Comparative studies have suggested that fruit traits, such as color, size, and protection, have evolved as covarying character complexes (''dispersal syndromes'') in response to selection by frugivorous dispersers. However, many comparative studies of disperser-specific syndromes have used species as sampling units, a method which implicitly assumes that character complexes evolve de novo in each species. This approach overestimates the number of times a character complex has evolved because covariation that results from common ancestry (plesiomorphy) is confounded with covariation across independent lineages (convergence). We compiled data on fleshy fruit traits from five regional floras to test the hypothesis that fruit traits form character complexes which covary independently of phylogeny (i.e. across lineages). Our results suggest that such character complexes are rare, and that analyses of covariation among these character complexes are extremely sensitive to the investigator's choice of sampling unit. When syndromes derived from observations of the foraging behavior of frugivores are analyzed using species as sampling units, our data show significant associations among traits at four out of five locations. In contrast, when genera are used as sampling units, there is no significant association among traits at any of the sites. Theories of coevolution between fruits and frugivores have proposed that fruit morphology is a response to selective pressures exerted by frugivorous birds and mammals. If these frugivores have shaped fruit morphology into dispersal-related character complexes or syndromes, then traits associated with a syndrome should be absent or much reduced in frequency in regions lacking that guild of dispersers. To test this prediction, we examined the flora of New Guinea, which lacks primates and other diurnal mammalian frugivores, and found no difference in the frequency of traits associated with dispersal by diurnal mammals.