Journal of Experimental Botany, Vol. 52, No. 357, pp. 821-827,
April 15, 2001
© 2001 Oxford University Press
Original Papers |
Towards optimization of growth via nutrient supply phasing: nitrogen supply phasing increases broccoli (Brassica oleracea var. italica) growth and yield
1 Centre de Recherche en Horticulture, Pavillon de l'Envirotron, Université Laval, Quebec (Qc) G1K 7P4, Canada
2 Agriculture Canada, CRDH/HRDC, St-Jean-sur-Richelieu, Quebec (Qc) J3B 3E6, Canada
Received 8 May 2000; Accepted 3 November 2000
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResults and discussionConclusionReferences |
|---|
A greenhouse experiment on broccoli (Brassica oleracea var. italica, cvs Windsor and Arcadia) was carried out in order to demonstrate that supplying nitrogen (N) to meet the nitrogen demands of plant growth stages, through N phasing, improves plant growth and yield, as compared to fertilizing at the conventional, optimal, constant N rate. Two broccoli cultivars and two rates of starter nitrogen fertilizer (optimum, 250 mg l-1 and sub-optimum, 150 mg l-1), were combined with three timings of fertigation change. Shifting N rate, at 60% and 75% of the market plant growth cycle significantly increased shoot dry weight and head fresh weight, compared to the constant-N rates treatments (controls). The highest yield and shoot dry weight were obtained when the N-rate was switched from the optimum level (250 mg l-1) to the sub-optimum level (150 mg l-1) at inflorescence initiation. The nitrogen-to-growth-stage-fitness effect was determined and partitioned into rate effect and phasing effect. The phasing effect was greatest, on both shoot dry weight and head fresh weight, at inflorescence initiation, and subsequently decreased until harvest time. None of the interactions was significant. The results demonstrated the superiority of nitrogen supply phasing over the conventional fixed-rate-supply method.
Key words: N-phasing, N-rate, N-use efficiency, fertigation.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionReferences |
|---|
Growth models
Knowledge of crop nitrogen demand is essential in predicting crop needs and, therefore, in developing reliable fertilizer recommendations for growers. Such recommendations are critical, for both economic and environmental reasons (Addiscott et al., 1991
). The N demand of a crop is defined as the N uptake over a set period, which allows the maximum dry matter (DM) growth rate under a given set of environmental conditions (Grindlay, 1997). The crop N demand was estimated as the product of potential DM growth rate, when N was non-limiting, and the critical %N, defined as the minimum N concentration in the plants needed for maximum growth rate at a given time (Greenwood, 1982).
The response of a crop to added N can be described by several models among which the quadratic (Fig. 1a), quadratic-plus-plateau and exponential models have been extensively used (Hunt, 1982; Ratkowsky, 1983; France and Thornley, 1984; Charles-Edwards et al., 1986; Tilman, 1988; Greenwood and Draycott, 1989; Tremblay, 1989; Cerrato and Blackmer, 1990; Hanks and Richie, 1991; Keen and Spain, 1992; Zebarth et al., 1995). The choice of one particular model over another is affected by such factors as the crop species and the range of N levels. However, the general pattern of crop growth involves the following phases from the lowest to the highest N levels: slow; rapid; maximum (or plateau); and decreasing growth. When expressed as a function of time, crop DM or yield exhibits a sigmoid behaviour (Fig. 1b). It is also known that N concentration, on a DM basis, varies with the developmental stage of the plant (Plénet and Lemaire, 1999) and, consequently, so does the N demand. Therefore, for each growth stage, there is probably a distinct N demand, which is best met by an optimum N rate corresponding to a critical %N in the dry matter of the plant at that growth stage.
|
Theoretical background
The conventional fertigation method of vegetables in the greenhouse provides N in the nutrient solution at a constant rate throughout the crop cycle. Several nutrient solutions have been used for either research or production of tomatoes, cucumber, lettuce, pepper, potato, and broccoli (CPVQ, 1990; Mohammad et al., 1999; Liu and Shelp, 1993). The conventional optimal N rate of supply (Nco) is selected as the one allowing just maximal yield at harvest. Since N demand is variable over time, each plant growth stage may have a specific N rate at which its growth rate is the fastest. Consequently, the conventional optimal N rate may either underestimate the instantaneous N demand (N uptake rate allowing just maximum growth), at the stages where it is the highest, or overestimate it, at the stages where it is the lowest (Fig. 1c
). The expected result is a reduced growth rate during those growth stages.
Specifically, in the early and late growth stages, growth rates are slow. For the former stage, this is attributable to the relatively low number of cells undergoing cell division, the insufficient leaf area available for light interception and photosynthesis, and the high proportion of photoassimilates being translocated to roots. Growth rate at the late growth stages is reduced due to self-shading and ageing (Ingestad and Àgren, 1992). At these stages, in a non-limiting N environment, the N uptake rate may exceed the N demand, leading to a situation of excess N accumulation in the plant. Unlike N deficiency, the state of excess N has been little documented. Osmotic problems were reported that were due to the disruption of the cation–anion balance and intracellular pH when nitrate reduction follows the termination of leaf cell expansion (Raven and Smith, 1976). Ammonium and, in particular, its equilibrium partner ammonia, are toxic at low concentrations. The main pathway of detoxification of ammonium ions taken up by the roots and ammonia derived from nitrate reduction, photorespiration or N2 fixation, is incorporation into amino acids, amides and related compounds (Marschner, 1995). When ammonium uptake exceeds ammonium assimilation, deleterious effects ensue. There exist several mechanisms for the removal of excess osmotic solutes from the shoot tissue. (a) Precipitation of excess solutes in an osmotically inactive form; synthesis of oxalic acid for charge compensation in nitrate reduction, and precipitation as calcium oxalate (Egmond and Breteler, 1972). (b) Retranslocation of reduced nitrogen (amino acids and amides) together with phloem-mobile cations, such as potassium and magnesium, to areas of new growth. (c) Retranslocation of organic acid anions, preferentially malate, together with potassium into the roots and release of an anion (OH- or 
) after decarboxylation. All these processes, which involve transport and/or synthesis, are energetically costly to the cell, and can be viewed as a diversion to its normal growth activities. A cell with excess osmotic solutes is considered to be in a state of excess solutes stress. Consequently, the conventional optimal N rate, which at times, in the growing period of a plant, brings about an excess N state, precludes maximal plant growth, at least from a theoretical standpoint. It was hypothesized that the adjustment of N supply to meet the N demands of different growth stages would lead to a better growth and yield as compared to the conventional, constant N fertilization rate. The maximum potential growth would be obtained by successively supplying optimal N concentrations of corresponding growth stages throughout the crop cycle. N phasing is defined here as the timely and successive supply of different rates of N. The N rate of 250 mg l-1 has been reported optimal or near optimal for the growth of broccoli (Liu and Shelp, 1993; Shelp et al., 1992). This rate is similar to that of Hoagland's solution culture (266 mg l-1). This study aimed at demonstrating that maximum yield and dry matter of broccoli, obtained with conventional constant N fertigation, could be improved through N phasing.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionReferences |
|---|
Plants and experiment management
Broccoli (Brassica oleracea var. italica cvs Arcadia and Windsor) seeds were sown in a greenhouse on 18 January 1999 in seedling trays containing rockwool cubes (Grodan AM 38/40) filled with vermiculite and supplied with one-quarter strength nutrient solution (described below). Four-week-old seedlings were transplanted, one per pot, into 6.0 l pots filled with fine sand. Pots were spaced 40 cm within rows, and 100 cm apart between rows in the greenhouse, resulting in a plant density of 25 000 plants ha-1. Natural lighting in the greenhouse was supplemented by high-pressure sodium lamps (HPS) that provided a photosynthetic photon flux density of 150 µmol m-2 s-1 at pot level. The plants were maintained on a 16/8 h light/dark period, at an average day and night temperature of 20/15 °C, and trickle-fertigated at a flow rate of 25 ml min-1, according to the following schedule: 1 min h-1 for the first 2 weeks; 3 min h-1 during the light period (16 h) and 3 min every 2 h during the dark period (8 h) from the third to the fifth week, and finally, 4 min h-1 during light periods and 3 min every 2 h during dark periods, from the sixth week until harvest (25 April 1999). Two modified Hoagland nutrient solutions (Hoagland and Arnon, 1950), differing only in their N levels (150 mg l-1 and 250 mg l-1) were used. They contained, in mg l-1, P, 77; K, 312; Ca, 60; Mg, 49; S, 129.41; Mn, 0.1; Cu, 0.03; Zn, 0.12; Co, 0.01; Mo, 0.018; B, 0.39; Cl, 0.1, and Fe, 2.1.
The sub-optimal N rate (150 mg l-1) was switched to the optimal one (250 mg l-1) and vice versa on 29 March 1999 (inflorescence initiation), in the first group of plants, and later on 8 April in the second group of plants. That is at 60% and 75%, of the market cycle, respectively. These two N rates were kept constant throughout the plant cycle, in the third group of plants.
Statistical analysis
The experimental design was a split-factorial with four replications, cultivar (2) as the main-plot factor, and, combined in the sub-plots, starter-nitrogen rates (150 mg l-1 and 250 mg l-1) and three fertigation switching periods: at 60%, 75% and 100% of the cultivars market cycle. To fit the quantification of the two controls (constant rates 150 and 250 mg l-1) into a shifting period scale, it was assumed that their rates were shifted just after the end of the market cycle, that is at 100% of it. Thus, the six sub-treatments consisted of four combinations of two nitrogen rates (150 mg l-1 and 250 mg l-1) switched at two growth stages, and two controls (150 mg l-1 and 250 mg l-1 at constant rates). The following sub-treatments were evaluated: (1) 150 mg l-1 N at constant rate; (2) 250 mg l-1 N at constant rate; (3) 150 mg l-1 N until 60% of the commercial cycle, followed by 250 mg l-1 until harvest (150;250)60; (4) 150 mg l-1 N until 75% of the commercial cycle, followed by 250 mg l-1 N until harvest (150;250)75; (5) 250 ppm N until 60% of the commercial cycle, followed by 150 mg l-1 N until harvest (250;150)60, and (6) 250 mg l-1 N until 75% of the commercial cycle, followed by 150 mg l-1 N until harvest (250;150)75. All statistical computations were done using the Statistical Analysis System (SAS Institute Inc., 1996). Analysis of variance, pooled over blocks, was performed for four variables recorded when inflorescences were at the stage of commercial maturity (13–15 cm): dry matter, leaf area, head weight, and head diameter. Significant means, computed using data from individual plants, were separated by orthogonal contrasts, and the main effect of the switching period was examined by fitting a response model.
|
Results and discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionReferences |
|---|
Shoot dry weight
Shoot dry weight varied significantly with respect to cultivar, nitrogen rates and switching periods at 0.05, 0.01 and 0.05 significance levels, respectively (Tables 1
, 2). None of the interactions was significant. All but one sub-treatment involving shifting N rates (N phasing) proved higher than the Nco. The highest dry weight was obtained when N rate was shifted from 250 mg l-1 to 150 mg l-1, at 60% of the market cycle (250;150)60. It was 10% greater than the conventional method using a constant rate of 250 mg l-1 N. Dry matter decreased linearly as N shifting periods occurred after the growth stage of 60% of the market cycle (Fig. 2). Arcadia out-yielded Windsor, and all the sub-treatments having 250 mg l-1 N as the beginning N level yielded more than those which used 150 mg l-1 N as the starter.
|
|
|
The symmetric sub-treatments (X,Y)Z and (Y,X)Z, for a given shifting period Z, represent the assessment of the two N rates, X and Y, at each of the two growth stages, before and after Z. Therefore, when the difference (X;Y)Z–(Y;X)Z is greater than zero, the former sub-treatment (X;Y)Z is better suited to both growth stages than is the latter (Y;X)Z, that is, it results in better growth. The corresponding effect can be termed the ‘nitrogen-to-growth stage-fitness effect’ or GSFE, and its value is relative. The GSFE decreased from the growth stage 60% (19 g) to commercial maturity (14 g), equalling the N rate effect (RE); obtained from the difference of the two constant N rates, (250;250)100 and (150;150)100 (Fig. 3
). This suggests that GSFE is a function of the rates effect and another effect resulting from shifting N rates: the shifting or phasing effect (PE). In fact, the phasing effect (PE) can be quantified as follows:
 |
 |
|
It can then be writen:
 |
Within the range of shifting periods (60% to 100%), PE varied from 5 g to 0. Practically translated, this means that each broccoli plant developed 5 g of dry matter more when switched from 250 mg l-1 N to 150 mg l-1 N, at 60% of the commercial maturity time, compared to the 250 mg l-1 N conventional constant rate. This gain progressively decreased as the shifting period occurred later in the cycle, eventually to reach zero (Table 3). Accounting for the plant density used in the experiment (25000 plants ha-1), switching N rates from 250 mg l-1 N to 150 mg l-1 at inflorescence initiation, resulted in a dry matter gain of 125 kg ha-1. When expressed as a percentage of GSFE, the phasing effect represented 26%. In other words, the difference between the two sequential formulae (250;150) mg l-1 and (150;250) mg l-1 is 74% due to the difference between the two N rates 250 and 150 mg l-1, and 26% to the shifting of these rates at 60% of the commercial maturity.
|
These results can be explained by the dynamics of nitrogen and dry matter accumulation and partitioning in broccoli. The rates of dry matter and nitrogen accumulation in two broccoli cultivars were particularly low within 2 weeks after transplantation, and the last 2 weeks of the commercial maturity (Bowen et al., 1998
). This suggests that N demand in broccoli is low in its early stages (about 2 weeks), then dramatically increases during the middle stages for about 4 weeks, and diminishes for about 2 weeks during the final stages of its growth. If it is assumed that inflorescence initiation (60% of the commercial cycle), is the threshold between the middle stage and the latest stage, then there is a reasonably physiologically sound explanation of these results. The sub-treatments (250;150)60,75 yielded more than the sub-treatments (150;250)60,75 because unlike the latter, they were supplied by the N optimum rate 250 mg l-1 throughout the longer middle stage during which the N demand is very high. Sub-treatment (250;150)60 yielded more dry weight than (250;150)75 because the decreasing N demand of the late stages adjusted right away to the decreased 150 mg l-1 N level, thus preventing the commonly observed adverse effect of excess N on plant growth. This was corroborated by field observations. Necrotic toxicity spots were observed on the leaves of the axillary buds of cultivars Arcadia, grown at N 250 mg l-1 constant rate, after inflorescence initiation, proving that a shift in N demand had led to a shift of the optimal N level. For the sub-treatment (250;150)75, the adjustment took place 10 d later resulting in a 4% decrease of dry matter which corresponds to a dry matter decrease of 460 kg ha-1. The superiority of sub-treatment (150;250)60 over sub-treatment (150;250)75 can be explained by the fact that under the sub-optimal condition of nitrogen, vegetative growth was limited; when the plants were supplied with the N optimum level 250 mg l-1 at inflorescence initiation, the vegetative growth increased steadily, possibly as a result of an enhanced induction of nitrate reductase activities and nitrate transporters genes (NRT1 and NRT2 families) whose expressions have been reported to be induced by nitrate (Crawford and Glass, 1998). The same inductive effect may explain the even performances of sub-treatments (150;250)60 and (250;250), both treatments received the same N rate during the last 40% of the growth cycle; one would have expected sub-treatment (150;250)60 to yield less dry weight than the other because of the N sub-optimal condition during 60% of the growth cycle, instead it did better, with a mean dry weight of 232.1 g, which was not significantly different from the 229.1 g of (250;250).
Leaf area
Shifting N rates at any time between inflorescence initiation and commercial harvest did not affect leaf area. Leaf area varied significantly with cultivars and with the order of N rates within combinations. The highest leaf areas were obtained when a solution culture of 250 mg l-1 N was used, irrespective of the switching periods. Cultivar Arcadia had the highest leaf area. None of the interactions was significant (Table 1, Table 2).
For a given sequence of N rates, the lack of variation of leaf area, regardless of the switching period, suggests that this character is already fixed at the stage of inflorescence initiation. This confirms some of the results of Bowen et al. (Bowen et al., 1998).
Head weight
Fresh head weight was significantly affected by switching N rates at any time between inflorescence initiation and commercial harvest. The highest yields were obtained when N rates were shifted at inflorescence initiation. Yields decreased as the shifting of N rates was delayed (Fig. 4). All the sub-treatments involving shifting N rates had higher head weights than those involving constant N rates. The highest yield was recorded when N rates in the culture solution was shifted from 250 mg l-1 to 150 mg l-1 at inflorescence initiation. This represented a mean increase in head weight of 58% more than that obtained using the N 250 mg l-1 conventional method. This 58% corresponded to a 5 t ha-1 of fresh weight. The two cultivars did not differ significantly in head fresh weight. Starter solution with 250 mg l-1 resulted in the plants with the highest yields. The three factors did not interact significantly (Table 1, 2). The GSFEs at 60% (inflorescence initiation) and 75% of the market cycle were, respectively, 99 g and 92 g. This indicates that, on the one hand, the adjustment of N supply from 250 mg l-1 to 150 mg l-1 was more effective at the 60% stage than at the 75% stage, and on the other hand, the sequence (250 mg l-1; 150 mg l-1) was better suited to the growth stages prior to and post inflorescence initiation than the sequence (150 mg l-1; 250 mg l-1). The respective PEs were 37 g and 30 g, indicating that the effect of N phasing was more marked at inflorescence initiation than at 75% of the post-transplantation commercial growth cycle (Table 3). Agronomically, this means that each individual broccoli plant head gained 37 g of fresh weight when N rate was shifted from 250 mg l-1 to 150 mg l-1 at inflorescence initiation, relative to the conventional (250;250) mg l-1.
|
The behaviour of broccoli fresh head weight with respect to N phasing supply, is similar to the trends observed in shoot dry weight. This is indicated by the good linear relationship between shoot dry weight and head weight at maturity (Fig. 5
). Thus, 86% of the weight accumulated in the florets can be attributed to shoot dry weight already accumulated by the broccoli plant before inflorescence initiation. The remaining variation, may be attributed to the loss of energy during various processes such as uptake, translocation, remobilization, maintenance, and transpiration. Under conditions of limiting N at the onset of floret development, translocation of N from other parts of the plants to the inflorescence tissues occurs in broccoli (Shelp and Liu, 1992; Liu and Shelp, 1995; Bowen et al., 1998). This fact explains the observations on head weight: using the optimal 250 mg l-1 N rate during the entire middle stage, where N demand was the highest, resulted in the greatest accumulation of dry matter before inflorescence initiation, that is necessary to increase head weight. Two possibilities may explain the observations at inflorescence initiation. Either N demand decreased as a result of dry matter accumulation decrease and ipso facto, fitted to the newly decreased 150 mg l-1 N rate, or it remained constant or was elevated and was compensated by the remobilization and translocation of N from leaves to the inflorescence. However, had N demand remained high, the sub-treatment (250;150)75 would likely have performed better than the sub-treatment (250;150)60 because it would have been supplied for a longer period with the optimal N rate. This suggests that N demand adjustment occurred rather than N remobilization.
|
Head diameter
Head diameter did not differ significantly for any of the factors cultivar, starter N rate, and N shifting period. Interactions between these factors were also not significant (Table 2). However, these data were not processed further because of the high percentage of the total variation attributable to the experimental error (65%). This can be explained by the variability observed within treatments with respect to the commercial maturity date. When the young inflorescence emerges, the florets are tightly clasped, and they progressively loosen as the inflorescence matures, contributing to the widening of the head. A random factor such as leaf blade shading of the inflorescence delayed the widening and commercial maturity of the head. This factor may have also contributed, although to a lesser extent, to the experimental error (20%) observed in analysing fresh head weight data. The inclusion of part of the already matured stem (15 cm) in the fresh head weight probably reduced the effect of maturity date variability on head weight.
|
Conclusion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionReferences |
|---|
This study demonstrated an hypothesis according to which the conventional ‘optimum’ N supply rate could be further optimized by matching N supply to N demand of distinct growth stages. The results showed that both shoot dry weight and yield improved by 10% and 58%, respectively, when N rates were shifted from 250 mg l-1 to 150 mg l-1 (N phasing) at inflorescence initiation, as compared to the conventional constant optimal N rate 250 mg l-1. Three different effects were detected: the rate, the phasing and the induction effects. The study did not investigate effects of N phasing at growth stages prior to inflorescence initiation. Therefore, the maximum phasing effect was not determined, nor was the maximum possible dry matter and yield under the prevailing environmental conditions. Potential growth and yield would be obtained by applying N at the concentration suited to each growth stage demand. In order to achieve this, the N demand at different growth stages must be known with much greater accuracy. To this end, study is needed to measure N demand and relate it to N uptake kinetics, N assimilating enzyme activities N and dry matter accumulation.
|
Notes |
|---|
3 To whom correspondence should be addressed. E-mail: Yves.Desjardin{at}plg.ulaval.ca
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionConclusionReferences |
|---|
Addiscot TM, Whitmore AP, Powlson DS.1991. Farming, fertilizers and the nitrate problem. Wallingford: CAB International.
Bowen PA, Zebarth BJ, Toivonen PMA.1998. Dynamics of nitrogen and dry matter partitioning and accumulation in broccoli (Brassica oleracea var. italica) in relation to extractable soil inorganic nitrogen. Canadian Journal of Plant Sciences 79, 277–286.
Cerrato ME, Blackmer AM.1990. Comparisons of models for describing corn yield response to nitrogen fertilizer. Agronomy Journal 82, 138–143.
Charles-Edwards DA, Doley D, Rimmington GM.1986. Modeling plant growth and development. Orlando, Florida: Academic Press.
Conseil des Productions Végétales du Quebec (CPVQ).1990. Légumes de serres: Culture sur film nutritif (NFT), sur laine de roche et en milieux tourbeux. Québec: MAPAQ, 21.
Crawford NM, Glass ADM.1998. Molecular and physiological aspects of nitrate uptake in plants. Trends in Plant Science 10, 389–395.
Egmond F van, Breteler H.1972. Nitrate reductase activity and oxalate content of sugar-beet leaves. Netherlands Journal of Agricultural Science 20, 193–198.
France J, Thornley JHM.1984. Mathematical models in agriculture. London: Butterworths.
Greenwood DJ.1982. Modelling of crop response to nitrogen fertilizer. Philosophical Transactions of the Royal Society, London Series B 296, 351–362.
Greenwood DJ, Draycott A.1989. Experimental validation of an N-response model for widely different crops. Fertilizer Research 18, 153–174.
Grindlay DJC.1997. Towards an explanation of crop nitrogen demand based on the optimization of leaf nitrogen per unit leaf area. Journal of Agricultural Science 128, 377–396.
Hanks J, Ritchie JT.1991. Modeling plant and soil systems. Madison, Wisconsin: American Society of Agronomy.
Hoagland DR, Arnon DI.1950. The water-culture method for growing plant without soil. California Agricultural Experiment Station, Circular No. 347.
Hunt R.1982. Plant growth curves: the functional approach to plant growth analysis. London: Edwards Arnold.
Ingestad T, Àgren GI.1992. Theories and methods on plant nutrition and growth. Physiologia Plantarum 84, 177–184.
Keen RE, Spain JD.1992. Computer simulation in biology: a basic introduction. New York: John Wiley.
Liu L, Shelp BJ.1993. Nitrogen partitioning in greenhouse-grown broccoli in response to varying NH4+:NO3- ratios. Communications in Soil Sciences and Plant Analysis 24, 45–60.
Liu L, Shelp BJ.1995. Mobilization of stored nitrate in broccoli (Brassica oleracea var. italica). Canadian Journal of Plant Sciences 75, 709–715.
Marschner H.1995. Mineral nutrition of higher plants. San Diego: Academic Press.
Mohammad MJ, Zuraiqi S, Quasmeh W, Papadopoulos I.1999. Yield response and nitrogen utilization efficiency by drip-irrigated potato. Nutrient Cycling in Agroecosystems 54, 243–249.
Plénet D, Lemaire G.1999. Relationships between dynamics of nitrogen uptake and dry matter accumulation in maize crops. Determination of critical N concentration. Plant and Soil 216, 65–82.
Ratkowsky DA.1983. Nonlinear regression modeling. New York: Marcel Dekker.
Raven JA, Smith FA.1976. Nitrogen assimilation and transport in vascular land plants in relation to intracellular pH regulation. New Phytologist 76, 415–431.[Web of Science]
SAS/STAT Institute Inc.1996. SAS/STAT User's Guide, Release 6.12 Edition. Cary, NC:SAS Institute Inc.
Shelp BJ, Penner R, Zhu Z.1992. Broccoli (Brassica oleracea var. italica) cultivar response to boron deficiency. Canadian Journal of Plant Sciences 72, 883–888.
Shelp BJ, Liu L.1992. Nutrient uptake by field-grown broccoli and net nutrient mobilization during inflorescence development. Plant and Soil 140, 151–155.
Tilman D.1988. Plant strategies and the dynamics and structures of plant communities. Princeton, New Jersey: Princeton University Press.
Tremblay N.1989. Effect of nitrogen sources and rates on yield and hollow stem development in broccoli. Canadian Journal of Plant Sciences 69, 1049–1053.
Zebarth BJ, Bowen PA, Toivonen PM.1995. Influence of nitrogen fertilization on broccoli yield, nitrogen accumulation and apparent fertilizer-nitrogen recovery. Canadian Journal of Plant Sciences 75, 717–725.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||