[Abstracted by M.K. Oliver:] Within the Great Lakes of Africa there exist the world's most diverse communities of freshwater fishes. In particular, the nearshore waters of Lakes Malawi and Tanganyika contain flocks of species, mostly members of the family Cichlidae, which may attain densities of up to 22 species and >500 individuals in a 50 m² area. While many of these cichlid species are very similar in general body morphology, there is high interspecific diversity in neurocranial morphology and dentition. This morphological diversity suggests that interspecific differences in feeding strategies may play an important role in reducing interspecific competition and maintaining high species diversity. Yet the results of studies investigating feeding habits of cichlids have been equivocal. Initial studies of the stomach contents of rock-dwelling fishes showed little or no evidence of food partitioning. This, along with laboratory studies, suggested that herbivorous cichlids, while being facultative specialists, are generally nondiscriminatory feeders. In reviewing the available literature on cichlid feeding habits, Greenwood (1981) concluded that "there is apparently complete interspecific overlap in environmental requirement." If this is true, then partitioning may occur only during periods of low food supply. More recent research has found significant variation of stomach contents among cichlid species. However, because stomach contents represent food consumed over a small time period and within a small area, these results do not conclusively demonstrate whether food partitioning is the exception or the rule. Other disadvantages of stomach content analyses include difficulty of identification and uncertainty over whether all observed stomach contents are assimilated to the same degree, or if some components, such as cyanobacteria, are indigestible. We present evidence for food resource partitioning among fishes of Lake Malawi, Africa, by comparing the stable isotope composition (delta ¹³C and delta ¹⁵N) of a number of nearshore fish species and their potential food sources. Because a fish's isotopic composition represents a spatio-temporal integration of the composition of assimilated food, this approach circumvents many of the problems in stomach analyses noted above.

We observed a strong distinction in delta ¹³C between nearshore benthic samples and planktonic samples; relative to the latter, benthic samples were much more enriched in ¹³C and displayed a wider range of delta ¹³C values. The wide range in Lake Malawi is likely due to differences in the severity of CO₂ limitation caused by large differences between pelagic and benthic photosynthetic rates. The delta ¹³C values measured for nearshore fishes covered a range nearly as broad as that measured for potential food sources. This broad range of isotopic values suggests that these nearshore species do not feed indiscriminately. Even within a closely related group of "herbivores," there are large differences in feeding strategies. Pseudotropheus tropheops is an obligate periphyton feeder, while its congener, P. zebra, has a strong preference for plankton. As expected, the semi-pelagic planktivore, Engraulicypris sardella, had a delta ¹³C similar to that measured for zooplankton and suspended particulate carbon. Other species filled the range between planktivore and obligate benthic feeder.

While species separation was greatest along the delta ¹³C axis, delta ¹⁵N also proved useful in elucidating dietary differences. Several species that were poorly separated on the delta ¹³C axis were well separated on the delta ¹⁵N axis. For example, the difference in delta ¹³C between Oreochromis squamipinnis and Taeniolethrinops praeorbitalis is < 0.3 per mille, but the delta ¹⁵N difference is 2.6 per mille.