This paper discusses some of the unique aspects of the interactions of livestock with the natural resources base. It argues the nature of these interactions make the use of a systems approach of particular importance when planning and executing livestock research if it is to be of any relevance to farmers in developing countries. The key issues are illustrated in two contrasting case studies.

In Nepal, shifts in land use patterns have led to marked changes in the availability and use of fodder resources. Farmers indicate that they are no longer able to use these optimally. Fodder from at least 90 different types of tree (some of which have yet to be properly classified botanically) are used to supplement the diets of buffalo, cattle and goats during dry seasons when food is in short supply. The nutritive value of each of these is affected by a wide range of management and environmental factors. Furthermore, diet selectivity means that fractions consumed differ markedly amongst both tree species and classes of livestock. It is clearly not feasible for researchers to evaluate this diversity using existing in vitro or in vivo indicators of nutritive value. Initial studies suggest that this variability is implicitly catered for in farmers' own assessments of relative nutritive values. A mechanistic understanding of this indigenous technical knowledge and the development of appropriate techniques for integrating it with models of the biological responses of animals to nutrients might, therefore, allow more effective assessment of strategies for tree fodder utilization.

Highland dairy systems in the humid tropics of Costa Rica (3000+ mm/year rainfall) could be very productive but suffer from a substantial dependence on imported inputs. Concentrate use is very high causing substantial underutilization of grazed pasture (the local resource), which is available year-round under these high-rainfall conditions. Over-supplementation with protein, which is the most expensive nutrient in the diet, at levels in excess of 200% of the requirement has been reported widely in these regions. Farmers want new strategies that enable them to manage their land and other resources in an alternative way, in order to be less dependent on grains and other imports. A decision-support system based on simulation and multiple criteria models representing a dairy farm has enabled the design and development of such strategies. Using this approach, farmers' objectives can be incorporated with a holistic understanding of the farming system in terms of management practices that will permit demand-driven animal production research at the farm level.