Sampling design
Twenty transects were set up along each road at approximately equal elevational steps following the MIREN protocol (see Haider et al. 2022 for the published version of the protocol). Each of these transect was then divided in three 2 m x 50 m plots organized in a T-shape. The first plot was laid out parallel to the road, encompassing the vegetation directly impacted by the road-associated disturbance, and both the second and third plots were set up perpendicular to the road, respectively from 2 to 52 m and from 52 m to 102 m, covering the less disturbed vegetation communities adjacent to the road. Henceforth, we will refer to these plots as “roadside plots” and “adjacent plots”. While anthropogenic disturbance was almost always higher in the roadside plot, some of the adjacent plots were also subject to anthropogenic disturbances, especially so at lower elevations. The presence and total cover of all vascular plant species present in each plot were then visually estimated and classified using the following semi-continuous scale: 1 = <1%; 2 = 1 to 5%; 3 = 5 to 25%; 4 = 25 to 50%; 5 = 50 to 75%; 6 = 75 to 95%; and 7 = 95–100%. Plant species were also categorized as either native or non-native following local and regional floras, with species introduced in the country or mountain region after AD 1500 considered to be non-native (Haider et al., 2022). Additionally, average yearly soil temperature was extracted from the SoilTemp global maps of soil temperature at a 1 km resolution (Lembrechts et al. 2022). In total, our dataset included 894 plots spread out across 62 separate mountain roads and 11 regions.