Two species of butterfly, Hamadryasferonia and H. amphinome, closely resemble each other in behavior, morphology, and upperside wing pattern, but differ in underside wing pattern: that of the former is mostly white, whereas that of the latter is mostly orangered. H. feronia is palatable; a caged male rufous-tailed jacamar (Galbula ruficauda) always attacked and ate it in feeding trials. In contrast, H. amphinome is unpalatable, and although it was sometimes attacked by the bird, it was always rejected after tasting (it was never eaten by the bird). However, the jacamar's response to these two butterfly species changed when their underside wing pattern was modified. Most individuals of H. feronia, whose white underside wing area was painted orange, were ignored or tasterejected by the jacamar. Thus, he responded to them as if they were H. amphinome. To investigate further the effect of color on prey selection in the jacamar, similar trials were performed with seven species of butterflies known to be palatable to the jacamar. Each butterfly was painted one of seven colors using felt-tipped marker pens. In addition to changing the color of the butterflies, the paint rendered them distasteful. After ingesting painted butterflies, the bird vomited, and tended to reject similarly painted individuals afterwards. Butterflies painted with black, orange, and red-colors generally associated with local unpalatable butterflies-were rejected more frequently than were those painted with green, blue, yellow, and brown-colors generally associated with local palatable butterflies. Individuals of naturally bright or light-colored butterfly species, on which the painted colors were better expressed, were rejected more consistently by the jacamar than were those of dull or dark-colored species. A maximum-likelihood model fitting the effects of species and color adequately describes the jacamar's response in these tests. The results show that this insectivorous bird, which frequently preys on butterflies, can associate the color of butterflies with their palatability. WARNING (APOSEMATIC) PATTERNS AND MIMICRY represent a major phenomenon of insect communities and may account in part for patterns of community diversification (Vane-Wright 1981; Gilbert 1983, 1984). Experimental studies on mimicry (Batesian & Mallerian) have shown that predators will learn to avoid distasteful prey models after unpleasant experiences with them, and can then generalize and reject other similar prey (mimics; reviewed by Rettenmeyer 1970; Ford 1975; Turner 1977; Brower 1984). Most demonstrated cases involve visual hunting predators and their prey, often birds and butterflies. It is suggested that the predators' generalization from model to mimic is based on prey species' shared physical attributes, especially the color pattern. The empirical studies that contribute to the present mimicry theory accounting for the resemblance among insects in nature have not simulated natural conditions in at least two important aspects. First, birds chosen for experiments designed to show that birds learn to discriminate against prey patterns associated with unpalatability are probably not representative of birds which constitute the most important selective agents in the field. Observational records (Marshall 1909; Carpenter 1942; Skutch 1983) and beak mark examinations (Carpenter 1933, 1937, 1939) suggest that a small subset of birds (such as bee-eaters, jacamars, and some flycatchers) that specialize in on-the-wing capture of mobile diurnal insects, prey more frequently on butterflies than do omnivorous foliage and ground gleaners such as chickens, quails, jays, tits, starlings, and tanagers. However, in past studies on this subject, the birds used in feeding trials have been taken almost exclusively from the latter group. Birds vary in their discriminatory and learning abilities as well as in their abilities to capture, handle, and physiologically tolerate a given prey type (Swynnerton 1915; Rothschild 1964; Rothschild & Kellett 1972; Alcock 1971, 1973; Fink & Brower 1981; Evans 1984). It is thus necessary to evaluate the role of specialized insectivorous birds as a major selective agent on flying insects in which mimicry is particularly well developed (Turner 1977). Second, prey items have typically been presented dead or immobilized, thus eliminating any behavioral element possibly involved in predator-prey interactions. But in nature both color pattern and behavior lead to the detection of prey by visual hunting predators, and prey behavior has repeatedly been demonstrated, in diverse predatorprey systems, to be an important cue initiating a predator's response (Curio 1976). The movement and behavior of prey not only indicate its presence but also convey signals which either attract (e.g., escape behavior) or discourage (e.g., normal and regular movement) the attention of potential predators. This paper investigates the response of a specialized insectivorous bird, G. ruficauda, to one of its major prey types, the local butterflies H. feronia and H. amphinome, using free-flying rather than immobile individuals as prey. I Received 27 February 1985, revision accepted 21 May 1986. 20 BIOTROPICA 20(1): 20-3