The effect of nitrogen and sulphur on the grain yield and quality of canola (Brassica napus L.) grown in the Western Cape, South Africa

Abstract

On average low grain yields of less than 1.5 t ha^-1 are obtained with canola in the Western Cape province of South Africa. Injudicious use of fertiliser has been suggested as a major cause, hence the study was done to determine the effect of soil and climatic conditions, as experienced at different localities, on grain yield and quality response of canola to nitrogen (N) and sulphur (S) applications in a three-year study. Treatments consisted of five N (0, 30, 60, 90 and 120 kg N ha^-1) and three S (0, 15 and 30 kg S ha^-1) rates laid out in a randomised complete block design with a factorial split-plot arrangement. The study showed that the effect of N supply increased with S supply and was greater in high rainfall areas and seasons than in drier conditions. Low rainfall during flowering and podding may limit yield responses to N applications, with applications of 120 kg N ha^-1 and 15–30 kg S ha^-1 needed in high rainfall areas to obtain yields of more than 2.0 t ha^-1. Application of S also improves the oil content of canola though this compromises protein content.

Keywords:

• canola yields
• nitrogen
• oil and protein content
• sulphur

Introduction

Canola (Brassica napus L.), an emerging oilseed crop, has become a plant of major economic importance, with global production increasing to almost 60 million metric tonnes worldwide in 2010 (Daku et al. 2012). The crop offers potential for use with production of high-quality edible oil, as a source of plant protein meal for feeds, or a potential alternative as biofuel (Rayner 2002; Abdallah et al. 2010).

Production of the crop in South Africa is still low compared to major global producers. The Monthly Food Security Bulletin of South Africa (DAFF 2012) reports on canola production statistics for the 2011 season with a total of 59 490 t on an area of 43 510 ha indicating a national average yield of 1.37 t ha^-1. Increasing land area for canola production within the Western Cape, where the crop is mostly planted, is restricted due to competition with other winter crops, hence yield increases should focus on an increase per land area. Fertilisation with nitrogen (N) has been reported to increase yield considerably in other canola-growing regions of the world (Hocking et al. 1997; Jan et al. 2002; Svečnjak and Rengel 2006; Balint et al. 2008) and in controlled environments (Ngezimana and Agenbag 2013), but showed poor and or variable responses in the Western Cape production area (Hardy et al. 2004).

In order to obtain high grain yields with good quality in canola, it is necessary to balance the nutritional supply and the synthesis of oil and crude protein in the seeds (Ahmad et al. 2007). Canola is a heavy nutrient feeder such that the requirements of various macronutrients (particularly N and sulphur [S]) are higher in canola compared to cereals (Oplinger et al. 2000; Gan et al. 2008). Sulphur plays a very crucial role in canola, affecting growth and yields, with its requirements about four times higher than that of wheat or maize (Abdallah et al. 2010).

McGrath and Zhao (1996) reported that low S supply suppresses the development of reproductive organs and may lead to silique abortion and decreased seed yield and oil content in canola. Sulphur forms an integral part of several amino acids and hence affects protein content of the seed (Jan et al. 2010). Malhi et al. (2007) has shown the effects of S deficiency and applied S to be more pronounced on seed than straw yield. Grain yield responses to applied S only occurred when N was applied and tended to increase as more N was applied (Breennon and Bolland 2008). Producing high-yielding canola therefore is only possible with sufficient supply of both N and S. Usually, addition of S within the range of 10–30 kg S ha^-1 has been found adequate for optimum seed and oil yield with 15–20 kg S ha^-1 suggested per tonne yield per year (McGrath and Zhao 1996). Jackson (2000) showed that about 20 kg S ha^-1 was adequate for optimum seed and oil yields, and total plant N, phosphorus, potassium and S uptake averaged 140, 25, 170 and 60 kg ha^-1, respectively. These results were obtained at the optimum levels of N and S to give a yield potential of 3 t ha^-1. The optimum S supply and uptake, however, depends on the N:S supply ratio of 7:1 suggested by Janzen and Bettany (1984), as well as soil and climatic conditions.

No research on the response of canola to S and N fertiliser applications has been done previously in the Western Cape canola-producing areas. The major aim of this research was to determine the effect of N and S interaction on the grain yield and quality of canola grown at different localities in the Western Cape province of South Africa.

Materials and methods

Locality

Field experiments were conducted at the Elsenburg (33°51′ S, 18°51′ E; 117 m above sea level [asl]), Langgewens (33°17′ S, 18°40′ E; 91 m asl) and Roodebloem (34°22′ S, 19°52′ E; 132 m asl) experimental stations of the Western Cape province of South Africa during the canola seasons of 2009, 2010 and 2011, as well as at Welgevallen (33°52′ S, 18°42′ E; 119 m asl) during 2010 and at Altona (33°42′ S, 18°37′ E; 42 m asl) during 2011. In 2009, Elsenburg received the highest total rainfall (May to Ngezimana and Agenbag October) of 568.3 mm, whilst Roodebloem and Langgewens received a total rainfall of 339.4 and 285.2 mm, respectively (Figure 1). In 2010, Elsenburg received more rain (442.9 mm) than Welgevallen (426.6 mm), Langgewens (366 mm) and Roodebloem (320.9 mm). During the canola growing season, the 2011 rainfall totals were far below that of previous years at most localities except at Langgewens (308.6 mm), which was higher than in 2009, although below that received in 2010 at the same locality. During 2011, Elsenburg received a higher total rainfall (325.8 mm) than Langgewens (308.6 mm), Altona (295.9 mm) and Roodebloem (241.5 mm). Temperatures were generally favourable for canola production even though the 2010 season was considerably warmer at all localities with temperatures experienced being above those of 2009 and 2011.

Figure 1: Rainfall data for Elsenburg (ELS), Langgewens (LG), Roodebloem (RB), Welgevallen (WLG) and Altona (ALT) during the (a) 2009, (b) 2010 and (c) 2011 growing seasons

Soil characteristics at the localities were variable and the N-mineralisation potential in the 0–200 mm soil profile was calculated and found to be on average 56.8, 59.9, 92.0, 93.5 and 111.3 kg N ha^-1 for the Langgewens (2009–2011), Roodebloem (2009–2011), Elsenburg (2009–2011), Welgevallen (2010 only) and Altona (2011 only) experimental sites, respectively. The present N-mineralisation potential is acceptable, provided that inorganic N fertiliser was added in production. These mineral-N values were calculated from nitrate and ammonium-N contents as measured at planting and mineralised during the first 40 d after planting, using the indophenol-blue (Keeney and Nelson 1982) and salicylic acid methods (Cataldo et al. 1975) to give an indication of the potential of the soil to supply N to the crops. Sulphur content of the soil decreased with depth at all localities and was on average below the required critical level of 6 mg S kg^-1 required for canola production (Peverill et al. 1999) at all localities except for Elsenburg, Langgewens and Welgevallen (Table 1). However, soils at Roodebloem had higher values for carbon (2%) than at Altona (1.27%), Welgevallen (1.08%), Elsenburg (0.79%) and Langgewens (0.48%), and hence are likely to have a higher cation exchange capacity (Rowell 1994).

Table 1: Soil sulphur content at planting at Altona, Elsenburg, Langgewens, Roodebloem and Welgevallen. Mean values for the 2009–2011 period are presented. Sulphur < 6.0 mg kg^-1 is the critical soil nutrient level at which the nutrient became deficient (Canola production manual, undated pamphlet sponsored by PNS and THRIP; Peverill et al. 1999)

Soil characteristic	Altona1			Elsenburg			Langgewens			Roodebloem			Welgevallen2
Depth (cm)	15	30	45	15	30	45	15	30	45	15	30	45	15	30	45
S (mg kg^-1)	7	3.3	2.1	8.7	5.89	3.97	6.63	4.35	4	4.01	2.18	2.07	11	6.2	7.2

1 2011 only

2 2010 only

Experimental design and treatments

The experiments were laid out in a randomised complete block design with a factorial split-plot arrangement. Treatments consisted of three S rates (0, 15 and 30 kg S ha^-1) as a main factor and five N rates (0, 30, 60, 90 and 120 kg N ha^-1) as a subfactor. All treatments were replicated four times. Nitrogen was applied in the form of limestone ammonium nitrate (LAN) with 28% N, whereas S was applied in the form of gypsum (CaSO₄.2H₂O) with 16% S. All S treatments were applied at planting whilst N was applied in 30 kg ha^-1 splits at planting and after every 30 days after planting (DAP) depending on treatment. At crop maturity, canola crops were harvested with a plot harvester in 5 m × 3 m plots and, after cleaning and weighing, grain yields were converted to kilograms per hectare. The canola grain protein and oil contents were analysed at the Department of Agriculture: Western Cape at Elsenburg using near-infra-red technology.

Data recorded was analysed, using split-plot design analysis of variance (ANOVA) with two factors as main plots and one subplot factor using Statistica 11 (StatSoft, Inc., Tulsa, OK, USA). To find the response of the treatments on different localities, locality was considered as a factor. Considering the fact that different cultivars were planted in the three different seasons and also that three localities were planted for each year (Roodebloem, Elsenburg and Langgewens), whilst Welgevallen and Altona were included in 2010 and 2011, respectively, grain yield data for the three seasons was analysed separately. For grain oil and protein content of canola, growing seasons were used as replications to do the statistical analysis (only Langgewens, Roodebloem and Elsenburg, where data was available for three years). Main and interaction effects were compared using the least significant difference (LSD) test at the 5% level of probability. Any treatment means found to be significantly different were separated using Fischer's protected LSD_0.05 and only statistically significant data is presented and discussed (Table 2). Means from the Statistica analyses were then subjected to best-fit regression curves plotted with Microsoft Office Excel, using polynomial equations to draw N and S response curves.

Table 2: Summary of results from ANOVA tables on yield data analysis

Factor	2009	2010	2011
Location	***	***	***
Sulphur (S)	***	***	***
Nitrogen (N)	***	***	***
Location × S	*	**	**
Location × N	*	***	ns
S × N	ns	***	*
Location × S × N	ns	**	ns

* P <0.05;

** P <0.01;

*** P <0.001; ns, not significant at P = 0.05

Results and discussion

Grain yield

During 2009, grain yield was significantly affected by N application rate, but responses differed between localities (Figure 2). Without any N fertiliser applied (N0), mean grain yield at Elsenburg (2 260 kg ha^-1) were significantly higher compared to Roodebloem (1 123 kg ha^-1) and Langgewens (890 kg ha^-1). The highest canola yields of 3 683 kg ha^-1 at Elsenburg and 1 631 kg ha^-1 at Roodebloem were measured with the fertilisation rate of 120 kg N ha^-1. However, yields at Langgewens (1 228 kg ha^-1) did not increase significantly when more than 30 kg N ha^-1 was applied (Figure 2). This difference in response at different localities is most probably due to differences in soil moisture (rainfall during the grain-filling period) and to a lesser extent differences in soil N content, because moisture availability greatly affects the yield response of canola to N fertilisation (Jackson 2000; Al-Barrak 2006).

Figure 2: Canola yields harvested at Langgewens, Elsenburg and Roodebloem localities as a result of different nitrogen fertilisation rates (0, 30, 60, 90 and 120 kg ha^-1) in the 2009 season

In 2009, Elsenburg received significantly more rain that was evenly distributed throughout the canola growing season compared with Langgewens where very low rainfall was experienced during the flowering and pod-filling stages (August, September and October) (Figure 1). Roodebloem also experienced low rainfall during August and September, but because of the later planting date yields probably benefit from the high rainfall during October. These results supported that of earlier studies (Mendham and Salisbury 1995; Champolivier and Merrien 1996), which found that most of the yield components were highly affected by water shortage occurring from flowering to the end of the seed setting stage.

Grain yield also differed significantly due to S application rates at all localities because of the generally low S contents (<6 mg kg^-1 soil) of the soil. In the 2009 season, the highest S application rate of 30 kg ha^-1 produced a mean yield of 2 115 kg ha^-1 compared to 1 805 kg ha^-1 when no S was applied. However, as shown in Table 3, application of S beyond 15 kg ha^-1 did not increase yield significantly at any of the localities except for Langgewens. Jackson (2000) reported similar findings of increasing grain yields with addition of S, and recommended that optimum seed yield occurred at 20 kg S ha^-1 in the western USA.

Table 3: Canola yields (kg ha^-1) harvested at Langgewens, Roodebloem, Elsenburg, Welgevallen and Altona localities at sulphur fertilisation rates (0, 15 and 30 kg ha^–1) in 2009, 2010 and 2011 season. Means for each season with at least one common superscript letter are not significantly different (LSD_0.05). (–) Denotes location not used in that year

Location	S rate (kg ha^-1)	2009	2010	2011
Langgewens	0	1 173^f	1 344^f	964^g
	15	1 197^f	1 462^e	1 124^f
	30	1 314^e	1 534^e	1 365^e
Roodebloem	0	1 535^d	1 293^g	2 552^c
	15	1 636^c	1 427^e	2 897^a
	30	1 720^c	1 471^e	3 032^a
Elsenburg	0	2 708^b	2 651^b	1 180^f
	15	3 089^a	2 943^a	1 339^e
	30	3 310^a	3 061^a	1 585^d
Welgevallen	0	–	2 104^d	–
	15	–	2 124^d	–
	30	–	2 474^c	–
Altona	0	–	–	2 412^c
	15	–	–	2 627^b
	30	–	–	2 861^a

In 2010, grain yield was significantly affected by N and S applications and differed between localities due to a significant locality × N × S interaction. This significant interaction generally showed that yields tended to increase with an increase in application rate of N and S differently between localities (Figure 3). If no S was applied, yields tended to level off or even decrease (Elsenburg) at N application rates of 60 and 90 kg N ha^-1. At sulphur rates of 15 and 30 kg S ha^-1, yields tended to level off with increasing N rates beyond 60 kg N ha^-1 at Welgevallen only. With S rates of 15 kg S ha^-1, yields at Elsenburg did not level off with even the highest N rates of 120 kg N ha^-1, while at S rates of 30 kg S ha^-1, yields at Langgewens did not level off at the highest N rate. These results clearly indicated that S requirements differed between localities tested due to soil and climatic differences in this year. There was, however, no L × N × S interactions in the other two years.

Figure 3: Canola yields harvested in 2010 at Elsenburg (ELS), Welgevallen (WLG), Langgewens (LG) and Roodebloem (RB) localities as a result of different sulphur (0, 15 and 30 kg ha^-1) and nitrogen (0, 30, 60, 90 and 120 kg ha^-1) fertilisation rates

In 2011 grain yield differed between Altona, Langgewens, Elsenburg and Roodebloem, with significant responses to N and S application rates as well as significant L × S and N × S interactions. Grain yields showed a significant L × S interaction in 2011 because, in contrast to the other localities, yields at Roodebloem did not increase with an increase in S rate from 15 to 30 kg S ha^-1 (Table 3). On average, grain yields were, however, significantly higher at Roodebloem (2 827 kg ha^-1) compared to Altona (2 634 kg ha^-1), Elsenburg (1 368 kg ha^-1) and Langgewens (1 152 kg ha^-1) during 2011. Significantly lower yields observed at Elsenburg in 2011 likely resulted from a combination of high rainfall during planting, which reduced the plant population, and lower rainfall received in July reducing the ability of the crop to compensate (Angadi et al. 2003) for the reduced population. At Langgewens sensitivity to water stress of canola during the flowering, pod-setting and grain-filling stages (Mendham and Salisbury 1995; Bagheri and Shahzad 2011) was clearly illustrated by lower grain yields obtained in all the seasons (Table 3), as this locality showed generally low rainfall from July to October in all seasons.

The significant N × S interaction in 2011 showed that in treatments where no S and N were added, yields increased from 1 575 kg ha^-1 to a maximum of 1 878 kg ha^-1 at 120 kg N ha^-1, but yield increases with applications of more than 30 kg N ha^-1 were not statistically significant (Figure 4). When S rates of 15 kg S ha^-1 and 30 kg S ha^-1 were applied, grain yields were significantly increased with increasing N applications to reach yields of 2 310 kg ha^-1 and 2 563 kg ha^-1, respectively, when 120 kg N ha^-1 was applied. This indicated that N applications >120 kg N ha^-1 may be needed at high yielding (rainfall) localities.

Figure 4: Effect of nitrogen and sulphur fertilisation rates on canola yields in the 2011 season

From the results it became clear that the grain yield responses to N were affected by S application rates. For this reason regression analyses were used to developed N × S response curves for each year. The regression curves (Figure 5) showed that, although N fertilisation increased canola grain yield at all S levels, the yields at 0 kg S ha^-1 tend to level off at lower N rates. In all years best fits with R² > 0.99 were obtained with S rates of 30 kg N ha^-1. This clearly illustrates the importance of S applications to achieve maximum yield response to N applications.

Figure 5: Grain yield responses of canola to nitrogen and sulphur fertilisation rates in the (a) 2009, (b) 2010 and (c) 2011 seasons

Grain oil and protein content

Grain oil and protein content differed significantly between localities with canola grain at Roodebloem (41.4%) having significantly higher oil contents than at Elsenburg (39.2%) and Langgewens (36.5%) localities (Table 4). A significant negative correlation (R² = 0.958) between the oil and protein content shows an inverse relationship such that increase in protein content lowers the oil content of the crop (Ahmad et al. 2007; Hassan et al. 2007).

Table 4: Effect of S-fertiliser rates on grain oil content (%) of canola grown at different localities. Mean values for the 2009–2011 period are presented. Means with at least one common superscript letter are not significantly different (LSD_0.05)

S rate (kg ha^-1)	Elsenburg	Langgewens	Roodebloem
0	39.1^b	36.2^d	41.3^a
15	39.3^b	36.6^c	41.4^a
30	39.3^b	36.7^c	41.5^a
Mean	39.2	36.5	41.4

Oil content was slightly increased at all localities as a result of applications of 30 kg S ha^-1, but such increase was only significant at Langgewens (36.2% to 36.7%) and not at Roodebloem (41.3% to 41.5%) and Elsenburg (39.1% to 39.3%) (Table 4). However, without any S applied, canola oil content at Roodebloem (41.3%) was even higher than with 30 kg S ha^-1 application rates at Langgewens (36.7%) and Elsenburg (39.3%). Sulphur application had no significant effect on the protein content at any of the three localities evaluated.

High N application rates lowered the oil content whilst increasing the protein content of canola grains (Table 5). Such effects of N on oil content were, however, highly dependent on locality such that application of 120 kg N ha^-1 resulted in decreasing canola grain oil content from 40.9% where no N was applied to 36.7% at Elsenburg. At the same locality, increasing N application rates resulted in an increase in protein content from 20.5% where no N was applied to 23.7% at an application rate of 120 kg N ha^-1 (Table 5). However, even though similar increases in protein and decreases in oil content with N fertilisation were shown at the other two localities (Langgewens and Roodebloem), these changes were not significant. This L × N interaction can therefore mainly be ascribed to the inherently different levels of N in the soil and effects of climatic conditions (especially soil moisture) on nutrient uptake and overall partitioning (Hocking et al. 1997; Jackson 2000; Malhi et al. 2007).

Table 5: Effect of N fertiliser rates on grain oil content (%) and grain protein content (%) of canola grown at different localities (Mean values for 2009–2011 period. From the table, oil and protein means with at least a common letter are not significantly different, LSD_0.05)

Location	N rate (kg ha^-1)	Oil (%)	Protein (%)
Langgewens	0	37.7^b	23.5^a
	30	37.4^b	24.6^a
	60	36.4^b	25.2^a
	90	35.6^b	25.3^a
	120	35.5^b	25.5^a
Roodebloem	0	42.8^a	18.5^b
	30	42.4^a	19.0^b
	60	41.7^a	20.2^b
	90	40.7^a	20.7^b
	120	39.5^a	22.5^b
Elsenburg	0	40.9^a	20.5^b
	30	40.3^a	21.6a^b
	60	39.8a^b	22.6^a
	90	38.3^ab	23.2^a
	120	36.7^b	23.7^a

Conclusions

The effect of N application on canola grain yields increased with availability of S and was greater in high rainfall localities. To obtain maximum responses to N applications, S applications of 15–30 kg ha^-1 may be needed on some localities if S content of the soil is <6 mg kg^-1. The N:S ratios that fall within 1:4 to 1:8 gave higher yields depending on inherent nutrient content and availability of moisture. It will, however, be worthy to investigate whether applications of >120 kg N ha^-1 will increase profit margins. Sulphur rates of 15–30 kg ha^-1 may also improve the canola grain oil content.

Acknowledgements

The authors are grateful to the Protein Research Foundation for financial support.