Vegetation structure

The results of the first Principal Component Analysis (PCA), including all trees (N=72) showed that vegetation structure starts to change soon after livestock is excluded. In just one year grass cover equalled that of long term exclosures, but its height was still shorter. The analyses retrieved two dimensions, which together explained 78% of the variance. PC1 (59% of the variance explained) was significantly correlated with all variables (Table S7) and defined a gradient between trees in farms with high vegetation cover and height (mostly grass) versus those with a higher percentage of short grass (i.e. ground). It showed that vegetation cover was higher beneath the canopies of oaks from which livestock had been excluded for one or ten years than in control trees. PC2 (variance explained of 16.7%) was correlated with four variables (Table S7) and separated areas with high cover of short and high grass (which corresponded to long-term exclosures) from those with a prevalence of medium-sized grass (Figure 2A).
The second PCA, performed including only the control oaks (N = 24) from Farms 1-3 (livestock present), showed that vegetation cover differed among them at their standard livestock densities (Figure 2B). PC1 (variance explained of 59.6%) was highly correlated with all variables, and defined a gradient between areas with short grass and those in which overall grass cover and height were higher. PC2 explained less variance (15.8%) and was correlated with only two variables that differentiated areas with high versus medium-sized grass (Table S7). Summarizing, in Farm 1 the grass beneath the canopies was scarce and shorter than in the rest, and in Farm 3 there was a high cover of grass which, on average, was also taller.