Résumé
It has been often shown that spatial distribution of species can be related to the main characteristics of their habitat. Usually, such relationships refer to point pattern analysis and try to determine whether the distribution of species is conditioned by their surroundings or not. They often seek correlations at a given place, date and scale and neglect the information potentially extracted from detailed surface pattern analyses that would require extended data. This study investigated the impact of landscape pattern and landscape context (of habitats) on carabid beetles assemblages in each place and on a continuous range of scales of a landscape. For this purpose, we develop a tool able to visualize and quantify the spatial variations of indices commonly used in point pattern analyses, such as contagion (heterogeneity) of a landscape and abundance (counts) of species. Using moving windows, we plotted contagion maps of two Brittany sites (France) and cross-correlation maps between the latter and carabid beetles spatial distributions, in order to study their local relationships. Associated landscape and insect index scaling profiles helped to interpret the correlations found. A rather good agreement with landscape ecology predictions over both sites has been found: significant relationships between land cover contagion and total carabid abundance on one hand, between agricultural intensification (mainly maize fields) and carabid body size on the other hand, have been observed. Nevertheless, detailed surface pattern analyses over a wide range of scales show quite high deviations from empirical results with no correlation, or even negative correlation, in some places between otherwise correlated indices.