Identification of keystone species
Keystone species are defined here as plant taxa that play a major role in providing a food source for the orthopteran herbivores (Millset al. 1993; Power et al. 1996). We identified keystone species using custom R scripts, submitting each network, preliminarily transformed into an igraph object (Csardi & Nepusz 2006), to a sequential and random removal of plant species. Insect species were considered to be extinct upon loss of all the plant species they feed on. Plant species were removed until all insect species became extinct. This was repeated for n*(n–1) simulations, with n equaling the total number of plant taxa in the network. The mean number of secondary extinctions caused by plant removal, to which we refer hereafter as the keystone score, was then calculated for each plant species. To examine the distribution and the keystone score of species within the plant functional space at the low (<1050 m a.s.l.) and high elevation (>2000 m a.s.l.), we performed a principal components analysis on plant traits with the function dudi.pca from the ade4 package (Thioulouse et al. 2018). We compared the distribution of species with different keystone scores in the functional space of plant traits of species at the lowest and highest elevation sites. For both elevation classes, we extracted plant species based on their presence within the corresponding elevation range. We further determined the 10 species with the highest keystone scores for each network, averaged their weights for each elevation class and placed them in the functional space of plant trait.