A mathematical programming model is developed and associated to a spatial pattern index (Ripley L-function) to analyse the optimal reserve design and implementation for the Little Bustard conservation in Plaine de Niort. The model structure corresponds to three spatial levels, fields, farm and landscape. Simple in terms of area representation it is detailed in term of farm behaviour and spatially explicit. The model is applied in a normative and in a positive way. The major findings of the normative approach relate to the trade-offs between the reserve pattern and its cost. It was found that the environmentally optimal reserve, which is randomly dispersed across the zone, is the most costly one. Within the positive approach, it is illustrated that the various reserve patterns generated within the normative approach can be obtained through relatively simple uniform contract structures. The most effective contract structure is a degressive set of two payments enabling all the farms to enrol at least a small share of their land.