Effects of nitrogen fertiliser application rate on nitrogen partitioning, nitrogen use efficiency and nutritive value of forage kale

Abstract

Winter grazing of forage kale (Brassica oleracea var. acephala L.) has been linked to environmental pollution and negative animal health issues. An experiment with seven nitrogen (N) fertiliser application rates (0–500 kg N ha⁻¹) was carried out at Lincoln, Canterbury, New Zealand to investigate N uptake, partitioning, apparent N use efficiency (aNUE) and nutritive value of forage kale. Final harvest dry matter (DM) yield increased four-fold to 25.5 t ha⁻¹ when 500 kg N ha⁻¹ was applied compared with the 0 kg N ha⁻¹ crops, and N uptake by five-fold to 425 kg ha⁻¹. The inability of the crops receiving ≤100 kg N ha⁻¹ to attain critical leaf area index and critical level of specific leaf nitrogen content resulted in lower DM yield compared with the higher N fertiliser application rates. The aNUE decreased from 111.6 kg DM kg⁻¹ N for the 50 kg N ha⁻¹ crops to 37.5 kg DM kg⁻¹ N for the 500 kg N ha⁻¹ crops. Nitrogen fertiliser application rate had no systematic effect on the crude fibre content of the kale crops. The crude protein content was unaffected by N supply, and ranged between 7.7% and 10.5%. Under conditions of optimal water and nutrient supply on deep soils with high water-holding capacity and long growing season (c. 7 months), it is recommended that N fertiliser is applied at 300 kg N ha⁻¹ to achieve a yield of about 22 t ha⁻¹ and aNUE of ≥51 kg DM kg⁻¹ N applied.

Keywords:

• Brassica oleracea ssp. acephala L.
• critical leaf area index (LAI_crit)
• critical specific leaf nitrogen (SLN_crit)
• nitrogen use efficiency (NUE)
• nutritive values

Introduction

Forage kale (Brassica oleracea var. acephala L.) is an important crop for winter feeding in temperate regions of the world, for example in New Zealand, the UK, continental Europe and the Pacific coast of America (Pelletier & Donefer 1973; Guillard et al. 1995). Forage kale crops have large nutrient requirements, and under conditions of adequate fertiliser and water availability, yields in New Zealand can exceed 20 t dry matter (DM) ha⁻¹ (Wilson & Maley 2006; Fletcher et al. 2007). Wilson et al. (2006) estimated that an 18 t DM ha⁻¹ kale crop removes up to 360 kg nitrogen (N) ha⁻¹ from the soil. Studies also indicate that a sub optimal N supply results in poor forage kale yields (Wilson et al. 2006; Fletcher et al. 2007; Chakwizira et al. 2009, 2012, 2014a). However, excess N applications can lead to accumulation of elevated nitrate-nitrogen in the plant (Fletcher & Chakwizira 2012; Chakwizira et al. 2015), leading to animal health issues, for example, nitrate poisoning (Nichol et al. 2003; Nichol 2007) and/or environmental pollution through nitrate leaching and/or denitrification (Smith et al. 2008; Smith et al. 2012; Monaghan et al. 2013) under winter grazed forage brassicas. Therefore, it is important that the nitrogen supply is closely matched with crop requirements, which must also consider the background soil fertility status (i.e. the amount of already existing soil N), soil moisture levels (Chakwizira et al. 2013) and potential crop yield (Wilson et al. 2006). These factors will inevitably differ from site to site. The environmental effects may be exacerbated by urine returns after grazing (Smith et al. 2012), especially under wet conditions when most of the kale crops are grazed, and also from excessive N fertiliser use.

The effect of N supply on kale DM production and/or N uptake is well understood (Wilson & Maley 2006; Wilson et al. 2006; Fletcher et al. 2007; Chakwizira et al. 2009, 2013). However, there is a gap in the knowledge of the effects of N supply on N use efficiency (NUE; kg DM kg⁻¹ N), feed quality and specific leaf N. Nitrogen supply needs to be closely matched to crop demand to achieve optimum NUE. To achieve high NUE, fertiliser N should be applied when it is required for growth (Binder et al. 2000). However, there are practical and economic limitations to the number and timing of N applications in large-scale operations (Chakwizira et al. 2014b). The development of best management practices is therefore required to achieve high productivity but low environmental impact through reduced N inputs and lower N losses.

The net photosynthesis of crop leaves and/or canopies is closely linked to the N status of the crop (Sinclair & Horie 1989), expressed at canopy level as specific leaf N (SLN; g N m⁻²) and can be represented by radiation use efficiency (RUE; g DM MJ⁻¹; Monteith 1977). RUE decreases when SLN falls below a critical threshold (Sinclair & Horie 1989). Chakwizira et al. (2014a) established a critical SLN (SLN_crit) of 2.6 g N m⁻² for forage kale. These authors also showed that crops growing at leaf N content below the SLN_crit had limited potential to maximise biomass accumulation. Low SLN has been associated with a reduction in the rate of leaf area index (LAI; m² m⁻²) expansion and subsequent reduction in radiation interception. For maximum DM production, crops have to attain a critical leaf area index (LAI_crit: the leaf area index at which the crop intercepted ≥95% of incoming photosynthetically active radiation) of 3.4 m² m⁻² (Chakwizira et al. 2011) earlier in the growing season and maintain full cover, particularly through summer, when radiation receipts are highest.

Generic feed quality values have been reported (Westwood & Mulcock 2012; Rugoho et al. 2014; Westwood et al. 2014) for ‘Gruner’ kale grown under non-limiting conditions of water and nutrient supply. For example, metabolisable energy (ME) values of 10–12 MJ kg⁻¹ DM and neutral detergent fibre (NDF) of 30%–35% DM are consistent with recommendations for optimum animal nutrition (NRC 2001). However, the crude protein (CP) content reported by the same authors varied widely from 7.2%–15% (Westwood & Mulcock 2012; Westwood et al. 2014) to 17.5% (Rugoho et al. 2014). The effects of N supply on feed value attributes have not been determined for forage kale. Moreover, these quality attributes have been shown to vary with N supply in other forage crops, for example, fodder beet (Albayrak & Yuksel 2010) and forage turnips (Turk et al. 2009). The objective of the research reported in this paper was to establish the effects of N application rates on nitrogen uptake and nitrogen use efficiency and determine feed quality attributes of forage kale grown under different N rates.

Materials and methods

The experiment was conducted at the New Zealand Institute for Plant & Food Research, Lincoln (43.83°S, 171.72°E), Canterbury, New Zealand. The site was situated on a deep (>1.6 m), well-drained Templeton silt loam (Immature Pallic soil [Hewitt 2010]; Udic Ustochrept [Soil Survey Staff 1998]) with an available water-holding capacity of about 190 mm m⁻¹ of depth (Jamieson et al. 1995). Physical characteristics of the soil were reported by Martin et al. (1992). The climate at Lincoln is temperate, with mild to cool winters and warm summers. Mean annual rainfall is 600 mm, distributed evenly throughout the year (NIWA 2014). Full weather details for the site during the study period and average long-term climate from 1970 to 2010 have been described by Chakwizira et al. (2014a,b).

The experiment was a randomised block design, with seven rates of N (0, 50, 100, 200, 300, 400 and 500 kg ha⁻¹) replicated four times. The estimated optimum N rate for the site, based on the forage brassica calculator (Wilson et al. 2006; Chakwizira et al. 2012) was 300 kg N ha⁻¹. Briefly, the forage brassica calculator determines the amount of fertiliser applied based on predetermined information such as site history (location and soil properties) and crop details (Chakwizira et al. 2012). The yield of the crop is based on average thermal time and solar radiation for specific locations in New Zealand. However, these calculators assume no water stress and therefore projected yield adjustments are required for rain-fed crops.

Cultivation involved deep ploughing followed by power harrowing. Soil samples to 150 mm depth were collected on 14 October 2011 and the average soil test results were: pH 6.1, P 19 mg kg⁻¹, K 160 mg kg⁻¹, Ca 1500 mg kg⁻¹, Mg 45 mg kg⁻¹, Na 40 mg kg⁻¹, sulphate-S 40 mg kg⁻¹ soil and available mineral N 55 kg ha⁻¹. The amounts of soil nutrients were determined as ‘MAF quick-test units’(Mountier et al. 1966) and converted into mg kg⁻¹ dry soil using the following conversion factors: P, ×1.1; Ca, ×125; K, ×20; and Mg, Na, ×5 (Chapman & Bannister 1994). Soil mineral N (nitrate and ammonium) tests were taken from individual plots before the different rates of fertiliser N were applied on 17 November 2011 (initial soil N) and again after the final harvest of the crops on 5 June 2012 (final soil N) to a depth of 1000 mm. The total soil N averaged per treatment for the initial and final soil N tests are shown in Table 1. The N fertiliser was applied as urea (46% N) and broadcast in equal amounts on 17 November (16 days after sowing [DAS]) and 29 December 2011 (58 DAS). Basal fertiliser comprising 250 kg ha⁻¹ triple superphosphate (TSP; P = 20.5%) and 10 kg ha⁻¹ borate 46 (B = 15%) was broadcast and incorporated into the soil before sowing.

Table 1 Background soil mineral nitrogen levels at the start of the experiment (initial) and at the final harvest (final), total dry matter yield, additional DM yield (difference between final DM and control crops), apparent NUE and stem and total N uptake.

	Mineral soil N (kg ha^−1)		Dry matter yield (t ha^−1)			N uptake (kg ha^−1)
Treatments (kg N ha^−1)	Initial	Final	Total	Additional^a	aNUE (kg DM kg^−1 N)	Total	Stem
0	84.9	32.8	6.7	–	–	85	49
50	110.7	26.6	12.3	5.6	111.6	144	91
100	91.6	33.0	15.1	8.4	83.6	153	102
200	68.8	27.2	19.1	12.5	61.8	226	155
300	103.5	38.4	21.9	15.2	50.8	281	186
400	75.5	46.1	21.4	14.7	36.7	343	244
500	84.6	47.5	25.5	18.7	37.5	425	282
LSD (5%)			3.2	3.5	38.0	66	53
P value			<0.001	<0.001	<0.01	<0.001	<0.001
d.f.			17	14	14	18	18

^aDifference between total DM and control yields for each treatment.

The cultivar ‘Gruner’, a giant kale (Westwood et al. 2014) was drilled on 2 November 2011 at 4 kg ha⁻¹ in 150 mm rows into a cultivated seedbed using a Taege drill with Oyjord cone. Plot size was 3.9 × 10.0 m (area = 39 m²), with 0.5 m between plots. Seeds were pelleted with ‘Superstrike^® Brassica’ (a.i thiamethoxam, iprodiane and thirum; PGG Wrightson, Christchurch, New Zealand) to guard against springtails (Bourletiella spp.) and fungal infections (Salmon & Dumbleton 2006). Insecticides Diazinon 800 EC and Lorsban™ 50 EC (a.i. 50% chlorpyrifos) were applied to kale seedling at 1 L ha⁻¹ on 8 November and 20 December 2011, respectively. The herbicide Banvel^® (a.i. 200 g/L dicamba) was applied on 20 December 2011.

Measurements

Dry matter yield

Dry matter harvests were taken at five-weekly intervals from 17 January to the final harvest on 5 June 2012. Quadrat size was 0.5 m² for the sequential harvests and 1 m² for the final harvest. The number of plants and total fresh mass per quadrat was determined in the field at each harvest. A representative five-plant subsample was retained on each harvest date to determine total DM yield. Each subsample was partitioned into leaf lamina, petiole and stem component. However, the DM partitioning was reported as leaf (combined petiole and lamina) and stem yield components. Dry mass was determined after drying at 60 ^oC to a constant mass. A subsample of leaf lamina was used to determine leaf area, using a leaf area meter (LI-COR model LI-3100; Lincoln, Nebraska, USA). The total leaf area per quadrat was determined and used to calculate the leaf area index. The accumulated LAI and a critical LAI (LAI_crit) of 3.4 m² m⁻² (Chakwizira et al. 2011) were used to interpret canopy development process during the season.

Nitrogen uptake and use efficiencies

Total N concentration for both the leaf and stem fractions were determined by Dumas combustion using a LECO CNS-200 analyser (LECO Corporation, St Joseph, MI, USA). Total N uptake was then calculated by multiplying plant component DM yield by the concentration of N in the harvested component. Nitrogen use efficiency was defined as the ratio of additional DM yield to fertiliser N input; often termed the agronomic efficiency (Singh et al. 1998; Asseng et al. 2001; Baligar et al. 2001). This was referred to as apparent NUE (aNUE) and was calculated as expressed in Equation (1) (Singh et al. 1998; Zemenchik & Albrecht 2002):

(1)

where Nx = N rate > 0 and N₀ is crop yield for the 0 kg N ha⁻¹ crops in the same replication.

It is important to note that aNUE calculations do not account for mineral N stored in the soil or derived from mineralisation of soil organic matter, and therefore variations in NUE per treatment are likely. This differs from the traditional calculations of NUE as a quotient of total DM (Table 1) and total N available (soil N plus fertiliser N) (Moll et al. 1982).

Specific leaf nitrogen (SLN; g N m⁻² leaf area) was calculated as the quotient of leaf nitrogen content and specific leaf area (Borrell & Hammer 2000; Massignam et al. 2001). The nutritive values were determined on a whole crop basis using near-infrared reflectance spectroscopy at feedTech laboratory at AgResearch, Palmerston North, New Zealand (Corson et al. 1999). Nutritive quality attributes measured included crude protein, acid detergent fibre and neutral detergent fibre (ADF, NDF) content, ash, lipid, soluble sugars and starch (SSS) and metabolisable energy (ME; MJ kg⁻¹ DM).

Data analyses

Data were analysed using analyses of variance (ANOVA) in GenStat v.14 (VSN International, Hemel Hempstead, UK). Significant interactions and main effects were separated using Fisher’s protected least significant difference (LSD) tests (α = 0.05). Where values show P < 0.1, a trend is indicated in the text. There was a wide variation in the background soil mineral N levels among the plots (Table 1). However, an analysis with background N as a covariate did not show a significant effect. As a result, analyses included N application rate as the independent variable.

Results

Total DM yield at the final harvest for the 0 kg N ha⁻¹ crops was 6.7 t ha⁻¹. When N was applied at 50 kg N ha⁻¹, DM yield increased (P < 0.001) by 84% to 12.3 t DM ha⁻¹, and was almost five-fold higher (P < 0.001) when 500 kg N ha⁻¹ was applied compared with the 0 kg N ha⁻¹ (Fig. 1A). However, the differences in incremental proportion of DM yield decreased with each successive rate of N. There were no yield differences between the 300 and 400 kg N ha⁻¹ rates at the final harvest. The DM yields for the 0 kg N ha⁻¹ crops were consistently lower (P < 0.001) than fertilised crops throughout the growing period (Fig. 1A).

Figure 1 Accumulation of: A, Leaf (closed symbols) and stem (open symbols) DM; B, leaf area index; C, N uptake; D, specific leaf N for kale crops grown under different N rates. ● control, ▼ 50, ■ 100, ◆ 200, ▲ 300, ⬣400 and * 500 kg N ha¹, at Lincoln, in 2011–2012. Vertical bars represent the least significant differences (LSD _0.05; d.f. = 18). LAI_crit = critical leaf area index; SLN_crit = critical specific leaf nitrogen.

The partitioning of DM to the leaves and stems showed a consistent pattern. Leaf DM increased with increasing total biomass up to 146 DAS. Stem DM, particularly for the ≥200 kg N ha⁻¹, continued to increase to the final harvest at 218 DAS. Stems contributed about 80%–86% of the total DM at the final harvest, across all treatments. At the final harvest, the maximum leaf and stem DM were 5.4 and 21.1 t ha⁻¹, respectively, both observed in the 500 kg N ha⁻¹ treatment.

The LAI increased (P < 0.001) with N application (Fig. 1B) and was consistently higher with increasing N supply throughout the growing season. For example, at 77 DAS, LAI for the 0 kg N ha⁻¹ crops was 1.6 m² m⁻² compared with 3.2 m² m⁻² when 500 kg N ha⁻¹ was applied. Similarly, at the end of the growing season, LAI was 1.3 m² m⁻² and 5.3 m² m⁻² for the 0 and 500 kg N ha⁻¹ crops, respectively. However, there were no differences in LAI between the 400 and 500 kg N ha⁻¹ crops. The maximum LAI of 6.1 m² m⁻² was attained 146 DAS for the 500 kg N ha⁻¹ crops. The LAI for the crops receiving ≤100 kg N ha⁻¹ were consistently below the LAI_crit throughout the growing season, compared with the crops receiving ≥300 kg N ha⁻¹ that maintained full cover through to the final harvest. The 200 kg N ha⁻¹ crops maintained full cover for 104 days.

The relationship among leaf attributes (Fig. 2A–D) showed strong positive correlations (R² ≥ 0.76). Leaf DM accounted for 84% of the variation in total DM, and for every 1 t DM ha⁻¹ increase in total DM, leaf DM increased by 150 kg DM ha⁻¹.

Figure 2 Correlation between: A, Total leaf DM and total DM yields (y = 0.15x + 0.27; R² = 0.84); B, leaf area index (LAI) and total leaf DM (y = 1.30x + 0.59; R² = 0.76); C, LAI and total leaf nitrogen uptake (y = 0.038x + 0.26; R² = 0.88); D, total leaf DM and total leaf nitrogen uptake (y = 0.026x + 0.93; R² = 0.89).

Total N uptake increased (P < 0.001) five-fold with N application, from 85 kg ha⁻¹ for the 0 kg N ha⁻¹ crops to 425 kg ha⁻¹ for the crops receiving 500 kg N ha⁻¹ (Table 1). Nitrogen uptake for the 0–400 kg N ha⁻¹ treatments levelled off at about 111 DAS on average (Fig. 1C). Nitrogen uptake maxima were markedly affected by N supply. When N was applied at 0 and 50 kg N ha⁻¹, N uptake plateaued at about 85 and 130 kg N ha⁻¹, respectively, compared with 300 kg N ha⁻¹ for the 300 and 400 kg N ha⁻¹ application rate. At least 70% of the total N uptake was within the first 111 DAS across all treatments. At the final harvest, the proportion of total N stored in the stems differed (P = 0.05) with N fertiliser application rate (Table 1), with 49 and 91 kg N ha⁻¹, for the 0 and 50 kg N ha⁻¹, respectively, stored in the stems. This represented c. 62% of the total N uptake. Similarly, stems stored 102 and 282 kg N ha⁻¹, which was c. 66% of the total N in the above-ground N when 100 and 500 kg N ha⁻¹ fertiliser was applied, respectively.

There was a significant (P = 0.002) interaction between the time of harvest and specific leaf N content (Fig. 1D). Specifically, there were treatment effects earlier in the season, up to 111 DAS, with no significant differences (P = 0.244) observed from then until final harvest at 218 DAS. Treatment effects were inconsistent for the ≤100 kg N ha⁻¹ treatments over time. Treatments receiving ≤100 kg N ha⁻¹ failed to achieve the SLN_crit throughout the growing season.

Nitrogen use efficiency

The apparent NUE at the final harvest decreased (P ≤ 0.01) with increasing N application rate (Table 1). For example, aNUE decreased from 111.6 kg DM kg⁻¹ N when 50 kg N ha⁻¹ was applied to 37.5 when 500 kg N ha⁻¹ was applied. However, there were no differences in aNUE among the treatments receiving ≥200 kg N ha⁻¹. Between 26.6 and 47.5 kg N ha⁻¹ of the total soil plus fertiliser N remained in the soil at the end of the season across the treatments, with higher values associated with crops receiving ≥300 kg N ha⁻¹.

At the final harvest, N fertiliser application rates had no consistent effect on nutritive value of ‘Gruner’ kale (Table 2). Nitrogen fertiliser rates were statistically significant (P ≤ 0.06) for ADF and ME, but the responses were inconsistent with respect to N treatment level. For example, ADF values were similar for the 0 kg N ha⁻¹ crops and other treatments, but the crops receiving 400 kg N ha⁻¹ had lower values than the 100 and 200 kg N ha⁻¹ crops. Moreover, CP was not affected (P = 0.48) by the application of N, and ranged between 7.7% and 10.5% across the treatments (Table 2). The crude fibre content averaged 29.1% for ADF and 42.7% for NDF. These were above the minimum requirements for animal production. The ME and SSS were at 12.2 MJ kg⁻¹ DM and 30.8%, respectively.

Table 2 Nutritive value of forage kale grown under seven nitrogen application rates at Lincoln, Canterbury in the 2011–2012 season and optimum range (NRC 2001; Nichol et al. 2003).

Treatment	Quality attribute*
kg N ha^−1	ADF	Ash	CP	Lipid	ME	NDF	SSS
0	28.7	6.8	7.7	1.9	12.4	41.9	33.2
50	27.5	7.5	10.2	1.8	12.7	41.9	31.3
100	30.6	6.8	8.6	1.6	11.9	43.5	29.8
200	30.8	6.5	7.8	1.6	11.8	46.0	30.0
300	28.1	7.3	9.4	2.1	12.3	41.1	32.0
400	27.4	7.3	10.5	1.7	12.7	42.0	30.7
500	30.3	7.2	9.3	1.8	11.9	42.3	28.8
LSD	2.6	0.7	3.4	1.1	0.7	4.7	6.0
P=	0.04	0.12	0.48	0.96	0.06	0.41	0.76
Mean	29.1	7.1	9.1	1.8	12.2	42.7	30.8
Optimum	>21	0.5	>12		10–11	>33	36–44

ADF, acid detergent fibre; CP, crude protein; ME, metabolisable energy; NDF, neutral detergent fibre, SSS, soluble sugars and starch.

*All units are % DM, except for ME (MJME kg⁻¹ DM).

Discussion

The increase in DM yield with N supply is consistent with reports in the literature (Fletcher et al. 2007; Chakwizira et al. 2009, 2013, 2014a, 2015). The low DM yield for the treatments receiving ≤100 kg N ha⁻¹ (Fig. 1A; Table 1) may be attributed either to: (1) the inability of these crops to close their canopies (Fig. 1B); and/or (2) the low SLN content for these crops earlier in the growing season (Fig. 1D). The ≤100 kg N ha⁻¹ crops did not attain the LAI_crit throughout the growing season (Fig. 1B), resulting in reduced radiation interception and the subsequent low DM yields. The crops receiving 200 kg N ha⁻¹ reached and maintained leaf area above LAI_crit for 104 days, while crops receiving ≥300 kg N ha⁻¹ maintained leaf area above LAI_crit for a period of about 141 days before final harvest. The implication was that both attainment of the LAI_crit (full cover) and the duration above LAI_crit are important as they determine the total amount of radiation intercepted and consequently total DM yield produced.

The SLN differed with N fertiliser rates (Fig. 1D), with differences more pronounced during the summer months (January–February). This is the period of maximum radiation load and therefore highest potential growth rate. During this period, the SLN of ≤2.4 g N m⁻² leaf for the treatments receiving ≤100 kg N ha⁻¹ (Fig. 1D) was lower than the critical level of SLN (SLN_crit) of 2.6 g N m⁻² established by Chakwizira et al. (2014a). These authors showed substantial reduction in potential biomass accumulation for crops grown below the SLN_crit. These observations of low SLN caused by reduced leaf area expansion and increased leaf senescence have also been observed in other crop species (Massignam et al. 2009). The net photosynthesis of crop leaves and/or canopies is closely linked to the SLN (Sinclair & Horie 1989). Furthermore, the net photosynthesis of a crop can be represented by radiation use efficiency (Monteith 1977), which decreases when SLN falls below a critical threshold (Sinclair & Horie 1989). Therefore, variation of SLN in this experiment may have contributed to the differences in DM yields reported.

Nitrogen uptake increased (P < 0.001) with N supply (Fig. 1C). At the end of the season (Table 1), the crops receiving 500 kg N ha⁻¹ had taken up a total of 425 kg N ha⁻¹, which was five-fold higher than the 0 kg N ha⁻¹ crops and double those receiving 200 kg N ha⁻¹. These results were consistent with those reported by Wilson et al. (2006) and Fletcher et al. (2007). Moreover, most of the N was taken up earlier in the season (111 DAS), which was consistent with reports for other forage crops, such as fodder beet (Chakwizira et al. 2014b). When factoring in the average background available soil N at the site of 89 kg ha⁻¹ (Table 1), the crops took up 64%–68% of the applied N for the crops receiving ≥100 kg N ha⁻¹. The deep, well-drained Templeton silt loam (Jamieson et al. 1995) soil can influence the way both soil and fertiliser N is used by the crops. These soils are able to keep the N in the root zone for longer, thus making it available to the crops, which may explain the high N uptake in Fig. 1C. This, coupled with the irrigation of the crops, meant the crops were able to take up most of the total N from the soil (85–285 kg N ha⁻¹; Fig. 1C) when 0–200 kg N ha⁻¹ was applied, compared with the 114–136 kg N ha⁻¹ reported by Stevens & Carruthers (2008) for similar levels of total N input for rain-fed forage kale. Kale crops grown under irrigation in deep soils have a higher yield potential (Wilson et al. 2006) than rain-fed crops in shallower soil (Stevens & Carruthers 2008; Chakwizira et al. 2009). The implication is that the rate of N application should be based on soil depth, water availability and length of growth season, as they all determine the yield potential for any site.

The amount of N remaining in the soil (Table 1) was higher for the crops receiving ≥300 kg N ha⁻¹ at 38–48 kg N ha⁻¹ than the 27–33 kg N ha⁻¹ for crops receiving ≤200 kg N ha⁻¹. Crops with high N inputs therefore present as high risk for winter nitrate leaching because of slow growth and low N uptake potential during winter. Moreover, bare land following in situ winter grazing is susceptible to high drainage losses and associated leaching of residual N. This, coupled with the fact that more than 70% of the ingested N is returned to the soil in the urine (Haynes & Williams 1993), means large amounts of the N taken up by the crops (Fig. 1C) can potentially be returned to the soil and exposed to leaching, causing environmental pollution. Recent studies show that annual nitrate leaching losses under winter grazed forage brassicas range between 52–81 kg N ha⁻¹ (Smith et al. 2008; Smith et al. 2012; Monaghan et al. 2013). This includes both residual and urine-returned N. The data we show for the residual N (Table 1) with no urine addition are comparatively high.

High crop N uptake may also contribute to animal health problems during grazing, for example, excess nitrates (Fletcher & Chakwizira 2012; Chakwizira et al. 2015) leading to nitrate poisoning (Nichol 2007). However, crops receiving ≤100 kg N ha⁻¹ took up more N than was applied; this implies there was some additional N in the soil, which could be attributed to the release of organic forms of soil N via mineralisation processes.

The inverse relationship between yield and aNUE (Table 1) was consistent with previous studies on maize (Islam & Garcia 2012) and fodder beet (Chakwizira et al. 2014b). The aNUE of 37–112 kg DM kg⁻¹ N (Table 1) in the current experiment was consistent with the range of 29–96 kg DM kg⁻¹ N applied (when recalculated excluding the yield of the 0 kg N ha⁻¹ crops) for ‘Gruner’ kale (Fletcher et al. 2007) grown under the same range of N application rates, at the same site.

The high stem proportion of total DM (84%; Fig. 1A) was consistent with previous reports (Fletcher & Chakwizira 2012; Westwood & Mulcock 2012) and has implications for the overall quality of the feed. Westwood & Mulcock (2012) reported low CP but high ADF and NDF for ‘Gruner’ kale compared with the short and medium height cultivars ‘Kestrel’ and ‘Regal,’ respectively, which are characterised by a high leaf to stem ratio (Adams et al. 2005; Chakwizira et al. 2010). Neither ADF nor NDF differed systematically with N application (Table 2), which is inconsistent with previous reports for other forage brassica crops, for example, forage turnips (Turk et al. 2009). These authors reported a decrease in ADF from a weighted mean of 18% for the 0 kg N ha⁻¹ crops to 13.3% for the 150 kg N ha⁻¹ crops, and NDF from 15% to 11% for the same treatments.

The CP content did not respond to N supply, with an average of 9.1% across the treatments. This was also inconsistent with previous reports for forage brassica crops (Pelletier et al. 1976; Turk et al. 2009; Rugoho et al. 2010) and other forage crops (e.g. fodder beet; Turk 2010) in which CP increased with N supply. Turk et al. (2009) reported a CP increase with N application, from 11% for the 0 kg N ha⁻¹ crops to 13.6% when 150 kg N ha⁻¹ was applied. The CP (or N concentration = CP/6.25) was higher for the leaves than stems (Pelletier et al. 1976) and therefore the proportion of the leaf and stem of total DM will determine the final nutritive value of the feed. The higher the stem component (Fig. 1A), the lower the CP% (Table 2). Westwood & Mulcock (2012) showed CP differed with kale cultivars, increasing in ascending order ‘Gruner,’ ‘Regal’ and ‘Kestrel; this is the same order in which the proportion of the leaf to total DM increases for the respective cultivars. These authors also showed that the nutritive values varied with brassica species, for example, swedes were high in ME compared with kale.

Conclusions

Dry matter yield increased four-fold to 25.5 t ha⁻¹ when 500 kg N ha⁻¹ was applied compared with the 0 kg N ha⁻¹ crops, and N uptake five-fold to 425 kg ha⁻¹. The inability of the crops receiving ≤100 kg N ha⁻¹ to attain critical leaf area index and critical specific leaf nitrogen content resulted in lower DM yield compared with the higher N fertiliser application rates. The aNUE decreased from 111.6 kg DM kg⁻¹ N for the 50 kg N ha⁻¹ crops to 37.5 kg DM kg⁻¹ N for the 500 kg N ha⁻¹ crops. Nitrogen application rate had no systematic effect on the nutritive value of kale crops. The average CP% across all treatments (about 9.1%) was lower than the optimum requirements for non-lactating animals during winter. The SLN averaged 2.4 (2.1–2.7) g N m⁻² leaf and was unaffected by N treatments. The absolute increase in DM yield with N supply decreased when N was applied at rates above 300 kg N ha⁻¹. Under similar conditions to this study—optimal irrigation, base fertilisers, deep soils with high water-holding capacity and a 7-month growing season—it is recommended that N is applied at 300 kg N ha⁻¹ to achieve a yield of about 22 t ha⁻¹ and NUE of ≥51 kg DM kg⁻¹ N applied.

Acknowledgements

This work was conducted through the Pastoral 21 Environment Programme (C10X1117; objective 7), jointly funded by MBIE, DairyNZ, Fonterra and Beef + Lamb New Zealand. We also thank Plant & Food Research employees who were involved in the experiment, particularly Mike George and Alexandre Michel.