- About this Journal ·
- Abstracting and Indexing ·
- Aims and Scope ·
- Annual Issues ·
- Article Processing Charges ·
- Articles in Press ·
- Author Guidelines ·
- Bibliographic Information ·
- Citations to this Journal ·
- Contact Information ·
- Editorial Board ·
- Editorial Workflow ·
- Free eTOC Alerts ·
- Publication Ethics ·
- Reviewers Acknowledgment ·
- Submit a Manuscript ·
- Subscription Information ·
- Table of Contents
Applied and Environmental Soil Science
Volume 2011 (2011), Article ID 161079, 16 pages
Predicting Mineral N Release during Decomposition of Organic Wastes in Soil by Use of the SOILN_NO Model
1Department of Plant and Environmental Sciences, The Norwegian University of Life Sciences, P.O. Box 5003, 1432 Aas, Norway
2Norwegian Water Resources and Energy Directorate, P.O. Box 5091 Majorstua, 0301 Oslo, Norway
Received 15 December 2010; Accepted 2 March 2011
Academic Editor: Silvana I. Torri
Copyright © 2011 Trine A. Sogn and Lars Egil Haugen. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
In order to predict the mineral N release associated with the use of organic waste as fertilizer in agricultural plant production, the adequacy of the SOILN_NO model has been evaluated. The original thought was that the model calibrated to data from simple incubation experiments could predict the mineral N release from organic waste products used as N fertilizer on agricultural land. First, the model was calibrated to mineral N data achieved in a laboratory experiment where different organic wastes were added to soil and incubated at 15°C for 8 weeks. Secondly, the calibrated model was tested by use of NO3 − leaching data from soil columns with barley growing in 4 different soil types, added organic waste and exposed to natural climatic conditions during three growing seasons. The SOILN_NO model reproduced relatively well the NO3 − leaching from some of the soils included in the outdoor experiment, but failed to reproduce others. Use of the calibrated model often induced underestimation of the observed NO3 − leaching. To achieve a satisfactory simulation of the NO3 − leaching, recalibration of the model had to be carried out. Thus, SOILN_NO calibrated to data from simple incubation experiments in the laboratory could not directly be used as a tool to predict the N-leaching following organic waste application in more natural agronomic plant production systems. The results emphasised the need for site- and system-specific data for model calibration before using a model for predictive purposes related to fertilizer N value of organic wastes applied to agricultural land.
In order to achieve sustainable food production there is a high focus on recycling of nutrients by utilisation of organic waste as a fertilizer in agricultural plant production. Effective and safe use of organic wastes in agricultural plant production requires risks and benefits to be weighed and documented. Heavy metal and organic contaminant accumulation, as well as transmission of pathogenic bacteria, are among the identified risks. These risks are continuously managed by advances in the waste processing technology. On the other hand, obvious positive effects of organic waste application on soil structure, porosity, water retention capability, cation exchange capacity, and biological activity have been identified, for example [1, 2]. The nutrient value of organic waste has shown to be dependent of type of waste, both origin and processing method . A general condition for using organic wastes as fertilizers is the ability to predict their mineralization dynamics during decomposition in soil and thereby the release of plant available nutrients. To know and be able to predict the production of mineral N compounds in soil during decomposition of organic waste are important in order to advice farmers. The farmers need to know the mineral N generation from the organic waste in order to adapt the application so maximum yield can be achieved and the potential risk for nitrate (NO3 −) leaching during the growth dormant periods minimized. The utilisation of organic waste products can benefit highly from assessment of individual waste characteristics . Several factors, for example, the biosolids treatment process, its C : N and lignin : N ratio, as well as prevailing climatic factors and soil characteristics at the site of application, may influence the fate of the waste organic N in soil [5–7]. In order to predict the N dynamics connected to organic waste decomposition in soil, mechanistic models where process dependencies of those factors mentioned are included may be useful tools. Several dynamic models describing the process of organic matter decomposition and mineralization of N in soil have been developed. Among the process-orientated N models are, for example, the one based on the simple and basic equation proposed by, Standford and Smith , to more complex models as those developed by for example, Molina et al.  and Johnsson et al. . They have in common that the organic matter decay is described by first-order kinetics. In most of the models the organic matter is fractionated into different stability pools, which each are assumed homogenous. The first-order rate coefficient determining the stability of the pool and thus the rate of at which pool decay. To be of value for prediction a model must be thoroughly evaluated towards replicated measurements from appropriate experiments.
It has been shown that simple laboratory incubation experiments give relatively precise predictions of mineral N released from recently incorporated organic matter during decomposition in soil [11–15]. Thus, it can be assumed that calibration of model parameters to data from routine incubation tests, and thereafter simulation of more realistic agronomic systems by the calibrated model could provide a simple predictive tool for advising farmers of the N Fertilizer value and potential nitrate (NO3 −) leaching loss of organic wastes on a site and waste type-specific basis. However, modelling studies at the field scale level in comparison with laboratory scale modelling is still scarce . Field scale modelling poses the challenge of including the influence of climate, soil texture, and plant growth on soil C- and N dynamics.
The objective of this study was to assess the transferability of the SOILN_NO model  tuned to data from an incubation experiment to a more realistic field scenario. The following two hypotheses were tested.(1)SOILN_NO reproduces the mineral N generation following decomposition of organic wastes in soil, studied in a laboratory incubation experiment by adjusting initial distribution of organic N in the different stability pools and parameters included in the management routine.(2)SOILN_NO calibrated to data from the incubation experiment, simulates the measured NO3 − leaching following organic waste application in a soil column experiment with barley growing in 4 different soils.
The hypothesis are connected to the target that a mechanistic model calibrated to data from simple incubation experiments in the laboratory, may provide a tool to predict mineral N release from decomposition of organic waste products in field.
2. Material and Methods
The data used for model calibration was generated in a laboratory incubation experiment, while the data used for model testing was produced in an outdoor soil column (lysimeter) experiment.
2.1. Incubation Experiment
The incubation experiment was conducted in the laboratory to measure transformation of organic N to ammonium (NH4 +) and NO3 − (here; NH4 + + NO3 − = mineral N) during decomposition of different organic wastes in soil. 10 g soil was prepared for the incubation experiment. The soil was sampled in an experimental field at Ås, south east Norway (10°45′ E, 59°40′ N), elevated 70 m above sea level. The soil parent material was a postglacial clay and the soil was a loam, classified as a typic haplaquept with 21% clay, 40% silt and 39% sand .
One set of the soil beakers was kept as unaffected controls (control). To the rest, different kinds of organic wastes were added. The organic wastes were (1)raw, primary (before stabilising or sanitary treatments) Sewage Sludge (SS1),(2)stabilised, sanitary treated, and digested (anaerobic) Sewage Sludge (SS2),(3)primary source sorted (not composted) Organic Waste (OW3),(4)easily composted (aerobic) Organic Waste (OW4),(5)rotator treated (composted, and mixed with bark and chips) Sewage Sludge compost (SS5).
Stabilised wastes have often been found to be highly resistant to microbial degradation in soils [12, 19, 20]. Decomposition of such stable material in soil is usually associated with a smooth and gentle path of mineral N production. It is relatively easy to tune a model to reproduce such a stable data set by calibration of parameters. However, a test of dynamic models to such data is generally of low value. In order to test the model dynamic properly, data including significant variation should be sought reproduced by the model. Securing an appropriate data set for model validation, organic matters high in energy were included in our experiment. These organic matters were assumed to intensively stimulate the microbial fauna and were thus expected to give significant effects on the N dynamic during decomposition in soil. Based on this expectation the following organic matters were included:(6) Meat Flour (MF),(7) Cow Fodder pellets (CF).
The soil probes were added SS2 in an amount corresponding to 0.30 mg total N/g dry soil, whereas the other organic materials were added in an amounts corresponding to 0.15 mg total N/g dry soil (Table 1). The beakers with soil/organic matter were incubated at 15°C for 8 weeks in a dark chamber. The soil moisture during the incubation was adjusted to 25.4 vol%, corresponding to 55% of the soil’s water holding capacity. The evapotranspiration loss was replenished by addition of deionised water along the inner surface of a loosely closed PVC bag surrounding the soil beakers.
Every week, beakers were removed and stored frozen until analysis of NH4 + and NO3 − concentration in the soil/organic matter mixture could be performed. The soil/organic matter probes were extracted according to a modified method of Bremner  by 2 mol L−1 KCl. The extracts were analysed for NH4 + and NO3 − by FIA (flow injection analysis). The number of replicates is two.
2.2. Lysimeter Experiment
In an outdoor lysimeter experiment at Ås 32 cylinders, 110 cm deep and 80 cm in diameter, were situated above an underground cellar. The bottom of the cylinders had a cross formed recession to the outlet which was filled with cleaned gravel in order to secure good drainage. In order to collect soil leachate from the bottom of the lysimeters, an outlet with a plastic pipe lead to a 35 L collecting can. The soils had been filled in the cylinders layer by layer and 4 different soils were included in the experiment. The different soils included (A–D) differed both in texture and organic matter content (Table 2). One of the soils, soil C, was from the same area as the soil sampled for the laboratory incubation experiment and was thus almost identical to that soil. The lysimeters received natural precipitation during a 2.5-year period. In spring inorganic fertilzer (NPK fertilizer) and organic wastes (Table 3) were added to the soil in the lysimeters. The treatment plan was factorial, included four different soil types and four Fertilizer/organic waste treatments, all combinations in two replicates. The control treatment within each type of soil received a yearly base fertilization of 12 g m−2 NPK Fertilizer. In the organic matter treatments, the OW4 or CF from the incubation experiment was applied in addition to a basic NPK fertilization (Table 3). A treatment combining application of OW4 and CF was also included (Table 3). Barley (Pernilla) was grown in all the three growing seasons (Table 3). Leachate was collected weekly or when measurable amount of leachate was available. When leachate was collected also the soil temperature in the upper 5 cm was measured. NO3 − concentration in the leachate was measured by FIA. Plant material was sampled at harvests and total N concentration in grain and straw was measured by a CHN-analyser (LECO EC-12).
2.3. Model Description
SOILN_NO is a mechanistic model simulating the transformations of C and N in soil. The model is based on the SOILN model by Johnsson et al.  but extended towards a more detailed description of mineralisation, immobilization, and denitrification. SOILN_NO is described in detail by Vold  and only a brief description is included here. In SOILN_NO soil organic matter is divided in three different stability pools: a readily decomposable fraction (litter 1), a more slowly decomposable one (litter 2), and a pool representing stable humus fraction. The microbial organic matter decomposition and the corresponding C and N transformations are described by first-order reaction rate kinetics, where each organic matter pool has a specific decay rate constant. The C and N flows are assumed proportional and the C : N ratios of the different organic matter pools which have to be given initially. Both mineralisation and assimilation of C and N are governed by a microbial growth yield factor, a humification coefficient and the C : N ratio of newly formed microbial biomass products.
The management routine, a submodel representing agronomic practices like addition of manure and litter material (green manure), was used in this model application and some details connected to this submodel need explanation. Manure total N content is partitioned into fractions called manure N “NH4 +”, “bedding,” and “faeces”. The manure N NH4 + fraction is added to the soil initial NH4 + content. The manure N bedding fraction is divided between the litter 1 and 2 pools according to a partition factor (LF_BEDDING), and the C flow from bedding decomposition is proportional to the N flow, specified by the bedding C : N ratio (CN_BEDDING). The manure N faeces fraction is treated as a separate pool of organic N. As the other organic matter pools decomposed by micro-organisms, also the faeces fraction has a separate set of parameters like a first-order decay rate coefficient, a microbial efficiency factor, and a humification coefficient which influence the rate of decomposition and translocation. Similar to the bedding fraction, the C flow from faeces decomposition is proportional to the N flow, specified by the faeces C : N ratio (CN_FAECES). Like the manure N bedding fraction, N in green manure is dispersed between litter 1 and 2 according a partition factor (LF_GREEN), and the C flow from decomposition of green manure is again proportional to the N flow, specified by the C : N ratio for the green manure (CN_GREEN).
The application of SOILN_NO to quantify NO3 − leaching from the outdoor lysimeter system with plants growing, requires water and heat transport data as input driving data and a plant uptake submodel to be activated. In the plant growth submodel, plant N uptake is described by a logistic growth function based on a priori information of plant N content at harvest . Daily soil water and temperature data at different soil depths must be given to the SOILN_NO model as driving variables. These data was generated by use of the COUP model [23, 24]. COUP is a mechanistic model simulating vertical water and heat transport in soil. The water and heat transport is mainly based on two coupled partial differential equations derived from Darcy’s and Fourier’s laws describing one-dimensional water and heat transport within a soil profile. As input data COUP requires standard daily meteorological data as air temperature, relative moisture, wind speed, precipitation, and global radiation. Soil characteristics are defined by a water retention curve and hydraulic conductivity as a function of water content and potential. To estimate water uptake by plants, COUP must be given a priori information of plant cover, leaf area, plant height, and root depth.
2.4. Model Applications
Previously SOILN_NO has been calibrated to reproduce observed mineral N generation and CO2-emission data following decomposition of different organic matters in soil during laboratory incubation . In that study a best-fit calibration of initial C and N values in the different organic matter pools defined in SOILN_NO was done. The model adequately reproduced the quite different C and N-dynamics following decompositions of a wide range of organic wastes (sewage sludge, household compost, etc.). However, to continue on this organic matter pool initial C and N value calibration seemed little adequate when subsequent growing seasons were to be simulated. For simulation of long-term N-dynamics during several growth seasons, the management routine included in SOILN_NO was presupposed to be more convenient than resetting the initial organic matter C and N pools for each new application of organic waste. Here, SOILN_NO was first calibrated to reproduce the mineral N data obtained for the control treatment during 8 weeks of incubation in the laboratory by adjustments of the initial C and N pools in the different soil organic matter fractions (litter 1, litter 2, and humus) (Figure 2). This calibrated initial data file, representing the soil organic matter C and N pools, was then used as initial data for all the different treatments. The additional N (and C) supplied to the soil via the different organic matters (Table 1) were taken into account by use of the management routine included in SOILN_NO. The different amount of N and the different stability of the organic matters were represented by the organic Fertilizer pools green manure and manure bedding, faeces, and NH4 + fractions. The characteristics of the organic additives were mimicked by calibration of the C : N ratios for the different pools and/or coefficients expressing the fractionation between easily decomposed (litter 1) and more stable (litter 2) fractions (e.g., LF_BEDDING), thus indirectly the rate of decay.
When SOILN_NO was applied to the soil lysimeter system the model needed driving variables for vertical water and heat transport. These data were generated by COUP. Given daily climatic data from Ås and a priori plant growth data as input, COUP was calibrated in order to fit observed water output from the bottom of the soil columns and temperature measurements in the upper 5 cm of the soils. When satisfactory match between observed and simulated figures was achieved, daily values for soil water, temperature and associated variables were stored as driving data files given as input to SOILN_NO. The SOILN_NO plant N uptake submodel was given the requested plant N content measured at harvest.
For the lysimeter experiment SOILN_NO was precalibrated to simulate the NO3 −-leaching in the control treatment for each of the 4 different soil types (Figure 1). This precalibration was done by adjustment of initial values for C and N in the litter 1, litter 2, and humus pools. The calibrated initial state data, representing each of the 4 soils were then used as initial values for the different organic matter application treatments for each of the soils. To test hypothesis 2 model simulations based on management routine pools and parameters calibrated to the laboratory, incubation data were tested towards flux of NO3 − out of the field lysimeters applied the same type of organic matter as used in the incubation experiment.
2.5. Statistical Analysis
Treatment effects and influence of soil type (only in lysimeter study) on mineral N generation, and in the lysimeters NO3 − flux as leachate and plant N uptake were tested by analysis of variance and multiple tests using the GLM and Student-Neuman-Keul’s (SNK) test in the SAS computer software system. The differences were considered significant at .
Model success in reproducing measured data was evaluated by use of the LOFIT (Lack Of FIT, (1)) statistic  where is the observed average at each time , is the predicted value at time , is the number of replicates within each experiment, and is the time step from 1 to . The relative sizes of LOFIT and the mean squared errors in measured data were compared using the -test suggested by Smith et al.  (2). denotes the individual observations, and the number of replicates. A value of calculated greater than the critical 10% value indicates that the total error between observed mean and simulated value was significantly greater than the error inherent in the observed data.
3.1. N Mineralization in the Laboratory Incubation Experiment
The added organic matters showed quite different C and N dynamics during decomposition in soil in the laboratory incubation experiment (Figure 2). As expected due to a high C : N ratio (29.83, Table 1), SS5 was a very stable organic waste with a slow generation of mineral N during decay (Figure 2). Although a C : N in the medium range (11.74 and 14.03, Table 1), SS2 and OW4 seemed also to be of the most stable organic wastes with a smooth mineral N release during decomposition (Figure 2). The decomposition of MF, which had the lowest C : N ratio (4.57, Table 1) was characterised by a rapid generation of mineral N during the first 4 weeks of incubation. During decay of SS1, OW3, and CF, although C : N ratios in medium range (Table 1), initial immobilization of N occurred prior to an increase in mineral N generation. For all the materials characterized by initial net assimilation of N, the period of N immobilisation was relatively short, presumably due to an effective burn out of the energy source. For the further discussion, it is worth noticing that OW4 and CF had almost the same C : N ratio (Table 1), however a quite different N dynamic during decay (Figure 2).
3.2. Calibration to Data from the Laboratory Incubation Experiment
The fractionation of the soil total organic matter into pools with different stability was set initially in SOILN_NO. The initial distribution of soil organic C and N content between the organic matter pools litter 1, litter 2, and humus was set according to best fit to data from the control treatment with soil only (Figure 2). Further simulation of the organic matter application effects on N dynamics were then based on these initial soil organic matter distribution data combined with adjustments of N pools and parameters in the SOILN_NO management routine (Table 8). The mineral N generation curves characterised by constantly increasing mineral N generation at different rates were generally well simulated by SOILN_NO (Figure 2, (OW4, SS5, and MF)), Table 4) by adjustment of the amount of N in different management routine compartments and the interdependent coefficients (Table 8). When the decay of organic matter was characterized by N immobilization initially, prior to the period with net N assimilation, the model tended to underestimate the amount of N immobilized (Figure 2 (SS1, CF)).
3.3. N Dynamics in the Outdoor Lysimeter Experiment
In the lysimeter experiment the study of mineral N generation following decomposition of the applied organic matters was approached by measurement of plant N uptake and NO3 − leaching. Within the control treatment the highest plant N uptake was seen in the moraine (soil A), and the lowest in the most sand rich soil (soil D) (Tables 5 and 6). In between were the plant N uptake in the clay rich soils (soil B and soil C), where the plant N uptake was not significantly different from each other. In all the soils, except for the sandy soil (soil D), a net loss of soil N was found in the control treatment (Table 5). The doses of N used in the control treatment were apparently not sufficient to cover the plants requirement as the plant uptake of N was higher than the amount added via NPK fertilizer. Application of OW4 resulted in a plant N uptake almost similar to the control treatment (Tables 5 and 6). As for the control treatment the highest plant N uptake was seen in the moraine (soil A), but when the organic matters were added a similar plant N uptake was also seen for the more clay rich soils (soil B and C, Table 6). In the first experimental year, when a relatively low dose of organic matters were applied (Table 3), addition of organic matters resulted generally in a lower plant N uptake that is in the control (Table 6). However, application of CF resulted in a significantly higher plant N uptake than in the control and OW4 treatments during the last two years of the experiment (Table 6). The combined treatment (OW4 + CF) resulted in a significantly higher plant uptake than when CF was applied solely. The energy rich CF apparently boosted the more stable OW4. When CF was applied either solely or combined with OW4, there were no differences in plant N uptake between the soils A, B, and C, but a significantly lower plant N uptake was found in the most sand rich soil (soil D) (Tables 5 and 6).
Independent of treatment the grain yield varied during the three growth periods (Table 6). The enhanced yield of the third growth season was partly a result of prevailing favourable climatic condition that year. However, it was also partly due to a residual effect of the organic matter applied in the earlier seasons since the increase in yield for the control treatment was less prominent than those receiving organic matter (Table 6).
The leaching of NO3 − was generally low in the growing seasons, while significant losses took place during the growth dormant periods. The winter periods had frequent changes between frost and thawing, and episodes with enhanced NO3 − concentrations in leachate occur throughout the winter periods. As a consequence of the somewhat depressed plant growth during the second growth season, particularly high NO3 − leaching was measured after that summer (Yr2_a and Yr3_s, Table 7). Independent of treatment, the highest leaching of NO3 − was observed from the moraine soil (soil A) (Figure 1). Application of CF, and CF combined with OW4, influenced generally a significantly higher NO3 − leaching than in the control, while application of OW4 had no or negative effect on the NO3 − leaching. For the control treatment, the leaching of NO3 − was almost similar for the soils B, C, and D (Figure 1). When OW4 was applied significantly lower leaching of NO3 − was observed from the two clay rich soils (soil B and C) during the autumn periods (Table 7). The boosting effect of CF combined with OW4 which was clearly seen on plant N-uptake, was not seen on NO3 − leaching. Though enhanced NO3 − leaching was seen in the moraine soil (Soil A) after harvest, the second experimental year (Yr2_a, Yr3_s, Table 7). Independent of treatment, substantially higher outflux of N (the sum of plant N uptake and NO3 − leaching) was found in the moraine (Soil A) than in the other soils. Application of OW4 gave no changes in NO3 − leaching relative to the control. Enhanced leaching of NO3 − was seen in autumn and spring after the depressed plant growth in the second experimental year. For the sandy soil (soil D) the N-leaching was particularly high in the autumn, while for the more clay rich soils a similar high N-leaching was seen in the spring (Table 7).
3.4. Testing SOILN_NO to Data from the Lysimeter Experiment
The management routine N pool size and parameter settings obtained by calibration to data from the OW4 and CF treatments in the laboratory incubation experiment (Table 8) were tested to measure NO3 − outfluxes from the soil columns. A relatively good fit (Figures 3 and 4: lines Sim_inc, Table 9) was obtained for the OW4 treatment, in soil A and B, while the degree of fit was dramatically reduced in soil C and D (Figures 5 and 6: lines Sim_inc, Table 9). In soil C and D SOILN_NO simulated too low NO3 − leaching compared to the observed. An almost similar trend was seen for the CF treatment (Figures 3–6, Line: Sim_inc). Use of the N pools sizes and parameters obtained from calibration to the CF treatment in the laboratory experiment (Table 8) made the model to some degree reproduce the NO3 − leaching observed in soil A and B (Figures 3 and 4). However, in the other soils SOILN_NO tended usually to underestimate the mineral N generation. The fact that SOILN_NO, in the calibration procedure, failed to reproduce the initial net assimilation of N seen for the CF treatment in the laboratory (Figure 2), seemed not to induce overestimation of the mineral N release when applied to field conditions. Thus, direct transfer of the pool sizes and parameters obtained by calibration to data from the laboratory experiment to field conditions failed for two of the soils included in this experiment. As already mentioned, a more rapid decomposition of organic matter in sandy soil (soil D) compared to the other more clay rich soil supports recalibration. However, soil C was from the same area as the soil used in the laboratory incubation experiment.
Disregarding the soil texture effect seen on organic matter decomposition and assuming the lack of applicability for the calibrated parameters and pool sizes to be due to general aspects as, for example, influence of plant growth and changes in scale, a new parameter and pool size setting, characteristic for each of the organic matters, but being suitable in all the soils, were searched (Figures 4–6, Line: Sim_n.c.A). However, no such general parameter and pool size settings were found.
The emphasis of this study was to evaluate if a mechanistic model calibrated to data from simple laboratory incubation experiments could be a useful tool estimating the N fertilizer value of organic waste in field. Both data from decomposition of organic matter in the laboratory incubation and the field soil column experiments were well suited for a modelling test as they reveal significant variations. Already from the incubation experiment it became clear that although the organic matter C : N ratio is one of the most considered quality criteria of organic matter , it is clear that the simple criteria alone is inadequate for predicting the rate of N mineralization [28, 29] and thereby the N fertilizer value of organic matter. Other explanatory factors have been suggested. Kirchmann and Lundvall  found microbial N immobilization to be governed by the presence of volatile fatty acids in a study with decomposition of cattle and pig slurry in soil. The fatty acids acted as an easily decomposable C source for the microorganisms and caused immobilization of Hattori and Mukai  claimed that high content of hemicellulose and lignin in sludge influenced low N mineralisation rates. Based on sequential analysis, Van Soest  and Van Soest et al.  fractionated organic matter into soluble, hemicellulose, cellulose, and lignin-like substances, all with different decomposition rates. There are several analytical approaches for characterising soil organic matter, for example, , however understanding the nature of organic matter decomposition in soil is still a challenge. By use of mechanistic modelling, different approaches have been used in order to quantitatively determine the decomposition of organic matter in soil. In the large group of compartment models the complex structure of the organic matter fractions with different stability are mimicked by fractionating the organic matter into different stability compartments. Henriksen and Breland [35–37] reproduced N dynamics during decay of crop residues by a model based on the fractionation by Van Soest . They separated the applied organic matter into three different compartments before further allocation into five soil organic matter pools. However, generally the partitioning of the total organic C and N into pools with different stability in the models is done by calibration of initial values and not related to any analytical or experimental method of fractionation. Based on data from an elution-incubation experiment combined with a modelling performance, Smith et al.  suggested that at least five different N fractions could be separated in waste water biosolids, but that the NO3 − production properties of the biosolids were adequately described by a twin-pool exponential model as proposed by Molina et al. . They assumed that it was because the rates of NO3 − formation of certain N fractions were sufficiently similar for some of them to be broadly grouped into either rapidly or slowly converted pools for practical and descriptive purposes. Based on these results they claimed that routine incubation tests combined with a twin-pooled exponential model could provide the basis for a simple predictive tool for advising farmers of biosolids N fertilizer value and how to minimize NO3 − leaching losses on a site and biosolid-specific basis, that is, the approach further analysed here. Relative to the study by Sogn and Bakken  the number of organic matter pools and stability parameters to be adjusted in SOILN_NO was in this study restricted to be those included in the SOILN_NO management routine only. This restriction reduced the merges between observed and simulated data relative to what achieved by the optimal fit adjustments carried out by Sogn and Bakken . Results from Sogn and Bakken  showed that use of first-order decay equations and a limited number of organic matter pools reproduce data on mineral N generation following organic matter decomposition. By adjusting only the pool sizes and parameters in the management routine, particularly the periods with N immobilisation were difficult to reproduce (Figure 2). However, the periods of N immobilization seen in the incubation experiment were generally so short that it was assumed that this immobilization would not affect negatively the plant N uptake and thus the plant growth during a real growth season. Thus for the aim of this model application the smoothened mineral N curve suggested (Figure 2) was evaluated as satisfactory for further employment to field conditions. Although reduced merge between observed and simulated data compared to the results shown by Sogn and Bakken , hypothesis no 1 was not rejected. By adjustments of the amount of N as well as some parameters included in the management routine (Table 8), SOILN_NO seemed for practical purposes to satisfactorily reproduce the observed mineral N dynamics following decomposition of different organic matters in a laboratory incubation experiment.
Also in the outdoor soil column experiment it was apparent the organic matter C : N ratio only, could not predict the rate of N mineralisation during decomposition in soil. The amount of total N added via OW4 was more than double of the amount added as CF (Table 3). Still, application of OW4 gave generally no or very small increases in both plant N uptake and NO3 − leaching relative to the control, while application of CF resulted generally in significantly enhanced plant N uptake as well as NO3 − leaching. Application of OW4 contributed mainly as surplus to the stable humus N in soil. However, by fractionating the organic matter C and N content into the different stability compartments also the observed N dynamics were approached. The significantly higher plant uptake in the combined treatment (OW4 + CF) than when CF was applied solely might be due to increased mineralization rate in soil if the overall nutrient availability to microbes increases . Boosting effects on yield which are achieved by applying energy rich materials together with more stable organic wastes are well known related to biogas production (e.g., ). Mixing manure and organic waste has also resulted in improved plant availability of the total N in the organic waste [41, 42]. Such relationship is, however, not included in the SOILN_NO, but was mimicked by individual best-fit adjustments of parameters, as well as initial N pools in the management routine (Figures 3–6, Line: Sim) in order to reflect the synergetic enhanced N mineralization.
The outdoor lysimeter experiment included four different soils having different soil reaction with mineral fertilizer (Figure 1) and organic matter applied. An influence of soil type on the fate of the waste organic N in soil has been found in several studies. Chae and Tabatabai  found the amount of N mineralised after application of different organic wastes to be significantly higher in clay rich than in sand rich soils. Also in a study by Lindemann and Cardenas  where sewage sludge was added to different soil and incubated for 32 weeks, the lowest percent mineralised was measured in a sandy soil. In contrast, some studies report just the opposite texture influence. Catroux et al.  and Hassink et al.  found that the mineralization of soil organic matter was more rapid in sandy soils than in clay soils. These results was explained by the fact that organic substances can be adsorbed in small pores and on surfaces of clays in clay rich soils and thereby rendered less available for microbial decomposition. In our column experiment there were also variations between the soil types with respect to N mineralization rate. It can hardly be interpreted as a unique effect of texture, but it seems as the mineralization also here was more rapid in sandy soils than in clay rich. Effects of soil texture are in COUP/SOILN_NO governed by description of porosity and water holding capacities. Soil texture specific heat and water flow variables were generated by the COUP model and given as driving variables to the calibrated SOILN_NO model when applied to the outdoor column experiment. The NO3 − leaching data indicates more rapid N mineralization in sandy soils compared to more clay rich (Table 7). Thus, the parameters and initial N pools for the same organic matter, have to be different for the sandy (Soil D) versus the more clay rich soils (Figure 6).
COUP is developed for Nordic conditions with winter periods with snow, frozen soil, and frequent episodes with rapid changes between frost and thawing. As shown in Figure 1 the high NO3 − leaching during autumn and spring was well simulated by COUP/SOILN_NO. However, the model failed to reproduce the measured NO3 − leaching after organic waste application from two (Soil C and D, Figures 5 and 6) of the four soils included in the experiment. Too low NO3 − leaching was simulated. For the sandy soil (soil D), pools and parameters were adjusted to reproduce the more rapid N mineralisation. Additionally, if winter periods had been included in the incubation experiment in the laboratory, the data used for model calibration, a more realistic parameterization of SOILN_NO parameters might have been achieved. A parameterization better reflecting the effects of freezing and thawing on organic matter decomposition in soil. Presupposing that experimental lysimeter artefacts were not the reason for the higher real NO3 − leaching than simulated, the practical consequence of the underestimated NO3 − leaching would be that making predictions based on simulation results from SOILN_NO calibrated to incubation data, would lead to underestimation of the NO3 − leaching risks in some soils. One reason for the discrepancy between measured and simulated values might be macropore flow which is not accounted for in the COUP model. Anyway, the test of the calibrated SOILN_NO model to data from the outdoor soil column experiment illustrated that SOILN_NO combined with data for laboratory experiments cannot directly be used as a tool to simulate NO3 − leaching from more realistic soil/plant systems at a larger scale. As mentioned in the introduction, several researchers, that is, [11–14] have claimed that simple laboratory incubation experiments give relatively precise predictions of mineral N released from recently incorporated organic matter during decomposition in real soil and plant systems. However, there are other studies showing results that oppose against such a tool. For example, by comparison of meadow and open field mineral N data, Paustian et al.  found the net mineralization of humus N to be strongly influenced by the plants activity. Additionally Jingguo and Bakken  found plants to influence the net mineralization significantly by their competition with micro-organisms for the available N. If not adequately described in a submodel with soil plant interactions, a strong plant influence on decomposition and N mineralization rates, may reduce the transferability of a model calibrated to mineral N data obtained in simple laboratory incubation experiments to real agronomic soil/plant systems. An additional factor here will also be the lack of winter periods in data used for model calibration. The effects of freezing and thawing on organic matter decomposition in soil was not included in the laboratory experiment, while these effects are included in the outdoor lysimeter experiment. Our hypothesis 2 was not verified. However, by minor adjustments of N pool sizes and parameters (Table 8) the model was capable of simulating the observed NO3 − leaching (Figures 3–6, Line : Sim). This means that use of the model may increase the understanding of how different factors influence the quantity of mineral N released during decomposition of organic matter in soil, but can so far only be used as a tool for exact prediction of plant-N availability and NO3 − leaching as a consequence of organic waste application when site specific recalibration is performed.
No rejection of hypothesis 1 and 2 means that the model calibrated to data from simple incubation experiments, adequately quantifies the influence of climate, soil texture, and plants on the N dynamics following decomposition of organic wastes in soil. The model may then be useful tool estimating the N fertilizer value of organic waste. The results from our study show that even though a model may successfully simulate data from an incubation study, results from such applications cannot be used directly to predict the fate of the organic N in organic waste products in more realistic soil/plant systems. However, by minor tuning of pools and parameters the model showed to be capable of reproducing observed data of NO3 − leaching also in field lysimeters. Need for site-specific recalibration has also been stressed in a model application study by Johnsson et al. . SOILNDB, a model based on the COUP and SOILN models combined with a parameter database and a parameter estimation algorithm was used. SOILNDB is an upscaling model system, and it is meant to be a tool for quantifying the effect of different agricultural management practices on NO3 − leaching from a region or whole country. It was found that more an application was limited in time and space, the higher was the risk for errors caused by use of simplified input data. These findings were related to changes in scale from large systems to smaller, that is, the opposite of ours. However, generally users were warned that the reliability of the SOILNDB results could be low if the model was used for applications it was not primary developed for. Anyway it was claimed that SOILNDB might be applicable to more limited area units and time periods provided that measured site-specific data were available. Challenging transmission of a model between different systems and scales often ends up in a debate of simplicity versus explanatory capability. Mechanistic models are often criticised for lack of generality and a high number of parameters, often with a vague physical definition. As stated by Thuries et al.  with reference to Christensen , the challenge of model makers is to “keep the balance between structural simplicity, explanatory capability and predictive power”. General models without need for site-specific parameter calibration for every new application are desired. Statistical models or semimechanistic models where several parameters and processes have been merged to represent some “key processes” are often preferred when doing system analysis. However, in an application of COUP/SOILN to predict effects of climatic changes on C, N, water, and heat dynamics in winter wheat production, Eckersten et al.  stressed the importance of using process-based modelling in such analysis in order to understand the complicated relationships between different processes and how processes interact when parameters and variables in a system change due to changes in external conditions.
By calibration of a restricted number of organic N pools and parameters in the SOILN_NO management routine the model reproduced the N dynamics observed during decomposition of different organic matters and waste products in soil in a laboratory incubation experiment. However, when upscaled and tested the calibrated model failed to reproduce observed NO3 − leaching from soil columns with barley plants exposed to natural climatic conditions when the same type of organic matters were applied to that system. The model significantly underestimated the NO3 − leaching from two of the soils. Recalibration of SOILN_NO was needed in order to obtain a satisfactory fit between observed and simulated NO3 − leaching. Application of the calibrated model to a system which includes different soil types, supported the need for recalibration of coefficients determining the rate of organic matter decomposition. Furthermore, if the data used for calibration had reflected effects of winter climate on organic matter decomposition, the need for recalibration would have been less. However, results from this study support the impression that the SOILN_NO is a flexible model, able to simulate quite different conditions and systems. However, this study emphasises the need for site-specific data as basis for model calibration when the model is upscaled and should be used for predictive purposes and does not support solely use of parameters obtained in laboratory incubation experiments for upscaling and practical use of the SOILN_NO model in agricultural management guidance.
- A. Shiralipour, D. B. McConnell, and W. H. Smith, “Physical and chemical properties of soils as affected by municipal solid waste compost application,” Biomass and Bioenergy, vol. 3, no. 3-4, pp. 261–266, 1992.
- S. Marinari, G. Masciandaro, B. Ceccanti, and S. Grego, “Influence of organic and mineral fertilisers on soil biological and physical properties,” Bioresource Technology, vol. 72, no. 1, pp. 9–17, 2000.
- S. R. Smith, V. Woods, and T. D. Evans, “Nitrate dynamics in biosolids-treated soils. I. Influence of biosolids type and soil type,” Bioresource Technology, vol. 66, no. 2, pp. 139–149, 1998.
- K. Debosz, S. O. Petersen, L. K. Kure, and P. Ambus, “Recycling of sewage sludge and household compost: effects on soil fertility under laboratory and field conditions,” Applied Soil Ecology, vol. 19, pp. 237–248, 2002.
- E. J. Rubins and F. Bear, “Carbo-nitrogen ratios in organic fertilizer materials in relation to the availability of their nitrogen,” Soil Science, vol. 54, pp. 411–423, 1942.
- J. M. Melillo, J. D. Aber, and J. F. Muratore, “Nitrogen and lignin control of hardwood leaf litter decomposition dynamics,” Ecology, vol. 63, no. 3, pp. 621–626, 1982.
- W. J. Parton, D. S. Schimel, C. V. Cole, and D. S. Ojima, “Analysis of factors controlling soil organic matter levels in Great Plains grasslands,” Soil Science Society of America Journal, vol. 51, no. 5, pp. 1173–1179, 1987.
- G. Standford and S. J. Smith, “Nitrogen mineralization potentials of soils,” Soil Science Society of America Journal, vol. 36, pp. 465–472, 1972.
- J. A. E. Molina, C. E. Clapp, and W. E. Larson, “Potentially mineralizable nitrogen in soil: the simple exponential model does not apply for the fist 12 weeks of incubation,” Soil Science Society of America Journal, vol. 44, pp. 442–443, 1980.
- H. Johnsson, L. Bergstrom, P. E. Jansson, and K. Paustian, “Simulated nitrogen dynamics and losses in a layered agricultural soil,” Agriculture, Ecosystems and Environment, vol. 18, no. 4, pp. 333–356, 1987.
- F. R. Magdoff and F. W. Chromec, “Nitrogen mineralization from sewage sludge,” Journal of Environmental Science and Health, Part A, vol. A12, no. 4-5, pp. 191–201, 1977.
- J. Z. Castellanos and P. F. Pratt, “Mineralization of manure nitrogen correlation with laboratory indexes,” Soil Science Society of America Journal, vol. 45, pp. 354–357, 1981.
- C. F. Parker and L. E. Sommers, “Mineralization of nitrogen in sewage sludges,” Journal of Environmental Quality, vol. 12, no. 1, pp. 150–156, 1983.
- H. Marstorp and H. Kirchmann, “Carbon and nitrogen mineralization and crop uptake of nitrogen from six green manure legumes decomposing in soil,” Acta Agriculturae Scandinavica, vol. 41, pp. 243–252, 1991.
- T. C. Flavel and D. V. Murphy, “Carbon and nitrogen mineralization rates after application of organic amendments to soil,” Journal of Environmental Quality, vol. 35, no. 1, pp. 183–193, 2006.
- B. Gabrielle, J. Da-Silveira, S. Houot, and J. Michelin, “Field-scale modelling of carbon and nitrogen dynamics in soils amended with urban waste composts,” Agriculture, Ecosystems and Environment, vol. 110, no. 3-4, pp. 289–299, 2005.
- A. Vold, Development and evaluation of a mathematical model for plant N-uptake and N-uptake and N.leaching from agricultural soils, Doctor Scientiarum thesis, The Agricultural University of Norway, 1997.
- T. Børresen, “The effect on soil physical properties of undersown cover crops in cereal production in southeastern Norway,” Norwegian Journal of Agricultural Sciences, vol. 7, pp. 369–379, 1993.
- N. Senesi, “Composted materials as organic fertilizers,” Science of the Total Environment, vol. 81-82, pp. 521–542, 1989.
- M. P. Bernal, A. F. Navarro, A. Roig, and J. Cegarra, “Carbon mineralization of composting organic materials in soil as a function of their age,” in Transactions of the 15th World Congress of Soil Science, vol. 9, pp. 314–315, Accapulco, Mexico, 1994.
- J. M. Bremner, “Inorganic forms of nitrogen. 1,” in Methods of Soil Analysis, Part 2. Agronomy, C. A. Black, et al., Ed., vol. 9, pp. 1179–1237, American Society of Agronomy, Madison, Wis, USA, 1965.
- A. Vold and J. S. Søreng, “Optimization of dynamic plant nitrogen uptake, using apriori information of plant nitrogen content,” Biometrical Journal, vol. 39, no. 6, pp. 707–718, 1997.
- P.-E. Jansson and S. Halldin, “Model for the annual water and energy flow in a layered soil,” in Comparison of Forest and Energy Exchange Models, S. Halldin, Ed., pp. 145–163, Society for Ecological Modelling, Copenhagen, Denmark, 1979.
- P. E. Jansson and D. S. Moon, “A coupled model of water, heat and mass transfer using object orientation to improve flexibility and functionality,” Environmental Modelling and Software, vol. 16, no. 1, pp. 37–46, 2001.
- T. A. Sogn and L. R. Bakken, “Prediction of mineral -N dynamic following additions of organic waste to agricultural soil,” DIAS report 13, Danish Institute of Agricultural Sciences, 1999.
- A. P. Whitmore, “A method for assessing the goodness of computer simulation of soil processes,” Journal of Soil Science, vol. 42, no. 2, pp. 289–299, 1991.
- P. Smith, J. U. Smith, D. S. Powlson et al., “A comparison of the performance of nine soil organic matter models using datasets from seven long-term experiments,” Geoderma, vol. 81, no. 1-2, pp. 153–225, 1997.
- S. Recous, D. Robin, D. Darwis, and B. Mary, “Soil inorganic N availability: effect on maize residue decomposition,” Soil Biology and Biochemistry, vol. 27, no. 12, pp. 1529–1538, 1995.
- T. Paré, H. Dinel, M. Schnitzer, and S. Dumontet, “Transformations of carbon and nitrogen during composting of animal manure and shredded paper,” Biology and Fertility of Soils, vol. 26, no. 3, pp. 173–178, 1998.
- H. Kirchmann and A. Lundvall, “Relationship between N immobilization and volatile fatty acids in soil after application of pig and cattle slurry,” Biology and Fertility of Soils, vol. 15, no. 3, pp. 161–164, 1993.
- H. Hattori and S. Mukai, “Decomposition of sewage sludges in soil as affected by their organic matter composition,” Soil Science and Plant Nutrition, vol. 32, no. 3, pp. 421–432, 1986.
- P. J. Van Soest, “Use of detergents in the analysis of fibrous feeds. II. A rapid method for the determination of fiber and lignin,” Journal of the Accosiation of Official Analytical Chemists, vol. 46, pp. 829–835, 1963.
- P. J. Van Soest, J. B. Robertson, and B. A. Lewis, “Methods for dietary fiber, neutral detergent fiber, and nonstarch polysaccharides in relation to animal nutrition,” Journal of Dairy Science, vol. 74, no. 10, pp. 3583–3597, 1991.
- I. Kögel-Knabner, “Analytical approaches for characterizing soil organic matter,” Organic Geochemistry, vol. 31, no. 7-8, pp. 609–625, 2000.
- T. M. Henriksen and T. A. Breland, “Nitrogen availability effects on carbon mineralization, fungal and bacterial growth, and enzyme activities during decomposition of wheat straw in soil,” Soil Biology and Biochemistry, vol. 31, no. 8, pp. 1121–1134, 1999.
- T. M. Henriksen and T. A. Breland, “Evaluation of criteria for describing crop residue degradability in a model of carbon and nitrogen turnover in soil,” Soil Biology and Biochemistry, vol. 31, no. 8, pp. 1135–1149, 1999.
- T. M. Henriksen and T. A. Breland, “Decomposition of crop residues in the field: evaluation of a simulation model developed from microcosm studies,” Soil Biology and Biochemistry, vol. 31, no. 10, pp. 1423–1434, 1999.
- S. R. Smith, V. Woods, and T. D. Evans, “Nitrate dynamics in biosolids-treated soils. III. Significance of the organic nitrogen, a twin-pool exponential model for nitrogen management and comparison with the nitrate production from animal wastes,” Bioresource Technology, vol. 66, no. 2, pp. 161–174, 1998.
- I. P. Hartley, D. W. Hopkins, M. Sommerkorn, and P. A. Wookey, “The response of organic matter mineralisation to nutrient and substrate additions in sub-arctic soils,” Soil Biology and Biochemistry, vol. 42, no. 1, pp. 92–100, 2010.
- Y. Li, S. Y. Park, and J. Zhu, “Solid-state anaerobic digestion for methane production from organic waste,” Renewable and Sustainable Energy Reviews, vol. 15, no. 1, pp. 821–826, 2011.
- F. Nevens and D. Reheul, “The application of vegetable, fruit and garden waste (VFG) compost in addition to cattle slurry in a silage maize monoculture: nitrogen availability and use,” European Journal of Agronomy, vol. 19, no. 2, pp. 189–203, 2003.
- T. Tontti, A. Nykänen, and M. Kuisma, “Waste composts as nitrogen fertilizers for forage leys,” Agricultural and Food Science, vol. 18, no. 1, pp. 57–75, 2009.
- Y. M. Chae and M. A. Tabatabai, “Mineralization of nitrogen in soils amended with organic wastes,” Journal of Environmental Quality, vol. 15, no. 2, pp. 193–198, 1986.
- W. C. Lindemann and M. Cardenas, “Nitrogen mineralization potential and nitrogen transformations of sludge-amended soil,” Soil Science Society of America Journal, vol. 48, no. 5, pp. 1072–1077, 1984.
- G. Catroux, R. Chaussod, and B. Nicolardot, “Assessment of nitrogen supply from the soil,” Comptes Rendus Academie Agriculture Francais, vol. 3, pp. 71–79, 1987.
- J. Hassink, L. A. Bouwman, K. B. Zwart, and L. Brussaard, “Relationships between habitable pore space, soil biota and mineralization rates in grassland soils,” Soil Biology and Biochemistry, vol. 25, no. 1, pp. 47–55, 1993.
- K. Paustian, O. Andren, M. Clarholm, et al., “Carbon and nitrogen budgets of four agro-ecosystems with annual and perennial crops, with and without N fertilization,” Journal of Applied Ecology, vol. 27, no. 1, pp. 60–84, 1990.
- W. Jingguo and L. R. Bakken, “Competition for nitrogen during decomposition of plant residues in soil: effect of spatial placement of N-rich and N-poor plant residues,” Soil Biology and Biochemistry, vol. 29, no. 2, pp. 153–162, 1997.
- H. Johnsson, M. Larsson, K. Mårtensson, and M. Hoffmann, “SOILNDB: a decision support tool for assessing nitrogen leaching losses from arable land,” Environmental Modelling and Software, vol. 17, no. 6, pp. 505–517, 2002.
- L. Thuries, M. Pansu, M.-C. Larré-Larrouy, and C. Feller, “Biochemical composition of organic materials in relation with their mineralization in sandy soil,” Soil Biology & Biochemistry, vol. 34, pp. 239–250, 2002.
- B. T. Christensen, “Matching measureable soil organic matter fractions with conceptual pools in simulation models of carbon turnover: revision of model structure,” in Evaluation of Soil Organic matter Models, D. S. Powlson, P. Smith, and J. U. Smith, Eds., pp. 143–159, Springer, Berlin, Germany, 1996.
- H. Eckersten, K. Blombäck, T. Kätterer, and P. Nyman, “Modelling C, N, water and heat dynamics in winter wheat under climate change in southern Sweden,” Agriculture, Ecosystems and Environment, vol. 86, no. 3, pp. 221–235, 2001.