Résumé : The cold-tolerant leaf beetle Gonioctena quinquepunctata displays a large but fragmented European distribution and is restricted to mountain regions in the southern part of its range. Using a RAD-seq-generated large single nucleotide polymorphism (SNP) data set (> 10,000 loci), we investigated the geographic distribution of genetic variation within the Vosges mountains (eastern France), where the species is common. To translate this pattern of variation into an estimate of its capacity to disperse, we simulated SNP data under a spatially explicit model of population evolution (essentially a grid overlapping a map, in which each cell is considered a different population) and compared the simulated and real data with an approximate Bayesian computation (ABC) approach. For this purpose, we assessed a new SNP statistic, the DSVSF (distribution of spatial variation in SNP frequencies) that summarizes genetic variation in a spatially explicit context, and compared its usefulness to standard statistics often used in population genetic analyses. A test of our overall strategy was conducted with simulated data and showed that it can provide a good estimate of the level of dispersal of an organism over its geographic range. The results of our analyses suggested that this insect disperses well within the Vosges mountains, much more than was initially expected given the current and probably past fragmentation of its habitat and given the results of previous studies on genetic variation in other mountain leaf beetles.