Demonstrating changes in the structure of plant and animal communities over geological time scales and linking these changes to human impacts in the Holocene epoch would be an important contribution to the fields of ecology and conservation biology1. Lyons et al.2 claim to provide such evidence based on a decrease in the proportion of spatially aggregated species pairs, using co-occurrence data from assemblages spanning the past 300 million years. However, we suggest that apparent flaws in their predictions, assumptions, methods and interpretations undermine this claim, and we question the conclusion that the structure of communities has fundamentally changed during the Holocene.