Birds use a remarkable variety of materials in constructing their nests and almost always employ materials readily available near the nest site. Many species include insulative feathers in the nest lining to improve energy budgets of adults and young (Møller 1991, Winkler 1993, Lombardo et al. 1995). Among the species that incorporate large numbers of feathers in nests is the Fork-tailed Palm- Swift (Tachornis squamata; hereafter “palm-swift”) of tropical South America, which uses “fresh body feathers…of middle-sized birds, especially pigeons” (Sick 1948:170), supposedly collected as airborne detritus (Carvalho 1962, Collias and Collias 1984, Chantler 1999). Made of feathers and a small amount of plant matter, the nest is glued with the bird's saliva to the inside of dead, folded, pendant leaves of the widespread palm Mauritia flexuosa (Sick 1948) or, in far northeastern Brazil where M. flexuosa is absent, the palm Copernicia prunifera (B. M. Whitney pers. obs.). Although palm-swifts have a closely corresponding, wide distribution, few nests have been examined, because these palms tend to grow in swamps and it is difficult to inspect the hanging leaves. Carvalho (1962) provided detailed descriptions of nesting behaviors and chick development of palm-swifts based on observations of several nests constructed on the undersides of live leaves of the palm Livistona chinensis in the city of Belém, Pará, Brazil. This palm is exotic in the New World, but possesses leaves morphologically similar to those of the above-mentioned, native palms.
In September 1995, I found two palm-swift nests in a pile of dead M. flexuosa leaves gathered on the grounds of the Tropical Hotel in Manaus, Amazonas, Brazil, and noted that, as had been reported (Sick 1948, Carvalho 1962), they were made mostly of contour feathers of several other species of birds. The palm-swift is airborne during most of its life, alighting only inside palm leaves to roost and nest, and it never travels far from palm groves. This led me to wonder how the birds could possibly gather such a large number of small feathers in a circumscribed area. A few hours of observation revealed the answer: palm-swifts attack other species of birds flying near palm groves, forcibly ripping feathers from their backs (see Fig. 1). Across the palm-swift's extensive range, its modus operandi appears to be well ingrained: on calm mornings (but also at other times of day), palm-swifts attain heights, often in excess of ∼100 m, near their palm grove and circle, waiting. When birds from the size of piping-guans to that of pigeons come by in level flight, or even some smaller species with undulating flight (e.g., some parrots, woodpeckers, flycatchers, and thrushes), the palm-swifts rapidly stoop from above and behind, sometimes in tandem, striking their victim in the middle of the back and tugging at feathers with the bill for about 1–3 s to dislodge a mouthful; the legs are used only to help stabilize the attack. The birds may then play with feathers, allowing some to float free to be picked up in a subsequent pass, or carry them directly to nest sites. Most of the victims react only slightly, but some individuals, perhaps those with prior experience and most small birds, initiate evasive flight behaviors at the first stoop of the palm- swifts or quickly dive to hide in trees.
Most of the attacks I have observed were directed at parrots and pigeons. Two nests I examined near Presidente Figueiredo, Amazonas, Brazil, consisted almost entirely of contour feathers of Dusky Parrots (Pionus fuscus) and pigeons (Patagioenas spp.), and all other nests described to date have contained feathers mostly from these two families of birds. Pigeons and ground-doves must make especially difficult, perhaps even dangerous, targets for palm-swifts, because they are some of the fastest flyers among Neotropical birds and also roll erratically in the course of flight. Taking feathers from most parrots, on the other hand, is probably relatively uncomplicated, because parrots fly more slowly than most other above-canopy Neotropical birds, and the larger species, in particular, have less erratic flight trajectories and wing movements. Several species of Amazonian columbids, psittacids, and icterids (especially oropendolas) form large evening roosts, commuting in straight-line flight-paths twice daily, with many individuals regularly passing groves of M. flexuosa, where they present dependable feather sources for palm-swifts.
The piratic behavior of palm-swifts probably evolved as the most efficient means of predictably obtaining insulative material that increased survivorship of naked nestlings; selective pressure certainly would have been heightened during glacial epochs and is maintained today in southern and central Amazonia by regular austral cold fronts that last for several days. It is noteworthy that the tent-like, pendant leaves of palms that are the exclusive nest sites of palm-swifts are perhaps the most exposed to the elements (especially cold and wind) of any apodine nest sites worldwide (others being inside caves, rock crevices, hollow trunks, chimneys, buildings, elaborate tubes of plant material, etc.). These pendant palm leaves offer the overwhelming advantage, however, of being practically inaccessible to vertebrate predators, including primates and snakes. Construction of nests inside these dead palm leaves may have required the use of ultra-lightweight, flexible material with high insulative properties that could be structured with saliva. Feathers from other birds, especially those of pigeons and parrots, emerged as the evolutionarily ideal resource for these fast-flying, agile swifts. Indeed, many apodine swifts, in their very narrow wings and long, forked tails appear to be built for aerial piracy, which is remarkably convergent with frigatebirds (Fregata spp.; Thomas and Balmford 1995), which obtain a fair amount of their food by chasing and out-maneuvering other birds.
The fact that it takes a pair of palm-swifts one to three months to construct a nest (Carvalho 1962) is probably a corollary of the time it takes to pirate a sufficient number of individual feathers from other birds; only small amounts of plant material have been reported from any nest. Having observed palm- swifts attacking ∼50 species of birds representing 21 families (Table 1), including characteristically aggressive kingbirds (Tyrannus spp.), I expect that almost any bird flying by can become an unwilling feather donor. I have never observed an attack on a perched bird.
Other reports of birds taking feathers for their nests from the bodies of other, live birds involve House Sparrows (Passer domesticus) plucking feathers from doves and other birds in cities (Summers-Smith 1963, Stidolph 1974, Leruth 1984, Bell 1994); these are probably opportunistic endeavors by this notoriously adaptive species. Similarly, diverse species of birds, not including swifts, have been observed taking hairs from live mammals, even humans, for nest material (e.g., Bent 1946, Goertz 1962, George 1985, Cody 1991). The palm-swift, however, has evolved a highly efficient strategy and procedure for harvesting feathers of other species of birds as a reliable source of nest insulation. It is to be expected that palm-swifts would also take advantage of any easy source of suitable, windblown feathers (e.g., preening birds dislodging some feathers, raptors tearing apart avian prey as suggested by Carvalho , etc.).
Feather thievery from live birds as practiced by palm-swifts, which I term “kleptoptily,” is a novel form of kleptoparasitism. In birds, kleptoparasitism has heretofore been limited to piracy of resources such as food (Campbell and Lack 1985) or, in the well-known case of Piratic Flycatcher (Legatus leu- cophaius), finished nests of other birds. I predict that the closest Neotropical relatives of T. squamata, and other species of swifts around the world, especially those with forked tails that construct nests containing feathers, such as the widespread African Palm-Swift (Cypsiurus parvus), Asian Palm-Swift (C. balasiensis), and other apodines, will soon be exposed as proficient kleptoptilists as well. For example, a report of two Alpine Swifts (Tachymarptis melba) attacking a Rock Dove (Columba livia), one on the back and one on the belly, and riding it to the ground (“causing the pigeon serious injury”), was interpreted as simple aggression (Chantler 1999) but was perhaps more likely a dramatic instance of kleptoptily.
I am grateful to B. Magnuson of Reseda, California, for help in locating references, and she, J. V. Remsen, Jr., D. Lane, and M. Cohn-Haft provided helpful suggestions. P. Stouffer, C. Collins, and an anonymous reviewer also provided constructive comments. Thanks to D. Lane for kindly depicting the palm- swift attack in Figure 1.