Confirmation bias leads to overestimation of losses of woody plant foliage to insect herbivores in tropical regions

Confirmation bias, i.e., the tendency of humans to seek out evidence in a manner that confirms their hypotheses, is almost overlooked in ecological studies. For decades, insect herbivory was commonly accepted to be highest in tropical regions. By comparing the data collected blindly (when the observer was not aware of the research hypothesis being tested) with the results of non-blind studies (when the observer knew what results could be expected), we tested the hypothesis that the records made in the tropics could have overestimated community-wide losses of plant foliage to insects due to the confirmation bias. The average loss of leaf area of woody plants to defoliating insects in Brazil, when measured by a blind method (1.11%), was significantly lower than the loss measured in non-blind studies, both original (5.14%) and published (6.37%). We attribute the overestimation of the community-wide losses of plant foliage to insects in non-blind studies to the unconsciously preconceived selection of study species with higher-than-average levels of herbivory. Based on our findings, we urge for caution in obtaining community-wide characteristics from the results of multiple single-species studies. Our data suggest that we may need to revise the paradigm of the highest level of background insect herbivory in the tropical regions. More generally, we argue that more attention should be paid by ecologists to the problem of biases occurring at the pre-publication phases of the scientific research and, consequently, to the development and the wide application of methods that avoid biases occurring due to unconscious psychological processes.