Figure 1. Ecosystem process rates in primary (green) and logged (orange) forest in response to experimental removal of invertebrates, fungi or vertebrates.

Ecosystem processes were quantified at three trophic levels represented by (a,b) leaf litter decomposition rate, (c,d) seed disturbance, defined as the combined removal and/or predation rate; and (e,f) invertebrate predation rate. Symbols indicate the taxa that contributed to the rates displayed. Absolute values (mean±95% CI) of the ecosystem process rates are presented in the left-hand column and are measured as (a) the slope of a linear regression model relating log_e-transformed litter mass (g) as a function of log_e-transformed time (number of days), (c) the proportion of experimental seeds removed or predated per day, and (e) the proportion of experimental mealworm larvae predated per day. Letters indicate habitat × treatment categories that did not significantly differ from each other (P<0.05). There was a significant treatment × habitat interaction for all three ecosystem processes, demonstrating that the role of invertebrates was stronger in primary than logged forests. In the right-hand column (panels b,d,f), values represent the proportional change in ecosystem process rates relative to control sites (calculated from data presented in the left-hand column). Values <1 (dashed line) indicate functions whose rate is reduced following the exclusion of a taxon; smaller values indicate larger reductions in the rate and hence a stronger contribution of that taxon to delivering the ecosystem process. Posthoc significance tests were used to examine the treatment × habitat interaction effects. For example, in panel b, the asterisk indicates that the effect on decomposition of excluding invertebrates was significantly (P<0.05) larger in primary forest than in logged forest. NS indicates nonsignificant interactions.