Design of farming systems for low input conditions: principles and implications based on scenario studies with feed allocation in livestock production

Abstract

This study addresses the issue of designing farming systems for low input conditions. By focusing on the problem of feed (resource) allocation in livestock systems it provides clues for the design of alternative agricultural systems and systems in general. Linear programming (LP) is used to examine system behaviour under conditions of varying feed resource qualities, for individual animal production levels that range from 0.75 to 3.00 times maintenance. Milk yield, animal numbers and amounts of feed offered or refused are measures of system performance in two hypothetical cases representative for actual farming systems. Case 1 considers available feed as one aggregated resource, with an average nutritive value ranging from that of straw to that of high quality forage. It establishes the individual animal output level that is required to achieve maximum total system output from a given feed resource. In Case 2, animals of different production levels are allowed to select between two feed resources that, through proper definition of the their proportions on offer, on average represent the same feed quality scale as in Case 1. The possibility of selection, however, allows the nutritive value of the actual intake to differ from the average nutritive value of feed offered. Better feed and higher (potential) individual animal output tend to increase total system output in terms of milk, by reducing the number of animals. Nevertheless, the term 'damning objective' is introduced to express that high targets for subsystem output reduces total system output if subsystem requirements exceed resource availability. Moreover, heterogeneity of (feed) resources can increase system output by using more production units with lower individual output. The results are tested against farmers' practice and situations reported in literature. Implications for the design of sustainable systems and further research are discussed.