Hans G. Wallraff Avian Navigation: Pigeon Homing as a Paradigm

Hans G. Wallraff Avian Navigation: Pigeon Homing as a Paradigm

Hans G. Wallraff Avian Navigation: Pigeon Homing as a Paradigm With 98 Figures

Dr. Hans G. Wallraff Max Planck Institute for Ornithology 82319 Seewiesen Germany e-mail: wallraff@orn.mpg.de ISBN 3-540-22385-1 Springer-Verlag Berlin Heidelberg New York Library of Congress Control Number: 2004110945 This work is subject to copyright. All rights are reserved, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication or parts thereof is permitted only under the provisions of the German Copyright Law of September 9, 1965, in its current version, and permissions for use must always be obtained from Springer-Verlag. Violations are liable for prosecution under the German Copyright Law. Springer is a part of Springer Science + Business Media springeronline.com Springer-Verlag Berlin Heidelberg 2005 Printed in Germany The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. Editor: Dr. Dieter Czeschlik, Heidelberg, Germany Desk editor: Anette Lindqvist, Heidelberg, Germany Cover design: Design & Production, Heidelberg Typesetting: perform electronic publishing GmbH, Heidelberg 31/3150WI 5 4 3 2 1 0 Printed on acid-free paper

Preface Bird migration includes homing performances in almost global dimensions. After having spent the winter in South Africa, a stork or a swallow returns in spring to its old nest in a village in Denmark. A redstart, a flycatcher or a bobolink arrives in its former breeding territory after nocturnal flights covering thousands of kilometres. In what way do the birds solve the problems of navigation behind these performances? Although this book deals neither with bird migration nor with storks and flycatchers, but focuses on a domestic animal, it nevertheless deals with a central problem implicated in migrations and other long-distance flights characterizing the life cycles of many avian species. It is well known that such birds can find their way home to their breeding sites over hundreds of kilometres of unfamiliar territory, but little is known about how they achieve this admirable feat. Experiments dealing with that question in detail, with navigational strategies and with use or non-use of various environmental signals have, for good methodological reasons, predominantly been conducted with carrier pigeons. Thus, what actually can be reported is mainly our present knowledge about pigeon homing. However, there is little doubt, and there are findings suggesting, that the results and conclusions obtained with pigeons can largely be generalized towards an understanding of the navigational capabilities of other birds as well. It is not the pigeon itself as a particular bird, as a companion of man over thousands of years, that is the focus of interest of this monograph, but specifically its role as a paradigm for the study of navigation mechanisms enabling a bird to find its familiar home site from distant unfamiliar areas. The book summarizes the outcome of about half a century of research. A final solution to the homing problem has not yet been achieved during this time; substantial questions still remain. However, it seems that now, after a long period of trial-and-error searches and accumulation of initially unexplainable observations, at least the four basic categories of exploited input signals have been identified: the sun and the geomagnetic field (both used to determine compass directions), the visual landscape (used to determine the current position inside a familiar area around home), and atmospheric trace gases perceived by olfaction (used to determine the position in relation to home even from far away in unfamiliar areas). Naturally, the last category was not among the candidate input signals hypothetically envisaged from the beginning. It entered the arena unexpectedly and achieved its central position, against understandable scepticism, under the pressure of more and more compelling empir-

VI Preface ical findings. Presentation of these findings, discussion of their significance and theoretical consequence, and a proper look at the atmosphere make up the longest chapter of the book. Other sections proved necessary in order to show that initially favoured physical cues, especially the earth s magnetic field or the sun, are apparently not used by birds to determine their current position with respect to home, but are nevertheless substantial components of the homing process as a whole. Further, it was necessary to describe the pigeons homing behaviour in some detail and to separate its navigationally relevant parts from concomitant peculiarities which must be considered for methodological reasons and whose disregard may lead to misinterpretation of results. The book tries to give a state-of-the-art survey for everyone interested, but also tries to meet the demands of a critical readership being more or less familiar with the field. Considering the former group, the main text does not aim to report comprehensively on all pigeon homing experiments conducted over five decades and on all debates and controversies connected with them. I focus on those findings and ideas that I consider sound and relevant and that formed my current picture of the field. Being selective in the interest of keeping the text sufficiently fluent and attaining comprehensible conclusions,however,implies the danger and the possible reproach that the presented picture has been made coherent at the cost of objectiveness. Therefore, I did not simply ignore findings and opinions that might confuse the picture, but deal with them in a separate section (Sect. 7.8), in a chapter on research history (Chap. 11) and in the Appendix at the end, in which selected particulars are discussed in some detail. Even readers who are not ready to accept my style of presentation, my arguments and conclusions may profit from the book, as it facilitates access to the extensive literature from which everybody can derive his or her own view of a field and a period of research that now possibly approaches its end phase. I may be wrong, but I feel that the types of pigeon releases, field glass observations and statistical analyses, as initiated by G. Kramer and G.V.T. Matthews in the 1950s and described in the following pages, have largely done their task. On the one hand, these experiments probably revealed the most fundamental peculiarities of pigeon homing (albeit many particulars are not yet clarified), while, on the other hand, future progress in solving the remaining most fundamental questions will require more sophisticated methods intruding deeper into the environment and into the animals neural mechanisms of signal processing. This monograph attempts to provide the material basis from which more refined research may commence. The book has a personal touch insofar as it includes a retrospect of my own research over a period of about 45 years, i.e. over almost the whole period during which pigeon homing has been intensely investigated. Naturally, my mode of looking at the accumulated material agrees most properly with the mode in which I have previously designed experiments aimed at achieving some understanding of the pigeons homing behaviour. Therefore, and because other

Preface VII corresponding data are often unavailable, many of the figures show results obtained by me and by colleagues with whom I worked.i thank all those who contributed to produce the related data files, in particular Silvano Benvenuti, Augusto Foà, Jakob Kiepenheuer, Michael Neumann, Ulrich Sinsch, Andrea Streng and my technical assistants Rainer Wahl and Karl Wielander. I remember most gratefully the productive collaborations I had over the years with Floriano Papi and his associates in Pisa, whose outcomes are included in the following pages. Last, but not least, I thank Floriano for his readiness to read a draft of the whole manuscript and for his helpful advice resulting in a number of improvements. Seewiesen / Starnberg, May 2004 Hans G. Wallraff

Contents 1 Introduction... 1 1.1 Avian Migration and Navigation... 1 1.2 Why Investigate Domestic Homing Pigeons?... 2 1.3 On Terms and Definitions... 4 2 Observation Data Used to Investigate Pigeon Homing... 7 2.1 Initial Orientation... 7 2.2 Homing Performance... 11 2.3 Homing Routes... 12 2.4 Recoveries... 15 2.5 Cage Experiments... 16 3 Basic Features of Pigeon Homing... 17 3.1 Homing of Inexperienced Pigeons... 18 3.1.1 Peculiarities of Initial Orientation... 18 3.1.2 The Problem of Motivation... 23 3.1.3 Limited Significance of Exercise Flights at Home... 26 3.2 Experience Gained by Homing Flights... 29 3.2.1 Number and Range of Previous Homing Flights... 30 3.2.2 Directions of Previous Homing Flights... 33 3.2.3 Familiarity with the Release Site or Area... 36 3.3 Temporal Variability... 38 3.3.1 Annual Periodicity... 38 3.3.2 Aperiodic Fluctuations... 41 3.4 Accuracy of Homeward Orientation... 42 3.4.1 Distance of Displacement... 42 3.4.2 Geographical Variability... 43 3.4.3 Distraction by Landscape Configurations... 44 3.4.4 Stochastic Noise... 46 3.5 Spatial Range of Pigeon Homing... 46 3.6 Multiple and Mobile Home Sites... 47 3.7 Conclusions and Perspectives... 48 3.7.1 Home-Related and Home-Independent Components of Initial Orientation... 48 3.7.2 Motivation, Selection and Homing Experience... 50 3.7.3 Inaccuracy and Variability of Homing Orientation... 51

X Contents 4 Potential Input Signals Exploitable for Home-Finding... 53 4.1 Directional References... 53 4.2 Indicators of Position... 54 4.2.1 Motion-Bound Signals... 54 4.2.2 Location-Bound Signals... 54 5 TheRoleoftheSun... 59 5.1 Effects of Shifting the Birds Circadian Clock... 59 5.2 Linkage Between Sun Compass and Map... 64 5.3 Orientation Under Overcast Skies... 66 5.4 Conclusions and Perspectives... 68 6 The Role of the Geomagnetic Field... 69 6.1 Effects of Artificial Magnetic Fields... 69 6.1.1 Magnets or Coils During Flight... 69 6.1.2 Magnetic Fields Before Release... 73 6.2 Geomagnetic Irregularities... 77 6.2.1 Spatial Magnetic Anomalies... 77 6.2.2 Temporal Magnetic Fluctuations... 78 6.3 Conclusions and Perspectives... 78 6.3.1 The Interference with the Stress Issue... 79 6.3.2 The Magnetic Compass Issue... 80 6.3.3 The Magnetic Map Issue... 81 6.3.4 The Path Integration Issue... 82 6.3.5 The Magnetic Sense Issue... 82 6.3.6 Overall Conclusions and Outlook... 85 7 The Role of the Chemical Atmosphere... 87 7.1 Effects of Olfactory Deprivation... 87 7.1.1 Olfactory Nerve Section... 88 7.1.2 Inactivation of the Olfactory Epithelium by Zinc Sulphate... 92 7.1.3 Nasal Anaesthesia... 93 7.1.4 Occlusion of Nostrils... 94 7.1.5 Insertion of Nasal Tubes... 96 7.2 Dependence on the Air the Pigeons Breathe... 96 7.2.1 Removal of Trace Gases by Filtration... 96 7.2.2 Air from Different Environments... 99 7.3 Olfactory Misguidance... 101 7.3.1 Spatial Separation of Sites of Air-Smelling and Release... 101 7.3.2 Outward-Journey Detours... 104 7.3.3 Natural Olfactory Misguidance... 105 7.4 Spatial Range and Variability of Olfaction-Based Homing... 107 7.4.1 Distance Range and Efficiency in Different Regions... 107 7.4.2 Temporal Variability... 110

Contents XI 7.5 Varying Home Site Conditions in Aviary Experiments... 111 7.5.1 Shielding, Deflecting and Reversing Winds... 111 7.5.2 Applying Artificial Odours and Winds... 118 7.5.3 A Digression to the Preferred Compass Direction (PCD)... 119 7.6 Spatial Structures in the Chemical Atmosphere... 120 7.6.1 A Working Hypothesis... 120 7.6.2 Gradients in Ratios of Trace Gases... 123 7.6.3 Applicability of Atmospheric Gradients to Navigation... 128 7.6.4 Assumed Dependence on Geographical Peculiarities... 131 7.6.5 The Short Distance/Long Distance Problem... 133 7.7 Avian Olfactory Perception: A Suitable Tool for Navigation?... 136 7.7.1 The Problem of Sensitivity... 136 7.7.2 The Problem of Odour Discrimination... 137 7.7.3 The Problem of Compound Selection... 137 7.7.4 The Problem of Adaptation/Habituation... 138 7.8 Arguments Against Olfactory Navigation and Replies... 138 7.8.1 Intuitive Incredibility... 138 7.8.2 Potential Non-Specific Side Effects... 139 7.8.3 Inconsistent Results... 140 7.9 Conclusions and Perspectives... 143 7.9.1 Experimental Evidence of Olfactory Navigation... 143 7.9.2 How Does Olfactory Navigation Operate?... 144 7.9.3 Do Pigeons Have an Olfactory Map?... 144 7.9.4 Unsolved Problems Remaining Challenges... 146 8 TheRoleoftheVisualLandscape... 149 8.1 The Landscape as a Home-Guiding Factor... 149 8.1.1 Non-Olfactory Homing... 149 8.1.2 Downgrading of Visual Signals... 152 8.1.3 Homing with Olfaction and Vision Impaired... 153 8.1.4 Interrelations Between Landscape and Sun Compass... 155 8.2 The Landscape as a Disturbing Factor... 157 8.3 Conclusions and Perspectives... 159 9 The Neural Bases of Pigeon Homing... 161 9.1 The Hippocampus and the Familiar-Landscape Map... 161 9.2 The Hippocampus and the Olfactory Map... 165 9.3 The Piriform Cortex and the Olfactory Map... 166 9.4 Roles of Other Brain Regions and Hemispheric Lateralization... 167 9.5 Conclusions and Perspectives... 167 10 Homing in Other Birds... 169 10.1 Experimental Approaches... 169 10.1.1 Brief Overview... 169

XII Contents 10.1.2 Compass Orientation and Preferred Compass Directions... 171 10.1.3 Goal-Oriented Navigation... 173 10.2 Homing in Natural Life... 176 10.2.1 Non-Migratory Excursions... 177 10.2.2 Homing as a Constituent of Bird Migration... 179 10.3 Conclusions and Perspectives... 181 11 Research History: Blind Alleys and an Unexpected Passage... 183 12 Overall Synthesis and Perspective... 187 12.1 Environmental Cues Involved in Home-Finding Processes... 187 12.2 Two Homing Mechanisms... 188 12.3 Challenges for Future Research... 189 12.3.1 Unsettled Basics of Pigeon Homing... 189 12.3.2 Problems of Visual Landscape Orientation... 190 12.3.3 Problems of Olfactory Navigation... 191 12.4 Outlook... 192 Appendix: Notes on Selected Particulars... 195 References... 205 Subject Index... 225

1 Introduction 1.1 Avian Migration and Navigation Among other aspects of bird migration, such as aerodynamics, energetics and associated migratory strategies, including remarkable ecological adaptations, the aspect of spatial orientation is certainly the most striking one. Consequently, migratory orientation has been investigated in great detail.it is known that birds can determine compass directions by means of the geomagnetic field, the stellar sky or the sun, which are globally available as directional references. Using their compasses, the birds can select genetically encoded directions which lead them to their winter quarters. An endogenous time programme gives the appropriate distance of flight in a temporal scale. Experimental data suggest that such bearing-and-distance programmes, which are roughly adapted to, and modified by, topographical and ecological conditions of the covered regions, and which in some species include even directional changes, are the general bases of many or most population-specific migratory routes. Applying in this way its endogenous intentions to the external world, even a young bird, leaving its birth place for the first time, eventually arrives in the far distant wintering range of its population. With larger birds,such as waterfowl or storks,which migrate in flocks during the daytime, also guidance of juveniles by adults along traditional routes is involved. Such guidance can be excluded in most of the smaller birds, particularly in the many passerine species migrating predominantly at night. Detailed reports on bird migration and migratory orientation can be found in other books (see Alerstam 1990; Berthold 1996, 2000, 2001; Berthold et al. 2003). In spring, the bearing-and-distance programme can be expected to operate analogously to its operation in autumn; only the directions need to be reversed. A bird may return thereby to its population-specific breeding range. It is hardly imaginable, however, that a redstart can reach, on this basis alone, the village and the garden in which it had bred the year before. Migrating at night and exposed to drift-causing winds, simply keeping the body axis on a genetically encoded compass course would be insufficient. In fact, the birds can perform better. It has been shown that birds of many species (e.g. starlings, cowbirds, blackbirds, swallows, swifts, terns, gulls, storks, shearwaters, petrels and albatrosses) are able to find a familiar home site by goal-related navigation from far distant unfamiliar areas to which they had been passively displaced (for lists of such experiments and for references see Matthews 1955; Wiltschko

2 1 Introduction 1992; Åkesson 2003). To achieve this performance, the animals must have more than a compass. Only if a bird is aware that it is north (and not south, east or west) of home, can it usefully apply a compass in order to steer a southward course guiding it home. The process of birds finding their way back from a position in a distant unfamiliar area (where the animals have never been before) to a familiar geographical position (which has been established, during longer-lasting previous presence, as a home site ) has often been called navigation in a narrow sense (Sect. 1.3). There is hardly any doubt that birds commonly apply goal-related navigation in the course of their migrations (Sect. 10.2.2) and that,without this capability, bird migration patterns would never have achieved their present overall character. Nevertheless, as substantial parts of migratory orientation, homing mechanisms are usually not closely inspected in books and articles dealing with bird migration. In fact, these mechanisms have not been closely investigated in migratory species. However, they have been intensely investigated in a non-migratory domesticated bird. I feel confident that the knowledge gained from pigeon homing can largely be considered as knowledge about avian homing in general and hence also of goal-finding in the course of bird migration. 1.2 Why Investigate Domestic Homing Pigeons? In research on avian navigation, the homing pigeon acts as a kind of laboratory rat. Without the experiments conducted with this domestic animal, only a small fraction of our present knowledge about the mechanisms used by birds to find their way home from remote areas would have been achieved. The advantages of homing pigeons, as compared to any wild species, are obvious. Hundreds or even thousands of pigeons can be kept in man-made lofts at man-selected locations. Over the whole year the birds are easily available for experimental use without laborious and stressful trapping procedures. Individual returns from releases can be precisely recorded, usually by only one observer. Homing pigeons are always motivated to return to their loft, not only during the short breeding season. Due to their strong homing drive,they fly off mostly immediately after release, thus giving opportunity to observe their initially selected courses, whereas most other birds tend to land on nearby trees or in meadows etc. before starting homeward movements some time later. Over short and median distances, many pigeons return in a non-stop flight, so that the time they need for homing reflects,by and large,the length of the route they have flown. Even many of the pigeons that fail to home provide information about their orientation: Due to their social impulses, they tend to enter one or another of the pigeon lofts which are, in some countries, quite densely scattered over the land. Reports by pigeon fanciers contribute instructive spatial distributions of recovery sites. Finally, pigeons are very handy, their size is neither too small nor too large. They are large enough to be observed with bin-

1.2 Why Investigate Domestic Homing Pigeons? 3 oculars over approximately 2 km and large enough to carry the weight of a small transmitter for telemetry or some other apparatus apt to record their flight paths. They are small and tame enough to be confined in large numbers together in manageable crates for transport. These outstanding advantages resulted in in-depth investigations on homing capabilities of birds being almost exclusively conducted with homing pigeons. Thus, the question arises whether pigeon homing can be seen as a model case of bird homing in general. At least there are good reasons to assume that it can. It appears extremely unlikely that, in the course of domestication from the rock pigeon (Columba livia) and selective breeding, an entirely novel navigation mechanism that did not exist before in the genus Columba, has been implanted into the strains of homing pigeons. The fact that the rock pigeon is usually sedentary implies no argument against the assumption that navigation mechanisms detected in homing pigeons can indicate mechanisms used by other birds during migration. Closely related species (e.g. Columba palumbus) do migrate, and Berthold et al. (1990) have shown that offspring of partially migratory birds of a given species can be genetically transformed, by means of selective breeding, to either full migrants or non-migrants within only a few generations. It is uncertain whether selection performed by man noteworthily affected the navigational capabilities of pigeons. Much more obvious is improvement of the homing drive, which appears less highly developed in the rock pigeon (Alleva et al. 1975; Visalberghi et al. 1978) and in other wild birds (e.g. Able et al. 1984). In pigeon races, the breeder is concerned with getting a maximum number of pigeons homed at maximum speed. The fastest homers are then preferentially selected for breeding. Homing speed in such races, however, is not primarily a result of outstanding navigation but rather of motivation, physical strength and persistence. The birds are usually trained to return in large flocks from always the same direction. Thus, the ability to find their way home from novel directions and sites by making individual decisions is not a criterion for selection. Of course, there may be differences among pigeons and other species in many details of homing behaviour, for instance in the motivation to fly home at all,in the hierarchy in which different navigational cues are used or in the effectiveness with which such cues are exploited. It seems most likely, however, that once we understand pigeon homing we are able to understand the principles of avian goal-oriented navigation in general. Nevertheless, this assumption remains merely an assumption unless it can be actually shown that at least the most substantial conclusions drawn from experiments with homing pigeons can in fact be generalized and are valid for other bird species as well. Some supporting evidence is already available (Chap. 10). Although the carrier pigeon is a unique subject for research on animal homing, the huge amount of data obtainable from millions of homing flights performed in the course of pigeon races has only a minor value as a source of information in the context of this research. As mentioned above, races take place in large flocks, usually comprising thousands of birds, so that, in a statistical

4 1 Introduction sense, an individual bird is not an independent unit. Moreover, the pigeons of a given region are usually displaced towards a standard compass direction with distances increasing in the course of the season. Only homing speeds are measured. By multiple correlation analyses using such data (e.g. Dornfeldt 1996) it is hardly possible to isolate causal relationships concerning navigational processes. To analyse these processes, it is necessary to conduct specifically designed experiments using individually flying pigeons released at sites in varying directions around home. Thus, researchers have no choice but to install their own scientific pigeon lofts in order to gather suitable data. 1.3 On Terms and Definitions Homing and navigation are the key terms used in this book. It is self-evident that a precondition for homing is the creation of a home to which an animal is willing to return from a distant site, whether it has moved to this site voluntarily or been displaced passively by man. Any adult bird usually makes its breeding territory a home site. Domestic pigeons feel bound to a loft in which they were living over several weeks postfledging (Sect. 3.1). Any act of return movement to this site may be called homing, but not any such act is commonly called navigation. If a pigeon or a starling returns from an accustomed feeding area some hundred metres or even a few kilometres away, we may hardly speak of navigation. However, if an albatross returns from a foraging flight covering some thousand kilometres, its homing to a small island in the midst of the ocean is undoubtedly an outcome of a remarkable navigational performance. So what is navigation? I refrain from giving a sharp definition. Attempts to do so, as attempted earlier, have never been successful. Usage of the term has even been broadened in recent years (navigation through the brain, the Internet etc.). Nevertheless, in the present context and in a narrower sense, the term true navigation is frequently used for goal-related orientation towards a given position on earth, usually a home site, towards which no direct sensory contact is available at the current position of an animal. In its strictest sense, the term is only applied if not even familiar landmarks surrounding the goal can be recognized, or are used, which potentially can guide to the goal. Such guidance by familiar landmarks is often called piloting. Anotherpair of terms tomention arecompass and map. These are tools used by human navigators and it is frequently implicated that animals also make use of such instruments. In fact, applying these terms in the field of animal orientation and navigation is very convenient, as they describe particular functions in a vivid and succinct way. While speaking about compasses and maps, however, we should be aware that we are using anthropomorphic metaphors. A flying bird might perhaps have an equivalent of a map exhibiting explored topographical features, but certainly it does not have a chart showing isolines of physical quantities that might correspond to a grid of coordinates. However, an animal navigator must have some representations in the brain referring to spa-

1.3 On Terms and Definitions 5 tial structures in the environment and must evaluate them in a way so that the outcome is comparable to the outcome of our use of such instruments. Thus, in order to avoid long-winded circumscriptions of neural processes which are largelyunknown, Ishallusethetermscompass and map as metaphoric shortcuts of such processes. As I do so with some reluctance, however, the terms should always be thought of as between quotation marks, although they are mostly not labelled that way because of their frequent occurrence. Further words of caution concern the attributes often preceding the word map. If I speak of a gradient map (Sect. 4.2.2), for instance, I do not presume that a pigeon is in any way aware of spatial gradients or that such gradients are mapped to some brain region. The term merely implies that the bird s navigational system exploits scalar values of relevant environmental parameters that increase or decrease more or less monotonically over some distance in a given direction. The bird is thought to make use of gradients that could be drawn (by man) on a map, but the bird is not thought to have such a map (Sect. 7.9.3). The attributes connected with a map may belong to quite different categories. They may refer to the physical or sensory nature of input signals (magnetic map, olfactory map), to the spatial structure of evaluated cues (gradient map, mosaic map, topographical map), to the application of true navigation as opposed to piloting (navigational map), to the spatial experience of an individual bird (familiar area map) or to a psychological category (cognitive map). Concerning the latter, I do not take a position on whether or not pigeons have a cognitive map (Sects. 4.2.2 and 9.1) (cf. Bennett 1996; Mackintosh 2002). Generally, I have no ambition to get into debates on terminology. Classifications of terms and definitions rarely find unrestricted acceptance by all researchers working on different topics within a given field. Lines of demarcation separating natural phenomena from each other may appear appropriate in one context but are often inappropriate when seen from another point of view or when thus far unconsidered circumstances have to be included. By and large, my notions are roughly, but not always literally, consistent with the terminological classifications as proposed by Papi (1992b) and Able (2000).

2 Observation Data Used to Investigate Pigeon Homing Many of the figures in this book show diagrams belonging to a small set of graph types referring to a small set of data types from which conclusions on the pigeons navigational performances and mechanisms are deduced. The reader should first become familiar with these types of data. Illustrating them by means of examples implies a kind of preview on the pigeon s general homing behaviour which will be inspected in more detail in the following chapter. An experiment on pigeon homing (in brief: a release) is usually conducted in the following way. A number of pigeons, individually labelled by numbered and coloured rings and selected according to age, homing experience, group membership, etc., are caught in their home loft (often at night, when they are sitting quietly) and confined in crates or baskets. In the morning they are transported (mostly inside a car, prevented from viewing the outside) to the release site where they remain sitting in the crates until each bird, after the vanishing of its foreflier, has individually been released. A simultaneous release of different groups of pigeons means actually that members of the different groups are released alternately in succession.release sites are,as far as possible, in open landscapes allowing unobstructed observation in all directions. Each bird is tossed into the air and observed with binoculars until it vanishes from sight. The time of release is recorded and another observer records the individual bird s time of arrival at the loft. Depending on the particular purpose and design of a given experiment, the procedural details are sometimes modified in one way or another. The following criteria are used to determine the pigeons performances. 2.1 Initial Orientation The most often reported data are the vanishing bearings of individually flying pigeons. At the time of vanishing, a pigeon is approximately 1.6 2.2 km distant from the starting point, the exact distance depending on atmospheric conditions, background (sky or landscape), colour of the pigeon and quality of the observer. From the individual bearings a mean vector can be calculated (Batschelet 1981) which gives the mean direction α (deviation clockwise from north = α N, from homeward = α H ) and, with its length a, a reciprocal measure of angular dispersion (Fig. 2.1). Length a is variable between 1 (all bearings in one direction) and 0 (bearings uniformly distributed).most important,partic-

8 2 Observation Data Used to Investigate Pigeon Homing Fig. 2.1. An example of initial bearings (peripheral dots) of individually released pigeons. The direction towards the home loft is marked by a double-headed arrow. From the ten bearings, a mean vector is calculated whose length a is indicated by the length of the central arrow (possible maximum 1=radius of the circle) and whose direction is α N, (deviation clockwise from north) and α H (deviation from home), respectively. The homeward component c H is the rectangular projection onto the axis pointing towards home. Values in the example are: home = 135 from north, αn = 86, αh = 49, a = 0. 84, c H = + 0. 54. Note for the following circular diagrams (if not otherwise stated): the double-headed arrow points always to the home loft. Peripheral symbols mark the vanishing bearings of individual pigeons or mean bearings computed from a sample of such primary bearings (in the case of second-order presentations; see Fig. 2.4, below). Mean vectors drawn as central arrows refer to a scale with value 1 equal to the radius of the circle ularly when a number of symmetrically arranged releases are combined and/or differently treated pigeons are to be compared, is the homeward component c H. It results from a rectangular projection of the mean vector onto the axis running through release site and home site (c H = a cos α H ). This component, variable between +1 and 1, depends on both direction and length of the mean vector. It is positive if the deviation from home is less than 90. Figure 2.2 shows examples of vanishing bearings of pigeons displaced from six different lofts to the same release site over distances of 191 345 km. The birds had never before been displaced from their home (so-called first-flight pigeons). It is obvious that those from the three northern lofts preferred northerly directions, whereas most birds coming from a southern home site vanished in the southern semicircle. It is also obvious, however, that the mean deviation from home as well as the angular dispersion of the bearings varied greatly. Such patterns of initial orientation are very typical. The various factors determining the shape of a particular pattern are discussed below. Low return rates in first-fliers displaced over such long distances are also typical. About a quarter of the initial flights are fairly straight, so that the pigeons vanish from sight within a minimally possible period of time of 1 2 min (Fig. 2.3). More frequently, the birds first fly some loops or circles (see insert in Fig. 2.3) and often they depart on a more or less meandering route. In rare cases, they remain more than 10 min within the range of visual observation. The vanishing in-

2.1 Initial Orientation 9 Fig. 2.2. Vanishing bearings of inexperienced pigeons from six home lofts in Germany released at two neighbouring sites near Giessen. Percentage of birds returned is given inside the circles. (Modified from Wallraff 1970a) Fig. 2.3. Vanishing intervals of 1,476 inexperienced (first-flight) pigeons from three Bavarian lofts observed in the course of 189 releases. Median = 50% and other percentiles are indicated along the top (unpublished summary from my data files). Insert shows a number of flight paths tracked by visual triangulation from three observation points (modified from Spott 1993)

10 2 Observation Data Used to Investigate Pigeon Homing terval represents an indirect measure of the decidedness of initial orientation. The median period of observation is around 3 min. These vanishing intervals are quite variable among individual pigeons and also their means are variable between releases. Experimental treatments, however, affect them rarely to a larger degree. Therefore vanishing times play merely a subordinate role as indicators of orientational performances. They are routinely recorded in every release, but usually cannot be used as a criterion for substantial conclusions. The directional distributions of the pigeons bearings at departure are much more conclusive and worthy of being analysed in more detail. Most useful for this purpose are so-called second-order analyses combining the results of a number of releases conducted at different sites. Figure 2.4 shows two kinds of second-order distributions. Each of the arrows in the upper graphs corresponds to Fig. 2.4. Initial orientation of inexperienced (first-flight) pigeons from seven lofts in Germany, from each loft at four symmetrically distributed sites at distances of 85 216 km. Direction towards home points upwards. Bearings of departing birds were determined 20 and 40 s after release and at time of vanishing from sight. Arrows in the upper diagrams show first-order mean vectors per release (vector length 1 = radius; cf. Fig. 2.1), whose second-order mean length (irrespective of direction) is given by = a. Ellipses indicate the 95 and 99% confidence ranges of the second-order bivariate mean (considering lengths and directions), which itself is represented by the centre of the ellipse and by the three numbers (angular deviation from home, vector length, homeward component). At the periphery of the lower diagrams, directions of the same vectors are shown, but now with equal weight independently of their lengths, i.e. of the angular scatter within each release. Arrows and numbers give the second-order mean vectors calculated from these 28 directions. (Data from Wallraff 1970a, supplemented by unpubl. data)

the central arrows shown in Figs. 2.1 and 2.2, i.e. it represents the mean vector calculatedfromasampleofbearingsobservedatonereleasesite(α H, a;all home directions upwards). The second-order mean vector ( α = H, a = H )resulting from the 28 first-order vectors is marked by the centre of the ellipses. In the bottom graphs, each first-order vector enters with equal weight irrespective of its length. Thus, it serves as one bearing as indicated by the peripheral symbols. A second-order vector of that kind (central arrows) is calculated analogously to the first-order vector in Fig. 2.1. Such second-order analyses including a number of release sites symmetrically distributed around the home loft are, for instance, necessary to test whether pigeons fly consistently homewards from various distant positions. At a single site, a direction preferred by a sample of pigeons could coincide with the homeward direction simply by chance. Figure 2.4 not only illustrates a statistical procedure, but also includes a message. Pigeons do not need some time of flying around after release to gain information about the direction of home. As little as 20 s after take-off they tend to prefer the semicircle including the homeward direction, though their bearings after 20 and 40 s are usually more scattered than at the time of vanishing.greater angular scatter results in shorter mean vectors obtained from different samples of birds (Fig. 2.4 top: = a = mean vector length per release independently of direction). The directions of these vectors, however, independent of their lengths, are well homeward oriented after 20 s as they are some 2 or 5 min later (Fig. 2.4, below). Obviously, the greater initial scatter is merely a technical outcome of the fact that the birds need some time to gain height and to orient their flight from an arbitrary starting course to an actively intended direction. Moreover, observed bearings are rarely identical with the bird s current flight direction. At a short distance, i.e. early after release, a flight path including a considerable lateral component with respect to the observer s view causes a much greater angular displacement than a linearly equally long path at a distance of 1 2 km. Thus, the pigeons directional tendencies are less clearly recognizable at the beginning, but the equal vector lengths and homeward components in the lower diagrams of Fig. 2.4 suggest that the pigeons do not need to circle around in order to obtain information about their position. Otherwise, also mean directions would initially be more scattered and less clearly homeward-oriented. More specific resultslaterleadtothesameconclusion(sect.7.2.1). 2.2 Homing Performance 2.2 Homing Performance 11 Another easily obtainable criterion is homing performance, i.e. homing duration or speed and homing rate. Figure 2.5 shows an example, presenting the same data in graphically different versions. Speed is expressed in units of beeline distance per time (km/h: Fig. 2.5A, B) and duration in time per unit of beeline distance (min/km: Fig. 2.5C, D). The histograms result in reversely skewed distributions. In either version, there are usually two classes of pigeons that do not fit to the scale. One class includes birds that homed very slowly,

12 2 Observation Data Used to Investigate Pigeon Homing Fig. 2.5. Homing performances of 486 inexperienced pigeons in 37 releases over 22 29 km distance in southern Bavaria. The same data are shown in four kinds of diagrams (unpubl. data) mostly on the day after the release or later. The second class concerns pigeons that never returned at all. With greater distances of displacement, and particularly with inexperienced pigeons, the majority of the birds may belong to these two classes. In more extreme cases, only the return rate may be available as a measure of homing performance. In the following sections, graphs of the type in Fig. 2.5D illustrate homing performance. Using this type, results obtained with different experimental treatments can be optimally compared within one diagram; the median and other percentiles of a sample can immediately be read (see abscissa values corresponding to ordinate values 25, 50 and 75% in Fig. 2.5D). As 1 min/km (= 60 km/h) indicates the approximate flight speed of pigeons, the abscissa roughly indicates the factor by which the routes flown exceed the beeline distance, provided that the birds did not interrupt their flight by some period(s) of rest. 2.3 Homing Routes Most of the presently available data do not include any information about the routes the pigeons have flown between vanishing a few minutes after release and arrival at the loft. In some experiments, however, the complete flight paths, or parts of them, were observed by one of the following four methods.

2.3 Homing Routes 13 1. Observation from airplane or helicopter. Using a slowly flying airplane (Hitchcock 1952) or a helicopter (Wagner 1970, 1972, 1974; Fiaschi et al. 1981), flocks of differently coloured pigeons could be visually observed while flying over considerable distances. The method is hardly usable to follow individually flying birds and is too laborious and expensive to collect data samples that are sufficiently large for statistical evaluations. Nevertheless, obvious responses to the local topography were observed (Sect. 3.4.3) and altitudes of flight were found to be mostly less than 100 m, and often less than 50 m, above ground. 2. Radio telemetry. Using ground-level receiving antennae, pigeons carrying small radio transmitters can be followed over distances of some 10 20 km (e.g. Windsor 1975; Walcott 1977, 1978). With one receiver at the release site, the range for vanishing bearings is thus correspondingly larger than the visual range. Using a number of receivers distributed over an extended area, flight paths can be determined by means of triangulation (Schmidt- Koenig and Walcott 1978; see Fig. 8.3). Basically unlimited recording of routes is possible when radio telemetry is combined with airplane tracking (Michener and Walcott 1978). Owing to the considerable expenditure of costs and time for each single bird, however, such techniques never found broader application in pigeon homing research. In more recent years, a much more elegant and efficient technique has been developed: that of satellite telemetry (e.g. Fig. 10.6). During its orbit around the earth, a satellite determines the position of a transmitter from time to time. As the transmitters necessary for this purpose are still fairly heavy, they have so far been used only with birds larger than pigeons. As miniaturization of devices will probably go on, we can expect that in several years this technique will become a useful tool also in pigeon homing research. However, for many purposes the meanwhile developed position recorder (see point 4 below) may be superior. 3. Direction recorder. Broader databases could be achieved by the development of a miniaturized device that measures and stores, at regular intervals, the compass direction of the pigeon s body axis during the whole homing flight (Bramanti et al. 1988; Dall Antonia et al. 1992). In its newer version it weighs only 13 g and hence can easily be carried by a pigeon. After the bird s arrival at home, the stored data can be extracted from the digital memory and flight paths reconstructed. A source of possible error and uncertainty in such reconstructions is wind drift. Nevertheless, carefully conducted experiments and evaluations have led to informative results, especially to findings illustrating topographical influences (e.g. Ioalè et al. 1994; Bonadonna et al. 1997, 2000). For an example of routes recorded with this device see Fig. 7.13B. 4. Position recorder. A most promising new technique has recently been described by von Hünerbein et al. (2000) and Steiner et al. (2000). Birds carry a recorder using the satellite-based global positioning system (GPS) which can store highly accurate position fixes at intervals of seconds. Data can then be downloaded and true flight paths plotted on a map (Fig. 2.6) and analysed. At

14 2 Observation Data Used to Investigate Pigeon Homing Fig. 2.6. Examples of homing flight routes of pigeons recorded by means of a GPS device. In this particular case, deviations to the east from the direct route appear to be typical. (Modified from von Hünerbein et al. 2000) thetimeofwriting,theweightofthedeviceis33 35gandhenceatthelimitof being suitable for pigeons. For first applications of GPS tracking beyond technical testing see Biro et al. (2002) and Guilford et al. (2004). As useful as they are and probably will be, a disadvantage of the two last-mentionedtechniquesisthefactthatthebirdsmustberecapturedinordertogain the recorded data and recover the fairly expensive device. Therefore, they are not very well suited for risky experiments in which many pigeons fail to return to their home loft. Just in the course of such experiments, however, which may, for instance,test the influence of a particular experimental treatment,it would often be particularly important to obtain information on the whereabouts of the non-homers. For their localization, either satellite telemetry (point 2) or a combination with recoveries (next section) may become a suitable future method. 2.4 Recoveries As regards observation of longer-distance movements, modern high-tech methods are presently just at the point of replacing classical primitive sources of information. Results shown in this book are still based on such sources. Homing pigeons are social and hence attracted by conspecifics they meet on their way. Thus, if they do not reach their home loft, they tend to enter another pigeon loft. In some countries, there is a quite dense network of such lofts.

2.4 Recoveries 15 Many of the fanciers have their pigeons under daily control and detect any foreigner among them.if such a foreigner has a conspicuous address label around its ring, possibly promising an award, the pigeon has a good chance of being reported to its owner. In Germany, about half of the non-returning pigeons, if labelled in such a way, have been reported (Wallraff 1989a). Recoveries show, Fig. 2.7. Vanishing bearings (circular diagrams) and recovery sites of pigeons from three lofts in the north (H Hohenkirchen; W Wilhelmshaven; O Osnabrück) and one loft in the south (F Freiburg) of a central release area near Giessen, Germany. Lines with arrowhead indicate pigeons that homed after re-release at one or two recovery sites (a.rep after report). Most of the birds were moreorlessexperiencedinhomingovershorterdistances(upto45km),theirlastpreviousflight not being from the direction used here. (After Pratt and Wallraff 1958, from Kramer 1959b)