High planting density benefits to mechanized harvest and nitrogen application rates of oilseed rape (Brassica napus L.)

Abstract

To evaluate the effects of planting density and nitrogen (N) application rate on agronomic characters, yield and N use of oilseed rape (Brassica napus L.), a split plot experiment was carried out. Two planting densities (1.5 × 10⁵ plant ha⁻¹ and 4.5 × 10⁵ plant ha⁻¹ as main plots) and four N rates (0, 90, 180 and 270 kg ha⁻¹ as sub plots) were designed for two cultivars of Brassica napus. Results indicate that increased planting density could lead to decreased plant height, branch number, canopy layer thickness, effective siliques per plant and harvest index, while branch height and N use efficiency increase. Under a single density, plant height, branch height, branch number, effective siliques per plant and yield increased with increased N application, but the oil content and N use efficiency decreased. For the same target yield (plateau yield) obtained in low-density planting, the N fertilizer requirement decreased by 22.8% and 25.4% in high-density planting in two experimental sites. On the whole, these results demonstrate that increased plant density can improve rape adaptability for mechanized harvest by regulating plant structure, decreasing the N requirement for reaching a target yield and increasing N use efficiency.

Key words:

• Rape (Brassica napus L.)
• plant density
• nitrogen rate
• yield
• nitrogen use efficiency

1. INTRODUCTION

Oilseed rape (Brassica napus L.) is one of the major oilseed crops in China, with the planting area and total yield accounting for 23.5% and 22.6% of world tallies, respectively (Yin and Wang 2011). Since 2000, in China, rape planting area has reached 7.0 × 10⁷ ha, and total yield has increased to more than 1.1 × 10⁷ t, which provides about 4.0 × 10⁶ t of cooking oil each year (Yin et al. 2009). However, influenced by increased costs for pesticides, fertilizer, and particularly labor, the planting area and yield of rape had fallen in recent years, especially 2006–2008. Although about 50% of domestic vegetable oils are derived from rape oil, more than 60% of oil consumption in China remains dependent on imports, and the self-sufficiency rate for production of edible oils has declined steadily (Zhang et al. 2011). According to Zhang et al. (2010), using manual labor (mainly for transplanting), the total cost to produce rape in China has reached 5183.5 Yuan (RMB, same below) ha⁻¹, with a net income of only 290.8 Yuan ha⁻¹. Nearly 50% of this cost was for labor, while labor costs were 244.8, 750.0 and 944.6 Yuan ha⁻¹ in Canada, USA and Germany, respectively, owing to 100% mechanization. Lower mechanization rates and greater labor force usage are the major factors restricting rape production in China (Yin and Wang 2012). However, the government of China has realized the importance of mechanized production and is advancing the combination of agricultural mechanization and improved agronomy in domestic rapeseed production (Zi 2007).

Uniformity of maturity at harvest and resistance to shattering are traits that can lead to better management of rape harvesting operations (Pari et al. 2012). However, under current conditions in China, rape planted by transplanting produces tall plants with many branches and easily shattered pods, which causes yield loss during harvest, as well as intertwining plants that are not suitable for mechanical harvest (Wang and He 2008; Li et al. 2010). Chemical ripening technology promotes rape pod maturity, reduces stem and pod water content, and thereby decreases yield losses, mechanical loads and machine jams (Ma et al. 2010). The tradeoff is that these treatments also cause changes in pod chlorophyll content, photosynthetic gas exchange and enzyme activity, which combine to negatively impact the 1000 seed weight, protein and oil content, and fatty acid composition of rape (Zhou et al. 2009).

As key cultivation factors, planting density and nitrogen (N) application rates regulate rape development and yield. Rathke et al. (2005, 2006) determined that fertilizer N rate strongly influences rape productivity. Ozer (2003) found that rape yield variation with N application rate was primarily due to changes in branch numbers, pod numbers per plant and 1000 seed weights. He also suggested that 160 kg N ha⁻¹ is adequate for rape to meet its N requirements. Zhang et al. (2012) reported that compared with 2.4 × 10⁵ plant ha⁻¹, the seed yields per plot of 3.6 × 10⁵ and 4.8 × 10⁵ plant ha⁻¹ treatments were significantly increased, and the seed oil content was significantly increased with the increase in plant densities. Zeng et al. (2012) demonstrated that rape yields increase with the increases of fertilizer application and planting density, with seed yields of more than 3000 kg ha⁻¹ possible under the optimum N application rate is 135–225 kg ha⁻¹, and the best planting density in the range of 4.5 × 10⁵–7.5 × 10⁵ plant ha⁻¹. If N application to rape is too low, then protein yield suffers, while if it is too high, then oil yield is not favored (Gao et al. 2010). Compared to transplanting, direct sowing of rape yields weaker growth of individual plants, with more plants reliant on community advantages in order to achieve higher yield, which makes building of appropriate population structures very important for this planting method (Song et al. 2010; Lei et al. 2011). In related work with rice (Oryza sativa L.), Zhou et al. (2010) reported that the N utilization ratio can increase from 10.1% to 45.7% by increasing rice planting density, and that the N application rate can be reduced by increasing the number of effective panicles.

Even with these published results, many aspects of planting density and N application rate influences on rape development and fertilizer utilization remain unclear. The objectives of this research were: (1) to study the effect of different planting densities and N application rates on rape yield, oil content, agronomic traits and N use efficiency, and (2) to explore the relationship between planting density and N application rate for direct sown rape. The overarching goal is to develop methods that will make rape production in China more amenable to mechanized practices.

2. MATERIALS AND METHODS

2.1. Site characteristics

Two field experiments were carried out at Wen’an town of Zhijiang city (2011–2012) and Yangluo town of Wuhan city (2012–2013), Hubei province, China. The Zhijiang experimental area (31°15′N, 110°32′E, 100 m above sea level) is in the hill and gully region on the western margin of the Jianghan Plain, which has a subtropical climate, with an average annual rainfall of 1042 mm falling mainly from May to August. The annual mean evaporation is 734 mm and annual temperature average is 16.5°C, with a frost-free period of 210 d. The Wuhan experimental area (30°37′N, 114°32′E, 16.5 m above sea level) lies in the east of Jianghan Plain, which has a subtropical climate, with an average annual rainfall of 1214 mm falling mainly from April to July. The annual mean evaporation is 1200 mm and annual temperature average is 16.6°C, with a frost-free period of 250 d.

Rice was the previous crop at both sites. Soil types were Anthrosols (Paddy soil) in Zhijiang and Wuhan, respectively. Soil samples were analyzed before planting (Bao 2000). For the Zhijiang site, the cultivated layer (0–20 cm) soil nutrient statuses were as follows: 2.37% organic matter, pH 5.93, and available N, phosphorus (P) and potassium (K) concentrations of 135.0, 3.6 and 118.4 mg kg⁻¹, respectively. For the Wuhan site, these values were 1.72% organic matter, pH 8.21, and available N, P and K concentrations of 88.5, 22.1 and 193.3 mg kg⁻¹, respectively.

2.2. Field trial management

The experimental was a split-plot design, including eight treatments in three replicates contained in 19.3 m² (10.0 m × 1.93 m) plots in Zhijiang and 20.0 m² (10.0 m × 2.00 m) plots in Wuhan. At each site, rape seeds were sown in 28 rows per plot with a row space of 35.7 cm and two planting densities, 1.5 × 10⁵ plant ha⁻¹ (low density, LD) and 4.5 × 10⁵ plant ha⁻¹ (high density, HD), were laid out as main plots, with four N rates, 0 (N0), 90 (N90), 180 (N180) and 270 (N270) kg ha⁻¹, included as subplots (Pearce 2005). Additionally, 90 kg ha⁻¹ phosphorus pentoxide (P₂O₅), 120 kg ha⁻¹ potassium oxide (K₂O) and 11.25 kg ha⁻¹ borax (10.7% boron, B) were applied to the Zhijiang site, while 60 kg ha⁻¹ P₂O₅, 90 kg ha⁻¹K₂O and 15 kg ha⁻¹ borax were applied to the Wuhan site. The total P, K and B and 70% of the N were applied as a basal dose before sowing, and the remaining 30% of N was applied at the seedling stage. The sources of N, P and K fertilizers were urea (46% N), superphosphate (12% P₂O₅) and potassium chloride (60% K₂O). Two cultivars were used, with Zhong Shuang 11 planted at the Zhijiang site, and Zhong You 5628 planted at the Wuhan site. Sowing took place on September 26, 2011 and September 28, 2012 at both experimental sites. The seed density was evaluated directly after seedling emergence and adjusted for precise planting density at the five-leaf growth stage for all plots. Field management followed local practices for rapeseed production. Rape was harvested at both sites on May 16, 2012 and May 6, 2013.

2.3. Measurements and harvest

Two days before harvest, five plants were randomly selected from each replicate and a total of 15 plants per treatment were sampled to determine biomass, N accumulation, and apparent recovery N use efficiency (ARNE) of plants and seeds. Agronomic traits such as plant height, effective branch height and effective branches per plant were investigated using the above mentioned samples before drying. Plants were cut just above the soil and put into nylon bags (0.5-mm mesh). After air drying in the greenhouse, shoots were fractionated into stems, pods and seeds, which were then weighed. Roots were carefully harvested by digging with a spade, then washed clean with water and stored in nylon bags (0.5-mm mesh). As with shoots, they were also air dried in the greenhouse and weighed.

Seed yield was measured at maturity for each plot and converted to kg per hectare, and then adjusted based upon a water content of 5.2% of fresh weight.

Plant N concentration was determined by the micro-Kjeldahl method (Ozer 2003), and seed oil content by near infrared spectroscopy (Foss, NIR system 5000) in the Ministry of Agriculture for Oilseeds and Products Quality Testing Center (Wuhan).

The amount of N absorbed by plants was calculated as follows:

(1)

ARNE was calculated by the following equation:

(2)

where NUfi: N uptake by plant or tissue with N application (kg ha⁻¹); NUf0: N uptake by plant or tissue with no N application (kg ha⁻¹); Nf: N rates (kg ha⁻¹)

2.4. Statistical analysis

Data were prepared in Excel 2003 (Microsoft, Redmond, WA, USA). Analysis of variance with means separated using Fisher’s protected least significant difference (LSD) test with a p < 0.05 threshold was conducted in DPS v. 8.1 (Tang and Zhang 2013). Rapeseed basic yield, plateau yield and N critical rate (Table 4) were determined by regression analysis in SAS (SAS Institute Inc.1999).

3. RESULTS

3.1. Agronomic characters of rape

As shown in Table 1, N application rates and planting densities had significant effects on plant height, effective branch height and effective pod number of rape. Compared with N0, increasing N application significantly increased plant height and effective branch height, with the highest plant height observed in the N270 treatment at both sites. N application significantly increased the number of branches and pods per plant. The highest numbers were obtained with 270 kg N ha⁻¹ in Wuhan, and 180 kg N ha⁻¹ in Zhijiang. Increasing planting density also significantly increased effective branch height, but plant height was reduced. Furthermore, plants in HD plots were thinner and carried fewer branches and pods. As a result, thickness of the rape pod canopy was reduced with increasing planting density. The interaction effects of planting density and N application were not significant in Wuhan site, but there were significant effect on height and effective branch in Zhijiang site (Table 1).

Table 1 Effect of planting density and nitrogen (N) application rate on rape agronomic characters in Wuhan and Zhijiang experiment sites

Density	N rate (kg ha^−1)	Plant height (cm)		Branch height (cm)		Effective branch (No. plant^−1)		Effective pod (No. plant^−1)		Thickness of pod canopy (cm)
		Wuhan	Zhijiang	Wuhan	Zhijiang	Wuhan	Zhijiang	Wuhan	Zhijiang	Wuhan	Zhijiang
LD	0	149.4 c	148.0 b	59.1 b	65.2 b	7.8 b	3.4 b	285.0 b	107.3 b	90.3 b	82.8 c
	90	169.7 b	177.0 a	65.4 b	69.8 b	8.0 b	5.6 a	355.5 a	175.3 a	104.4 a	107.2 a
	180	178.4 ab	178.4 a	77.0 a	70.2 b	9.1 a	6.0 a	377.8 a	176.0 a	101.4 a	108.2 a
	270	182.4 a	179.1 a	75.9 a	86.8 a	9.5 a	6.0 a	425.4 a	176.1 a	106.5 a	92.3 b
HD	0	128.9 c	148.8 b	64.3 b	83.2 b	5.1 b	1.8 c	117.6 b	44.4 a	64.6 b	65.6 ab
	90	153.0 b	157.6 a	69.9 b	85.0 b	6.4 a	2.5 b	208.6 a	59.6 a	83.1 a	72.6 a
	180	170.9 a	161.4 a	90.4 a	89.0 b	6.5 a	3.3 a	229.4 a	68.6 a	80.5 a	72.4 a
	270	172.6 a	163.3 a	89.6 a	102.9 a	7.1 a	2.8 ab	278.8 a	65.4 a	83.0 a	60.4 b
	Significance of
Density		< 0.0001	< 0.0001	0.0031	< 0.0001	< 0.0001	< 0.0001	< 0.0001	< 0.0001	< 0.0001	< 0.0001
N		< 0.0001	< 0.0001	< 0.0001	< 0.0001	0.0015	< 0.0001	0.0002	< 0.0001	0.0490	< 0.0001
Density × N		0.4072	0.0002	0.5419	0.8994	0.6827	0.0036	0.9856	0.0582	0.9716	0.0391

LD, low density (1.5 × 10⁵ plants ha⁻¹); HD, high density (4.5 × 10⁵ plants ha⁻¹); mean values (n = 15) followed by the same letters are not significantly different at the 0.05 level by least significant difference (LSD); Wuhan site in 2013, cultivar as Zhong You No. 5628; Zhijiang site in 2012, cultivar as Zhong Shuang No. 11.

3.2. Seed yield and oil content

Table 2 shows that N application had significant effects on yield. Rapeseed yield increased significantly from 2170 to 3266 kg ha⁻¹ and from 1512 to 2260 kg ha⁻¹ when N application increased from 0 to 180 kg ha⁻¹ in the Wuhan site and Zhijiang site, respectively. There were no significant differences when more N was applied.

Table 2 Effect of planting density and nitrogen (N) application rate on rape (Brassica napus L.) seed yield, oil content and harvest index in Wuhan and Zhijiang field experiment sites

Density	N rate (kg ha^−1)	Seed yield (kg ha^−1)		Oil content (%)		Harvest index
		Wuhan	Zhijiang	Wuhan	Zhijiang	Wuhan	Zhijiang
LD	0	2170 c	1512 c	46.8 a	50.0 a	0.291	0.305
	90	2704 b	1909 b	45.8 b	49.2 b	0.276	0.22
	180	3173 a	2173 a	44.3 c	48.7 bc	0.293	0.232
	270	3183 a	2166 a	43.8 c	48.4 c	0.295	0.233
HD	0	2274 c	1553 c	46.9 a	50.2 a	0.276	0.261
	90	2916 b	2035 b	45.9 b	49.3 b	0.281	0.186
	180	3266 a	2260 a	44.8 c	49.0 b	0.278	0.199
	270	3234 a	2203 a	43.9 d	48.3 c	0.284	0.194
Significance of
Density		0.0184	0.1411	0.2903	0.4830
N		< 0.0001	< 0.0001	< 0.0001	< 0.0001
Density × N		0.6201	0.8914	0.5115	0.7600

LD, low density (1.5 × 10⁵ plants ha⁻¹); HD, high density (4.5 × 10⁵ plants ha⁻¹); mean values (n = 3) followed by the same letters are not significantly different at the 0.05 level by least significant difference (LSD); Wuhan site in 2013, cultivar as Zhong You No. 5628; Zhijiang site in 2012, cultivar as Zhong Shuang No. 11.

Within N application treatments, HD planting produced higher yields than LD planting, and the largest yield difference was between N0 and N90, regardless of site. This represents increases of 7.8% and 6.6% in Wuhan and Zhijiang, respectively. Among all treatments, N180 coupled with HD planting produced the largest gains in yield, as well as the highest yields, of 3266 and 2260 kg ha⁻¹ in Wuhan and Zhijiang, respectively. Within all sites, rape yield did not vary significantly between the interaction of planting density and N application (Table 2), but HD planting with moderate N (N90 to N180) application led to higher yields in comparison to LD planting with N0 application.

Table 2 shows that seed oil contents were mainly affected by N application. When N application rose from 0 to 270 kg ha⁻¹, seed oil content decreased by 6.4% for both planting densities at the Wuhan site, while it decreased by 3.3% and 3.8% for LD and HD planting densities in Zhijiang, respectively. There were no significant differences between the two planting densities and the interaction effects of planting density and N application. Conversely, planting density had a main effect on harvest index of rape; it decreased with increased planting density, and was especially obvious in Zhijiang (Table 2).

3.3. Apparent recovery N use efficiency

N application rates and planting density affected total N uptake significantly (Table 3). Increasing N rates resulted in significantly higher total N uptake in rapeseed. In N0, there was little difference in N uptake between planting densities. With N application, N uptake was promoted not only between N treatments, but also between planting density treatments.

Table 3 Effect of planting density and nitrogen (N) application rate on rape (Brassica napus L.) N accumulation and apparent recovery N use efficiency in Wuhan and Zhijiang field experiment sites

Density	N rate (kg ha^−1)	Total N accumulation (kg ha^−1)		N use efficiency (%)
				Whole plant		Rape seed
		Wuhan	Zhijiang	Wuhan	Zhijiang	Wuhan	Zhijiang
LD	0	76.4	91.4	–	–	–	–
	90	109.3	146.9	36.6	61.7	19.7	17.9
	180	135.1	178.2	32.6	48.2	20.6	15.3
	270	147.5	190.7	26.3	36.8	16.9	10.4
HD	0	78.6	91.5	–	–	–	–
	90	114.2	147.8	39.6	62.7	32.1	22.4
	180	141.3	181.4	34.9	50.0	24.1	19.5
	270	150.0	197.4	26.5	39.2	18.4	12.9

LD, low density (1.5 × 10⁵ plants ha⁻¹); HD, high density (4.5 × 10⁵ plants ha⁻¹); Wuhan site in 2013, cultivar as Zhong You No. 5628; Zhijiang site in 2012, cultivar as Zhong Shuang No. 11.

ARNE (Table 3) decreased with an increase of N application. Compared to N90, it was slightly reduced with N180 and more so with N270 application. ARNE increased in HD planting compared with LD planting for all N application rates, yet was more significant with moderate N (N90 to N180) application in Wuhan, and with high N application in Zhijiang.

3.4. Models of N application rate and yield

Table 4 shows linear plus plateau fertilization models based on N application rates and rape yields in two planting densities. It shows that with LD planting, the plateau yields were 3178 and 2170 kg ha ⁻¹, which correspond to 169.9 and 149.2 kg N ha ^–1 for the Wuhan and Zhijiang sites, respectively. Meanwhile, with HD planting, the plateau yields reached 3250 and 2232 kg ha ⁻¹, with N application rates calculated as only 136.8 and 126.6 kg ha⁻¹ in Wuhan and Zhijiang, respectively. To obtain the same yield with HD planting as observed with LD planting required 126.7 and 115.2 kg N ha ⁻¹, which reduced N fertilization by 25.4% and 22.8% in Wuhan and Zhjiang, respectively (Table 5).

Table 4 Linear plus plateau models of rape (Brassica napus L.) seed yield regressed against nitrogen (N) application rate in two planting densities for Wuhan and Zhijiang field experiment sites

Site	Density	Fertilization model	R^2	Basic yield (kg ha^−1)	Plateau yield (kg ha^−1)	N critical rate (kg ha^−1)
Wuhan	Low	y = 5.933x + 2170, x < 169.9	0.999**	2170	3178	169.9
		y = 3178, x ≥ 169.9
	High	y = 7.133x + 2274, x < 136.8	0.999**	2274	3250	136.8
		y = 3250, x ≥ 136.8
Zhijiang	Low	y = 4.403x + 1512, x < 149.2	0.999**	1512	2170	149.2
		y = 2170, x ≥ 149.2
	High	y = 5.356x + 1553, x < 126.6	0.995**	1553	2232	126.6
		y = 2232, x ≥ 126.6

**, significant at P < 0.01 level; Wuhan site in 2013, cultivar as Zhong You No. 5628; Zhijiang site in 2012, cultivar as Zhong Shuang No. 11.

4. DISCUSSION

4.1. Cultivar effect on agronomic characters and seed yield of rape

Rape yield response to increasing N rate varied with different cultivars and environmental variables, including weather, soil type, residual fertility (especially nitrate) and soil moisture (Rathke et al. 2006). Rape biomass or stand density in the field is a prerequisite for optimized N application rate (Thoren and Schmidhalter 2009). Balint and Rengel (2008) reported that rape dry matter of stems, leaves, siliques and seeds were strongly influenced by genotypic differences, but only stems and seeds were affected by the treatment of N application. Mahli et al. (2007) indicated that hybrid cultivars generally produced more biomass and seed yield than open-pollinated cultivars, and provided greater net economic returns under both moist and relatively dry conditions. In the present study, rape agronomic characters (especially the number of branches and pods per plant) and yield differed significantly between the two sites. Whether for LD or HD planting, the maximum branch number, pod number and seed yield of rape were obtained in the hybrid cultivar “Zhong You 5628” (Wuhan site), which were all significantly higher than the conventional cultivar “Zhong Shuang 11” (Zhijiang site). As an increased supply of nutrients is typically required to support higher yields, Karamanos et al. (2005) showed that N requirements for optimum seed yield were higher for hybrid than for conventional cultivars. In the present study, N requirements for plateau yield were 20.7 kg ha⁻¹ and 10.2 kg ha⁻¹ higher for “Zhong You 5628” than for “Zhong Shuang 11” under LD and HD planting conditions (Table 4). Therefore, the “Zhong You 5628” cultivar had better growth, and the highest number of branches and pods per plant were obtained in the highest N application.

4.2. Optimum planting density contributes to mechanized harvesting of rape

Mechanization is an important and efficient tool to enhance crop yield; it also helps to reduce labor drudgery and ultimately increases farmers’ prosperity (Shahid et al. 2010). Yue et al. (2009) proposed that rape suitable for mechanization should have a tight plant type, be lodging resistant, flower and mature collectively, and produce more pods on the main stem. Zhang et al. (2006) also reported that the average height of rape suitable for mechanized harvesting was about 160 cm. LD planting promoted plant growth and development, which led to taller and bigger plants with more branches and pods per plant, and required more N than HD planting (Sidlauskas and Tarakanovas 2004; Leach et al. 1999). Plants grown at higher densities are thinner, carry fewer branches and fewer pods, and are more tidy in plant structure type than those grown in low densities (Mobasser et al. 2008). The results herein demonstrate that rape plant structures may vary significantly under two growing densities. With N application, rape average height of HD were 166 cm and 161 cm, which significantly lower than those in LD, at 177 cm and 178 cm in Wuhan and Zhijiang, respectively, and the thickness of rape pod canopy was reduced significantly with increasing planting density. These features make rape grown in HD fields more suitable for mechanized harvest than that grown at LD. Rape obtained the maximum yield in 10.5 × 10⁵ plants ha⁻¹, but with increasing planting density, stem breaking strength decreased significantly. In order to reduce the high loss rate in the mechanized rape harvesting process, the appropriate planting density of 4.5 × 10⁵ plants ha⁻¹ was advantageous to mechanized harvesting, with thinner stems, fewer branches and high stem-lodging resistance (Dong et al. 2012).

4.3. Optimum planting density and N application rate increases rape N use efficiency and decreases N requirement to reach target yield

Planting density is a major factor in the determination of the ability of crops to capture resources. Lloveras et al. (2004) reported that under irrigation, a high wheat (Triticum aestivum L.) seeding rate was required to obtain high yields in Mediterranean areas. Conversely, Elfadl et al. (2009) showed that in a low-input farming system, safflower (Helianthus annuus L.) can compensate for low plant density by producing substantially more heads plant^–1 and seeds plant^–1, while using residual soil N more efficiently when compared to HD plants. The combination of these results demonstrates that a complex set of interactions and tradeoffs in plant development and resource acquisition and utilization needs to be elucidated to fully understand how planting density affects crop yield.

The amounts of N fertilizer required for maximum yield of rape in previous work depended on environmental conditions (Gan et al. 2007). Prasad and Shakla (1991) concluded that rape yield is affected by the interaction between planting density and N fertilization, whereby the optimal seed yield could be achieved by increasing planting density and N levels over levels commonly in use at that time. Kazemeini et al. (2010) reported that increasing N fertilizer and planting density caused a boost in seed yield in rape and the highest yield resulted with application of 150 kg N ha⁻¹ and a planting density of 9.0 × 10⁵ plants ha⁻¹.

The current work shows that ARNE increases with increasing plant density (Table 3). Moreover, adjusting N application rates from 0 kg ha⁻¹ to 180 kg ha⁻¹, coupled with growing 4.5 × 10⁵ plants ha⁻¹, produced the largest yield difference between treatments (Table 2), with increases of 7.8% and 6.6% between the N0 and N90 treatments in Wuhan and Zhijiang, respectively. The N180 treatment combined with HD planting also led to the highest yields (32662260 kg ha ⁻¹ and 2260 kg ha⁻¹ in Wuhan and Zhijiang, respectively). Therefore, for a given target yield, increasing planting density may reduce N fertilizer requirements significantly (Table 5). In order to fully optimize yield per plant and throughout a field, growers need a complete set of field management and fertilizer input guidelines. However, in actual production in China, a lack of farm labor, insufficient fertilizer inputs and inadequate nutrient management combine to restrict production of LD rape (Zhang et al. 2011; Li et al. 2012). This study suggests that under this production system, increasing planting density can effectively improve N use efficiency and decrease the N required to reach the same target yield. Consequently, based on the low comparative benefit of producing rape and a lack of farmer enthusiasm to produce rape, development of high density direct sowing rape production systems is one of the most important ways to maximize rape production efficiency in China.

Table 5 Calculated nitrogen (N) application rates for rape (Brassica napus L.) based on target yields under different planting densities in Wuhan and Zhijiang field experiment sites

Site	Target yield^1) (kg ha^−1)	N rate (kg ha^−1)			Decrease (%)
		Low density	High density	Decrement
Wuhan	3178	169.9	126.7	−43.2	−25.4
Zhijiang	2170	149.2	115.2	−34.0	−22.8

Target yields were the highest yield in low density planting within a site. Wuhan site in 2013, cultivar as Zhong You No. 5628; Zhijiang site in 2012, cultivar as Zhong Shuang No. 11.

4.4. Possible mechanisms for high planting density effecting decreased N application rates

Raising yield per unit of input is a very important factor to keep in mind in attempts to obtain high yield of rape. In previous work, close planting increased the high photosynthetic efficiency portion of the group structure, which might form a foundation for improving yield per unit of input (Bilgili et al. 2003). Increasing planting density to a certain extent inhibits plant economic characters, such as branch number per plant and pod number, yet leaf area index and light energy utilization in a high planting density community increases and contributes to increases in seed yield (Kuchtová and Vasak 2004; Mehrnet 2008; Kazemeini et al. 2010). Hosseini et al. (2006) stated that increasing planting density and N application rates resulted in increasing rape leaf area index and decreased weed dry matter, suggesting that optimum planting density and N fertilization may increase the competitive ability of canola against weeds. Su et al. (2011) suggested that HD planting of rape can minimize nutrient losses and make up for a lack of individual growth through development of population dominance. Crop yield formation relies on source-sink coordination of canopy photosynthetic production and root acquisition of available soil nutrients (source), along with canopy nutrient absorption and redistribution (sink) (Zhao et al. 1995). With adaptation of a canopy group to the local soil environment, a well-developed root system improves photosynthetic productivity and increases yield (Yamazaki and Harada 1982). Our research also shows that HD planting of rape yields plants with more total root length, surface area and volume per unit area than LD planting at all growth stages (unpublished data), and thereby enhances the ability of rape plants to acquire nutrients. In short, there are many possible explanations for the relation between ARNE and planting density observed in this study. As of yet, the mechanism for improving N fertilizer utilization efficiency with high planting density remains unclear. Future work, such as studies of root architecture, soil enzyme activities and plant N metabolism, will clarify these mechanisms. Nevertheless, this study is an important step towards improving rape yield and decreasing costs, while also suggesting methods to make the crop more uniform for mechanized production.

5. CONCLUSION

Compared with low planting density, high planting density of rape may improve field level production capacity and N use efficiency, especially under low N regimens. Therefore, with one target yield, increasing planting density allows for significant reductions in N application. Increasing planting density significantly decreases plant height, branch number and effective pod number per plant, and thereby makes the canopy more uniform and suitable for mechanical harvest.

ACKNOWLEDGMENTS

This work was supported by the Central Public Interest Scientific Institution Basal Research Fund (1610172009003, 2014ZL029) and the National Scientific Support Program of China (2010BAD01B05, 2010BAD01B09). The authors would like to thank the staff of Yichang Academy of Agricultural Sciences for their excellent support in carrying out field experiments, the Root Biology Center, South China Agricultural University for help in writing the manuscript and Golden Fidelity LLC for help in checking the English.