Re-examining Kozlov et al.: Estimation of biases in museum specimen data requires robust sampling

Kozlov and colleagues1 call into question the application of herbarium specimens to quantify historical patterns of herbivory2-5. It is already widely appreciated that collectors of herbarium specimens may tend to avoid insect damage, thus making herbivory estimates from herbarium specimens potentially down-biased2. However, Kozlov et al. additionally suggest that variation in sampling selectivity among collectors and curators may lead herbarium specimens to misrepresent patterns of herbivory in nature. The authors sought to quantify these biases by collecting and contrasting insect herbivory data across 17 plant species from herbarium versus standard field ecological sampling procedures, and then assessed the selection of these specimens by curators. They concluded that herbivory estimates from herbarium specimens are highly variable, rendering them an inaccurate representation of herbivory in nature. Our re-analysis of Kozlov et al.s data suggests that, in contrast with their results, herbarium specimens indeed provide a useful record of herbivory as long as sample sizes are appropriate. In addition, we assert that by arguing that herbarium specimens are "distorting mirrors", Kozlov et al.s conclusions fundamentally overstep their data, which narrowly assesses biases across species. Kozlov et al. argue that herbarium specimens are inaccurate data sources, but fail to characterize the specific circumstances under which assumed biases would apply. Thus, Kozlov et al.s data do not support their main premise, and the authors extrapolate beyond the specific biases investigated in their study; we believe their contribution does a disservice to researchers interested in exploring the potential value of herbarium specimens for studying herbivory through time.