Olfactory Navigation in Homing Pigeons: Are the Current Models Atmospherically Realistic?

--True olfactory navigation necessitates that homing pigeons (Columba livia) employ an odor "map" when determining their position relative to the home loft. Two models exist currently to explain the nature of this olfactory map. The "mosaic" map model suggests that pigeons learn a patchwork of olfactory cues by direct experience with local odors and determine displacement from the home loft by recognizing these familiar odors upon release. The "gradient" map model postulates the existence of stable odor gradients that extend over distances of up to 1,000 km. Pigeons learn the home value of an olfactory gradient while growing up, and then determine their positions with respect to home by comparing this learned value with the gradient value detected at the point of release. I analyzed both the mosaic and gradient olfactory map models in light of new data on the long-range transport of atmospheric aerosols. These data, obtained in 1983 as part of the Cross-Appalachian Tracer Experiment (CAPTEX '83), cover the region of New York state where extensive experiments on pigeon homing have been conducted from the Cornell University pigeon loft during the past 18 yr. The CAPTEX '83 data provide the first opportunity to assess accurately the atmospheric plausibility of an odor-based navigation system for homing pigeons flown in the northeastern United States. I conclude that the observed patterns of long-range aerosol transport place important limits on the type of olfactory navigation system potentially utilized by homing pigeons. Moreover, a comparison of the CAPTEX '83 data with meteorological information relevant to atmospheric conditions in other geographic locations suggests that regionally specific atmospheric patterns may dictate whether pigeons can reliably employ odors during navigation. This regional specificity is proposed as an explanation for the observed discrepancy between results obtained in olfactory navigation tests conducted in Italy and those performed in the northeastern United States. Received 29 September 1986, accepted 30 January 1987. OLFACTORY sensitivity in birds has traditionally been considered inferior to that of most other vertebrates (early work reviewed by Bang and Wenzel 1985). Recent evidence, however, indicates that several avian species possess highly functional olfactory systems. Studies of avian olfactory anatomy (Bang and Cobb 1968, Bang 1971), physiology (Wenzel and Sieck 1972, Macadar et al. 1980, Smith and Paselk 1986), and behavior (Stager 1964; Wenzel 1968, 1971; Goldsmith and Goldsmith 1982; Buitron and Nuechterlein 1985) support the possibility that odors play a significant role in the life of at least some bird species. With the exception of work on the Turkey Vulture (Cathartes aura; Stager 1964), most demonstrations of olfactory orientation in birds involve studies of foraging at short distances, usually a few meters or less. Recently, Sooty Shearwaters (Puffinus griseus), Pink-footed Shearwaters (P. creatopus), Northern Fulmars (Fulrnarus glacialis), Black-footed Albatrosses (Diornedea nigripes), and Ashy Storm-Petrels (Oceanodrorna hornochroa) have been observed to 369 The Auk 104: 369-379. July 1987 370 JERRY A. WALDVOGEL [Auk, Vol. 104 make downwind approaches toward experimental releasing wicks scented with the odors of fish oils, squid, or krill (Hutchison and Wenzel 1980, Hutchison et al. 1984). Although the range over which these odors were effective in attracting procellariid species was not determined precisely, distances of at least a few hundred meters probably were involved. The most impressive claim for long-range olfactory orientation in birds has been made in an effort to explain the homing of pigeons (Columba livia; Papi 1976). Questions about the validity of this theory have generated considerable debate (e.g. see Papi et al. 1978, Gould 1982), and research groups in Italy and Germany continue to report results that support the possibility of olfactory navigation in pigeons. Despite the fact that similar tests conducted from other lofts in Germany and the United States generally do not support the olfactory theory (reviewed by Able 1980), some investigators suggest that the bulk of the evidence permits the conclusion that olfaction forms an essential component of the pigeon's navigation system (Wallraft 1980, 1984; Papi 1982). Unlike the simple olfactory orientation response observed in procellariids (where the birds orient directly upwind toward the source of an odor), pigeons are thought to develop a two-dimensional odor "map" of the area surrounding their home loft. This map allows navigation from any direction with respect o home without having to detect odors emanating from the loft itself. An olfactory map can thus potentially be used over tens or hundreds of kilometers to determine the correct homeward direction from any given release point. Wallraft (1981) discussed two forms that this navigational odor map may take. The first model, the "mosaic" map, is based on the original model proposed by Papi et al. (1972). The mosaic map is learned by young pigeons while growing up at their home loft, and develops from direct experience with the odors encountered during exercise and training flights. More distant odors brought to the loft by prevailing winds also may be incorporated into the mosaic map. While the mosaic map is thought to operate at distances of up to 200 km, its use is necessarily limited to areas where the bird has had direct experience with local or wind-borne odors. A second model, the "gradient" map, overcomes this distance limitation by assuming the existence of very long-range atmospheric odor gradients that are stable in space and time. Pigeons would use such gradients to fix their positions at distances well beyond their previous flight experience by comparing the observed odorant concentration at the release point with the remembered value of the gradient at the home loft. The concept of a gradient map helps to explain the rapid homing of pigeons from release points up to 1,000 km from their loft (Wallraft 1981). Numerous examples exist that document olfactory homing to the source of an odor by means of upstream orientation in a well-defined directional medium (e.g. the river phase of adult salmon migration and the attraction of male insects to female sex pheromones). Yet the use of an odor map for true navigation has never been demonstrated conclusively for any species. A major problem for airborne olfactory navigation is that odors that contribute to the formation of a map are subject to spatial and temporal distortions created by atmospheric turbulence and changes in synoptic flow patterns. Modifications of wind flow due to orographic (terrain) features are also important factors in determining odor directionality. Orographic features are particularly important for the pigeon, which generally flies at altitudes of <30 m (i.e. treetop level). Thus, the relative influence of several atmospheric mixing parameters that contribute to wind and odor directionality ultimately determine the usefulness of olfaction as a basis for true navigation in any low-flying animal. The phenomena of dispersion and long-range transport of particulate and aerosol matter in the upper atmosphere has been well documented and discussed (references cited by Darzi and Winchester 1982, Partington et al. 1983, Ottar et al. 1984, Robinson 1984). Such upperair transport probably is not important to avian navigation because of the substantial altitudes involved (generally several thousand meters or more). To date, however, only one attempt has been made to address formally the influence of lower atmospheric dynamics on odor navigation in birds. Becker and van Raden (1986) concluded that the concept of avian olfactory navigation has important shortcomings based on the European meteorological examples they analyzed, primarily because of differences between observed air currents and mean wind directions for a given locality; temporal and July 1987] Olfactory Navigation i Pigeons 371 Fig. 1. Maximum observed tracer concentrations measured in femtoliters per liter (fl/1) for each of the CAPTEX '83 releases from Dayton, Ohio (labeled R). Each panel (a-d) represents the results of one 3-h release. Tracer concentrations were measured over the 52 h following each release. Isolines of equal concentration indicate the shape and trajectory of each tracer plume. (Redrawn from Ferber et al. 1986.) spatial instability of different odor sources; and lack of evidence for long-lived, strong, and stable olfactory gradients in the lower atmosphere. A similar analysis of lower atmospheric dynamics has not been possible for the area surrounding the Cornell University pigeon loft (Ithaca, New York), where detailed observations of homing pigeon behavior have been made during the past 18 yr. This was due primarily to the lack of adequate information on long-range, lower atmospheric flow patterns over the North American continent. Recently, data collected in a study of long-range atmospheric transport in the northeastern United States (the Cross-Appalachian Tracer Experiment; CAPTEX '83) have become available, and provide the first comprehensive analysis of surface-level flow patterns in regions where Cornell pigeons are flown regularly. This combination of atmospheric and navigation data presents a unique opportunity to determine empirically whether observed atmospheric transport patterns support the use of odors in the homing pigeon's navigational map as suggested by the mosaic and gradient map models. THE CAPTEX '83 STUDY Objectives and methods.--CAPTEX '83 was a major field study designed to simulate the long-range transport of pollutants in the atmosphere (Ferber et al. 1986). The study was directed by the National Oceanic and Atmospheric Administration, in cooperation with several other federal and private agencies in the United States and Canada. From the perspective of research