Nitrogen fertilization to broccoli cultivars at different planting times: Yield and nitrogen use

Abstract

The effects of three nitrogen (N) fertilizer rates (0, 120, and 240 kg N ha⁻¹) and two planting times (May or late June/July) on yield and N use of the early cultivar ‘Milady’ and the late cultivar ‘Marathon’ of broccoli (Brassica oleracea var. italica) were investigated on three silty loam soils varying in soil mineral N (Nmin) in the southernmost part of Norway during 1999 and 2001. In all crops receiving fertilizer, rapid uptake of N started about three weeks after planting. The relative yield of broccoli heads increased with increasing soil available N (fertilizer N plus Nmin) at planting to 200–250 kg N ha⁻¹ and then levelled off. The two lower fertilizer rates were more restrictive to yields in early-planted than in late-planted crops. A general increase in harvest index with increasing N rate reflected a stronger effect of N on the head yield than on the total above ground biomass production. The apparent recovery of fertilizer N decreased with increasing N rate and was on average 74% in total above ground biomass and 25% in broccoli heads. Despite a higher N uptake, the average soil mineral N level at harvest increased from 12 kg N ha⁻¹ on unfertilized plots to 27 and 78 kg N ha⁻¹ on plots receiving 120 and 240 kg N ha⁻¹, respectively; this increase was stronger in early than in late plantings and stronger in ‘Milady’ than in ‘Marathon’. The yield of broccoli heads was similar in the two cultivars, but ‘Milady’ had a lower total biomass production and thus a higher harvest index, presumably due to earlier head initiation.

Keywords:

• Brassica oleracea var. italica
• harvest index
• head initiation
• nitrogen uptake
• nitrogen recovery
• residues
• soil mineral N
• vegetable

Introduction

Broccoli (Brassica oleracea L. var. italica Plenck) is a nitrogen demanding vegetable. Several workers have reported higher yields and increased weight of individual broccoli heads with enhanced nitrogen application rates (Dufault & Waters, 1985; Kowalenko & Hall, 1987; Everaarts & de Willigen, 1999a; Bélec et al., 2001). The amount of nitrogen required for maximum yield varied greatly, due to soil properties, climatic conditions and product standards.

Information is scarce about the dynamics of the plants’ nitrogen uptake during the 6–8 week growing period. Bowen et al. (1999) studied partitioning and accumulation of nitrogen in broccoli related to soil mineral nitrogen throughout the growing season in Canada. They showed that plant dry matter production and N accumulation were very low during the first two weeks after transplanting, but increased substantially during the next two weeks. During the final two weeks, N accumulation in stem and broccoli heads continued, while the N content of the leaves decreased at the lowest N rates (0 and 125 kg N ha⁻¹), but increased at higher N rates. This indicates that when N supply was limited, N was translocated from the leaves to support the growth of broccoli heads.

To achieve a high yield of broccoli, the crop must be supplied with an adequate amount of nitrogen at the right stages of development. The supply of N should be sufficient for a maximum growth rate, yet low enough to avoid loss to the environment and to minimize fertilizer costs. Thus, the optimum level of N supply must be one that combines a high saleable yield with a high efficiency of fertilizer nitrogen.

The efficiency of fertilizer nitrogen can be expressed by the apparent fertilizer N recovery (REC),

1

where U _F and U ₀ are the amounts of crop N taken up in the presence and absence of fertilizer N, respectively, and N _F is the amount of fertilizer N applied (Greenwood & Draycott, 1988). The general trend in broccoli, as in other vegetable crops, is that REC declines more or less linearly with increasing N application. Zebarth et al. (1995) and Riley and Vågen (2003) calculated REC according to equation 1, and found REC in both crop biomass and in the harvested product to decrease with increasing N. Letey et al. (1983) also found the ratio of N in crop biomass to applied N to decrease with increasing N application. In contrast, Kowalenko and Hall (1987) found REC in crop biomass to increase from 50 to 63% when N application increased from 125 to 250 kg ha⁻¹, while REC in broccoli heads remained 24% irrespective of N rate.

For environmental as well as economic reasons, the aim is a highest possible fertilizer N recovery in the crop. In many vegetable species, only a fraction of the total biomass is harvested for consumption. Large amounts of N-rich plant residues are incorporated into the soil after harvest, with increased risk of leaching. Thus, agronomic practices increasing the harvest index (HI) would reduce the risk for N leaching.

Broccoli cultivars differ in yield potential, length of growing period and in morphological traits such as shape, colour, bud size, etc. There is, however, little knowledge on how different broccoli cultivars respond to nitrogen, and whether there are cultivar differences in the proportion of nitrogen recovered in the crop during the growing season.

The objective of the present research was to investigate how varying N rates affect N uptake, yield and recovery of fertilizer N in an early and a later maturing cultivar of broccoli, and whether these characters respond differently when the crop is grown early or late in the growing season.

Materials and methods

Experiments

The early cultivar ‘Milady’ F1, and the late cultivar ‘Marathon’ F1, the latter being widely used in Europe, were grown in three fields (silt loam soils ∼15% clay, ∼2.5% soil organic matter) during the seasons 1999 and 2001 (Table I). The preceding crops in the year before planting were for field 1 annual ryegrass, field 2 potatoes, and field 3 red clover seed production. Three out of six experiments were planted in May (referred to as ‘early’), and the remaining ones in June/July (‘late’). In either case, the soil was fallowed from spring until planting. The experiments were conducted at Apelsvoll Research Centre, Division Landvik, in Grimstad (58°20'N) in the southernmost part of Norway.

Table I. Planting dates, time to harvest (average of N levels), mean temperature during the growing period (recorded at the local weather station), and soil mineral N in the 0–30 and 30–60 cm soil layers at planting for each experiment.

			Days from planting to harvest			Nmin in soil before planting, kg ha^−1
Experimental field	Year / Planting time	Planting date	Milady	Marathon	Average temp. °C	0–30 cm	30–60 cm
1	1999-early	26 May	47	53	14.3	29	nd
1	1999-late	7 July	70	71	16.1	49	nd
2	2001-early 1	11 May	56	59	14.0	37	13
2	2001-late 1	28 June	62	62	16.5	65	11
3	2001-early 2	30 May	56	59	15.2	72	9
3	2001-late 2	5 July	59	61	16.3	116	19
nd: not determined.

A basal dressing of 40 kg P ha⁻¹ and 220 kg K ha⁻¹ was applied before planting. Soil pH at planting was 5.8 in the 1999 trial, while in 2001 it ranged from 6.1 to 6.6. Nitrogen was supplied at rates of 0, 120, and 240 kg N ha⁻¹ in addition to soil mineral N at planting (Table I). The highest N fertilizer rate was about 10% higher than the current recommendation for commercial broccoli production in Norway, which is 220 kg N ha⁻¹ (Anonymous, 1999). To ensure a continuous nitrogen supply to the plants throughout the growing period, the amounts were split into four equal dressings: at planting, and two, four, and six weeks later. Nitrogen was applied as calcium nitrate (Ca(NO₃)₂), 15.5% N (14.4% NO₃ ⁻ and 1.1% NH₄ ⁺). The first application at planting was broadcast, while the following ones were surface row applications.

All six experiments were arranged according to a factorial complete block design with three replications and free randomization of N levels and cultivars. Broccoli plants were raised in a greenhouse, and transplanted when they had developed 3–4 foliage leaves. In 1999, the first planting was delayed because of the weather, and the transplants grew somewhat larger than in the other experiments.

In 1999 the average distance between rows was 0.5 m (0.7 – 0.4 – 0.4 m) with 0.35 m between plants within a row. In 2001, the average row distance was changed to 0.58 m (0.75 – 0.5 – 0.5 m) for practical reasons. To obtain the same plant density as in the previous year (5.7 plants per m²), plant distance within the rows was reduced to 0.3 m. Plot size was 5×10 m in 1999 and 5.25×10 m in 2001, which gave 10 rows per plot in 1999 and 9 rows per plot in 2001. All plant samples were collected from five rows in the central part of each plot. This means that the borders to neighbouring plots were always at least two rows.

The nitrogen content of soil (nitrate and ammonium) and plant dry matter was determined by sampling at planting, one week after planting, and thereafter every other week until harvesting was completed. On each occasion, soil samples were made up of 10 subsamples taken with a soil auger to a depth of 30 (1999) or 60 cm (2001) in each plot. The samples were frozen and later analysed by means of an Aquatec autoanalyser after extraction in 1 M KCl. Plant samples consisted of the above ground parts of 10 plants from each plot, collected from previously defined parts of the plot for each sampling. These were finely chopped and immediately frozen for later determination of dry matter content and total nitrogen (Kjeldahl-N) in plant dry matter. At final harvest, broccoli heads (inflorescences) were harvested when they had reached their maximum size, but before the flower buds had begun to swell. The number of days from planting to harvesting was recorded for 30 heads harvested from a defined area within each plot.

Weather conditions

The mean temperature of the experimental periods ranged from 14.0 to 16.5°C (Table I), and the average daily global radiation from 16.7 to 21.9 MJ m⁻². Water was a non-limiting factor, as the crops were irrigated when the estimated water deficit to the soil field capacity reached about 30 mm. There were a few incidents of heavy rainfall, which may potentially have caused nitrogen leaching from the soil. These occurred on 10 August (73 mm), and 15 and 16 August (50 + 22 mm) in 1999, and on 3 August (49 mm) and 8 and 9 August (40 + 8 mm) in 2001.

Calculations

The nitrogen recovery in the above ground plant biomass was calculated according to equation 1. The nitrogen recovery in the harvested product was calculated in the same way, only substituting the amounts of crop N by the amounts of N in the harvested broccoli heads.

The nitrogen balance (kg N ha⁻¹) was calculated as the sum of soil Nmin and N in above ground biomass at harvest minus applied fertilizer N, soil Nmin at planting, and N in above ground biomass at planting. Only the soil Nmin values for the 0–30 cm soil layer were used for this calculation. Data for the 30–60 cm soil layer are not shown.

The harvest index (HI) was calculated as the harvested broccoli heads’ fraction of the total above ground biomass on a dry weight basis. The fraction of N located in the heads was calculated correspondingly.

Relative yield Y/Ymax was calculated as the individual yields (Y) divided by the average maximum yield (Ymax) of each cultivar in each experiment.

Analysis of variance was carried out by the SAS procedure ANOVA (SAS, 1987). The main effects of planting time, cultivar, nitrogen rate and their interactions were tested against their respective interactions with field number, which was considered a random variable comprising both soil and climatic year-to-year variation (Table II). Significant differences between treatments were identified by Duncan's multiple range test at the 5% probability level.

Table II. Analysis of variance for various characters, showing main effects and interactions between variables. Significance levels p≤0.05, 0.01, and 0.001 are denoted by the symbols *, **, and ***, respectively. Tendencies, defined as 0.05 < p≤0.15 are shown by their respective p-values.

	Main effects			Interactions
Characters	N rate	Planting time	Cultivar	N rate×planting time	N rate×cultivar	Planting time×cultivar	N rate×planting time×cultivar
Yield of heads, fresh weight	*	0.08	–	**	–	0.12	–
Biomass at harvest, dry weight	**	–	*	0.10	*	–	–
N in biomass	***	–	–	–	–	–	–
N in heads	*	0.12	–	0.14	–	*	–
N in crop residues	***	–	–	–	0.13	–	–
Nmin in soil at harvest, 0–30 cm	***	*	–	*	*	–	0.07
Nitrogen balance	–	–	–	0.08	0.09	–	0.10
Harvest index	0.09	–	*	0.07	–	–	–
N recovery in biomass	*	–	–	–	–	–	–
N recovery in heads	0.09	–	–	–	–	–	0.11
Fraction of N in heads	–	–	–	0.10	–	–	–
***: p≤0.001, **: p≤0.01, *: p≤0.05

Results

Yield

On average for cultivars and planting dates, the fresh weight yield of broccoli heads more than doubled and almost tripled by increasing the N rate from zero to 120 and 240 kg N ha⁻¹, respectively (Table IIIa). The yield reduction with reduced N supply was significantly greater in the early than in the late plantings. There was also a tendency (p = 0.08) that the late plantings gave higher yields than the early ones. There was no significant difference in yield of heads between the two cultivars, either when compared on fresh weight (Table IIIa) or on a dry weight basis (not shown). However, on average, ‘Marathon’ seemed to give the higher yield in the early plantings, while ‘Milady’ was slightly superior in the late plantings (Table IIIa).

Table III. Fresh weight yield of heads (a), dry matter yield of above ground biomass (b), and nitrogen uptake to harvested product (c) at three N fertilization rates in early and late plantings of two broccoli cultivars. n=9 (3 exps×3 reps). Different letters indicate significant main effects of N rate or planting time at the 5% probability level.

	Early planting			Late planting
kg N ha^−1	Milady	Marathon	Mean	Milady	Marathon	Mean	Mean
a) Fresh weight yield of heads (Mg ha^−1)
0	4.33	4.39	4.36	7.97	8.21	8.09	6.23 a
120	12.00	13.41	12.71	16.46	15.88	16.17	14.44 b
240	16.97	18.98	17.97	19.35	18.12	18.74	18.36 b
Mean	11.10	12.26	11.68 a	14.59	14.07	14.33 a	13.01
b) Above ground dry matter biomass (Mg ha^−1)
0	2.66	4.54	3.60	3.25	4.66	3.96	3.78 a
120	4.51	6.31	5.41	5.42	6.55	5.99	5.70 b
240	4.95	6.47	5.71	6.39	7.09	6.74	6.23 b
Mean	4.04	5.78	4.91 a	5.02	6.10	5.56 b	5.24
c) Nitrogen uptake to broccoli heads (kg N ha^−1)
0	15	14	14	28	25	26	20 a
120	47	47	47	63	58	61	54 b
240	72	69	70	78	72	75	73 b
Mean	45	43	44 a	56	52	54 a	49

The relative yield of broccoli heads (calculated as a percentage of the maximum yield within each experiment) increased almost linearly with increasing amount of available N up to ∼230 kg N ha⁻¹ (Figure 1). Beyond this level, there was no further response, maximum yields being achieved over a range of available N levels.

Figure 1.  Relative yield (Y/Ymax) as a function of the sum of Nmin content of the soil (0–30 cm depth) at planting and applied fertilizer nitrogen, in six experiments with broccoli during the years 1999 and 2001. The relative yield Y/Ymax was calculated as the individual yields (Y) divided by the maximum yield (Ymax) of each cultivar in each experiment. n = 3 for each data point (3 reps).

Total dry matter production

The total above ground biomass at the time of harvest increased with N rate. ‘Marathon’ produced a significantly higher biomass than ‘Milady’, and this was more pronounced with rising N rate (Table IIIb). The tendency to a stronger response in dry matter biomass to increasing N rate at early planting was in line with the corresponding significant interaction for the yield of heads (Table II).

Nitrogen in plant dry matter

N rate was the only factor to influence the total nitrogen uptake in above ground biomass at harvest (Table II). The increase with N rate was observed in the amount of N accumulated in heads as well as in crop residues (Table IIIc and IV, respectively). Additionally, the crop residues of ‘Marathon’ tended to contain more nitrogen with increasing N rate than those of ‘Milady’ (6%, 11%, and 19% higher for N rates of 0, 120 and 240 kg ha⁻¹, respectively (Table IV).

Table IV. Nitrogen uptake to plant residues (kg N ha⁻¹) of two broccoli cultivars at three N fertilization rates. n=18 (3 exps×3 reps×2 planting times). Different letters indicate significant main effects of N rate or cultivar at the 5% probability level.

	Cultivar
kg N ha^−1	Milady	Marathon	Mean
0	56	60	58 a
120	117	130	123 b
240	148	176	162 c
Mean	107 a	122 a	115

For the whole biomass there was no tendency to a difference between the N uptake curves of the two cultivars, and thus, Figure 2 presents the average values. However, ‘Milady’ had an 8% higher N uptake than ‘Marathon’ to the heads in the late plantings, in contrast to only 5% in the early ones (Table IIIc).

Figure 2.  Nitrogen uptake to the above ground plant biomass (upper part of figures), and the Nmin content of the soil in 0–30 cm depth shown inversely (lower part of figures), from planting to harvest in six experiments with broccoli at N fertilization rates of 0, 120, and 240 kg ha⁻¹ during the years 1999 and 2001. The early plantings (a, c, e) to the left, and the late plantings (b, d, f) to the right. n = 6 for each data point (2 cvs×3 reps).

Furthermore, a few tendencies to additional effects could be seen (Table II). There was a tendency that the harvested product (broccoli heads) accumulated less nitrogen in the early plantings than in the late ones (Table IIIc). This was related to a relatively lower N uptake at 0 and 120 kg N ha⁻¹ in the early plantings (20% and 67% of the N uptake at the highest N rate) compared with the late ones (35% and 81%). This tendency compared quite well with the corresponding and significant differences in broccoli head yield (Table IIIa).

Nitrogen in the soil

The content of mineral nitrogen (Nmin) in the soil reached its maximum from three to five weeks after planting (Figure 2). Subsequently there was a continuous depletion at the two higher fertilization rates. At no N application the depletion started about one week from planting.

The highest N rate, 240 kg ha⁻¹, left significantly higher amounts of Nmin in the soil at harvest than did the 0 and 120 kg ha⁻¹ treatments (Table Va). The early plantings left considerably higher amounts of Nmin in the soil at harvest time than did the late ones. However, while Nmin of the unfertilized treatment was similar for early and late plantings, it had risen about ten times at the highest N rate in early plantings, but only about three times in the late plantings. On average for early and late crops, ‘Milady’ left more than two times as much Nmin in the soil as did ‘Marathon’ at the highest N rate, but only 8–40% more at the two lower rates. This difference between cultivars tended to be more pronounced in early than in late plantings.

Table V. Mineral nitrogen left in soil (0–30 cm) after harvest (a) and nitrogen balance (kg N ha⁻¹, calculated as soil Nmin plus N in biomass at harvest minus: applied fertilizer N, soil Nmin at planting, and N in biomass at planting) at three N fertilization rates in early and late plantings of two broccoli cultivars. n=9 (3 exps×3 reps). Different letters indicate significant main effects of N rate or planting time at the 5% probability level.

	Early			Late
kg N ha^−1	Milady	Marathon	Mean	Milady	Marathon	Mean	Mean
a) Mineral nitrogen left in soil after harvest (kg N ha^−1)
0	14	12	13	14	8	11	12 a
120	44	42	43	12	10	11	27 b
240	183	70	126	32	29	30	78 c
Mean	80	41	60 a	19	16	17 b	39
b) Nitrogen balance (kg N ha^−1)
0	21	31	26	37	24	30	28 a
120	23	44	33	12	8	10	22 a
240	98	14	56	−41	−27	−34	11 a
Mean	47	30	38 a	3	1	2 a	20

Nitrogen balance

N balances of the unfertilized plots were similar in early and late plantings, which indicated a mean net mineralization of 28 kg N ha⁻¹ (Table Vb). In the early plantings, the N balance increased with N rate up to an average of 56 kg N ha⁻¹. In the late plantings the opposite was the case, as the N balance decreased with N rate, and the highest N rate showed a mean deficit of 34 kg N ha⁻¹. This differential response to N rate at early and late plantings was mainly due to ‘Milady's large increase in soil Nmin at the harvest of the early plantings (Table Va), while ‘Marathon’ seemed to produce a lower N balance at the highest N rate also at the early plantings.

Harvest index and nitrogen recovery

The dry matter harvest index (HI) of the early cultivar ‘Milady’ was 20% higher than that of ‘Marathon’ (Table VI). A higher N rate tended to increase the harvest index, and more in the early plantings than in the late ones.

Table VI. Harvest index based on dry matter yields at three N fertilization rates of two broccoli cultivars (a) or at two planting times (b). n=18 (3 exps×3 reps×either 2 planting times or 2 cultivars). Different letters indicate significant main effects of cultivar, planting time or N rate at the 5% probability level.

	a) Cultivar		b) Planting time
kg N ha^−1	Milady	Marathon	Early	Late	Mean
0	0.19	0.16	0.17	0.18	0.17 a
120	0.25	0.21	0.24	0.22	0.23 ab
240	0.29	0.24	0.29	0.24	0.26 b
Mean	0.24 a	0.20 b	0.23 a	0.21 a	0.22

The nitrogen recovery in the above ground crop biomass at harvest was significantly higher for the 120 than for the 240 kg N ha⁻¹ treatment (Table VII). When only the harvested product was considered, there was a similar tendency (Table II). There was no significant difference in N recovery between the two cultivars or between early and late plantings. The three-factor interaction between N rate, planting time and cultivar (Table II) indicated that the reduction in N fertilizer recovery ratio in heads with increased N rate was less in ‘Milady’ than in ‘Marathon’ with early planting, while in late planting it was the same in both cultivars (data not shown).

Table VII. Apparent fertilizer N recovery in total above ground biomass and in the harvested product (agronomic efficiency) at two N fertilization rates. n=36 (3 exps×2 cvs×3 reps×2 planting times). Different letters indicate significant main effect of N rates at the 5% probability level.

kg N ha^−1	Biomass	Harvested product
120	0.82 a	0.28 a
240	0.65 b	0.22 a
Mean	0.74	0.25

The fraction of N located in broccoli heads tended to increase with increasing N rates in the early plantings, but showed no consistent tendency in the late plantings (Table VIII).

Discussion

Nitrogen

The broccoli yield increased approximately three times from the zero to the maximum N rate (Table IIIa). Since some of the experimental fields had high levels of soil mineral N at planting (Table I), and the applied N rates were fixed, the available N values (Nmin at planting plus applied fertilizer N) covered a wide range, from 29 to 375 kg ha⁻¹. As can be seen in Figure 1, there tended to be a slight decrease in yield at the highest available N values. This may indicate that the optimum N supply was in the range 200 to 250 kg N ha⁻¹, which is consistent with current recommendations for broccoli production in Norway (Anonymous, 1999).

Several workers have, in accordance with our findings, reported increasing yields of broccoli with increasing N rates (Letey et al., 1983; Tremblay, 1989; Zebarth et al., 1995; Everaarts et al., 1996; Gutezeit, 1996; Everaarts & de Willigen, 1999a), and quite a number of reports recommend fertilizer rates between 220 and 270 kg N ha⁻¹ (Cutcliffe et al., 1968; Greenwood et al., 1980; Kowalenko & Hall, 1987; Everaarts & de Willigen, 1999a; Vågen, 2003). Bowen et al. (1999) estimated the maximum marketable yield to be produced at an application of 436–558 kg N ha⁻¹, which is about twice the highest amount of N applied in our experiments. However, their experiments were made with a 36% higher plant density (77,500 plants ha⁻¹), which may partly account for the higher N requirement.

There was a strong relationship between broccoli yield and available N in the soil (Figure 1). Thus, the available N in the soil appears to be a useful predictor of potential broccoli yield. Sampling for Nmin shortly before planting would enable the grower to adjust the N rate to establish an available N level within the recommended range.

The harvest index (HI), or the fraction of the biomass harvested for consumption, was approximately 50% higher at the maximum N rate than on unfertilized plots (Table VI). Kowalenko and Hall (1987) also reported such an increase with N rate. Other studies found no effect of N rate on HI (Everaarts, 1994; Everaarts & de Willigen, 1999b). While other factors, such as plant density and the mean temperature prior to head initiation, may also have an impact on harvest index, our clear impression is that N rate is important for the allocation of assimilates to broccoli heads.

The nitrogen uptake to the above ground crop biomass showed a similar time course throughout the growing period in all experiments (Figure 2). The rapid N uptake started about three weeks from planting. Prior to that the root system is small but expanding, and the capacity for N uptake is limited (Greenwood et al., 1989). The relative figures for N taken up in the above ground plant biomass at an input of 0, 120, and 240 kg N ha⁻¹ were 100, 227, and 301, respectively (Table IIIc and Table IV). These relative figures are very similar to the corresponding figures for the fresh weight yield of heads, which were 100, 232, and 295 (Table IIIa). In contrast, they are strikingly different from the relative figures for total above ground DM accumulation, which were 100, 151, and 165, respectively (Table IIIb). This suggests that the actual N uptake of the crop is more determining for harvestable yield than is the total biomass produced to support broccoli heads. When the N supply is low, plants can still produce a reasonable amount of biomass with a lowered tissue N concentration, but the yield of broccoli heads will be dramatically reduced.

The fraction of N located in the heads ranged from 24 to 37%, with an average of 30% (Table VIII), which is comparable with the range 22 to 37% found by Everaarts and de Willigen (1999b). The difference between the average harvest index (Table VI) and the average fraction of N located to the heads reflects that broccoli heads are nitrogen sinks during inflorescence development (Shelp & Liu, 1992; Bowen et al., 1999; Everaarts & Willigen, 1999b). However, the effect of increasing N rate was generally much stronger on dry matter HI than on N fraction located in the heads, indicating that an abundant supply of N during the latter part of the growing period may also prevent the natural relocation of N from stems and leaves to heads. In this context, Everaarts and de Willigen (1999b) even found the fraction of N located in the heads to decrease with increasing N rate.

Table VIII. The fraction of N located in the harvested product at three N fertilization rates in early and late plantings. n=9 (3 exps×2 cvs×3 reps). Letters indicate lack of significant main effects of planting time and N rate at the 5% probability level.

	Planting time
kg N ha^−1	Early	Late	Mean
0	0.27	0.28	0.28 a
120	0.31	0.31	0.31 a
240	0.34	0.29	0.32 a
Mean	0.31 a	0.29 a	0.30

The soil mineral N levels first increased, due to nitrogen fertilizer application and mineralization, and then decreased as plant N uptake accelerated (Figure 2). In the late plantings, soil Nmin values decreased more rapidly than could be expected from the contemporary N uptake curves. These declines correspond well with incidents of heavy rainfall and explain the negative N balances at the highest N rate in the late plantings (Table Vb). Feller and Fink (2002) reported that a Nmin residue in the soil at harvest of 40 kg N ha⁻¹ was required to achieve optimum yield of broccoli.

The Nmin content of the 30–60 cm soil layer was only determined in 2001, and the results have not been shown except for the initial values in Table I. However, in the last planting on field 3, the 30–60 cm soil layer at the highest N rate actually contained more N than the 0–30 cm soil layer by harvest time. This is consistent with the heavy rainfall incidents that occurred about halfway through the growing period.

Planting time

The broccoli yield, total biomass production, and N content in broccoli heads all tended to be lower in the early than in the late plantings (Table III). A likely reason is the lower temperature during the first half of the growing period, leading to an earlier head initiation in early than in late plantings (Grevsen & Olesen, 1999). Thus, the assimilate flow would be directed to head rather than leaf development at an earlier growth stage, causing a lower total biomass production and a lower yield potential. This could partly be counterbalanced by increased fertilization rate, causing a higher relative growth rate, leaf area index, and radiation use efficiency in early plantings (Vågen et al., 2004). Another explanation for the stronger effect of N rate on head yield in the early than in the late plantings could be that N mineralization was lower due to lower soil temperature (Fink & Scharpf, 2000). This, however, appeared not to be the case in our study, as the N balance of the unfertilized plots was practically equal regardless of planting time (Table Vb).

Cultivar

On average for N rates, ‘Milady’ produced 9% lower yield than ‘Marathon’ in the early plantings, but 4% higher yield than ‘Marathon’ in the late plantings (Table IIIa). The lower yield of ‘Milady’ in the early plantings was probably related to earliness, as ‘Milady’ was harvested several days before ‘Marathon’ (Table I). It seems that ‘Milady’ has a lower time requirement for head initiation than ‘Marathon’, which may also explain the difference in harvest index between the two cultivars (Table VI). Grevsen and Olesen (1999) estimated thermal time requirements from planting to head initiation in three broccoli cultivars and found it to vary from 38 to 55 day-degrees centigrade, with a base temperature of 9.9°C.

The similarity of the nitrogen uptake patterns of the two cultivars indicates that ‘Milady’ did not reach an earlier harvest time in the early plantings because of a higher uptake rate of N. In contrast, it seems that ‘Milady’ was less capable than ‘Marathon’ of taking up high amounts of N for biomass production, at least within the required growing period for the two cultivars. This was also reflected in a higher amount of soil mineral N left by ‘Milady’ at harvest, especially at the highest N rate and in the early plantings (Table Va). In order to minimize the leaching potential from broccoli crops, it therefore seems important that fertilizer rates are adjusted to the expected growing period and N uptake capacity of the cultivar being grown.

Conclusion

The present research substantiates the current Norwegian recommendation of a total supply of 200 to 250 kg N ha⁻¹ in broccoli production. Fertilizer rates should be adjusted for soil mineral N to meet this requirement.

We gratefully acknowledge the technical assistance of Mr. Peter Stanton. Thanks are also expressed to Dr. Hugh Riley for valuable comments on the manuscript at an early stage.