Aquatic Mammals 1999, 25.2, 79–90 Consumption rate, food preferences and transit time of captive giant otters Pteronura brasiliensis: Implications for the study of wild populations Sarah K. Carter 1 , Fernando C. W. Rosas 2 , Andrew B. Cooper 3 and A. C. Cordeiro-Duarte Instituto Nacional de Pesquisas da Amazônia, Laboratório de Mamíferos Aquáticos, Caixa Postal 478, Manaus, AM 69011-970, Brazil 1 Current address: Washington Cooperative Fish and Wildlife Research Unit, School of Fisheries, Box 357980, University of Washington, Seattle, WA 98195-7980, USA 2 Current address: Centro de Estudos do Mar, Universidade Federal de Paraná, Av. Beira Mar s/n, Pontal do Sul, Pontal do Paraná, PR 83255-000, Brazil 3 Quantitative Ecology and Resource Management, Box 357980, University of Washington, Seattle, WA 98195-7980, USA Abstract Food consumption, food preferences and transit time of digesta were determined in captive giant otters, Pteronura brasiliensis, at the National Institute of Amazon Research (INPA), Manaus, Brazil. Food consumption of an adult female was 0.0997 kg 1 day 1 . Giant otters showed sig- nificant and varied preferences for the single Siluriformes (catfish) and various Characiformes species offered. The adult female preferred Anostomidae and Hypophthalmus (catfish), fol- lowed by Triportheus. Myleinae (pacu) were the least preferred, and other species of Characiformes offered were intermediate between Triportheus and Myleinae but not different from one another. The subadult male preferred Psectrogaster, Potamorhina and Semaprochilodus, followed by Hypophthalmus and finally Hemiodontidae. Within species, larger fish are chosen significantly more often than smaller fish (78.7% of trials, P =0.002). Species, size and individual otter significantly affected the percentage of times fish offered were consumed completely (P<0.016). Overall, most fish were consumed com- pletely beginning with the anterior end. Characoids were consumed completely more often than sil- uroids (77.8% vs. 38.6% of trials, P<0.0001), but the percentage of times different characoids were consumed completely varied (range 0–100%). Small fish (5–15 cm) are more likely to be consumed completely than medium (15–25 cm) fish (84.9% vs. 80.2% of trials, P<0.02). Transit time of particulate markers averaged 3.13 h. Captive preferences are compared to diets of wild otters in the region (central Amazônia), and implications of study results for determination of food habits in wild otters using scat analysis are discussed. Introduction The giant otter, Pteronura brasiliensis, is endemic to South America, inhabiting remote areas of the Amazon, Orinoco and La Plata river basins from Venezuela south to southern Brazil (Carter & Rosas, 1997). Giant otters are classified as vulner- able by the World Conservation Union (IUCN), and are subject to increasing threats to the quantity and quality of their habitat from development, deforestation and exploitation of natural resources such as gold and oil in the Amazon and other regions of South America (Carter & Rosas, 1997). Giant otters are primarily piscivores, fishing in the margins of rivers and lakes (reviewed in Carter & Rosas, 1997). Food habits studies of giant otters are most often based on identification and quanti- fication of hard parts of prey species found in the feces (e.g. Laidler, 1984; Rosas et al., in press). Occasionally direct observation is possible (e.g. Duplaix, 1980; Brecht-Munn & Munn, 1988), but this method is logistically difficult, costly and time consuming. Studies which utilize scats for diet determination are less intrusive and less costly. Uncertainty is introduced into results from such studies because hard parts found in scats depend on the consumption method of the prey, the rate at which parts pass through the animal’s digestive system, the rate at which prey remains decay and other factors (Pierce & Boyle, 1991). Few giant otters are maintained in captivity worldwide, and even fewer in regions within their  1999 EAAM