Katja Poveda, Daniel S. Karp, Rebecca Chaplin‐Kramer, Tim Luttermoser, Ricardo Pérez-Álvarez, Megan E. O’Rourke, Emily A. Martin, Heather Grab
Land-use change has altered the composition of our landscapes to favour agriculture, negatively affecting biodiversity and ecosystem services. However, the links between landscape composition, pest control and yield remain unclear. Using a global structural equation model of 116 studies from 28 countries, we tested three hypotheses: the 'natural enemy hypothesis', that natural areas increase natural enemies and suppress pests; the 'resource concentration hypothesis', that more agriculture increases pests; and the 'agronomic quality hypothesis', that agriculture-dominated landscapes occur in high-yielding areas. Results show landscape composition affects yield directly and indirectly through crop, herbivore and natural enemy traits. At larger scales (~1250 m), natural habitats increase yield, but at smaller scales (~250 m), yields decline with more surrounding natural habitat, likely due to lower agronomic quality or edge effects. Our findings suggest that we can harness the positive effects of natural areas at large scales while mitigating drawbacks at smaller scales to promote sustainable production.