Development of a diagnostic model of the interactions between landscape structure, land transport infrastructures and emblematic species: the case of lynx in the Jura Massif
The aim of the Lynx project was to study the interactions between the landscape, the land transport infrastructures and the lynx population of the Jura mountains. It was organized around three components: a "biostatistics" component, an "expert" component and a component combining the two previous approaches.
The combination of the "biostatistical" approach and the "expert" approach has highlighted one important result: the collision risk map produced on the basis of expert opinion is less accurate than the one based on a purely analytical approach, based on the observed collision data. In addition, the protocol for collecting data from the expert panel (placing collision points on a map) has made it possible to apply methods of analyzing data from ecology and thus typology of the experts met, then refining the interpretations of the interviews practiced.
The analysis of experts is complementary to the biostatistical analysis because it allows, on the one hand, to target certain variables that could be ignored by the analytical approach, and on the other hand, to highlight certain areas. at a very specific risk. Indeed, because of the more or less good resolution of the data available for the biostatistical analysis, it is possible that some variables escape this analysis. The more pragmatic vision of the experts makes it possible to provide details that are sometimes difficult to take into account by an analytical approach (for example the engrillagement of motorways, the probable obstacle to the crossing of this road by a lynx). In addition, the experts are able to define areas of high risk of collision, which reflects their ability to select sites with a particular combination of ecological and structural factors that increase the risk of collision with a lynx.
As such, promoting a biostatistical approach to the detriment of an expert approach, or vice versa, is not a recommended strategy. Rather, we recommend combining these two approaches that seem complementary rather than exclusive to each other.
However, since the risk map derived from the expert points is less precise than that obtained from the collisions observed, and the experts use in a quarter of the cases the location of existing collisions (and at the same time provide the same information as the data). observed), we suggest establishing the risk map from only the collisions data observed, but to integrate previously defined variables with the help of problem experts. It is therefore important to create the panel of experts to meet. Indeed, the variability of the experts (their knowledge of the environment, the species, their involvement in the exercise that is asked of them, etc.) suggests that the very constitution of this panel can have consequences on the quality and relevance final results. For example, experts facing the field will provide more relevant and accurate information than others about habitats
conducive to collisions. In addition, the exercise asking experts to point out areas at risk of collision is relevant to the validation of the panel of experts, but also to give more or less weight to the interpretation of their interviews.
Lastly, the risk zones pointed out by the experts coupled with the results of the analytical approach may allow to set up a validation in the field of the results obtained. This validation is necessary to rule out possible forgetting of important variables that no expert would have noted during the interviews.
- Category: Researches 2008 - 2009