Oceanic currents strongly affect the movement of larvae within marine metapopulations.  Consequently, ecologists attempting to identify source subpopulations that are net exporters of larvae often attempt to simulate oceanographically-driven patterns of larval exchange.  Such efforts commonly assume that within-population processes (mortality, fecundity) are spatially constant, so source-sink status is largely a function of disperal processes.  However, the predators of the species being modelled usually also have planktonic larvae which are affected by the same oceanic currents, potentially producing correlations in the recruitment (and thus abundance) of prey and their predators.  In the Virgin Islands, the recruitment of a wrasse and one of its major predators, a grouper, show just such a correlation.  Moreover, this correlation leads to substantial spatial variation in wrasse mortality: wrasse at high-recruitment (= high-predator) sites experience more intense density-dependent predation.  I explored the potential consequences of this phenomenon using simple metapopulation models.  When prey and predator larval supply are not correlated (the standard assumption in most metapopulation models),  patches supplied with a larger proportion of the larval pool make a greater contribution to the metapopulation than patches receiving fewer larvae.  When prey and predator settlement are strongly coupled, the opposite is true: subpopulations with lower rates of larval supply (above some minimum) are more essential to the metapopulation.   Future attempts to identify source populations (such as when selecting sites for protection in marine reserves) could profit from a multispecies perspective that accounts for spatial correlations in the dispersal of predators and their prey.