Evaluating an environmental indicator: Case study of MERLIN, a method for assessing the risk of nitrate leaching

摘要

The need to achieve acceptable levels of nitrate in drinking water has led to the development of simulation models and indicators for assessing the environmental performance of agricultural practices. These indicators are necessarily based on simplifications in order to meet the practical constraints of feasibility, but they should nevertheless meet scientific standards, especially as regards their validation. The overall objective of this paper is to evaluate the MERLIN indicator and sub-indicators, an assessment method developed by French agricultural advisors. This tool estimates and classes the risks of water pollution by nitrates, integrating farmer practices both during crop cultivation and in between two successive crops, as well as soil sensitivity to leaching. The evaluation was performed according to the methodological framework proposed by Bockstaller and Girardin [Bockstaller, C., Girardin, P., 2003. How to validate environmental indicators? Agric. Syst. 76, 639-653]: design, output and end-use validation. Design validation involved submission of the method to experts and checking the sub-indicators against literature. Output validation was carried out by comparing real values to indicator outputs. End-use validation was based on information gathered by users. In particular we compared output data from the original method with that of users that had adapted the method to their situation, in order to assess the consequences of these changes. The first step confirmed that the assumptions of MERLIN and its sub-indicators are scientifically sound. However, the weighting of the different sub-indicators raised questions. The second output validation step gave acceptable results for the EQUIF sub-indicator but the MERLIN test highlighted the need for additional experimental data before validation. This approach also showed that improvements in the precision of parameters do not necessarily increase the accuracy of the classification. The last step confirmed that the indicator is considered useful by decision-makers but also reveals that in some cases users adapt parameter values to their situation. This has lead to the production of a user guide which defines the method more clearly to avoid numerous adaptations by users.