Towards improved soil organic matter balance models in plant production systems: a focus on aspects of parameter survey

Abstract

The increasing world population and advancing climate change are major challenges for current and future agriculture. Agricultural systems must be designed in such a way that they increase productivity without causing additional environmental pollution, and reduce expansion of agricultural area. Soils play an important role in this objective, in particular soil organic matter (SOM), which is ascribed a variety of functions (e.g. provisioning of nutrients for plant growth and CO2 carbon sink). Farmers must therefore be able to organise their management in such a way that SOM stocks are safeguarded and, ideally, promoted. To facilitate this aim, so-called “humus balances” have been developed as simple practice-applicable models to assess SOM supply in arable farming. These models relate to pools and fluxes of either organic carbon and/or nitrogen in the soil-plant system. Both soil organic carbon (SOC) and soil total nitrogen (STN) are important components and indicators of SOM that are closely related to each other. As the stoichiometry of carbon and nitrogen is an important factor in the development of SOM stocks, it is important to record both parameters when assessing SOM dynamics. However, quantification of SOC, STN as well as total SOM stocks is difficult, as their relatively small amounts are very heterogeneous in the soil and the possibilities of their measurement are limited (detectability of short- term/low quantity changes, lack of standards). SOM balance models can be useful planning tools and are also being discussed as political instruments against the background of carbon sequestration and carbon farming. The suitability of SOM balance models is often criticised, though. In particular, poor model validation is highlighted, but there is also room for improvement in the parameterization and calibration of the models. In addition, there are many possibilities to further develop the models, for example to better estimate the SOM dynamic in the subsoil or to evaluate new management practices (e.g. new crops, fertilizers, tillage). The aim is to develop models that are reliable and easy to use. Farmers and agricultural advisors are also interested in models that make it possible to balance the supply of nutrients to crops and provide information on carbon sequestration. The aim of this thesis is to contribute to the improvement and development of SOM balance models, with a focus on the HU-MOD model that has been developed at the Chair of Organic Farming at the Justus Liebig University Giessen. One objective was to estimate the uptake of nitrogen from SOM mineralization by a crop that is fertilized with mineral nitrogen. This information is important for the calculation of SOM loss and the demand for organic matter supply. For this purpose, a greenhouse experiment with sorghum-sudangrass (Sorghum bicolor X Sorghum sudanense) and maize (Zea mays) was conducted. Plants were grown in bags and fertilized withlSN-labelled ammonium nitrate. The aim was to contribute to the scarce data on sorghum-sudangrass as an energy crop with regards to nitrogen derived from fertilizer (NdfF) in the plant’s biomass and fertilizer nitrogen utilization (FNU). It should also clarify whether it is advisable to use parameters of maize as proxies for sorghum-sudangrass if field data is missing for the latter. Results showed that FNU of sorghum-sudangrass (6S %) was significantly higher than that of maize (49 %), if grown in a greenhouse. Both crops accumulated more soil N than fertilizer N. The share of fertilizer N on total N uptake was also higher with sorghum sudangrass (NdfF = 38 %) compared to maize (NdfF = 34 %). This leads to the conclusion that parameters of maize should not be used as proxies for sorghum-sudangrass in SOM balance models. Parametrization of the crops in the model could not be validated though, as there was no sufficient data base (no long-term field experiment) for this purpose. With regard to model parametrization, it can be stated that the approach with lSN fertilisation under clearly defined growth conditions and system boundaries is overall suitable for obtaining data. However, the conditions in the greenhouse differ significantly from those in the field and direct transfer of the results is difficult. Future studies could be carried out in a lysimeter experiment to provide a better comparison with real conditions. In addition, gas measurements and the detection of leachate could provide a more complete representation of the N pathways. Another question was to what extent short-term field experiments can be used to assess SOM change under cropping systems as a basis for the validation of SOM balances and other models. To date, the validation of SOM balance models - if there is one - has been based on data from long-term field experiments. Although this is the most reliable approach, it is also time-consuming and expensive. The possibility of assessing parameters for new crops or fertilisers in short-term field experiments is therefore of great interest. To study this, a short-term field experiment (two series, each one year) was carried out at the experimental station Gladbacherhof of the Justus Liebig University Giessen. Target crops were winter wheat, potatoes, red clover for fodder (removed) and red clover for green manure (not removed). No additional fertilizer was applied. For SOC and STN assessment in the topsoil (0-30 cm), six subsamples per plot were analysed at the beginning and after harvest. Different outlier determination procedures were then applied to calculate plot means and subsequently measured data was compared to predictions by the HU-MOD balance model for each crop. The soil sampling design, paired with outlier determination, resulted in a decrease of minimum detectable differences (MDD) by a factor of 0.S3 for SOC and 0.63 for STN masses. This allowed for detection of changes in the magnitude of 3.7 % and 2.6% of background SOC and STN levels, respectively. SOC and STN changes were significant with treatments that had the highest effects (potatoes and mulched red clover). The comparison of apparent and modelled changes with HU-MOD proved the suitability of this data collection for the purpose of parameter validation. In addition, the observation period needed for SOM change detection can be reduced noticeably. Field experiments that aim to develop SOM balance methods can profit from this sampling procedure and new management practices, fertilizers or crops can be validated in a shorter time period. Further improvements to the sampling design could include additional series, consideration of variable spatial impacts on the soil in crop stands and bulk density estimation parallel to each sampling procedure. The third objective of this thesis was to clarify whether it is necessary to assess and consider SOM quantity changes in the subsoil in order to estimate and validate C and N balances in the soil. The rationale for this objective is that SOM balance models usually do not refer to a defined soil depth, but to organic matter inputs and outputs in general. For this study, mean SOC and STN changes were assessed in 30 cm depth increments down to 90 cm in the organic arable long-term field experiment Gladbacherhof over a l7-year observation period (three crop rotations). The experiment comprises three different farming types (mixed and stockless with either rotational ley or cash crops). Each farming type included four tillage treatments (full inversion, two-layer plough, reduced inversion and non-inversion). While differences in farming types could be recorded for topsoil SOC and STN, tillage treatments had effects on SOC and STN in 30-60 cm, where full inversion tillage resulted in an increase compared to reduced tillage treatments. This effect even led to a significant differentiation of these treatments with regards to the whole soil profile (0-90 cm). The results suggest that sampling depth should be extended to at least 60 cm in order to improve validation performance of SOM models for tillage effects on soil organic matter (SOM). The three experiments of this thesis show that sampling procedure and experimental setup have a great potential to further develop SOM balance models in such a way that they become more reliable planning and monitoring tools and can react in a more flexible way to new developments in agricultural practice. The results of this thesis contribute to the standardised improvement and refinement of SOM balance models.