Scale and system dependencies of indirect effects of large herbivores on phytophagous insects: a meta-analysis

An increasing number of studies are being conducted to examine the density- or trait-mediated indirect effects of large herbivores on phytophagous insects. However, these effects are highly context dependent and no general trends have been made clear. We conducted a meta-analysis focusing on three factors capable of affecting detection of the indirect effects of large herbivores on phytophagous insects: type of response variable, experimental scale, and characteristics of study organisms. Overall, large herbivores exerted a negative effect on insects, a trend that was prominent in studies using insect abundance as a response variable. No particular trends were observed in studies using herbivory rate as a variable, and these studies often focused on plant trait-mediated effects more than density-mediated ones. Experimental scale affected the strength of indirect effects: within-year or individual tree level experiments did not follow any trends, whereas 1–10 year experiments or 0–10 ha scale experiments show a negative impact on insects. Characteristics of large herbivores and growth forms of transmitter plants also contributed to variations in the observed effect size; negative effects were reported in livestock-grassland ecosystems and neutral effects in tree-dominated systems. There was a close association between response variable, experimental scale, and characteristics of study organisms, and these effects jointly contributed to the apparent trends. To predict the impacts of large herbivores at ecosystem level, it is necessary to eliminate these biases arising from study design and to evaluate the effect on insect densities at large spatial and temporal scales.