Summary 1.Remote-sensing measurements of marine primary productivity are widely used to predict the distribution and movements of marine top predators, despite the fact that predators do not feed directly on phytoplankton but several trophic levels higher up the food chain. 2. To test for potential links between primary productivity and top-predator feeding hotspots, we assessed spatial match-mismatch across four trophic levels of the Benguela upwelling zone (south-east Atlantic). The food chain studied consisted of phytoplankton, zooplankton (copepods), pelagic fish (anchovies and sardines) and two top predators (Cape gannets Morus capensis and human fisheries). 3. Remote-sensing data of sea-surface temperature (SST) and chlorophyll_a concentration were used as indices of phytoplankton abundance throughout the study area. Copepod biomass and pelagic fish density were determined during at-sea surveys in the South African section of the Benguela using net tows and hydro-acoustics, respectively. Seabird (Cape gannet) home ranges and foraging zones were assessed from two Namibian breeding colonies (Mercury and Ichaboe) and two South African colonies (Lambert's Bay and Malgas) using global positioning system (GPS) tracking. Industrial fishing for anchovies and sardines was investigated using South African landing statistics and logbooks. 4. Our spatial analyses revealed a strong match of seabird at-sea habitat and zones of high primary productivity throughout the southern Benguela. Conversely, there was a marked spatial mismatch between copepods and pelagic fish, as well as between pelagic fish, seabirds and human fisheries: copepods were present in the southern Benguela but pelagic fish usually feeding upon them were located further east (Indian Ocean), outside of the Benguela sensu stricto . Consequently, the majority of these pelagic fish were out of reach for seabirds and fisheries confined to the southern Benguela. 5. Synthesis and applications . Our study demonstrates the impact of an ecosystem shift across one of the world's most productive marine ecosystems and highlights the limitations of using remotesensed patterns of primary productivity to interpret the foraging behaviour of marine top predators. These findings underline the importance of a better knowledge of food web spatial dynamics to support ecosystem-based fisheries management and the conservation of marine top predators.

No abstract