An overview was given on how statisticians can help researchers to use better experimental designs

and to improve interpretation of the data obtained. The following questions were discussed:

N response:

what range to cover

how many levels to use

equal spacing between levels

equal replication of levels

what form of response curve to fit

sampling

Growth curves and critical N:

how often to sample crop and how many plants to sample

what form of response curve to fit

fitting families of growth curves to data from different treatments

relating growth rate to N content

how many application rates

N application:

rates, application methods, application forms

factorial experimentation

numbers of factor levels

total number of treatment combinations: fractional factorials, confounded designs

response curves

Soil/environment sampling:

how many samples

where to sample: spatial variability

when to sample: temporal variability

what form of response curve to fit

relationships between response variables

A fuller description of a number of these factors will be made available later. If there are any questions

please contact the ENVEG administrator at enveg.enveg@hri.ac.uk.

A definition of crop quality depends very strongly on user requirements. There are numerous criteria:

• farmer:
• disease resistance, marketable yield
• market:
• visual aspects of crop
• user:
• visual, taste, nutritional aspects, absence of toxic substances or residues

Most quality aspects are difficult to quantify and empirical scales are mostly used. The overall quality has to be quantified as well, which is still more difficult.

For example, the relationship between N application and head quality for broccoli. Some factors, e.g. head size, increase with increasing N supply up to a very high level of N. Most other quality factors reach an optimum at moderate N supply and decreases with higher supplies. The optimum changes for different parameters, and the problem is to delineate an appropriate zone of N supply, or availability, which gives the highest overall quality.

A quality model for one vegetable crop can not be extrapolated to another vegetable crop, contrary to the situation in cereals, where extrapolation of quality models is feasible. This stresses the need for further research for vegetable crops.

In the 'EC Quality Standards for Horticulture Produce' there are 7-8 pages per vegetable about quality aspects. In the UK this is published by MAFF.

Influence of N on Chemical quality of vegetables

With increasing N level in soil, there is an increase in NO₃^-, oxalic acid and free amino acids content in the crop. Vegetables are responsible for most of the human nitrate intake. The EC has set NO₃^- limits for lettuce and spinach and will do so for other vegetable species as well. Various factors influence nitrate content, e.g.:

variability from country to country: climatological conditions

variability from season to season

uptake of nitrate is controlled genetically

nitrate is inversely related to DM

fertilizer form: NO₃^- ó NH₄⁺

time of the day (large diurnal variations in NO₃^- content)

interaction with fertilizer form (NO₃^- ó NH₄⁺) 

Vegetables can be divided into two categories at harvesting, single or multiple harvests. Root and bulb crops are normally harvested only once, however most brassica's and lettuce benefit from multiple harvests. Work by Burns has shown that multiple harvests give greater yields and that the crop also shows a smaller yield response to N. Cauliflower's should be harvested every three or four days to achieve the maximum yield.

At harvest, we try to assess everything! Certainly the total, marketable and residue fresh and dry weights. For larger and higher value crops such as Cauliflower's and Cabbage, individual heads are measured, weighed and graded.

In the UK, quality standards, e.g. size, colour, shape and appearance for all vegetable, salad and fruit crops are specified in a MAFF publication entitled 'EC Quality Standards for Horticulture Produce'. All produce destined for sale to the general public must meet those standards, which are strictly enforced by the main buyers, who are the big multiple supermarkets.

As a general rule, marketable yield shows an increase in response to increasing N. However excessive N can have a toxic effect leading to a decline in yield. Even if there is no decline, it is no guarantee that the quality remains unaffected. Excess N can improve the appearance of a crop but can also lessen the resistance to mechanical damage during harvest, lead to discolouration, a loss of flavour and a decline in post harvest keeping quality.

In Germany, response curves are fitted to data using various equations: Mitscherlich, quadratic, linear + plateau.

There can be as much as 100 kg/ha N difference in determination of optimum using different curves. The optimum is assessed visually first. Fertiliser is usually applied as two dressings, a base dressing followed three to four weeks later by a top dressing.

It is a common German practice when doing field experiments to produce response curves, to add the data from different years together and mean the results. This is done by allocating the optimum yield in any one year at a value of 100, and making all other results a proportion of that. The assumption is that the N level is the only important factor.

The German ‘rule of thumb’ is to ensure that there is 100 kg/ha N available to young roots.

Determination of optimal N fertilizer amounts for vegetables and assessment of nitrate contents: experiments on chernozems. Four levels of N was used in the experiment. Measurements included nitrate content, total N, yield, amounts of residues in some cases.

Quadratic response curves were fitted to yield data (excluding extremely low yields). All species showed a statistically significant yield increase with increasing N, with spinach, Chinese cabbage and savoy cabbage showed the strongest response, and had a maximum yield above the highest N level. N fertilization is heavily influenced by market expectations. Chinese cabbage in Austria is heavily fertilized to produce big heads. Spinach has to be harvested the Thursday before Easter, and needs to have a dark green color, with a minimum of yellow leaves. Obviously maximum yield is not the only important factor for N recommendations, aspects such as nitrate content at harvest, the consumer’s demands, and economical optimum have to be taken into account.

High amounts of N were left on the fields, mainly after cabbages. To estimate the leaching risk of nitrogen from these residues, research was initiated to determine N content in the residues, but as of yet, only limited amounts of data are available. From the first data it is obvious that the amount of N in the residues increases with increasing N fertilization. For white cabbage and savoy cabbage half the N of the total plant remained on the field as residue N. Data on nitrogen in crop residues should best be given with a range of values, since the N in residues heavily depends on N supply.

An overview of current coring methods and spacing of sampling was given. There is no clear or good method for measurement or root length or density. Questions were raised as to the validity that only 10% of the roots are required to support the nutritional requirements of the crop.

A common method for assessing root density is to separate the roots from soil and put the roots in tray with grids. The number of roots intersecting each gridline is then counted, giving the root length density. The method is very time consuming. Using an areameter can reduce the work.

Alternative methods:

soil cores ("core break method")

"Profile wall method"

"Glass wall method". Advantages: non destructive method, repeated measurements; disadvantages: the different soil bulk density near glass will affect the root density.

Root observation in box: soil is disturbed.

Minirhizotron: acrylic tubes are dug into the soil in a 30° angle; at the time of measurement a light source and a camera are used to count roots. This method counts less roots in the topsoil and more in the subsoil as compared to the core break method. The reason is that older (inactive) roots are not accounted for.

The question was raised, whether we need three-dimensional information, in time, for modeling purposes.

Unpublished data and data from literature on nutrient demand were gathered. Only where fertilization was carried out according to official recommendations or good fertilizer practice, was the data included (to avoid cases of luxury N consumption or N shortage). Minimum number of datasets per crop was five, but very little data is available. Data included fresh matter, N content and N demand (kg/ha). Results from the Netherlands and Denmark are in the same range.

Growth curves are not needed in simple Nmin system but are needed in KNS, also in Swiss system (max. 60 kg N per application).

Standard KNS system uses three different growth curves, standard curve based on experimental data, and a curve below and above. The farmer visually assesses his crop’s growth against a standard, chooses the curve that matches his yield expectation and adjusts his N application to suit.

N uptake curves in Germany are plotted against time and gives much variation, using day degrees may give a more uniform picture.The use of phenological stage of the crop improved variability significantly.

An understanding of crops growth and its associated nitrogen demand is vital before any sensible conclusion can be drawn about fertilizer requirements.

Many measurements can be made at harvest to obtain the overall amount of fertilizer required. The only sensible way of doing this is to measure total dry matter yield for at least 7 levels of N fertilizer including a nil control. Determine the minimum level of N for maximum dry matter yield and determine the crop N content.

Given this amount of N, providing there are no other limitations to yield such as disease or poor growing conditions, it will be useful to examine the relationships between markeable yield, total N offtake and the amount of N left in crop residues. In experiments at HRI, dry matter yield is determined by harvesting:

root crops: 2 rows x 10m/replicate, lettuce, brassica: 30 plants/replicate

Multiple harvest crops are problematic in terms of when to determine final level of N uptake. There are different approaches possible:

Use of 2 different harvest areas, one of which is completely harvested to determine total DM yield at 50% marketable

Mark and grade all single plants as they become marketable. At the end of the harvesting period the entire crop is then harvested for determination of dry matter yield.

 Assessments of fertilizer efficiency are also possible where of ¹⁵N labeled fertilizer has been applied. However, the cost of labelled fertilizer remains a limitation, and normally it will only be applied to microplots. The following microplot sizes have been used:

1.2 x 2.1 m for Brussels sprouts, 1.2 x 1.6 m for lettuce, onion, leek, swede, redbeet

The fraction of N derived from soil and fertilizer, respectively, allows calculation of fertilizer efficiency.

Assessing the relationships between growth & N requirement during growth are important for the timing of fertilizer application to N demand. There are few shortcuts to this sort of data. Tei et al. (1996) sampled at weekly intervals:

2 rows x 1.02 m lettuce (10 plants), 2 rows x 0.43 m onions (10 plants), 2 rows x 0.65 m red beets (10 plants)

This procedure involves a huge amount of work. However, if the shape of the growth curve is known, e.g. if the crop is known to have an exponential growth phase followed by a linear, it is only necessary to sample enough point to be able to establish the parameters of the growth curve.

These assessments should also be made at several N levels in order to determine the critical N concentration. For certain crops, e.g. Cabbage, application of all the N (up to 400 kg ha^-1 N) at planting leads to greater growth and yield. Environmental considerations may demand that e.g. split fertilization or fertilizer banding is made.

Why are we using critical N concentration? Do we want to assess optimum growth or yield, or assess the overall requirement for N? Maybe we should be assessing overall plant demand for N and to what extent the soil supplies the plant’s demand. Choosing which part of the plant to assess and when to assess it, is very important. In Portugal the youngest petiole is normally used.

Critical N concentration is the minimum concentration of total N that produces the maximum yield at a given time and situation (Justes et al., 1994). It is used to assess: 

• yield response: potential yield, potential seed production, N concentration at 90% maximum production
• plant status - quality: NO₃^- content, skin colour (fruits), protein content in cereals

Critical N concentration is measured by means of NO₃^-- or total N content. The latter gives a more accurate image. The problem remains which part of the plant to sample. One can analyze the youngest mature leaf petiole or use the preside dress nitrate soil testing to measure critical N in the soil in order to predict fertilizer needs. The critical N concentration allows for: 

• evaluation of fertilizer efficiency
• increased efficiency of fertilizers
• economic use of fertilizer
• minimization of nitrate leaching losses
• greater control over crop quality

There followed a discussion of the methods by which nitrate is measured in plant material. Water, both hot and cold, copper sulphate and also water:methanol mix. Some people asked whether there is or should be an official EC method.

When measuring the N status of plants, we have a choice of the following methods: 

• Total N
• Reduced N (Kjeldahl, organic N)
• Total nitrate N (dry matter basis)
• Cell sap nitrate N (volume basis)
• Indirect methods (eg. chlorophyl)

 The information can be used to:

• diagnose deficiencies directly in the field, and subsequently apply additional N dressings in the growing season
• model growth response to N in fertilizer planning models (the most common measurements are total N and reduced N)

 A. Laboratory methods

• Nitrate-N is usually measured either in whole plants (e.g. lettuce, cabbage) or in particular parts (e.g. potato or cereal leaves) - both involve fine-chopping and boiling (usually in copper sulphate, CuSO₄, potassium chloride, KCl, or water).
• Results are usually expressed on DM-basis (involves drying), but sometimes on fresh weight basis (in Germany?)
• Conflicting evidence and opinions about whether to extract from fresh or dried material. Fresh/frozen samples mostly used.

 B. Field methods

• Cell-sap extracted from specific part of plant (e.g. leaf petioles), mixed with reagents and measured by colour comparison with chart or by simple hand-held calorimeter.
• Results expressed on volume basis (mg/1).

 Interpretation:

• Nitrate contents decline rapidly as plants develop. Critical values are therefore usually expressed as linear function of time. Examples: potatoes (Swe.), cabbage (Nor.), lettuce (Ger.)

Advantages & Problems:

• Rapid assessment, but nevertheless somewhat laborious. Time does not always reflect plant development accurately. Large variability in experimental basis for critical values

• Most commonly used in relation to arable crops, but also in use for vegetables (e.g. WELL_N/N_ABLE models from HRI)
• Total-N usually measured in whole plant (leaf+stem), but sometimes in separate fractions (e.g. potato leaf and tuber)
• Values usually expressed on dry weight basis, and exhibit a typical decline as growth proceeds (dilution effect)
• Interpreted in various ways by different workers:

• in relation to time from planting (Siman, Sweden)
• in relation to temperature sum (Hansen et al., Denmark)
• in relation to total DM-mass, excluding fibrous roots (Greenwood et al., UK, Justes et al., France)

• General equation for both arable and horticultural crops proposed by Greenwood et al. (1986). Later modifications have been made for some vegetable crops (Greenwood & Draycott, 1989; Greenwood et al. 1996), for C3/C4 crops (Greenwood et al. 1990) and at early growth stages of wheat (Justes et al. 1994, France)
• Such equations are used in models to calculate the plants' requirement for nitrogen (none when actual N >critical N, otherwise a linear growth decline with actual N/critical N)
• Unresolved questions:

• Is greater differentiation between crops required?
• What are the critical-N values during early growth?
• How to consider nitrate-N in vegetables (often high)?

Critical N concentrations as they are used e.g. in the N_ABLE model seem to be very (too?) high.

Maximum of luxury comsumption in the N_ABLE model is twice that of critical N, and this was received with some doubt; the suggestion was 'critical N plus fifty percent'. An explanation could be that in England farmers rely heavily on luxury consumption in the beginning of growth to supply the crop with enough N later in the growth stage (split application could e.g. be difficult if the soil becomes too dry later in the growing season). In this way perhaps N concentrations 50% higher than the critical level as a maximum could be possible.

Nitrate sap testing is questionable: there are too many influencing factors (see e.g. discussions in EUROPP concerted action)

Results from experiment testing the different forms of N sources. Wide range of e.g. organic household waste compost, sugar beet waste etc. Compost proved ineffective, but hornmeal provided good results even on short term crops. Ammonium nitrate, (NH₄)NO₃ and urea gave high nitrate levels, while slow release fertilisers gave good results on sandy soils.

DCD is used to reduce denitrification, but the only positive effect on yield was recorded from a maize trial on a light, sandy soil, and DCD may leave higher amounts of NO₃^- in the soil at harvest. Urea is used as a top dressing in Portugal, mainly calcareous soils.

Ammonium sulphate/nitrate and incorporated urea is used both in Greece and Spain, but not in Austria, mainly because soil pH is too high and the risk of ammonia losses are too high.

Very high concentrations of ammonia were placed in depots to slow down denitrification. Roots growing near the depot were killed, but presumably nitrification is inhibited in the depot so the risk of leaching can be reduced. The amount of fertilizer applied can be reduced by 20% but plants appear to have problems with the system.. It may however be useful in certain cases, e.g. where the soil is sandy and well drained, and where rows are widely spaced.

In timing experiments, better yields were achieved when there was high N at the beginning of the experiment. Banding only works with some crops, and needs further work to compare the methods.

The presentation included examples of growth curves of lettuce, leeks, gherkins, white cabbage, etc. N requirements are assessed by comparing growth visually against a standard curve, N amount required to reach marketable yield is then estimated from the curve.

Application N = Expected plant N – Mineral N.

N uptakes are measured at optimum N for maximum yield.

With fertigation, the crop can be supplied with exactly the nutrients it needs, but the N_min cannot then be used, because where does one take the samples?

In a hydroponic system, a temporary break in N supply reduced yield markedly in Lettuce and also depressed yield in Cauliflower. This shows possibilities of using fertigation as a tool to study effect of different fertilization practices.

Water and fertilizer placed closed to roots has been showed to be an influence in the early growth stages by overcoming problems of poor root development. Fertigation provides a high frequency, from 2-3 times per week to almost continuous supply of water and N, making it easy to match supply and demand. Amounts and frequency of feeding can influence the timing in leafy vs. fruit producing vegetables.

Fertigation is assumed to give uniform application of water and fertiliser: However some recent data has contradicted this assumption (data from California: AU = 63%, furrow irrigation: AU = 80%). Three-dimensional distribution patterns if drippers are more than 50 cm apart: this has important consequences for soil sampling strategy!

Mineralization: unknown effect of different soil moisture regimes; mineralization has to be calculated for the limited soil volume that contains roots only.

Matching the supply of N with the demand allows cropping on very poor soils (comparable to hydroponics). This would allow the possibility for studying N demand and effect of deficiencies. We have no data for vegetables of the effect of fertigation on root distribution.

Investigation on tomato rooting system in a sandy soil under 4 different irrigation systems: use of the trench method to count roots 

• root density decreased with drought
• most roots were concentrated in the top 40 cm (due to higher BD layer) 

We do not know the effects of ridges on N distribution in soils, on Nmin target values or on the distribution of roots, H2O, or fertiliser N. Other opportunities offered by fertigation are lower N buffer values because roots are more concentrated in irrigation furrows, it is easy to adjust nutrient ratios and it leaves a more constant NO₃- concentration in the soil.

When and why do we sample? In Norway available N at the start of the growing season is determined on a number of fields for a certain area and averaged. Fertilizer advice for that region is then based on this average value. Considerations to be taken into account: sampling depth, number of samples required, correction for bulk density, suitable methods for pretreatment and extraction.

[Tables: Nmin target values for vegetables at begin of growing season (data from Germany), at different growing stages (Switzerland) and as a function of expected N mineralization rate (Belgium).]

There are vast differences in bulk densities and N mineralisation which occur across Europe because of different soil types and temperature ranges, e.g. between Norway and Greece, or even on very short distances. This makes interpretation of measurements difficult. It also has important consequences for the number of samples required to obtain a certain accuracy.

Highlighted the different mineralisation rates of organic fertilisers, sawdust 20% p.a., slurry 50% p.a., organic manure’s 80% p.a. Examples of N mineralization from pig slurry. Complete mineralisation of slurry may take up to eight years.

Biomass production and N uptake by green manure planted at different dates (18/8, 1/9 and 19/9). Use of cover crops/green manures is becoming standard practice across Europe. Green manure was incorporated in November. Mineralization takes place over winter and in February all study objects had very similar Nmin contents. The N accumulation in catch crops as a function of sowing date is much lower than the theoretical value.

There followed a quick discussion on how soil samples are analysed. Huebert always analyses his soils fresh, Hugh freezes at once, and then mills frozen in a special mill.

The main purpose of measuring N mineralization is to improve our prediction of fertilizer nitrogen requirements. Sources of plant material in vegetable crop rotations include: harvest residues, green manures, mulches. Supplying organic material to the soil is important to maintain a good soil fertility and soil structure, and it allows to take advantage of the N fixing capacity of legumes. Which measurements have to be done to follow N mineralization, and at what time will depend on production system and geographical location. Factors influencing N mineralization: 

• soil type (NH₄⁺ fixation, CEC)
• immobilisation capacity
• C:N ratio
• climatic conditions
• soil temperature (No decomposition when soil temperature below 5^oC)
• depth of incorporation
• amounts incorporated (denitrification)
• the time that the plant residues remain on the soil surface
• age of incorporated material (lignin, cellulose, hemicellulose content)
• size of plant material
• handling of plant material before analysis (number of fractions considered, drying temperature) 

Many different methods of measuring N mineralization exist:  

• incubation studies without a crop
• pot experiments with a crop
• incorporating plant material in bags in the soil (pros and cons of litter bags)
• use of ¹⁵N
• use of models
• chemical methods: total N, C/N, EUF method, strong acids/bases, etc. (most of these chemical methods show very poor correlation with actual field N mineralization) 

A discussion about what temperature plant material should be dried at for chemical characterisation followed. Answers varied between air drying (problems of rotting) to drying at 80^oC. Loss of volatiles occurred if the residues were dried at too high a temperature.

J.N. Sørensen presented a review paper from Jarvis et al (1996), and a new easy pot incubation method for measuring nitrogen mineralization from easily decomposable organic material.

Two types of litter bags were also introduced; nylon (similar to what is used at HRI) for decomposition and enclosed polythene for mineralisation.

Many studies deal with N mineralization, but most are about agricultural crops, and very little about vegetable crops.

Chemical composition of the residues are known to affect the release of Nitrogen: 

• Polyphenols are important in legume crop residues, but not for most other vegetable crops.
• soil temperature (including freezing/thawing cycles)
• soil moisture
• soil density
• soil texture
• soil tillage 

For temperate Western European conditions soil temperature is usually the main influencing factor on mineralisation. Oxidation of ammonium to nitrate is temperature dependant. In southern Europe the limiting factor to N mineralization is often the moisture content. The knowledge of the interactions between all these factors is limited, and also very little is known about the effect of consecutive wetting and drying cycles, as in irrigated agriculture.

The results of a two year study of N leaching under field grown lettuce were presented (Water transport and leaching of nitrogen over the year). A tunnel underneath a vegetable field allowed investigation of water and nitrogen transport from the soil surface to the groundwater. Nitrogen losses due to leaching amounted to 100 kg N/ha/yr under the vegetable field compared to 300 kg N/ha/yr under a permanent fallow field. Mineralization of the organic matter contributed two thirds, mineral fertilizer one third of the available nitrogen. An improved fertilizer recommendation thus has to rely upon measurements of available nitrogen assessed by rapid tests (N-min).

In Switzerland, 50 kg/ha N leachate is acceptable, however the concentration of nitrate in suction cups can reach 100ppm.

Christian also showed the results from a complicated field experiment, which was assessing general water flow through the soil. This showed preferential water flow and nitrate losses.

The most important factors controlling nitrate leaching are: 

• amount of rainfall plus irrigation
• soil texture and structure
• amount of N applied
• chemical form of N applied
• time of N application
• organic N mineralization; amount, time and N content of the organic manures
• nitrate losses in runoff water 

In Portugal, there is only a short rainfall period, fertiliser must sometimes be applied in less than ideal conditions. There is no fertiliser recommendation system in Portugal, and no real incorporation of crop residues. Urea is used extensively for top dressing. Measures are taken to reduce N leaching, and fertilizer recommendations are being made, based on: 

• mineral and organic N in the soil
• supply of N by crop residues
• crop N requirement
• soil type
• climate (rainfall and temperature)
• choice of fertilizer
• time of application
• application of manures (N leaching from animal manures varies from year to year)
• amount of N in irrigation water 

Split fertilizer applications can be done at any time in the growing season because they are often hand applied, or applied continuously, through fertigation. It is common practice to have a base dressing and up to two top dressings. Cover crops as a means of reducing nitrate leaching are used mainly on small family farms in the North. There is also some use of "Leaching Inhibitors": low solubility fertilizers, gradually soluble fertilizers, coated fertilizers, controllers of microbial activity (bio inhibitors) and water retaining substances (polymers). 

Current research on nitrate leaching in USA: This research includes pot and lysimeter experiments to study nitrate leaching and incubation trials, pot experiments and field trials to study N mineralization. Leaching experiments are conducted in pots watered from the bottom, experiments from the USA have shown this to be more efficient.

Factors influencing soil mineral N pool. Many bacterial factors at work on ammonium. Biological processes reduce/fix nitrogen into benign compounds – however the risk of formation of environmentally unfriendly products is a problem.

Some of the measurements that have been done in Gent:  

• in situ: closed box method to accessing gaseous losses
• in lab: acethylene inhibition method 

Measurements of immobilization (using ¹⁵N) have been done in wheat. Very little data is available on gaseous losses from vegetables, an experiment with lettuce is planned.

There are lots of data from grassland: N application increases denitrification, but not significantly: 2.56 kg/ha loss after application of 90 kg/ha N. TDR has been used to measure soil moisture.

 by Matthias Fink

The institute is not equipped to monitor these processes on a fundamental basis, therefore an applied approach to N losses is used. Many fertilizer experiments all over the country were analyzed and calculation of the N balance (Nmin + fertiliser – uptake) always revealed N losses, which in most cases could not be due to leaching. Regression lines were fitted to the data and higher losses were seen to be associated with longer growing periods and higher N application rates. An example with cauliflower over a ten week period shows 50 kg/ha N mineralisation.

Other apparent losses: the well known temporary immobilization of freshly applied fertilizer N. A graph of soil mineral N against time shows soil mineral N declining sharply in the middle of the period even with high rates of N fertiliser. It was suggested that it was immobilisation followed by mineralisation. According to new research this strong immobilization could be due to algae. It was suggested by Dr. Hofman that it was a sampling error – perhaps caused by missing the very concentrated top 2-3 cms of soil. There followed a discussion about sampling and the errors involved.

Water and N interactions are best assessed through the use of models. There are conflicting results in literature concerning interactions. Some experiments (eg. on tomatoes in coastal dunes in California) suggest that the limiting factor principle applies. Other experiments (eg. on broccoli and cauliflower in California) indicate that there is interaction between water and N use. A quadratic response surface was fitted to the yield data. The highest N gifts resulted in a yield reduction (leaching of nitrate). There were also effects on cauliflower quality, namely the incidence of hollow stem.

Scaife irrigation trails – irrigation always increases yield, N does to a certain level.

Interactions depend on soil type, rainfall, amount of irrigation and N level. Single and integrated reactions depend on all these factors. Again there are no conclusive results on possible interactions between N and water. In some years interactive effects can be seen, in other years not. An overview was given of experiments on N and water interaction planned over the next 4 years.

Some general remarks concerning these interactions: is there a difference in type of N fertilizer between irrigated and non irrigated crops in some countries? For potatoes in the Netherlands, an extra yield of 10 tons/ha of tubers is needed to make irrigation pay. In the near future environmentally based restrictions on irrigation are likely to be introduced. Do N recommendations change if crops are irrigated? Greater yield can be obtained through irrigation, but rentability has to be considered.

Principles of the Swiss model were explained. The data that were used to develop the model (the tunnel experiment under field grown lettuce) are available on request. Some growers/farmers do not believe that the recommendations based on the Nmin system can be improved. Also, the Swiss government does not believe in models and the possibility of improving the N fertilizer advice, especially not for vegetable crops. N balances at farm level is currently preferred. Farmers are subsidized to grow a green manure rather than another cash crop in autumn, thereby extensifying agriculture.

A sensitivity analysis was done on Christian’s simple model, rainfall proved to be the most important parameter.

Models are a means of collecting knowledge. It is a way of putting existing knowledge together on paper, in formulas. If a model gives wrong predictions, it merely reveals that there are still important gaps in our knowledge and it gives us an opportunity to improve our knowledge. Therefore it is important that that the different model subroutines can easily be accessed and improved.

Care should be taken with the number of model parameters, since additional parameters can always make the model fit in a particular situation (hence importance of thorough model validation). Many earlier experiments can not be used in model validation because not all needed data are available.

Models are very useful tools to calculate the outcome of different scenarios or management strategies and compare them. An example is given for a 3 year crop rotation.The N_ABLE model is available on the internet at http://www.qpais.co.uk/nable/nitrogen.htm

A very important aspect is to get the growers interested in using the models. The process of integrating models under one shell can be very helpful in this respect, because inputs used by different prediction models (diease, pests or nutrition) have to be input only once.