Fig. 4. Observed responses of individual traits at the guild-level (CWM) to land-use intensity, shown as estimated parameter +/− 95% confidence intervals.

The responses shown were extracted for each trait individually as the slope ± confidence interval of a linear model with the CWM trait as a response to land-use intensity after correction for other environmental covariates (see Figure S1). P-values (two-sided t-tests) were adjusted for multiple testing. Land use intensity explained between 0 (non-significant traits) and 40% (Aboveground primary consumer arthropods: voltinism) of the total variation in individual trait CWM (Table S2). Belowground guilds are shown over a brown background, aboveground guilds over blue. A standardised response below 0 (blue lines) indicate a negative response to land-use intensity, in accordance with hypotheses (expected higher trait values in ‘slow’ communities). Responses above 0 (green line) indicate a positive response to land-use intensity, in accordance to hypotheses (expected higher trait values in ‘fast’ communities). Red lines indicate traits with opposite responses to that hypothesised. 95% intervals crossing the 0 line indicate no significant response (grey lines). The number of replicates (sites with available data) are shown for each guild. Collembola: 138 for voltinism, 140 for the other traits. Arthropods belowground (primary consumers): 111 for feeding generalism, 136 for other traits. Icons were acquired and adapted from Phylopic.org (artists: M. Dahirel, B. Lang, M. Crook, J. A. Venter, H. H. T. Prins, D. A. Balfour, R. Slotow, T. M. Keesey, A. A. Farke, Y. Wong, G. Monger).