Detection of direct versus indirect effects: were experiments long enough?

To evaluate the hypothesis that indirect effects generally take much longer to become evident in manipulative studies of community regulation than do direct effects and thus may often be missed, I studied the effect of experiment duration in a survey of marine intertidal interaction webs. Contrary to expectation, indirect effects appeared either simultaneously with direct effects or shortly after direct effects were evident. While experiment durations varied greatly, on average most direct and indirect effects became statistically significant within the first 20%-40% of the total experiment duration. Further, the duration of most experiments appeared sufficient so that most indirect effects that would be generated by the manipulation could be observed. On average, a period of "constancy" (i.e., of no further change) lasting roughly 20%-60% of the total experiment duration occurred after the last indirect effect was observed. Experiment duration did not vary with web species richness, which suggests no tendency to perform manipulations for more (or less) time in more complex webs. The number of indirect effects per species did not increase with increasing experiment duration, nor did the number of longer interaction chains (four species vs. three species), which suggests no trends for increased complexity of indirect effects with longer experiments. Ecological theory states that, in interaction webs whose dynamics are imperfectly known, indirect effects may compromise the predictability of species manipulations. However, empirical results suggest that, despite incomplete knowledge of indirect effects, community dynamics may be more predictable than expected.