Figure legends:
Figure 1: a. Correlation between number of songbird species and individuals along an elevational gradient in the eastern Himalayab. Correlation between number of songbird species and arthropod abundance along the same gradient. Site at 200m and 2000m elevation examined further in this study shown in blue and yellow respectively. Based on data in (Price et al. 2014)
Figure 2 Venn diagram showing diet overlap at multiple taxonomic levels as assessed with molecular metabarcoding between weaver ants, birds at low elevations and birds at mid-elevations. Bird and ant silhouettes taken from phylopic.org. Anthony Caravaggi produced the bird silhouette.
Figure 3: (a-b) Data from 34 trees paired for size and species, where one tree had weaver ants and the other did not.a. Number of arthropods belonging to the orders Lepidoptera and Coleoptera b. Leaf damage estimated on 10 leaves distributed around the tree.
Figure 4: Number of arthropods (excluding ants and suborder Homoptera) before and one month after weaver ant removal and exclusion experiment. Data from 30 trees paired for size and species, with one member of each pair subject to weaver ant removal and exclusion. Grey lines and dots show control trees and purple lines and dots show treatment trees. Darker dots and lines indicate multiple overlapping points.
Figure 5 a. Difference in mean abundance of major arthropod orders in control and treatment trees one month after ant exclusion and removal (derived from the data in Fig. S7) b.Frequency of different arthropod orders in bird diets at low elevations (Proportion of individual birds with at least one sequence from that order, N = 18 individuals)