The evolution of resource use in herbivores has been conceptualized as an analog of the theory of island biogeography, assuming that plant species are islands separated by phylogenetic distances. Despite its usefulness, this analogy has paradoxically led to neglecting real biogeographical processes in the study of macroevolutionary patterns of herbivore-plant interactions. Here we show that host use is mostly determined by the geographical cooccurrence of hosts and parasites in spider mites (Tetranychidae), a globally distributed group of plant parasites. Strikingly, geography accounts for most of the phylogenetic signal in host use by these parasites. Beyond geography, only evolutionary transitions among major plant lineages (i.e., gymnosperms, commelinids, and eudicots) shape resource use patterns in these herbivores. Still, even these barriers have been repeatedly overcome in evolutionary time, resulting in phylogenetically diverse parasite communities feeding on similar hosts. Therefore, our results imply that patterns of apparent evolutionary conservatism may largely be a byproduct of the geographic cooccurrence of hosts and parasites.evolutionary conservatism | geographic signal | host use evolution | parasite-host interactions | spider mites E cological interactions are commonly thought to be evolutionarily conserved (1). This idea roots back to Darwin and is based on the assumption that traits regulating interactions are mainly an ancestor's legacy. Following this rationale, it is expected that phylogenetically related species will also interact with partners that are phylogenetically clustered. Plants and the herbivores that feed on them constitute a large amount of the Earth's terrestrial biodiversity. However, the processes behind the origin of such diversity remain controversial (2). Related plant hosts tend to harbor similar communities of both herbivores (3-5) and pathogens (6, 7). Still, host shifts may be favored when they involve distantly related plants producing similar chemical compounds as defense against parasites (8). In fact, macroevolutionary evidence shows that host shifts of parasites to unrelated plants are common (9-13), questioning the ubiquity of evolutionary conservatism.These patterns of host use in parasites have led to the formulation of the "parasite paradox," which states that although parasites are generally specialized on a given (set of) host plant(s), host shifts to distantly related hosts occur frequently (14, 15). Further, the frequent shifts between evolutionarily distinct hosts observed in parasites have led to the analogy of considering plant species as resource islands separated by evolutionary distances (16)(17)(18). In this metaphor, related plants constitute resource archipelagos within which colonization of new resource islands entails few costs. Among such archipelagos, however, colonization processes may be costly, potentially leading to disruptive selection and hence to speciation via host shift (19,20). Paradoxically, framing the evolution of plant host use ...