An agro-ecological framework for integrated nutrient management, with special reference to Kenya

This thesis provides a framework for integrated nutrient management in agricultural land use systems, with particular reference to its impact on productivity, fertilizer use efficiency, and sustainability in well-delimited tracts of land (agro-ecological units), characterized by a specific set of soil and climatic properties. Most of the research was conducted in Kenya, but methodology and results can be applied to any tropical region.Quantitative assessments are made of the nitrogen, phosphorus and potassium balance in the root zone of the arable land in sub-Saharan Africa. Land use systems are characterized by nutrient inputs (mineral fertilizer, manure, atmospheric deposition, biological nitrogen fixation, sedimentation) and nutrient outputs (removal of harvested crop parts and residues, leaching, denitrification, erosion), and the balance between the two. It is shown that outputs exceed inputs all over the continent. As scale-inherent simplifications were inevitable, a more detailed study is presented for the Kisii District in Kenya, with similar results. The alarming figures call for agronomic and policy interventions in the soil nutrient balance. A scaleneutral decision- support model of this nature is described, in which scenarios for improved nutrient management are worked out.Mineral fertilizers are, with the present abolishment of subsidies in many African countries, increasingly expensive, and it is evident that they must be used efficiently. With this in mind, a network of 70 researcher-managed, but farm-based factorial fertilizer trials was established in rainfed agricultural Kenya. The trial sites were chosen such that they represent wider ranges of similar environments. As such, the results form a basis for fertilizer recommendations that are not just crop-specific, but also specific for agro- ecological units. The site selection procedure is demonstrated, based on a profound scrutiny of soil and climate maps in Kenya, and followed by results of four years study on the response of maize to fertilizers and manure in three agro-ecological units.The above approach leads to recommendations which apply to a district scale, but do not account for spatial variability between and within individual farmers' fields. Field-level heterogeneity, however, affected some trials as it was beyond the researcher's control. Two examples are dealt with in detail, i.e. (i) bypass flow and associated nitrogen leaching in a cracking clay soil, and (ii) the spotty emergence of the parasitic weed Striga hermonthica.As the establishment and maintenance of a trial network is costly and timeconsuming, computer models are increasingly used as an alternative. The QUantitative Evaluation of the Fertility of Tropical Soils (QUEFTS) model describes relations between (i) soil chemical parameters, (ii) potential supply of N, P and K from soils and fertilizers, (iii) actual uptake of N, P and K by maize, and (iv) maize grain yield. Characteristic features of the model are the inclusion of all macronutrients, acknowledging interactions between them, and the low input requirements enabling agronomists in tropical environments to test the model. The four steps of QUEFTS are calibrated with input data from fertilizer trials in different agro- ecological units. A complete run of the modified version shows a high correlation between measured and calculated yield, but although new relations are found, the basic structure and theoretical concepts of the original QUEFTS still stand