Modelling dairy grazing systems: an integrated approach

This thesis describes a management décision-support system for dairy grazing systems based on simulation and multiple criteria decision-making (MCDM) models.Appropriate selection of holistic management strategies for livestock farming systems requires: 1) understanding of the behaviour of, and interrelations between, the different parts of the system, 2) knowledge of the basic objectives of the decision-maker managing such enterprise, and 3) understanding of the system as a whole in its agro-ecoregional context.Increasing economic and environmental pressures on livestock production systems have created the need to re-evaluate current management practices and to study new alternatives to ensure their sustainability. As a consequence, the demand for décision-support systems based on mathematical models has increased in the past years. Validated simulation models provide cost-effective means to represent the dynamics of the system and its components, while MCDM models allow for appropriate selection of resource allocation strategies depending on the different objectives and management 'styles' of particular individuals. Integration of both mechanisms provides the necessary elements for efficient décision- support at farm or ecoregional level.A décision-support system based on these techniques has been built to represent pastoral dairy production systems. The biological aspects (grass growth; grazing; digestion and metabolism; animal performance, and herd dynamics) are represented by simulation studies under a variety of management regimes. The outputs from the simulation runs (such as pasture utilisation, stocking rates, milk yields, fertilizer use, etc.) are used as data input to the MCDM models, and the latter have been used to select the management strategies which make the most efficient use of the farm's resources (i.e. land, animals, pastures).Examples are given with reference to highland dairy farming in Costa Rica. Nevertheless, the model frameworks are generic and can be adapted to different farming systems or ruminant species. The effect of model formulation and sensitivity, different decision-maker objectives, and/or activity or constraint definitions on management strategy selection are analysed. Future areas of research to expand this work to other livestock farming systems and to integrate other related disciplines into this décision-support framework are also discussed