Modelling Nitrogen and Phosphorus Cycling in Agricultural Systems at Field and Regional Scales

Abstract

Models evolve together with the evolution of our notion and perception of reality. Models can be narratives, graphical or mathematical descriptions, or computer simulations. More than two millennia ago, Chinese and Greek phi-losophers already had the notion that the environment was composed of the interacting elements earth, air, water, life and metals, but the complex relation-ships between these factors could only be understood after the birth of modern chemistry, at the end of the eighteenth century. The chemist Justus von Liebig (1803–1873) played an important role in unravelling how plants acquire nu-trients from soil, air and water, but other chemists and microbiologists in the eighteenth and nineteenth centuries also contributed to improving the under-standing of nutrient cycling processes (Smil 2001). Since that time, numerous (long-term) field experiments have been carried out to test Liebig's mineral theory and its modifications. For over a century and a half, dose-response ex-periments have addressed one or more of the following five basic questions (Van Noordwijk 1999): (1) to what extent do nutrients limit crop yield and quality?; (2) what is the quantity of nutrients supplied by the soil?; (3) what constitutes an effective fertiliser?; (4) how much fertiliser should be applied?; and (5) what are the environmental consequences of fertiliser use? With the increased availability of computers, increases in labour costs, and the increased awareness of the complexities involved in nutrient cycling as a Peter de Willigen, Oene Oenema, Wim de Vries 362 result of spatial and temporal heterogeneity, computer simulation models have become increasingly important, replacing, to some extent, long-term field ex-periments. For the nitrogen (N) cycle, Clark (1981) distinguished four stages of model development, namely (1) the process diagram stage, showing the connec-tion between the possible biogeochemical transformation processes for a nutri-ent; (2) the process and compartmental stage, showing the flow and transfor-mation of a nutrient between various sites or compartments; (3) the budgeted flows and compartmental stage, with quantified pools and flows; and (4) the simulation stage, i.e. the representation of the 'real' world via a computer model. Clark (1981) noted that published process diagrams show great similarity and durability in the literature, but that compartmental and ecosystem diagrams and simulation models tend to be highly individual and to show a great propensity for change. Hence, whereas nutrient transformation processes are considered to be ubiquitous, and there appears to be a common notion about these pro-cesses, (agro)ecosystems are notoriously diverse, with great differences in the relative importance of compartments and the transfer of nutrients between compartments. The early simulation models of the 1970s and 1980s – the so-called research models – focussed on increasing an understanding of the mechanisms of nu-trient cycling and testing of ideas and hypotheses, e.g. Beek and Frissel 1973; DeCoursey et al. 1989; Dutt et al. 1972. A range of applications, including quantification of nutrient flows and losses, scenario and policy analyses, and management guidance, has evolved. Such applications of simulation models are increasingly used as discussion-and decision-support tools by policy makers, managers and farmers. Along with their increased use, there is also a continuing debate over how correct and appropriate such models are as well as over the reli-ability and accuracy of their results. This ongoing debate fuels model improve-ment, testing and comparison. The increased use of nutrient modelling tools in policy and management decision-making also reflects increased concerns in today's society about nutrient losses from agro-ecosystems and their role in en-vironmental sustainability. This chapter deals with computer models simulating nutrient cycles in (agro)ecosystems. We limit the discussion to models simulating the cycling of nitrogen (N) and phosphorus (P), because these nutrients have the largest influ-ence on crop production and the environment, although aspects of the carbon (C) cycle that are relevant for these cycles are also discussed. With respect to spatial scale, we restrict this chapter to plot-scale models and regional mod-els because these models (with some modifications) are usually also the core of models at catchment level (plot-scale models) and at national, continental or even global levels (regional models). Models dealing with nutrient cycling in farming systems, the agricultural sector, and food chains in natural ecosystems and human societies are special cases and are not discussed here. For further information on these topics, the reader is referred to McCown (2005) and refer-ences therein.