LAUSR

Infectious diseases are rapidly emerging and many are increasing in incidence across the globe. Processes of land‐use change, notably habitat loss and fragmentation, have been widely implicated in the emergence and spread of zoonoses such as Lyme disease, yet evidence remains equivocal. Here, we discuss and apply an innovative approach from the social sciences; instrumental variables, which seeks to tease out causality from observational data. Using this approach, we revisit the effect of forest fragmentation on Lyme disease incidence, focusing on human interaction with fragmented landscapes. Although human interaction with infected ticks is of clear and fundamental importance to human disease incidence, human activities that influence exposure have been largely overlooked in ecology literature. Using county‐level land‐use and Lyme disease incidence data for ~800 counties from the northeastern United States over the span of a decade, we illustrate (a) that human interaction with fragmented forest landscapes reliably predicts Lyme disease incidence, while ecological measures of forest fragmentation alone are unreliable predictors and (b) that identifying the effect of forest fragmentation on human disease entails addressing the feedback between Lyme disease risk and human decisions to avoid interaction with high‐risk landscapes. Synthesis and applications. Our innovative approach and novel results help to clarify the equivocal literature on the effects of forest fragmentation on Lyme disease and illustrate the key role that human behaviour may be playing in the ecology of Lyme disease in North America. Accounting for human activity and behaviour in the ecology of disease more broadly may result in improved understanding of both the ecological drivers of disease, as well as actionable intervention strategies to reduce disease burden in a changing world. For example, our model results indicate that forest fragmentation by human settlement increases Lyme disease incidence, which has practical implications for land‐use policy aimed at disease reduction. Specifically, our model suggests land‐use regulations that reduce parcel size would be an actionable approach to reduce Lyme disease transmission for policymakers concerned about increasing Lyme disease incidence in the northeastern United States.