Extended Data Fig. 2: Sensitivity analysis of the canopy fogging undersampling.

Effect of canopy arthropod undersampling on the energy fluxes in four land-use systems (dark green – forest, light green – jungle rubber, orange – rubber, yellow – oil palm; n = 8 sites per system, standard error is shown as the variation measure). We tested how the maximum tree height for efficient fogging affected our results. Average and 90th quantile (Q90) of tree height were calculated for each plot (a). The quantiles were further used to estimate bias in energy flux estimations aboveground (solid line) and belowground (dashed line), assuming that fogging works efficiently to assess canopy arthropods up to a certain tree height (models were run for each additional metre starting from 14 m as a ‘large bias’ and up to 22 m as a ‘small bias’; b). Land-use effect was more pronounced if we assumed a large bias due to higher trees in rainforest, than in plantations (n = 8, 1 SD variation; c).