Laboratory feeding assays employing the common Canbbean wrasse Thalassoma bifasciatum were undertaken to determine the palatability of food pellets containing natural concentrations of crude organic extracts of 71 species of Caribbean demosponges from reef, mangrove, and grassbed habitats. The majority of sponge species (69%) yielded deterrent extracts, but there was considerable inter-and intraspecific vanability in deterrency. Most of the sponges of the aspiculate orders Verongida and Dictyoceratida yielded highly deterrent extracts, as did all the species in the orders Homosclerophorida and Axinellida. Palatable extracts were common among species in the orders Hadromerida, Poecilosclerida and Haplosclerida. Intraspecific variability was evident, suggesting that, for some species, some individuals (or portions thereof) may be chemically undefended. Reef sponges generally yielded more deterrent extracts than sponges from mangrove or grassbed habitats, but 4 of the 10 most common sponges on reefs yielded palatable extracts (Callyspongia vaginalis, Mycale laevis, Niphates erecta, Iotrochota birotulata), including those most commonly eaten by sponge-eating reef fish. The presence of symbiotic zoanthid cnidarians of the genus Parazoanthus in the tissues of otherwise palatable sponges had little effect on the deterrency of tissue extracts, indicating that these commensal polyps do not confer a chemical defense by association There was no relationship between sponge color and deterrency, suggesting that sponges are not aposematic and that color variation is the result of other factors. There was also no relationship between the toxicity of sponge extracts (as determined in previous studies) and deterrency, confirming the invalidity of previous assessments of chemical defense based on toxicity. Although chemical antipredatory defenses are important strategies for most Caribbean sponges, some common species appear to rely on other tactics.

Laboratory and field feeding experiments were conducted to assess the palatability to predatory reef fish of prepared foods containing natural concentrations of glass spicules from 8 specles of Caribbean reef sponges. Sponge species with high concentrations of spicules in their tissues, and with variable spicule morphologies, were chosen for the experiments. The presence of spicules did not alter food palatability relative to controls for any of the sponges tested. Analyses of ash content, tensile strength, protein, carbohydrate, and lipid content, and total energy content were conducted on tissue samples from 71 species of Caribbean demosponges from reef, mangrove, and grassbed habitats, and compared to previously reported data on the chemical defenses of the same species. There was no evidence to support the hypothesis that sponge species with palatable extracts have higher concentrations of inorganic structural elements, as measured by the mean ash content of their tissues. In addition, the tissues of palatable sponges were not different from those of chemically deterrent species with regard to mean tensile strength, protein content, carbohydrate content, and total energy content, but the tissues of chemically defended species did have a higher mean lipid content than those of palatable species. Sponges that lack chemical antipredatory defenses do not appear to compensate with structural or nutritional defenses, but may instead direct energy otherwise used for the production and storage of secondary metabolites to increased growth and reproduction.