Hummel et al. (1988) hypothesised the concomitant existence of two separate food chains in the Westerschelde: a photo-autotrophic coastal food chain in the marine part and a heterotrophic chain in the brackish part. The present study intends to re-examine the hypothesis on the basis of recently published data. Biomass gradients of the important functional units along an estuarine transect were observed to differ from those reported by Hummel et al. (1988) in some important aspects. The biomodal primary production gradient reported by Spaendonck et al. (in press) does not resemble the phytoplankton biomass curve, gradually increasing from the sea to Antwerp proposed by Hummel et al. (1988). Estimates of mesozooplankton biomass were found to be about an order of magnitude lower than reported by Hummel et al. (1988) and to display a completely different and more complex spatial pattern. However, the new gradient found is more in line with the hypothesis of two food chains than the gradient reported by Hummel et al. (1988). In the macrobenthos the biomass peak in the brackish part reported by Hummel et al. (1988) could not be confirmed. This finding does not falsify the original hypothesis as the function of this detritus dependent macrobenthic fauna is largely taken over by the hyperbenthic mysids, a group of previously unknown importance in the system. The existence of two food chains is also supported by the gradients observed in fish and epibenthic invertebrates, functional units not addressed by Hummel et al. (1988). In the zone between the two different food chains the dominant animal groups of the pelagic system have only a low biomass, this is the nutrient rich desert of the title. The zone upstream of the Dutch-Belgian border supports no hyperbenthos, no epibenthos and no mesozooplankton because of the low dissolved oxygen concentrations (less than 40% saturation), but there is a prominent peak in the microzooplankton. Clearly, in the brackish part, the richness of most functional units can only be ...