Europ. J. Agronomy 52 (2014) 291–296
Contents lists available at ScienceDirect
European Journal of Agronomy journal homepage: www.elsevier.com/locate/eja
Effect of nitrogen fertilization on nitrate leaching in relation to grain yield response on loamy sand in Sweden S. Delin ∗ , M. Stenberg Swedish University of Agricultural Sciences, Department of Soil and Environment, P.O. Box 234, SE-532 23 Skara, Sweden
a r t i c l e
i n f o
Article history: Received 30 May 2012 Received in revised form 16 August 2013 Accepted 26 August 2013 Keywords: Nitrate leaching Nitrogen fertilization Site-specific fertilization
a b s t r a c t High rates of nitrogen (N) fertilizer may increase N leaching with drainage, especially when there is no further crop response. It is often discussed whether leaching is affected only at levels that no longer give an economic return, or whether reducing fertilization below the economic optimum could reduce leaching further. To study nitrate leaching with different fertilizer N rates (0–135 kg N ha−1 ) and grain yield responses, field experiments in spring oats were conducted in 2007, 2008 and 2009 on loamy sand in south-west Sweden. Nitrate leaching was determined from nitrate concentrations in soil water sampled with ceramic suction cups and measured discharge at a nearby measuring station. The results showed that nitrate leaching per kg grain produced had its minimum around the economic optimum, here defined as the fertilization level where each extra kg of fertilizer N resulted in a 10 kg increase in grain yield (85% DM). There were no statistically significant differences in leaching between treatments fertilized below this level. However, N leaching was significantly elevated in some of the treatments with higher fertilization rates and the increase in nitrate leaching from increased N fertilization could be described with an exponential function. According to this function, the increase was <0.04 kg kg−1 fertilizer N at and below the economic optimum. Above this fertilization level, the nitrate leaching response gradually increased as the yield response ceased and the increase amounted to 0.1 and 0.5 kg kg−1 when the economic optimum was exceeded by 35 and 100 kg N ha−1 , respectively. The economic optimum fertilization level depends on the price relationship between grain and fertilizer, which in Sweden can vary between 5:1 and 15:1. In other words, precision fertilization that provides no more or no less than a 10 kg increase in grain yield per kg extra N fertilizer can be optimal for both crop profitability and the environment. To predict this level already at fertilization is a great challenge, and it could be argued that rates should be kept down further to ensure that they are not exceeded due to overestimation of the optimum rate. However, the development of precision agriculture with new tools for prediction may reduce this risk. © 2013 Elsevier B.V. All rights reserved.
1. Introduction In Sweden, as in many other countries, farmers are encouraged by the authorities to fertilize no more than the economic optimum, in order to minimize nitrogen (N) leaching and subsequent pressure on the environment. This involves choosing the right source, right place, right timing and right application method, and can be referred to as best management practice (Goulding, 2000; Roy et al., 2006). The optimum N fertilization rates vary between sites and years, due to differences in yield potential and soil nitrogen supply. To meet the requirement on each farm or field, general recommendations for each crop and area should be adjusted depending on previous crop, sowing date, soil type and soil and plant N analyses. However, optimum N fertilization rate may vary
∗ Corresponding author. Tel.: +46 511 67235; fax: +46 511 67134. E-mail address: sofi
[email protected] (S. Delin). 1161-0301/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.eja.2013.08.007
considerably even within individual fields (Delin et al., 2005) and site-specific fertilization with respect to variations within fields could therefore reduce N leaching further (Basso et al., 2011). For this there are tools such as the Yara N-sensor (Reusch, 1997), which has been used on a limited number of farms in Sweden and other countries during the past 15 years. This type of equipment may lead to reduced average fertilizer rates, since farmers may otherwise tend to adjust their fertilization to the best yielding parts of the field. However, the main effect is likely to be that the N is better distributed within fields and that the average rate is similar. The effect on leaching would then depend on the difference in leaching between fertilization above and below the optimum. According to some empirical models on N leaching response to N fertilization rate (Simmelsgaard and Djurhuus, 1998), the effect on leaching is similar above and below the economic optimum. If such a model is used for calculating the benefit of site-specific fertilization compared with uniform fertilization the decreased leaching in areas where fertilization is reduced would be cancelled out by increased
292
S. Delin, M. Stenberg / Europ. J. Agronomy 52 (2014) 291–296
leaching where fertilization is increased. However, if leaching is much less affected below the optimum, as reported by Lord and Mitchell (1998), site-specific fertilization within individual fields has the potential to reduce leaching regardless of whether the average fertilization rate is reduced or not. The extent to which N leaching is affected above and below the economic optimum varies in literature. Bergström and Brink (1986) studied leaching at different N fertilization levels in a 10year study on a clay soil in Sweden and found a gradual increase at levels above 100 kg N ha−1 . In their study fertilization levels were the same in each plot between years, regardless of crop and expected fertilization demand. The gradual increase may therefore have been a consequence of accumulated effects from several years where one plot may have been fertilized above the optimum one year and below it the next. However, Simmelsgaard and Djurhuus (1998) compiled data from several different experiments and described the dependence of leaching on fertilization by an exponential function rising considerably already at small fertilization rates, meaning that the relationship was close to linear around the economic optimum, increasing by 0.25–0.35 kg N ha−1 per kg N applied at differences of up to 20 kg N above or below the recommended rate (i.e. as practised by farmers in Denmark around 1980). Lord and Mitchell (1998) present British results on N leaching at different N inputs in relation to the economic optimum calculated for a known yield response to fertilization. In contrast to the studies cited above, they found that leaching was only affected very slightly (<0.05 kg kg−1 N applied) at rates below the economic optimum, but on average by 0.52 kg kg−1 above economic optimum rates. Engström et al. (2010) studied nitrate N leaching at different N fertilization levels in oilseed rape and also found a steeper response in terms of leaching above the optimum (0.5 kg kg−1 applied) than below (0–0.2 kg kg−1 ), with less effect of fertilization on N leaching when the winter was cold. The effect of fertilization on leaching is described in a number of models for simulating N dynamics, in Sweden mainly the SoilN and SoilNDB models (Johnsson, 1990; Eckersten and Jansson, 1991; Larsson et al., 2002) or in the advisory model STANK in MIND (Aronsson and Torstensson, 2004). The way in which leaching is affected by fertilization differs between models. In some models leaching is significantly increased already at low fertilization rates (Eckersten and Jansson, 1991; Simmelsgaard and Djurhuus, 1998), while in others leaching only begins to be affected by fertilization first at or slightly below economic optimum fertilization (Brentrup et al., 2004; Larsson et al., 2002; Aronsson and Torstensson, 2004; Beaudoin et al., 2005). The objective of the present investigation was to study the effect of N fertilization on nitrate N leaching depending on grain yield response, i.e. above and below the economic optimum, in a cereal crop grown on loamy sand under Swedish weather conditions. The hypothesis was that N leaching response is dependent on crop N removal from the field with harvest, and that N leaching is significantly affected by N fertilization rate only at fertilization levels with a weak grain yield response, i.e. above a certain level that could coincide with the economic optimum, depending on the price ratio between grain and fertilizer N.
Table 1 Nitrogen fertilization levels used in oats in the trials. Percentage of expected economic optimum fertilization level
Actual fertilization rate, kg N ha−1
A B C D E F G
0 45 70 90 110 135 60 + (30, 0 and 40 in 2007, 2008 and 2009, respectively)
0% 50% 75% 100% 125% 150% 100% (adjusted after crop emergence)
had a larger fraction of coarse sand in the 30–60 cm layer (12% clay, 17% silt and 71% sand) and the 60–90 cm layer (13% clay, 20% silt and 67% sand). Bi-annual field trials were conducted with spring oats (Avena sativa L.) as the first crop in three consecutive years (2007, 2008 and 2009). Each trial had the first year seven N fertilization treatments (Table 1) distributed randomly within each of four blocks. The following crop was winter wheat (Triticum aestivum L.) (2008 and 2010) or spring barley (Hordeum vulgare L.) (2009), which received the same rate throughout the experiments, according to local recommendations for each crop (160 and 90 kg N ha−1 for winter wheat and spring barley, respectively). Nitrogen was applied as granulated ammonium nitrate on the soil surface at the time of sowing of spring cereals, except in 2008, when fertilization was delayed until after crop emergence. In the subsequent winter wheat N was applied in spring, at stem elongation (GS 30–32; Zadoks et al., 1974). Phosphorus and potassium were applied at uniform rates according to current recommendations for the crop and area. Grain yield was measured plot-wise by combine harvester and reported at 85% dry matter. Grain samples were analysed for N content with Near-Infrared Transmittance detector (NIT, Infratech 1240) and used for calculating N removal with the harvested grain (N offtake). 2.2. Weather The growing season in 2007 was favourable, with adequate amounts of precipitation and normal temperature for the area, and was followed by a mild winter (Fig. 1). The spring of 2008 was very dry, followed by heavy precipitation in August and then a rather mild winter. In 2009, the spring was also dry, but the growing season was followed by a cold winter.
2. Materials and methods 2.1. Experimental setup Nitrate N leaching in response to different fertilizer N doses was investigated on an Inceptisol (USDA Soil Taxonomy) in south-west Sweden (58◦ 22 N, 13◦ 29 E) with loamy sand soil (14% clay, 22% silt and 64% sand) with pH (H2 O) 6.4 and 2.8% soil organic matter (1.6% C and 0.14% N) in the 0–30 cm layer. Cation exchange capacity (CEC) was 130 mmolc kg−1 dry soil and base saturation 78%. The subsoil
Fig. 1. Weather data for the three years and long-term averages for the period 1961–1990.
S. Delin, M. Stenberg / Europ. J. Agronomy 52 (2014) 291–296
293
Fig. 2. (a) Grain nitrogen offtake and (b) grain yield as a function of N fertilizer rate, with quadratic polynomials fitted to the data and error bars indicating standard error.
2.3. Crop measurements
2.5. Residual soil mineral nitrogen
Above-ground crop biomass and plant N during the growing season were estimated based on remote sensing techniques combined with crop sampling. The oat crop was scanned with a hand-held radiometer (Yara N-sensor) measuring crop reflectance within different wavelength bands two or three times between stem elongation and ear emergence, i.e. at growth development stages GS 30–59 (Zadoks et al., 1974). From these measurements four reflectance indices (IR/R, IR/G, Si1 and Si2) were derived from infrared (IR), red (R) and green (G) wavelength bands, where Si1 and Si2 are indices used by Yara for estimating chlorophyll content and biomass, respectively (Söderström et al., 2004). On the same occasions, plant samples were taken by cutting the crop at the soil surface from four 0.25 m2 areas within each plot in one of the blocks. Crop biomass was determined and N content analysed with Dumas elemental analysis on a LECO CNS-2000 analyser (LECO Corporation, St. Joseph, MI, USA). The values of above-ground plant N obtained from plant cuttings were used for interpretation of N-sensor data. The index with the strongest correlation was then used to estimate above-ground plant N in all plots.
Soil samples were taken just after harvest (August) for determination of nitrate and ammonium N. For this, 12 cores from 0 to 30 cm soil level and 8 cores from 30 to 60 cm depth were mixed to composite samples for each area and depth. Subsamples of 30 g were extracted with 100 ml 2 M KCl and N analyses were carried out using colorimetric methods on a Technicon autoanalyser. From sampling until N determination, soil samples were stored frozen.
2.4. Nitrate leaching Soil water was sampled with ceramic suction cups (Djurhuus and Jacobsen, 1995) installed in triplicate at 80 cm depth in each plot. The suction cups were mounted on 25 cm long and 2.5 cm wide PVC tubes. Ditches were excavated from the positions of the suction cups in the field plots to sampling stations along the side of the field experiment. Each cup was installed in holes of the same diameter made vertically from the ditches. The holes were sealed with bentonite to prevent movement of water along the PVC tubes and the ditches were refilled with soil. Hoses from the cups were assembled in the sampling stations. Sampling was carried out by applying a suction of 60–70 kPa 24 h before the sampling was performed. This was done every second week during periods with drainage water runoff, from the time of fertilization (April) until June the following year, accounting for both direct and residual effects. Periods with drainage runoff were considered to be those when runoff could be measured in a leaching measurement facility at Lanna research station, 25 km west of the study site, where continuous measurements of drainage runoff are made with tipping bucket equipment. The sampled water was analysed for nitrate by flow injection analysis (Tecator AB, Höganäs, Sweden) according to the colorimetric cadmium reduction method . Nitrate N leaching was determined from nitrate concentrations in soil water and discharge measured at Lanna research station.
2.6. Data analysis Grain yield and N offtake was plotted against N fertilization rate. To these scatter-plots, second order polynomials were fitted in the software SigmaPlot 12 (Systat Software, Inc., Richmond, CA, USA). The curves were then used to estimate economic optimum N fertilization rates by identifying the points where the slope of the functions equalled the price ratio of grain to fertilizer, which was estimated here to be 10:1, based on average Swedish prices for fertilizer N and cereal grain in the past 10 years. In order to compare the influence of fertilization on leaching above and below the optimum for all years in the same plot, the difference in leaching and residual soil mineral N from that in the unfertilized treatment was plotted against the deviation in fertilization rate from the economic optimum. To these data an exponential function (y = y0 + a exp(bx)) was fitted. The nitrate leaching per unit grain yield (g NO3 -N kg−1 ) was also plotted against the deviation in fertilization rate from the economic optimum in order to identify the fertilization rate at which leaching per unit harvested grain was at a minimum. Statistical analysis of differences in grain yield, soil mineral N and nitrate leaching between treatments was performed with GLM ANOVA in the software SAS 9.2 (SAS Institute Inc., Cary, NC, USA). For yield and soil data, Fisher’s LSD was calculated for all variables except nitrate leaching, where data were missing from one plot. 3. Results 3.1. Yield, N offtake, nitrate leaching and residual mineral N Nitrogen offtake and grain yield responded positively to fertilization within the whole range of rates applied in 2007 (Fig. 2 and Table 2). The optimum fertilizer N rate was estimated to be 104 kg N ha−1 , since at that rate the yield response was 10 kg grain per kg N applied. The positive yield response resulted in few and small differences in nitrate leaching between treatments in 2007 (Table 3), and leaching was only slightly elevated in treatment F (highest N rate) and treatment G (split N dose). These two treatments also had a lower yield return per kg fertilizer applied (Fig. 2b)
294
S. Delin, M. Stenberg / Europ. J. Agronomy 52 (2014) 291–296
Table 2 Yield (kg ha−1 , 85% dry matter), N offtake with grain (kg N ha−1 ) and residual soil N after harvest (kg N ha−1 ) in the three study years. 2007 N dose A B C D E F G a
0 45 70 90 110 135 –a LSD0.05
2008
Yield
N offtake
3427 4683 5011 5515 5554 5808 5129 386
51 75 84 94 97 101 91 6.4
Soil N 22 21 18 20 22 29 20 3.6
2009
Yield
N offtake
Soil N
Yield
N offtake
Soil N
2647 3157 3138 3447 2990 2645 3554 396
54 68 71 77 66 60 78 8.4
23 22 27 28 31 41 26 9.1
2430 3208 3466 3635 3437 3697 3792 395
39 54 59 63 60 65 66 6.9
15 17 16 17 18 17 17 1.0
60 kg N at sowing + 30, 0 and 40 in 2007, 2008 and 2009, respectively.
and treatment F had significantly higher residual soil mineral N than the other treatments (Table 2). In 2008, when dry weather (Fig. 1) occurred after delayed fertilization, N offtake and yield did not respond well to fertilization (Fig. 2 and Table 2). The optimum fertilizer N rate in 2008 was only 12 kg N ha−1 and leaching increased more significantly with fertilization in that year than in the other two years when optimum fertilization rates were higher (Table 3). In 2009, yield responded positively to fertilization, but at a much lower level than in 2007 (Fig. 2 and Table 2). The optimum fertilization rate was only 61 kg N ha−1 . However, there was no clear response in terms of residual mineral N and nitrate leaching (Fig. 2), and few statistically significant differences between treatments (Tables 2 and 3). Yield in the succeeding crops did not differ significantly between treatments in any of the years (p = 0.12–0.53; data not shown). The nitrate leaching response above and below the economic fertilizer optimum from all years combined, mirrored the yield response in those years (Fig. 3), and could be described with an exponential function, y = y0 + 1.4 exp(0.027x) (r2 = 0.84; p = 0.001). The overall conclusion from this was that fertilization did not significantly affect nitrate leaching (i.e. less than 0.04 kg NO3 -N kg N−1 ) as long as each extra kg of fertilizer N resulted in at least a 10 kg increase in grain yield (85% DM). At this level, nitrate N leaching per kg grain produced was at its minimum (Fig. 4). However, above this level, leaching increased exponentially and amounted to around 0.1–0.4 kg N leached for every extra kg N applied (Fig. 3). Residual soil mineral N and nitrate N leaching were correlated (r2 = 0.75) and every additional kg of residual soil mineral N corresponded to 2 additional kg of leached nitrate N (Fig. 5). 3.2. N-sensor measurements and crop N uptake
were similar between years at later development stages (GS 45–59; Fig. 6). However, while there were no differences between treatments at GS 41 in 2008, there were significant differences already at GS 37 in 2007 (Fig. 6). The differences between treatments at GS 37 in 2009 were intermediate.
Fig. 3. Deviation in nitrate N leaching from that in the unfertilized treatment and deviation in grain yield from that at the economic optimum for fertilization rates above and below economic optimum fertilization. A quadratic polynomial fitted the yield data and an exponential function fitted the leaching data. Error bars indicate standard error.
The Yara index Si1 (Söderström et al., 2004) correlated most strongly to the above-ground N determined from the cut crop (r2 = 0.85–0.97). This relationship was therefore used for calculating aboveground N in all blocks (Fig. 6), except when there was no such relationship (GS 37; 2009) or when N-sensor measurements were missing (GS 41; 2008) and therefore only values from the crop cuttings in one block are presented. The N uptake levels recorded Table 3 Nitrate-N leaching losses (kg N ha−1 ) in the different treatments (A–G, see Table 2) in the three years. Different letters indicate significant differences between treatments and years (˛ = 0.05). 2007 A B C D E F G
2008 48ab 50ab 44a 45a 48ab 53b 54b
A B C D E F G
2009 37a 41ab 49ab 55b 50ab 80c 42ab
A B C D E F G
23a 25ab 24a 24ab 29b 27ab 27ab
Fig. 4. Nitrate leaching per kg harvested grain at different fertilization rates above and below the economic optimum.
S. Delin, M. Stenberg / Europ. J. Agronomy 52 (2014) 291–296
Fig. 5. Nitrate leaching as a function of soil mineral nitrogen at harvest.
4. Discussion 4.1. Degree of leaching response The impact of fertilization on leaching was much more significant above the economic optimum than below it, confirming findings by Lord and Mitchell (1998). A similar picture was presented by Brentrup et al. (2004) in a Life Cycle Assessment of agricultural practices, where the effect of fertilization on nitrate leaching was based on N balances, and leaching was assumed to be zero as long as the N removal by the crop exceeded the input with fertilizer. At fertilizer doses above this level the leaching effect increased as crop N removal ceased. In our experiments N offtake equalled N input at fertilizer rates of 90, 70 and 60 kg N ha−1 in 2007, 2008 and 2009, respectively (Table 2), which coincided fairly well with the optimum N rate in 2007 and 2009. However, the contribution of N from the soil also needs to be considered. The net effect of fertilization on crop N removal balanced 30–50% of the fertilizer N input at fertilization rates below the optimum, but <20% at fertilization rates above the optimum. That meant a net N surplus of 0.5–0.7 kg N kg−1 per extra kg fertilizer N applied below the optimum, which did not seem to lead to any elevation in nitrate leaching. Much of this N was probably taken up by the crop and later added to the soil organic N pool with the crop residues. This is likely to be mineralized later on, and risks being leached if not taken up by a crop. Obviously this did not happen in the short term. Some of the surplus may also have been lost by gaseous emissions through denitrification. The magnitude of these losses will depend on the aeration of the soil and the soil nitrate concentration, and may therefore be highly related to fertilization N rate if the soil becomes waterlogged shortly after fertilization. Above the optimum, the surplus was higher, 0.8–1.0 kg N kg−1 per extra kg fertilizer N applied, which is approximately 0.3 kg N kg−1 more than below the optimum. The effect on nitrate leaching was about
295
5–7 kg NO3 -N ha−1 on applying 50 kg N kg−1 above the optimum and 20 kg NO3 -N ha−1 on applying 100 kg N kg−1 above the optimum. This means 0.1–0.2 kg NO3 -N leaching kg−1 N applied in this range, i.e. only around 10–25% of the net N surplus or about 50% of the extra net surplus compared with below the optimum. The amount of N stored as soil organic N is likely to be dependent on the crop residue biomass and should cease, as biomass production does not respond to fertilization. The rest of the extra surplus is therefore more likely to be lost by denitrification. The leaching response of around 0.1 kg N kg−1 when the economic optimum was exceeded by 30–40 kg N ha−1 was similar to values reported by Bergström and Brink (1986) and Simmelsgaard and Djurhuus (1998), but slightly less than that found by Lord and Mitchell (1998). The differences in effects are likely to be the result of different soil types and amounts of runoff, as described in several models (Addiscott and Whitmore, 1991; Larsson et al., 2002; Aronsson and Torstensson, 2004). When setting up the advisory model STANK in MIND (Aronsson and Torstensson, 2004) with the soil type and climate area of the experimental site (sandy loam and Skara community), the leaching response was found to be rather similar, but with a significant response already when fertilizing at 30 kg N ha−1 below the economic optimum (0.1 kg N kg−1 N applied) and a linear instead of an exponential response above the optimum (0.3 kg N kg−1 N applied). The response below the optimum could be relevant in years with heavy rainfall just after fertilization, which was not the case during our study.
4.2. Fertilization strategy The results show that N leaching is affected by both fertilization rate and timing of fertilizer application. For instance, late application appeared to be unfavourable in 2007, when split application in treatment G caused lower yield (n.s., Table 2) and significantly higher nitrate leaching (Table 3) than when the whole dose was applied early. The very small response in yield in 2008 was because of late fertilizer application followed by a long period of warm, dry weather during which the crop was developing fast but the fertilizer N was unavailable to the crop due to lack of moisture in the soil surface on which it was placed. This resulted in later crop N uptake in that year (Fig. 6). If fertilization had been performed at sowing according to plan, N utilization by the crop would probably have been better. However, the key result is that leaching increases when fertilization no longer gives a yield response, i.e. leads to less than 10–15 kg grain per extra kg N added. However, this level varies between sites and years and predicting the appropriate fertilization rate before fertilization is a great challenge. Splitting the N dose allows the fertilizer rate to be adjusted to crop demand later in the season, when the optimum N fertilizer rate is easier to predict from the N status of the crop (Fig. 6). This is already routine practice in many areas and for many crops. However, there is a need to develop
Fig. 6. Aboveground crop nitrogen in treatments with different N fertilization rates at different crop development stages (GS; Zadoks et al., 1974).
296
S. Delin, M. Stenberg / Europ. J. Agronomy 52 (2014) 291–296
it further for all crops and climates, as they require different strategies. For instance, most of the fertilizer needs to be applied early in dry areas, in order to avoid much of the fertilizer remaining unavailable for the crop after application, whereas in wet areas early application of most of the N may increase N leaching losses (Goulding, 2000) or emissions due to denitrification (Bakken, 1988). 4.3. Precision fertilization to reduce leaching The difference in effect on leaching above and below the economic optimum indicates that site-specific N fertilization has the potential to reduce N leaching if it means that fertilization rate is kept at or below the optimum on the majority of the area. The variation in fertilizer requirement within a single field may amount to a standard deviation of around 30 kg N ha−1 (Delin and Lindén, 2002; Wetterlind et al., 2008). The exponential function in Fig. 3 suggest that leaching could be reduced by 2 kg N ha−1 if fertilization is adjusted to achieve the economic optimum compared with exceeding the optimum by 30 kg ha−1 , or by 6 kg N ha−1 compared with exceeding the optimum by 60 kg ha−1 . This means that even in fields where fertilization is adjusted to the actual mean requirement for that specific field, leaching can be further reduced if areas with lower requirements within the field can be identified. This can be done using tools that measure crop reflectance to estimate crop N status, such as the Yara N-sensor (Söderström et al., 2004). The field could also be divided into zones with different levels of expected N fertilizer requirements (Basso et al., 2011), based for instance on historical yield and protein maps (Delin, 2005) or soil data (Delin and Berglund, 2005). However, much can probably be gained simply by meeting the average demand of the field. Requirements vary between fields and years and there is always a risk of farmers overestimating the fertilizer requirements in order to obtain the best possible yields. 5. Conclusions As long as the yield response amounted to at least 10 kg grain per kg extra fertilizer N, nitrate leaching was not significantly affected by fertilization rate. At this level the nitrate leaching per kg grain produced was at its minimum, but at higher levels nitrate leaching increased exponentially with fertilization. Thus the development of N fertilization strategies and tools for precision N fertilization to maximize N use efficiency could be interesting from both an economic and environmental point of view. Acknowledgements This study was funded by the Swedish Foundation for Agricultural Research. The authors also wish to thank the staff at the Rural Economy and Agricultural Society of Skaraborg and at the Lanna Research Station, SLU, for technical assistance in the field experiments. References Addiscott., T.M., Whitmore, A.P., 1991. Simulation of solute leaching in soils of differing permeability. Soil Use Manage. 7, 94–102.
Aronsson, H., Torstensson, G., 2004. Beräkning av olika odlingsåtgärders inverkan på kväveutlakningen. SLU. Uppsala. Avd. för vattenvårdslära. Ekohydrologi 78 (in Swedish). Bakken, L.R., 1988. Denitrification under different cultivated plants: effects of soil moisture tension, nitrate concentration, and photosynthetic activity. Biol. Fertil. Soils 6, 271–278. Basso, B., Ritchie, J.T., Cammarano, D., Sartori, L., 2011. A strategic and tactical management approach to select optimal N fertilizer rates for wheat in a spatially variable field. Eur. J. Agron. 35, 215–222. Beaudoin, N., Saad, J.K., Van Laethem, C., Machet, J.M., Maucorps, J., Mary, B., 2005. Nitrate leaching in intensive agriculture in Northern France: effect of farming practices, soils and crop rotations. Agric. Ecosys. Environ. 111, 292–310. Bergström, L., Brink, N., 1986. Effects of differentiated applications of fertilizer N on leaching losses and distribution of organic N in the soil. Plant Soil 93, 333–345. Brentrup, F., Küsters, J., Lammel, J., Barraclough, P., Kuhlmann, H., 2004. Environmental impact assessment of agricultural production systems using the life cycle assessment (LCA) methodology II. The application to N fertilizer use in winter wheat production systems. Eur. J. Agron. 20, 265–279. Delin, S., 2005. Within-field variations in grain protein content – relationships to yield and soil nitrogen and consistency in maps between years. Precis. Agric. 5, 565–577. Delin, S., Berglund, K., 2005. Management zones classified with respect to drought and waterlogging. Precis. Agric. 6, 321–340. Delin, S., Lindén, B., 2002. Relations between net nitrogen mineralization and soil characteristics within an arable field. Acta Agr. Scand. Section B: Plant Soil Sci. 52, 78–85. Delin, S., Lindén, B., Berglund, K., 2005. Yield and protein response to fertilizer nitrogen in different parts of a cereal field: potential of site-specific fertilization. Eur. J. Agron. 22, 325–336. Djurhuus, J., Jacobsen, O.H., 1995. Comparison of ceramic suction cups and KCl extraction for the. determination of nitrate in soil. Eur. J. Soil Sci. 46, 387–395. Eckersten, H., Jansson, P.E., 1991. Modelling water flow, nitrogen uptake and production for wheat. Fert. Res. 27, 313–329. Engström, L., Stenberg, M., Aronsson, H., Lindén, B., 2010. Reducing nitrate leaching after winter oilseed rape and peas in mild and cold winters. Agron. Sustain. Dev. 31, 337–347. Goulding, K., 2000. Nitrate leaching arable and horticultural land. Soil Use Manage. 16, 145–151. Johnsson, H., 1990. Nitrogen and Water dynamics in Arable Soil. A Modelling Approach Emphasizing Nitrogen Losses. Dept. of Soil Sciences, Reports and Dissertations 6, 36. Larsson, M., Johnsson, H., Hoffmann, M., Mårtensson, K., 2002. Technical description of SOILNDB (V. 1.0) Teknisk rapport 64, Avdelningen för vattenvårdslära, SLU. 30 pp. (in Swedish). Lord, I.E., Mitchell, R.D.J., 1998. Effect of nitrogen inputs to cereals on nitrate leaching from sandy soils. Soil Use Manage. 14, 78–83. Reusch, S. 1997. Entwicklung eines reflexionsoptischen Sensors zur Erfassung der Stickstoffversorgung landwirtschaftlicher Kulturpflanzen (Development of a Reflectance Sensor for the Establishment of Nitrogen Support of Agricultural Crops). Dissertation. Forschungsbericht Agrartechnik des Arbeitskreises Forschung und Lehre der Max-Eyth-Gesellschaft Agrartechnik im VDI (VDIMEG) 303, Kiel, 157 pp (in German). Roy, R.N., Finck, A., Blair, G.J., Tandon, H.S., 2006. Plant Nutrition for Food Security: A Guide to Integrated Nutrient Management. FAO Fertilizer and Plant Nutrition Bull. No. 16. Food and Agriculture Organization of the United Nations, Rome, Italy. Simmelsgaard, S.E:, Djurhuus, J., 1998. An empirical model for estimation nitrate leaching as affected by crop type and the long-term N fertilizer rate. Soil Use Manage. 14, 37–43. Söderström, M., Nissen, K., Gustafsson, K., Börjesson, T., Jonsson, A., Wijkmark, L., 2004. Swedish farmers’ experiences of the Yara N-Sensor 1998–2003. In: Mulla, D.J. (Ed.), Proceedings of the 7th International Conference on Precision Agriculture and Other Precision Resources Management. Minneapolis, USA (published on CD-ROM). Wetterlind, J., Stenberg, B., Jonsson, A., 2008. Near infrared reflectance spectroscopy compared with soil clay and organic matter content for estimating within-filed variation in N uptake in cereals. Plant Soil 302, 317–327. Zadoks, J.C., Chang, T.T., Konzak, C.F., 1974. A decimal code for the growth stages of cereals. Weed Res. 14, 415–421.