This paper investigates impacts of the creation of Marine Protected Areas (MPAs), in both economic and biological perspectives. The economic indicator is defined as the sum of discounted benefits derived from exploitation of the resource in the fishery sector, assumed to be optimally managed. The biological indicator is taken as the stock density of the resource. The basic fishery model (C.W. Clark, Mathematical Bioeconomics: The Optimal Management of Renewable Resources, second ed., John Wiley and Sons, New York, 1990) will serve as a convenient benchmark in comparing results with those that are derived from a model of two patchy populations (cf. R. Hannesson, Marine reserves: what would they accomplish, Mar. Resour. Econ. 13 (1998) 159). In the latter, a crucial characteristic is the migration coefficient with describes biological linkages between protected and unprotected areas. A set of situations where both economic and biological criteria are enhanced, after introducing a MPA, is presented. These results are obtained with the help of numerical simulations. To cite this article: D. Ami et al., C. R. Biologies 328 (2005).

[fre] L'évaluation contingente est couramment utilisée pour donner un prix aux actifs naturels. Le traitement économétrique des valeurs obtenues à l'aide de cette méthode directe de révélation des préférences pose des problèmes du fait du nombre élevé (généralement autour de 50%) de valeurs nulles. Différentes méthodes de traitement des réponses égales à zéro, dans le cas d'une question ouverte, peuvent être envisagées. Trois modèles sont étudiés, testés économétriquement et comparés : le modèle de régression linéaire, le modèle de Heckman et le modèle de Tobin. On s'interroge également sur les comportements micro-économiques sous-jacents à chacun de ces modèles. L'objectif de cet article est de montrer qu'il n'est pas pertinent d'utiliser le modèle de Tobin car les valeurs nulles ne peuvent pas être traitées uniformément comme des valeurs censurées. [eng] Analysing Responses at Zero in Contingent Valuation by Dominique Ami and Brigitte Desaigues Contingent valuation is often used to put a price on natural assets. Econometric analysis of values obtained by this direct method of preference revelation raises problems due to the high proportion (generally around 50%) of zero values. Various methods of analysing responses at zero can be considered in the case of open-ended questions. Here three models are studied, tested econometrically and compared: the linear regression model, the Heckman model, and the Tobin model. The microeconomic attitudes underlying each of these models are also examined. The object of the paper is to show that use of the Tobin model is not relevant since zero values cannot be treated uniformly in the same way as censored values.