Agri-environmental indicators to assess cropping and farming systems. A review

International audience

anglais. Environmental impacts of agriculture cannot be always assessed by using direct measurements. Since the 1990s, numerous agri-environmental indicators were developed to assess the adverse effects of cropping and farming systems in the environment, such as water pollution, soil erosion, and emission of greenhouse gases. Here we present the different types of indicators developed during the last decade and review the progress of the methods used for their development. The application of different groups of indicators is discussed and illustrated by examples in the fields of nitrogen losses and pesticide risk: (1) indicators based on a single or a combination of variables related to farmer practices, (2) indicators derived from operational or more complex simulation models assessing emissions of pollutants, and (3) measured indicators linked directly to environmental impacts. The nitrogen indicator (IN) of the INDIGO method and the MERLIN indicator will be presented and used to illustrate the methodological discussion. We show that a good identification of the end-users, of the practical objectives of the indicator, and of the spatial and temporal scales is essential and should be done at a preliminary step before designing the indicator itself. The possibilities of deriving an indicator from a model and of setting a reference value are discussed. Several methods are also presented to study the sensitivity and the validity of agri-environmental indicators. Finally, several practical recommendations are made. As only few data are usually available at the regional level, several simple indicators should be used for assessing a given impact at this level. When more detailed information is available, indicators based on operational models can be useful to analyse the effects of several factors related to soil, climate, and cropping system on an environmental impact. In experimental studies, we suggest using both measured indicators and model-based indicators.