Population dynamics
Confirming the first prediction, the landscape metrics indicate that the cushion individual distributions in low-elevation communities (i.e. , experiencing warmer conditions) are more fragmented than that in high-elevation communities (i.e. , experiencing colder conditions; Figure 2). Specifically, we found that four key metrics areca. 40–90% lower in warmer communities than in colder communities (Table S1). Two are the total cushion patch area (CA) and the mean cushion patch area (Area_Mn), the other two are the percentage of core area of the landscape containing cushions (CPLAND) and percentage of like adjacencies of patch types (PLADJ). However, the number of cushion patches (NP) and edge density (ED) in warmer communities are ca. 30% - 1000% higher than these parameters in colder communities (Table S1). Additionally, a large proportion of the area of the warmest (i.e. , lowest elevation) community was previously occupied by individual cushions (now traces of dead cushions), but this proportion decreases with decreases in temperature (i.e. , elevation) and no cushion traces were detected in the coldest (i.e. , highest elevation) community (Figure 1). All above results suggest that a large number of individual cushions will be excluded due to ongoing climate warming (Figure 1), resulting in more rapid population turnover.