Alpine environments are mosaics of micro-habitats, and as such are associated with high levels of genetic heterogeneity and local adaptation. Plants growing in such complex habitats are interesting to investigate the drivers of natural selection, particularly due to their dependence on extrinsic forces at fine spatial scales. Landscape genomic analyses are powerful tools for studying local adaptation to environments, where genetic variation in populations is associated with environmental variables describing local site conditions to detect candidate genes under selection using Genotype–Environment Associations (GEA). The development of this field has been spurred by improvements in resolutions of genomic and environmental datasets, increasing the power to search for putative genes under selection and to study local adaptation in non-model organisms. However, most studies do not fully account for the appropriate spatial scale of environmental predictor variables, with potential impacts on downstream analyses and subsequent interpretations of adaptations to environments. To address this issue, we investigate the spatial scale of natural selection, implementing multiscale frameworks into GEA analyses. Using a case study from the herbaceous plant Arabis alpina collected across four valleys in the western Swiss Alps, we demonstrate how to perform multiscale landscape genomic analyses using multivariate redundancy analyses, investigating the impact of different spatial resolutions of predictor variables on the detection of candidate SNPs involved with local adaptation. Our results highlight the importance of using a multiscale framework in landscape genomic studies to obtain a more complete understanding of how the environment is inducing local adaptation of alpine plants.