Sampling design
Sampling in the PASTUS database was designed according to a stratified random scheme, where samples were selected at random within strata. This process was done using the software ArcMap 10 (ESRI, Redlands, CA, USA). The basis for randomization was the map of habitats of Catalonia 1:50000 (Carreras, J. & Diego, 2006) for the Eastern and Central sectors, the map of habitats of Madres-Coronat 1:10000 (Penin, 1997) for the North-Eastern sector and the land use map of Navarra 1:25000 (Gobierno de Navarra, 2003) for the Western sectors. Four variables were initially considered for sampling stratification within each sector: altitude (< 1800 m; 1800-2300 m; > 2300 m), slope (0-20º; 20-30º; > 30º), macrotopography (mountain top/southern-facing slope; valley bottom/northern-facing slope) and grazing management (sheep grazing; cattle grazing; mixed grazing). Accordingly, we determined a set of homogeneous grassland patches by crossing the stratification variable layers. Grassland patches were then listed by type and arranged within each list randomly to determine sampling priority. At least one to two replicates of each patch type were sampled.
In each sampled grassland patch, a 10 x 10 m2 plot was systematically placed in the middle of each homogeneous grassland patch, including a particular plant community. Soils and vegetation were sampled inside this 100 m2 plot, and environmental variables assessed (see Rodríguez et al., (2018) for additional details about sampling design).