Towards improved nitrogen management in silage maize production on sandy soils

Maize has become a highly appreciated crop in Dutch dairy farming during the last 25 years. The current cropping technique, however, is associated with a low recovery of soil mineral nitrogen (N) and serious losses of N to the environment. This gave rise to the research described in this thesis which begins by reviewing the literature on the role of N in crop production, on N efficiency and on N losses ( Chapter 1) . The research reported in subsequent chapters had three objectives:to identify and to increase our understanding of factors that determine the partitioning of N inputs over crop and losses, with emphasis on maize grown on sandy soils,to provide techniques to improve the N efficiency of maize crops,to integrate these techniques into a consistent, comprehensive and environmentally sound N management system .The results of the experiments to meet these objectives are presented in Chapters 2-6 and synthesised in Chapters 7-9.As placement of nutrients close to the root system can improve the efficiency of nitrogen (N) use, four experiments on placement were conducted in the Wageningen Rhizolab. In these experiments the temporal and spatial distribution of maize roots was studied and N fertiliser placement options were compared ( Chapter 2 ). The observations revealed clear vertical and lateral root length density gradients during the first 9 weeks after emergence. Root length density (Lrv), as determined in core samples 9 weeks after emergence, was found to be positively related to the number of roots counted concurrently on minirhizotron walls (n). Lrv/n ratios were 1.13, 1.76, 0.99 and 1.21 cm cm -1in the successive experiments. Root numbers counted in each experiment from emergence until 9 weeks after emergence, were converted into root length density values and related to thermal time to describe Lrv as a function of the temperature sum. The average vertical root extension rates were 0.7 and 1.1 cm d -1at temperatures of 13 and 16°C, respectively. The corresponding values for the lateral extension rate were 1.0 and 1.6 cm d -1.Calculations performed at the data indicated that the N content of a 9 week old maize crop could generally not be explained by mass flow alone. Transport distances between roots and mineral N in the soil may thus have restricted the availability of N as suggested by preferential uptake of mineral N from soil compartments with a high root length density. Positioning fertiliser N close to the plant increased the recovery by a factor of 1.3 to 1.4. Dry matter (DM) yields of maize were not significantly affected by the application method of N, however. It seems that the root extension rate and the availability of N from sources other than fertiliser, were sufficient to cover the shoot demand under the prevailing circumstances.The effect of the placement of cattle slurry on the DM yield of silage maize was studied in five experiments in which slurry was injected in spring at a rate of circa 120 kg slurry-N ha -1in slots 25 cm apart ('standard injection') or in slots 75 cm apart ('banded injection'). Subsequently, maize was planted in rows 75 cm apart, parallel to the slots, either at random lateral positions in the 'standard injection' treatment or 10 cm next to the injection slots of the 'banded injection' treatment ( Chapter 3 ). All treatments, including a control without slurry, were combined with 0 and 20-31 kg ha -1subsurface banded phosphorus (P) starter fertiliser. It was found that the DM yields of silage maize were reduced by 8% on average when conventionally injected slurry ('standard injected') was not supplemented with a P starter.However, the yield reduction was limited to 2% when slurry was banded ('banded injection'). Observations on the distribution of soil mineral N (SMN) and roots in two of the experiments indicated that during the first 5-7 weeks after planting, nutrients were predominantly obtained from the soil close to the plant row. This may explain why the positive response of maize to placement was strongest and significant on P-responsive sites, indicating that placement mainly improved the availability of slurry P. Improvement of the availability of slurry N may have played a secondary role. Placement improved the recovery of slurry-N by a factor of 1.2. The results of these experiments thus suggest that slurry placement can minimise the risk of yield loss associated with reduced fertiliser inputs and contribute to a better nutrient balance between fertiliser inputs and removal in crop products.The effects of the application time of slurry were studied in eight experiments including treatments in which circa 250 kg slurry-N ha -1yr -1was applied in autumn or in spring, with or without a nitrification inhibitor. The results, reported in Chapter 4 , showed that adding the nitrification inhibitor dicyandiamide (DCD) to autumn-applied cattle slurry retarded nitrification and, consequently, reduced nitrate losses during winter. However, spring-applied slurry without DCD was on average associated with even lower N losses and higher maize DM yields. Nevertheless, considerable N losses were observed even with spring application of slurry. This was reflected by a large variation of the apparent N mineralisation during summer, ranging from 0.36 to 0.94 kg ha -1d -1. On average, 40 percent of SMN present in spring was lost during the growing season. Hence, the amounts of residual soil mineral N (RSMN) were lower than expected. Multiple regression analysis with SMN in spring, N crop uptake and cumulative rainfall as explanatory variables, could account for 79% of the variation in RSMN. Postponing slurry applications to spring and adjusting N inputs to crop requirements were insufficient to keep the nitrate concentration in groundwater below the EC threshold for drinking water.To test the hypothesis that the recovery of SMN by crops and its subsequent utilisation for DM production may increase when the application of N is postponed until after crop emergence, nine field experiments were conducted ( Chapter 5 ). In five experiments the effect of slurry applied at a rate of circa 340 kg N ha -1before planting silage maize, was compared to the effect of a similar rate applied as a split dressings: half before planting and half at the 4-6 leaf stage. For the latter applications the slurry was either injected or banded. The remaining four experiments investigated the effect of splitting mineral fertiliser N. In these experiments, circa 150 kg slurry-N ha -1, applied before planting maize, was supplemented with mineral fertiliser N at rates ranging from 40 to 160 kg ha -1, either applied before crop emergence, or split. When split, 40 kg ha -1of the mineral fertiliser N rate was banded at the 4-6 leaf stage.Split applications of cattle slurry had a significant positive effect on the DM yield in two of the five experiments compared with the conventional non-split application, but only when the post-emergence slurry application was banded. Banding, however, is not in accordance with present legislation. Split applications of mineral fertiliser N had a significant positive effect in one experiment where rainfall was excessive but not in the others. The results provide insufficient evidence to justify recommending farmers to split applications. SMN sampling at the 4-6 leaf stage should hence be considered a check of the appropriateness of early N applications followed by exceptional weather conditions rather than a routine observation on which the post-emergence N dressing is to be based in a deliberate splitting strategy. The data suggest that the financial return from a 40 kg ha -1supplementation with mineral fertiliser N, is questionable if there is more than 175 kg N ha -1in the upper 0.6 m soil layer at the 4-6 leaf stage.The presence of RSMN after the harvest of maize seems inevitable to a certain extent. Rye and grass cover crops can potentially intercept RSMN, reduce overwinter leaching, transfer SMN to next growing seasons and reduce the fertiliser need of subsequent crops. These aspects were studied for 6 years in continuous silage maize production systems with N input levels ranging from 20 to 304 kg total N ha -1, applied as mineral fertiliser and/or slurry ( Chapter 6 ). It was found that rye and grass cover crops were able to absorb on average 40 kg N ha -1in the aboveground plant parts. The actual N uptake was largely determined by winter temperatures and hardly by residual SMN. At low N input levels, cover crops reduced N leaching in accordance with their N uptake. At high N input levels, however, the reduction of leaching losses exceeded the sequestering capacity of the cover crop, suggesting that cover cropping stimulated the immobilisation of N or the loss of N via denitrification.Cover crops had no positive effect on maize yields at larger N rates and under these conditions cover crops did not improve the conversion of SMN into crop N. This was only partly reflected by an increase in residual SMN on plots where cover crops had been incorporated, as much of the excess N on maize had already been lost during the growing season. In N deficient maize production systems, however, cover crops increased the DM yield of maize. Their effect was equivalent to the effect of fertiliser N rates amounting to 105% and 44% of the aboveground N in rye and grass, respectively. In the first few years, cover crops decomposed incompletely during the growing season following their incorporation. Over the years, however, their effects on subsequent maize crops increased. This supports the hypothesis that the effects of cover crops can accumulate when grown year after year. Averaged over the 6 years, 115% and 73% of the aboveground rye N and grass N, respectively, were recovered in the crop-soil system.The residual effect of manuring has to be quantified if the financial returns of farming systems are to be maximised and contamination of the environment is to be avoided. It is especially important to quantify the residual effect on maize land in The Netherlands, since manure has been dumped on this land for over 25 years. A simulation model was therefore calibrated with data from a long-term field experiment and used to estimate the effects of Dutch manuring practice on maize land. The time course of the N mineralisation rate was estimated for three scenarios: i) following actual manure applications, which have declined with time (A scenario); ii) assuming continuous applications in accordance with the present and anticipated legislation (P scenario); iii) assuming applications of 200 kg mineral fertiliser N ha -1yr -1only (M scenario). The results of this scenario study are presented in Chapter 7 . They show that the estimated mineralisation rate (following the A scenario) for 1995 is 23-31 kg N ha -1yr -1higher than when manure is applied at moderate rates (following the P scenario). The corresponding estimates for the year 2005 still amount to 18 - 19 kg N ha -1yr -1.These calculations suggest that it may be extremely difficult to maintain soil organic N pools with mineral fertiliser only. Consequently, the mineralisation rate following the M scenario decreases with time, as do the yields of silage maize. The magnitude of the residual effect found in the present study, indicates that recommendations need to be fine-tuned and that there is scope to do so. The distinct long-term impact of previous manure applications also underlines the need to interpret results from short-term experiments with caution.The current use of N in silage maize production in The Netherlands leads to considerable N losses to the environment. However, maize growers fear that reducing N inputs so as to minimise N losses, might depress yields. The study described in Chapter 8 was conducted to investigate this. It aimed to quantify: 1) the response of silage maize DM yields to N, 2) the economically optimal N reserve, and 3) the trade-off between silage maize DM yield and N losses. The indicators of N losses used in this study were the difference between N input and N uptake and the post-harvest residual soil mineral N. In the study, regression models were used to fit DM yields and N uptakes of silage maize measured in 25 experiments on sandy soils in The Netherlands to the sum (SUMN) of the soil mineral N reserve (SMN early ) in March-April, plus mineral N in fertiliser, plus ammonium N in spring-applied slurry. The values obtained for the economically optimal SUMN in the upper 30 and 60 cm of soil were respectively 173 and 195 kg N ha -1(both with standard errors of 15 kg N ha -1). The economically optimal SUMN was not significantly related to the attainable DM yield.The DM yield and N uptake were also fitted to the measured soil mineral N reserve (SMN late ) in May-June which was determined in 10 of the experiments. The values obtained for the economically optimal SMN late amounted to 147 and 229 kg N ha -1(standard errors of 14 and 15 kg N ha -1, respectively) for sampling depths of 30 and 60 cm, respectively.The apparent N recovery (ANR) of maize averaged 53% at the economically optimal SUMN. The ANR rose considerably, however, when N was applied at lower rates, indicating that N losses may be much smaller in less intensive maize cropping. When maize was fertilised at 100 kg N ha ha -1below the economic optimum, the ANR was 73%, the difference between the mineral N input and the N crop uptake decreased by 57 kg N ha ha -1and the soil mineral N residue at the end of the growing season (0-60 cm) decreased by 24 kg N ha ha -1. The associated reduction in DM yield averaged 16%.The general conclusion from this study is that adjusting the N input to a level below the economically optimal rate can reduce the risks for N losses to the environment associated with conventional maize production, with a limited effect on silage yields.In Chapter 9 the results from previous chapters are drawn together and generalised. It was shown that N losses in the production of maize appear to originate from a combination of soil, climate and crop characteristics. In The Netherlands, maize is mainly grown on sandy soils. These soils have a limited capacity to retain water and therefore N may easily be lost to greater depths when it is available or applied too early. Nitrogen deeper in the soil can be intercepted in time less easily by the maize root system. It is possible to minimise the risks for an insufficient synlocalisation of N and roots by postponing N applications. However, the N from postponed applications may also be recovered poorly, as the application may coincide with quantities of N becoming available from mineralisation, making the application redundant. Moreover, if applied after crop emergence, the late-applied N may be positioned inappropriately in relation to the positioning of functioning roots. All in all, the rationale of applying mineral N (including that from manure) is to supply N when and where the maize needs N for its biomass production, at a time that the maize root system is still unable to obtain that N sufficiently from soil reserves. These aspects limit the time span during which the application of mineral N is effective. N losses resulting from the current technique for cropping maize on Dutch sandy soils can thus be attributed to a combination of factors :too high application rates of N due to regional slurry surpluses and a tendency for maize growers to compensate a priori for unforeseen N losses,inaccessibility of the soil N reserves for roots in a cold and wet spring,too early or too late applications of N,application of N in non-rooted or poorly rooted soil compartments,N mineralisation outside the period that maize takes up N.From the research described in this thesis it is concluded that maize growers can reduce N losses by using the following cropping techniques . These techniques will only become effective, however, if they are used concomitantly in a consistent and comprehensive system :drastically reducing the N application rate (including manure N),fine-tuning the application rate of N by taking account of site characteristics such as the intrinsic soil mineral N reserves, the manuring history and the N contribution from previous crops including cover crops,confining the application of N to those parts of the soil profile where a high root length density is anticipated,postponing the application of N until shortly before sowing,restricting post-emergence applications of N to situations where N deficiency is indicated by the soil mineral N reserves,growing cover crops after maize.These cropping techniques are currently insufficiently exploited in conventional maize cropping sytems. Consequently, conventional cropping systems of maize will not yield a quality of the shallow groundwater under maize land that meets the EC Nitrate Directive for drinking water.Although it appears to be technically feasible to introduce 'nitrate-safe' cropping systems for maize, consisting of a combination of measures including reduced manure and fertiliser inputs, placement of manure and of fertilisers and growing cover crops during winter, this will inevitably cost money. The way the current conventional cropping technique is laid out and implemented makes it difficult but not impossible to introduce these alternative systems. Without the suggested combination of additional measures, the anticipated Dutch legislation on the use of manure and fertilisers will probably insufficiently improve the quality of the groundwater under maize land in many regions with sandy soils.