Sea surface temperature (SST) is a useful proxy for phytoplankton growth rates22, phytoplankton community structure31 and total global primary production3233, as it integrates many of the variables influencing phytoplankton growth. Dissolved carbonate concentrations also vary with SST; therefore, as all of the variables influencing Ep are either directly or indirectly associated with SST, SST should be a useful proxy for Ep and thus plankton δ13C values, as seen in the strong latitudinal δ13C gradients14. The algebraic relationship between SST and δ13C values is complex (e.g.21), and varies spatially with climatic and oceanographic conditions. In any single area, however, temporal variations in SST will result in temporal fluctuations in Ep and therefore will be coincident with variations in δ13C values of plankton. Thus the regions occupied by feeding animals are more likely to be regions where there is high co-variance between time series of tissue δ13C values and SST than those regions showing little or no covariance3435. This approach assumes that animals return to the same area over the duration of the time series, but has advantages over the fixed isoscape approach in that no prior knowledge of baseline isotope values is needed, and the arithmetic relationship linking the controlling variable to the tissue isotope value is free to vary with location.