Herbivory through the ages: Bias assessments to aid in widespread use of collections-based data

Herbarium specimens were historically collected to document the identity and location of plant species worldwide. Increasingly, massive digitization efforts have brought attention to these specimens as repositories for unique, largescale datasets describing plant-insect interactions and how they have shifted during the Anthropocene. In particular, herbarium specimens are increasingly harnessed for investigating the effects of climate on insect damage to plants. While herbarium specimens are poised to revolutionize the study of changing species interactions over decades to centuries, they also harbor biases associated with collector and curator preferences that may add noise to estimates derived from statistical models, and, perhaps more challenging, introduce systemic biases across the very axes that researchers aim to investigate. Here, we test for systemic biases in herbivory data gathered from herbarium specimens across two axes of interest to global change ecologists: taxa and time. We find that while estimates of herbivory are down-biased by collectors, who uniformly try to minimize herbivory, herbarium specimens reliably capture herbivory differences between species and across the plant phylogeny. Using a unique historical dataset of herbivory on oaks, we additionally show that herbarium specimens capture changes in herbivory over decades. Together, these assessments add to the growing evidence that herbarium specimens are reliable sources of herbivory data that can expand our understanding of plant-insect interactions and how they respond to anthropogenic change.