|dc.description.abstracteng||There is a general consensus that world food production has to be increased significantly to fulfill the growing demand until 2050. However, at the same time resource use efficiency has to be improved due to declining resource bases (oil, phosphor) and the environmental effects of inputs (pesticides, nitrogen). Solving this paradox of “producing more with less” will rely, for example, on applying inputs according to attainable yield levels. However, attainable differs from year to year due to different weather conditions (attainable yield and risk relationships). Furthermore, soil conditions affect the water storage for plant uptake. Traditionally, field trials have been conducted to assess attainable yield in a certain regions. In regions with well-developed extension service it was possible to develop simple empirical yield response relationship to climate and fertilizer based on a range of field trials. Such data rich environments are usually restricted to the developed world and mostly lacking in the developing countries. However, simple empirical yield models like for example the French and Schultz approach for southeastern cannot capture the complex interaction between the factors, which determine attainable yield (water, solar radiation, rainfall, nitrogen, soil properties). So that it is not possible to develop site specific management recommendations, which are necessary to improve resource use efficiency and closing the yield gap.
Therefore, crop modelling, mainly in modern simulation frameworks like APSIM or DSSAT has been widely used in the scientific community for setting such yield targets, in particular for wheat and maize. However, for other crops and certain conditions such as soil constraints there is less information found in the literature (Chapter I). With this in mind, the main objective of this thesis was to develop/adapt crop model approaches for oil palm plantations in Indonesia, oilseed rape production in Europe and wheat cropping in southeastern Australia to set yield targets applicable for agronomists, farmers or plantation managers i.e. taking data availability and the socioeconomic context of the crop production system into account. After selection and adaption all model approaches presented in this study were evaluated against field trial data. Finally, challenging the idea of crop usage for sustainable intensification, all three models were applied to typical problems in the respective production systems.
In the first research chapter (chapter II) a new physiological based oil palm model (PALMSIM) is presented. Assessing potential yield in oil palm based on crop modelling depicts a challenge. First of all there are few models available, which are rarely tested against field trial data. Secondly, these models are data input intensive in parameterization and in terms of running them. The high data demand for oil palm modeling is lacking, such as the essential basic and necessary data such as soil information, cultivar parameter and long-term weather records. This makes the application of standard modelling approaches (daily time step, detailed water balance etc.) unlikely. PALMSIM therefore follows the idea of being both simple enough in terms of data and at the same time incorporates sufficient plant physiological knowledge to be generally applicable across sites with different growing conditions. The version presented in chapter II simulates potential yield based on incoming solar radiation only and therefore only gives realistic yield levels for optimal growing conditions. Nevertheless, it was possible to evaluate the model against field trial data from Indonesia and Malaysia. In the next chapter (chapter III), the PALMSIM model is extended by incorporating a simple water balance. This is often used in oil palm cultivation to assess water deficiency. The improved PALMSIM version is then used to exemplify and illustrate the use of crop modelling in oil palm sustainable intensification, the extension into marginal degraded sites, and the increase of productivity in existing plantations. One case study presented in this chapter makes use of a recent report by the World Resource Institute, which aims to identify degraded sites in Kalimantan. PALMSIM was run for water-limited potential on a 0.1°grid for Kalimantan and overlaid with the suitability map produced in the above-mentioned report. Results show that 8.1% of the suitable land has a potential productivity of more than 40 Mg FFB /ha. The largest proportion (35.6% of the suitable land or 115,300 km2) falls into the category between 35 and 40 Mg FFB ha. In the second case study presented in the paper, PALMSIM was setup for six plantation sites in Indonesia. Long-term weather data was derived using WorldClim data in the stochastic weather generator MarkSim. In all six sites, best management practices were introduced in five blocks. As a comparison, similar blocks were selected and managed following standard practice in the plantation. The potential and water-limited yield was then simulated for each plantation. This shows that potential yields are generally higher in Sumatra than in Kalimantan due to higher solar radiation. Water deficiency was a problem at two sites, either due to low rainfall or soil constraints. The gap between water-limited yield and actual yield differs from location to location, and therefore requires a site-specific analysis of the factors causing the yield gap. To sum up, in the two case studies the scope for sustainable intensification at regional and at plantation level was explored in a quantitative manner - a novel approach to oil palm production.
While the scale of decision making for oil palm is often regional, plantation or the smallest unit the block level, the scale and the challenges for German oilseed rape production is field scale (typically 1-4 ha) and needs a more powerful approach in terms of factors which are taken into account. Winter oilseed rape production is typically characterised by low nitrogen (N) use efficiency. Defining site-specific fertiliser strategies based on field trials and crop modelling may help to improve the ecological efficiency of this crop. However, no model has been evaluated for winter oilseed rape that simulates the growth of the plant as limited by the interaction of water and N. In this chapter the APSIM canola model, originally developed for the temperate regions of Southeast Australia, was adapted for conditions in Germany and tested successfully against measured data (biomass, grain yield, leaf area index, N-uptake and soil mineral N) from three sites around Göttingen and with different N-fertiliser rates. In the second part of the study the evaluated model was used in a simulation experiment to explore site specific climate and soil related production limitations to match fertiliser rates to these yield targets. Simulation results indicates that water supply plays a critical role when maintaining high N use efficiency and simultaneously grain yields of 4000 kg ha-1.
In the last chapter ASPIM was again used to develop site-specific recommendations; here for a case study from southeastern Australia. (annual average rainfall of 250-300 mm). Field productivity shows enormous spatial variation. Since these differences are largely related to soil variation in fertility, subsoil constraints (high salt, Boron levels) and plant available water capacity, three distinctively different zones - low subsoil constrained sandy zone, moderately subsoil constrained zone, and severe subsoil constrained, clay soil - were defined for one field at five sites in the Mallee. To assess the scope of zone-specific management, zone specific yield and soil properties were surveyed for each site in 2006 and 2007. Additionally, the crop model APSIM was parameterised for these challenging soils (taking subsoil constraints into account), successfully tested against the observed yield data and finally used to carry out a long term simulation experiment investigating the response of the three zones to nitrogen fertilisation over multiple seasons (50 years). For the severe constrained zone, simulated and observed yields were well related to rainfall, indicating that this soil zone is limited by water. Nitrogen fertilisation above the standard rate (30 kg/ha) should be avoided, especially in low rainfall years. Simulated and observed yields for the low constrained zone showed a weaker relationship with rainfall. Simulation analysis suggested a potential increase of production on these sandy soils due to higher N-input as evaporation rate and the organic matter content (lower N-supply) are lower than for the other two soil zones.
Across the three case studies, crop modelling has provided useful insights for setting yield ceilings. However, the development and the application of crop models have to be system specific. Currently, we are not able to simulate tropical plantation crops in a similar manner to annual crops like maize and rice, due to missing data in terms of validation, but even more so in terms of input data (soil and especially weather data). A compromise might be for the current situation the PALMSIM model, which still gives useful information despite its low input demand. However, in contrast it was relatively easy to develop for the annuals oilseed rape and wheat site-specific simulation analysis, which can serve as blueprint to improve perennial crop modelling.||de