 2015). In the Farmax scenarios, the quantity of plantain in the diet was increased between March and May by feeding PR/WC + PL pasture silage harvested earlier in the season. However, any benefit from this strategy was not apparent in the Overseer results, as the proportion of silage with plantain was too small to have an effect. Later versions of Overseer did not reflect the effect of plantain in silage, because more recent research was less convincing (Shepherd 2020).

In a Waikato-based analysis by Romera et al. (2017), it was predicted that 40% reductions in N leaching could be achieved and milk production maintained by farming with diverse pastures (which included plantain making up an unspecified proportion of the sward) as opposed to PR/WC pastures. Based on recent research (Box et al. 2017), it can be assumed that plantain was an important driver for the reported reductions in N leaching, with an additional effect (beyond that on animal urinary N excretion) via soil processes. The latter effect has not been accounted for in the present study. However, while plantain is predicted to have the capacity to noticeably reduce N leaching, its presence alone is not enough to maintain the lower N leaching; a sufficient proportion is required to have a significant impact on urinary N excretion (Minnée et al. 2020).

Plantain persistence

The persistence of plantain populations in PR/WC + PL pastures affects the quantity of plantain in the diet and therefore affects N leaching as seen in the results of this study. For example, including plantain without maintenance in pastures, while using the original persistence curve, had no effect on N leaching. Indeed, it is necessary to maintain the plantain content in pasture through periodic direct drilling to achieve noticeable reductions in leaching. The same held true for the sensitivity analysis, though the targeted 20% reduction in N leaching was not achieved. When plantain was most intensively maintained through direct-drilling every second year, the reductions in N leaching ranged from 15–19% less than the Control.

Long-term data of plantain persistence were not available for PR/WC + PL pastures. Bryant et al. (2019) reported that it was rare for plantain seed sown into existing PR/WC + PL to result in an increase of plantain to more than 30% of pasture dry matter. This is less than the 50% assumed in original persistence curve for this case study. The persistence of plantain is not well understood. We consider the assumptions made based on personal communication with experts were not unrealistic but require checking.

Romera et al. (2017) reported that without N leaching constraints the economic incentives of diverse pastures alone would not be enough for large-scale adoption. Regardless, the environmental benefits and maintenance of milk production via diverse pastures, e.g. Box et al. (2017), are worth further consideration.

Methods for establishing and maintaining plantain (Bryant et al. 2019) and associated costs (Edwards and Pinxterhuis 2018) in mixed swards have been explored, but in practice these factors still are a major limitation of the use of plantain.

Financial performance

Cost of cropping. Growing fodder beet on the milking platform was more profitable than importing barley to achieve the same amount of metabolisable energy offered. Growing an oat catch crop on the milking platform and ensiling this was also more profitable than importing pasture silage, as occurred in the NC and FB scenarios. For the Control (NC₀) and other NC treatments, the cost of imported feed and harvested pasture silage was $1285 ha⁻¹ (Appendix, Table A4), compared with the total feed costs (imported feed, homegrown crops and harvested pasture silage) of $1122 ha⁻¹ for the FB scenarios and $1000 ha⁻¹ for the BC scenarios. Income did not change, so the cost differences were directly reflected in operating profit.

The cost of plantain maintenance. Direct drilling was shown to be the most effective method for establishing new plantain plants into existing pastures (Bryant et al. 2019). Before and after drilling, pastures could be grazed as part of the normal rotation as the new plantain plants could establish between grazings (Bryant et al. 2019). While direct drilling had better establishment rates than broadcasting plantain seed, according to Bryant et al. (2019), broadcasting is a cheaper option, especially when mixed with fertiliser. Therefore, this might be a more attractive option for farmers and could be implemented more frequently to achieve the same plantain contents as infrequent direct drilling. More research into establishment methods and pasture management is needed to maintain plantain contents over more than a few years in pastures on commercial farms, without the need to frequently resow pastures.

Irrigation and drainage

The only difference in irrigation between treatments was where fodder beet crops were grown. The presence of an oat catch crop had a slight mitigating effect on the total drainage from the crop block (6 mm difference between FB and BC treatments Table 5). Overall, less drainage occurred under crop blocks than the average pasture block. This is probably due to less irrigation on the crop blocks.

Comparisons with other modelling studies

Farms differ in numerous ways and are subject to significant environmental and financial variations (Doole et al. 2013), therefore farm management solutions to reduce N leaching need to be tailored to suit individual farms and their context. Beukes et al. (2017) and Beukes et al. (2018) showed that the potential for N leaching reductions from farm systems depends on the farm system, its management and the mitigation strategies applied. Despite this, their results align well with ours.

Beukes et al. (2017) modelled the 2014/2015 dairy season for the milking platform and support block as the Baseline for a mid-Canterbury dairy farm consisting of a 335 ha milking platform and 155 ha support block on a Lismore soil. A hypothetical scenario was modelled with a feed pad, increased effluent area, and FRNL options: fodder beet, which was lifted and fed on the feed pad rather than grazed in situ as for the case study farm, an oat catch crop and one-third of the milking platform sown in mixed pastures (“grasses, herbs and legumes”). This hypothetical scenario averaged a 19% decrease in N leaching from 65 to 53 kg N ha⁻¹ yr⁻¹ for the whole farm.

Beukes et al. (2018) investigated the effects of mixed pastures, catch crops and the replacement of kale crops with fodder beet crops on N leaching and profitability of a North-Canterbury case study farm consisting of a 310 ha milking platform, a 210 ha support block and a 255 ha beef block on a Lismore soil. When FRNL solutions were applied to the entire farm system (plantain in pasture on a third of the milking platform and replacing kale with fodder beet followed by an oat catch crop), N leaching was reduced by 19% (from 31 to 25 kg N ha⁻¹ yr⁻¹) if urinary N effects of plantain were included in the modelling. The reduction was only 8% when using the then publicly available version of Overseer, which did not reflect these effects of plantain. The greatest point of difference between our case study and these previous studies was the inclusion of plantain persistence and its impact on N leaching and profitability. No other studies were found to investigate the role of plantain persistence on N leaching from pasture-based dairy farms.

Further research or modelling of mitigation options such as plantain, fodder beet and catch crops, needs to assess the potential for unwanted consequences, such as soil carbon loss due to frequent pasture renewal (e.g. to maintain plantain content) and cropping in rotation with pasture (Wall et al. 2021). Potential effects such as a different seasonality of pasture production and pasture quality have not been taken into account in our modelling exercise either.

Conclusions

Modelling showed that N leaching could be reduced by 27% with an increase in profitability of 2% for a modelled Canterbury dairy farm by including three feed forages (plantain in pasture and a fodder beet/oat catch crop combination on 4% of the milking platform area), but only if plantain was direct-drilled every second year to maintain plantain content in pasture (scenario BC₄). Increasing the frequency of direct drilling reduced N leaching, but this reduced profitability when compared with the Control, except when both fodder beet and oats were cropped on the milking platform (BC₄).

The use of plantain as a mitigation tool is relatively new and the persistence of plantain in PR/WC + PL pastures remains a major research gap. Analysis of the plantain treatments showed that intensive management was necessary to maintain populations high enough to promote a substantial reduction in N leaching. This leads us to a new critical research question: how can we maintain a large enough population of plantain in PR/WC + PL pastures to achieve substantial, cost-effective reductions in N leaching without unintended consequences such as soil C loss?

Acknowledgements

Special thanks to Chris Rathgen of St Andrews Dairy Ltd. For his willingness to share his farm data for this case study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was co-funded by the dairy farmers of New Zealand through a DairyNZ Master’s Scholarship. The University of Waikato also contributed funds and supervision. This project was also part of the Forages for Reduced Nitrate Leaching programme (FRNL). FRNL’s main funder was the New Zealand Ministry of Business, Innovation and Employment, with co-funding from research partners DairyNZ Inc., AgResearch, Plant & Food Research, Lincoln University, The Foundation for Arable Research and Manaaki Whenua – Landcare Research.

Appendix:

methods

Table A1. Feeds and default values used in Overseer.

Overseer feed (substitutes)	Feed characteristics			Quantity fed (t DM yr^-1)
Feed	DM (%)	N (%)	Energy (MJ ME/kg DM)	NC	FB	BC
Good quality pasture silage (grown)	22.5	2.9	11	230	190	190
Good quality pasture silage (imported)	22.5	2.9	11	160	240	101
Fodder beet forage crop	0.17	0.8 (root) 1.5 (stover) 1.8 (top)	12.5		315	315
Triticale silage (oat silage grown)	33	1.6	10			126
Soya bean meal (protein pellets)	88	8.8	13.3	57	57	57
Wheat grain	88	1.92	13.5	373	373	373
Barley grain	88	2	13	253

Notes: Feed characteristics are from Wheeler (2018) and Wheeler and Watkins (2018). DM = dry matter, MJ ME = megajoules of metabolizable energy, N = nitrogen, NC = No Crops treatments, FB = Fodder Beet treatments, BC = Both Crops treatments.

Table A2. Cost of plantain maintenance via direct drilling plantain seed.

Maintenance expense	Modelled cost
	$/ha
Plantain seed	100
Direct drilling (includes all contractor costs)	100
Maintenance cost ($ ha^−1 maintained area)	200
Maintenance cost ($ ha^−1 total effective area)	20

Notes: Total annual maintenance costs are for 31.3 ha (10% of the farm). Costs are rounded to the nearest $10.

Table A3. Post-model processing for the impact of plantain on nitrogen (N) leaching for treatment NC₀.

Pasture class, based on age	Proportion of plantain in the diet (%PL in diet)	Urinary N load relative to Control (no plantain)	Estimated urine patch N loading rate	Estimated annual N leached for affected area (Overseer output)	Proportion of milking platform affected	Contribution to N leached at the farm level
			kg N ha^−1 yr^−1	kg N affected ha^−1 yr^−1	%	kg N leached effective ha^−1 yr^−1
Control (NC0)		1	750	26
	A = %PL in pasture*0.84	B = −0.6205* A+1	C = 750*B	D	E	F =  D*E
1st year pasture	42%	0.74	555	17	0%	26
2nd year pasture	34%	0.79	594	19	0%	0
3rd year pasture	17%	0.9	672	22	0%	0
Supplements/4+ year old pasture	0%	1	750	26	100%	0
Total					100%	26

Notes: Eighty-four percent of the diet comes from pasture containing plantain. Note that the proportion of plantain (%PL) in the Supplements/4+ year old pasture category is set at 0%, even though the 4+ year old pastures contain 10% PL. Annual N leaching from the milking platform was determined by estimating the urinary N load. This estimate was used in Overseer alongside post-model processing to overcome Overseer’s limitations related to plantain. The values shown are for treatment NC₀ (Control, plantain sown at renewal without further population maintenance). First the proportion of plantain in the diet (%PL in diet; A) was estimated by multiplying the proportion of plantain in the pasture by the proportion of the diet that contained plantain (pasture and harvested pasture silage) for the appropriate class of pasture (Table 1). It was assumed that pasture silage was harvested from across all levels of plantain on the farm, therefore the plantain content of silage was 18%. The relative impact on urinary N of each class of pasture (B) was estimated using Equation (1). Then the urine patch N loading rate (C) was estimated and entered into OverseerScience to estimate the average N leaching affected ha⁻¹ (D) for each category. This was multiplied by the proportion of the milking platform covered under each class (E, see ‘Treatments’ in the Methods section) to calculate the contribution of each class to the whole farm N leaching (F).

Table A4. Expenses database for the Farmax milking platform Control NC₀. 2017/2018 average Marlborough-Canterbury regional data was used.

Category	Expense	$ Total	$ ha^−1	$ cow^-1	$ kg^−1 MS to factory
			312.8 ha	1124 cows	509,633 kg
Wages	Wages	393,400	1258	350	0.772
Stock	Animal health	94,416	302	84	0.185
	Breeding	65,192	208	58	0.128
	Farm dairy	26,976	86	24	0.053
	Electricity	44,960	144	40	0.088
		231,544	740	206	0.454
Feed	Pasture conserved	32,200	103	29	0.063
	Feed crop	0	0	0	0.000
	Bought feed	369,755	1182	329	0.726
	Calf feed	7960	25	7	0.016
		409,914	1310	365	0.804
Other farm working expenses	Fertiliser (excl. N)	115,110	368	102	0.226
	Nitrogen fertiliser	95,124	304	85	0.187
	Irrigation	96,342	308	86	0.189
	Regrassing	31,593	101	28	0.062
	Weed & pest	10,322	33	9	0.020
	Vehicles	50,963	163	45	0.100
	Fuel	30,578	98	27	0.060
	R&M land & buildings	140,760	450	125	0.276
	Freight	20,385	65	18	0.040
		591,179	1890	526	1.160
Overheads	Administration	50,674	162	45	0.099
	Insurance	29,090	93	26	0.057
	ACC	8992	29	8	0.018
	Rates	25,024	80	22	0.049
		113,780	364	101	0.223
	Depreciation	249,720	798	222	0.490
Total Operating Expenses		1,989,537	6360	1770	3.904

Notes: Bolded numbers indicate the cost that was used for each expense category. MS = milksolids; N = nitrogen; R&M = repairs and maintenance. Some expenses are not reported e.g. management wage and cash crop because they do not apply and/or they were incorporated into another category.

Table A5. Expenses database for the 180 ha Farmax Control support block. Average 2017/2018 Marlborough-Canterbury regional data was used.

Category	Expense	$ Total	$ ha^−1
Feed	Pasture conserved	47,600	264
	Feed crop	82,170	457
		129,770	721
Grazing	Owned run-off adjustment	144,000	800
Other farm working expenses	Fertiliser (excl. N)	16,560	92
	Nitrogen	7338	41
	Regrassing	18,180	101
	Weed & pest	5940	33
	R&M land & buildings	29,160	162
		77,178	429
Overheads	Administration	2880	16
	Insurance	4140	23
	Rates	7200	40
		14,220	79
Total operating expenses		365,168	2029

Notes: Bolded numbers indicate the cost that was used for each expense category. N = nitrogen; R&M = repairs and maintenance.