HBW 13 - Foreword on bird migration by Ian Newton

View foreword in PDF format: 

The term migration, as used by ornithologists, is usually understood as a regular, large-scale return movement of individual birds twice each year between restricted breeding and wintering areas. Non-migratory birds are usually described as sedentary or resident because they occupy the same areas year round, and their populations show no obvious seasonal shifts in distribution. However, closer examination of bird behaviour reveals a more complex and variable array of movement patterns, which for convenience can be divided in to six main types (after Newton 2008):

•  Routine movements centred on the place of residence. As part of everyday life, such movements occur in all birds, whether classed as resident or migratory. They include the flights from nesting or roosting sites to feeding sites, or from one feeding site to another. In most birds these movements are short and localised, restricted to a circumscribed home range, and extending over distances of metres or kilo­metres. But in other species (notably pelagic birds) regular foraging movements can extend over hundreds of kilometres out from the nesting colony.

•  One-way dispersal movements also occur in both sedentary and migratory bird species. After becoming independent of their parents, young birds typically disperse in various directions from their natal sites. Individual young seem to have no specific inherent directional preferences, so within a population, dispersal movements occur randomly in all directions, unless constrained by local topography. In most bird species, dispersal distances can be measured in metres, kilometres or tens of kilometres, but in a few species (notably pelagic birds) such distances can be much greater. Post-fledging dispersal of this type does not usually involve a return journey (see below), but in any case most surviving young subsequently settle to breed at some distance from their hatch-sites (called natal dispersal). In addition, some adults may change their nesting locations from year to year (breeding dispersal), or their non-breeding locations from year to year (non-breeding or wintering dispersal). In general, dispersal movements cause no seasonal change in the centre of gravity of a population or in its overall geographical distribution.

•  Migration in which individuals make regular return movements, at about the same dates each year, often to specific destinations. Compared with the above movements, migration usually involves a longer journey over tens, hundreds or thousands of kilometres and in much more restricted and fixed directions. Most birds spend their annual non-breeding period at lower latitudes than their breeding period. Such migration occurs primarily in association with seasonal changes in food availability, resulting from the alternation of warm and cold seasons at high latitudes, or of wet and dry seasons in the tropics. Overall, directional migration causes a massive movement of birds twice each year between regular breeding and wintering ranges, and a general shift of populations from higher to lower latitudes for the non-breeding season.

•  Dispersive migration in which post-breeding movements can occur in any direction from the breeding site (like dispersal), but still involve a return journey (like other migration). Although these movements occur seasonally between breeding and non-breeding areas, they do not necessarily involve any change in the latitudinal distribution of the population, or any change in its centre of gravity. They are evident in some landbird species usually regarded as ‘resident’, but also include altitudinal movements in which montane birds shift in various directions from higher to lower ground for the non-breeding season. In addition, many seabirds can disperse long distances in various directions from their nesting colonies to over-winter in distant areas rich in food, but returning to their colonies the following spring.

•  Irruptions (or invasions) are like other seasonal migrations, except that the proportions of birds that leave the breeding range, and the distances they travel, vary greatly from year to year (the directions are roughly the same but often more variable between individuals than in regular migration). Such movements are usually towards lower latitudes, and occur in association with annual, as well as with seasonal, fluctuations in food supplies. In consequence, populations may concentrate in different parts of their non-breeding ranges in different years. Examples include some boreal finches that depend on sporadic tree-seed crops, and some owls that depend on cyclic rodent populations.

•  Nomadism in which birds move from one area to another, residing for a time wherever food is temporarily plentiful, and breeding if possible. The areas successively occupied may lie in various directions from one another. No one area is necessarily used every year, and some areas may be used only at intervals of several years, but for months or years at a time, whenever conditions permit. The population may thus be concentrated in largely different places in different years. This kind of movement occurs among some rodent-eating owls and raptors of tundra and boreal regions, and among many birds that live in desert regions, where infrequent and sporadic rainfall leads to local changes in habitats and food supplies. Because these changes are unpredictable from year to year, individual birds do not necessarily return to areas they have used previously, and may breed in widely separated areas in different years.

These different kinds of bird movements intergrade, and all have variants, but in any bird population, one or two kinds usually prevail. Almost all bird species show post-fledging dispersal movements, in addition to any other types of movement shown at other times of year, and some species show both nomadic and irruptive movements. Through migration, irruption and nomadism, birds exploit the resources of mainly different regions at different times. The birds thereby maintain greater survival or reproductive success (and hence greater numbers) than if they remained permanently in the same place, and adopted a more sedentary lifestyle. In other words, because of their movements, populations probably persist at considerably larger size than they otherwise would.

The main variables in these different types of bird movements include: (1) the directions or spread of directions; (2) the distances or spread of distances; (3) the calendar dates or spread of dates; and (4) whether or not they involve a return journey. They also differ in whether they occur in direct response to prevailing conditions, or in an ‘anticipatory’ manner, in adaptation to conditions that can be expected to occur in the coming weeks, and leading birds to leave areas before their local survival would be compromised or arrive in other areas in time to breed when conditions there are suitable. Each of these aspects of bird movement behaviour can be independently influenced by natural selection, giving overall the great diversity of movement patterns found among birds, related to the different circumstances in which birds live. In the rest of this chapter, however, I shall be concerned primarily with regular seasonal migration —that large-scale seasonal shift in bird populations which has fascinated people throughout history, raising obvious questions such as why birds do it, where they come from or go, and how they find their way.

Migration as a product of natural selection

Migration might be expected to occur wherever individuals benefit more, in terms of survival or reproduction, if they move seasonally between different areas than if they remain in the same area year-round (Lack 1954). The usual reason why breeding areas become unsuitable during part of the year is lack of food. Such food shortages occur for many birds because plant growth stops for part of the year, and many kinds of invertebrates die or hibernate or become inaccessible under snow and ice. At high latitudes, daylengths also shorten in winter to such an extent that many diurnal birds would have too little time to get enough food, even if it were available. Hence, the purpose of the autumn exodus from high latitudes is fairly obvious.

The reason why birds leave their wintering areas to return in spring is sometimes less obvious, because many wintering areas seem able to support the birds during the rest of the year. But if no birds migrated to higher latitudes in spring, these latitudes would remain almost empty of many species, and a large seasonal surplus of food would go largely unexploited. Under these circumstances, any individuals that moved to higher latitudes, with increasing food and long days, might raise more offspring than if they stayed at lower latitudes and competed with the birds resident there. So whereas the advantage of autumn migration can be seen as improved winter survival, dependent on better food supplies in winter quarters, the main advantage of spring migration can be seen as improved breeding success, dependent on better food supplies in summer quarters. Compared to survival, reproduction often has more stringent requirements in terms of specific food needs and predation avoidance.

In effect, migration reduces the seasonal fluctuations in food supplies to which a breeding population could otherwise be exposed. This does not rule out an influence of other factors, such as reduced predation, parasitism or competition, all of which have at one time or another been proposed as contributing to the evolution of migration, but on scant evidence. Whatever the main selective forces, however, the migratory habit ensures in the long term that species in seasonal environments adopt and maintain movement patterns that allow individuals to survive and breed better than if they remained in the same area year-round.

Global patterns

Throughout the world, migration is most apparent wherever the contrast between summer and winter (or wet season and dry season) conditions is great. Migration thus allows individual birds to exploit different areas at different times of year, whether to benefit from seasonal flushes of food or to avoid seasonal shortages. In fact, some migratory birds occupy habitats over winter that they could not use for breeding, and then occupy breeding areas that would not support them in winter. This applies to all arctic nesting shorebirds which spend the winter on sea coasts where, due to tidal flooding, nesting would be impossible, and then migrate north to breed on the arctic tundra which is frozen and snow-covered for the rest of the year. Thus some bird species exist over much or all of their range only by exploiting widely separated habitats at different seasons.

Although most marked at high latitudes, migration also occurs in the tropics, especially in the savannahs and grasslands exposed to regular wet and dry seasons. In the northern tropics, for example, many species move south for the dry (non-breeding) season, some crossing the equator, while in the other half of the year many species of the southern hemisphere move north. In contrast, birds confined to lowland equatorial rainforest are probably the least migratory, especially the small insectivores of the understorey where conditions remain relatively stable and suitable year-round. This year-round consistency in the rainforest environment removes any advantage in moving, and many individuals may remain within the same few hectares throughout their adult lives. In the same forests, however, some nectar-eaters and fruit-eaters move within small latitudinal or altitudinal bands in response to flowering and fruiting patterns, while other birds from higher latitudes move in for their ‘winter’.

In general, the proportions of species which are migratory increase from equatorial to polar regions, as the contrast between summer and winter conditions widens (Newton & Dale 1996a, 1996b). Progressing northward up the western seaboard of North Africa and Europe, for example, the proportion of breeding bird species which move out totally to winter further south increases from 29% of species at 30°N (North Africa) to 83% of species at 80°N (Svalbard), a mean increase of 1.3% of breeding species for every degree of latitude (Figure 1).

This relationship holds, it has been suggested, because at high latitudes the numbers of resident birds (species and individuals) are held at low level by winter severity. The flush of food in summer is greater than the small number of resident species can exploit, leaving a surplus available for summer migrants. The latter therefore increase in proportion with latitude, as the severity of the winters increase, and the numbers of year-round resident species decline. At lower latitudes, a large proportion of breeding species can remain year-round, leaving fewer openings for summer visitors (Herrera 1978, O’Connor 1985).

In the largely different avifauna of eastern North America, the proportion of migrants among breeding species also increases with distance northwards from 12% at 25°N to 87% at 80°N, a mean increase of 1.4% per degree of latitude (Figure 2; Newton & Dale 1996b). The difference between eastern North America and western Europe (Figure 2) reflects the climatic shift between east and west sides of the Atlantic: over most of the latitudinal range, at any given latitude winters are colder in eastern North America than in western Europe. Correspondingly, at any given latitude, a greater proportion (on average about 17% more) of breeding species leaves eastern North America for the winter than leaves western Europe. The slopes of the two linear regression lines calculated from the data in Figure 2 do not differ significantly, but their positions do, reflecting this climatic difference.

On both continents, this northward trend in the prevalence of migration is easily understood in terms of winter conditions. In Europe, during much of the 20th century, mean January temperatures exceeded 10°C only in southern Spain and North Africa; they lay within the range 0–5°C in much of western Europe, but fell below freezing and as low as –15°C in most of Fennoscandia, plunging to –20°C in Novaya Zemlya in the far north. Minimum winter daylengths were around 11 hours at 35°N in southern Europe but decreased to zero at the Arctic Circle. The season of plant growth lasted 6–9 months at 35–50°N, but shrank to less than three months in Svalbard, a mean decline in growing season of about one month for every 11° of latitude. In continental western Europe, most fresh waters north of 55°N froze in winter, although they mostly remained open in Britain and Ireland. Much of the Baltic and Barents Seas also iced over during the course of the winter, closing these areas for seabirds. In North America, similar latitudinal trends occurred, but were more marked because the continent spans a wider range of latitude than Europe. Throughout much of these areas, winter temperatures are now rising, as part of ‘global warming’, and bird migration patterns are changing accordingly, but they still relate to gradients in prevailing conditions.

The few species that remain to winter in the far north include the Common Raven (Corvus corax), Rock Ptarmigan (Lagopus mutus), Gyrfalcon (Falco rusticolus) and Snowy Owl (Nyctea scandiaca) among landbirds, and the Northern Fulmar (Fulmarus glacialis), Ivory Gull (Pagophila eburnea) and Glaucous Gull (Larus hyperboreus) among seabirds. The most northerly seabirds depend in winter on the open water provided by polynyas, and some of the gulls also scavenge the remains of seals killed by Polar Bears (Ursus maritimus). However, some individuals of these species may move south to some extent in the darkest weeks of winter.

The relationship between the proportion of migrants and latitude, established above for Europe and North America, continues southwards towards the equator. There are insufficient data to examine the trend in detail, but by 8° latitude in Panama, only five (0.6%) out of 807 breeding species are wholly summer migrants (Ridgely & Gwynne 1989). This is consistent with the regression line between proportion of migrants and latitude derived from the data for North America in Figure 2. The five migratory species found at Panama are all insectivores which leave for the winter dry season and head further south, namely American Swallow-tailed Kite (Elanoides forficatus), Plumbeous Kite (Ictinia plumbea), Common Nighthawk (Chordeiles minor), Piratic Flycatcher (Legatus leucophaius), and Yellow-green Vireo (Vireo flavoviridis).

The proportions of all bird species that are migratory are correlated not only with latitude, but also with various climatic factors that vary with latitude, such as the temperatures of the hottest or coldest months or the temperature difference between the hottest and coldest months (Newton & Dale 1996a, 1996b). These various measures are of course interrelated, but what really matters is the degree of climatic difference between summer and winter. It is this difference that, for many birds, governs the difference in food supply between summer and winter at particular latitudes, and hence the difference in environmental carrying capacity between the two seasons.

The seasonal difference in carrying capacity may also vary from west to east, according to changes in climate (as between west and east sides of the Atlantic). From west to east across Europe, summer climates become warmer and drier, and winter climates become colder. In consequence, progressing eastward through Europe into Asia, increasing proportions of the local bird breeding species become migratory. This is especially obvious in comparing populations of coastal areas that live under mild oceanic climates with those further inland that live under more extreme continental climates. For example, Common Starlings (Sturnus vulgaris) live year-round on the Shetland Islands at 60°N, while at the same latitude in Russia (and for 10–15° south of it) Common Starlings are wholly migratory.

Superimposed on the overall latitudinal trend is another related to diet. Broadly speaking, those species that are resident year-round in a particular region exploit food sources that remain available there all year, whereas those that leave after breeding exploit foods that disappear about that time. In the northern coniferous forests, for example, residents include mainly species that feed directly from trees, on bark-dwelling arthropods (tits, treecreepers), fruits and seeds (some corvids, finches, tits), buds or other dormant vegetation (grouse), or that eat mammals and other birds (some corvids, raptors and owls). Almost the entire resident landbird fauna at high northern latitudes falls into one or other of these dietary categories. In contrast, species that depart for the winter include those which eat active leaf-dwelling or aerial insects (warblers, hirundines) or which eat foods that become inaccessible under snow or ice (ground feeding finches and thrushes, some raptors, waterfowl and waders). Towards the equator, as winters become less severe, the range of bird dietary types that remain for the winter increases, as a wider range of food-types remains available year-round.

This relationship between migration and diet means that, in some mid-latitude areas, similar numbers of species may be present in summer and winter, but species composition changes somewhat between seasons, as some species from lower latitudes are present only in summer and other species from higher latitudes only in winter (Newton & Dale 1996a, 1996b). In southern England, for example, insectivorous swallows and warblers arrive from the tropics for the summer, whereas fruit-eating thrushes from further north arrive for the winter. Seasonal changes in bird communities in particular regions are thus tied to seasonal changes in the types of food available. This emphasises the point that migrants often exploit seasonal abundances in both their breeding and non-breeding areas. It is a strategy that, for obvious reasons, is much more developed in birds than in most other animals.

The overall effect of bird migration is to alter the latitudinal distribution of birds between summer and winter, so that species numbers in the northern hemisphere are at their greatest in the northern summer and in the southern hemisphere in the austral summer (northern winter). Take the west European migrants as an example. Some species move relatively short distances within Europe, but others move longer distances to Africa or southern Asia (Newton 1995). But the net result, each autumn and spring, is a huge latitudinal shift in avifaunal distribution (Figure 3). In summer, the whole European assemblage of breeding birds is (by definition) concentrated north of 25°N, but in winter the same assemblage extends southwards as far as the southern tip of Africa (35°S) and beyond. Forty-eight species of Palearctic birds reach the southern Cape of South Africa (Harrison et al. 1997), and some seabirds extend into the seas further south. When they are in their wintering areas, the migrants add to the local species, increasing the overall species numbers, especially in the tropics.

The above analyses (Figures 1-3) were based on the presence or absence of species at particular latitudes in winter. They were therefore based only on complete migrants, in which all individuals move out for the winter, while for purposes of analysis partial migrants (in which only a proportion of individuals leave for the winter) were counted as year-round residents. However, in many species that breed over wide areas, a greater proportion of individuals migrate from higher than from lower latitudes. Thus, some such species in the northern hemisphere are completely migratory in the north of their breeding range and completely sedentary in the south, while in intervening areas some individuals leave and others stay (partial migration). European examples include Common Blackbird (Turdus merula) and Peregrine Falcon (Falco peregrinus), and North American examples include American Robin (Turdus migratorius) and Red-tailed Hawk (Buteo jamaicensis). In general, therefore, the extent to which any population migrates for the winter broadly corresponds to the degree of seasonal reduction in food supplies. Taking account of partial as well as complete migrants, the latitudinal trends discussed above would be even more marked.

Altitudinal migration

By moving a few hundred metres down the sides of a mountain, birds can achieve as much climatic benefit as by moving several hundred kilometres to lower latitudes, but without the extra winter daylength. Mirroring the latitudinal trend, with rising altitude, increasing proportions of breeding species move out for the winter, but in contrast to latitudinal migration, altitudinal movements can be in any directions that reach lower ground. They occur on mountain ranges worldwide, from the Himalayas to the Andes, and can involve a large proportion of local montane species. Examples include the Citril Finch (Serinus citrinella) and Water Pipit (Anthus spinoletta) in Europe, and the Rosy Finch (Leucosticte arctoa) and White-tailed Ptarmigan (Lagopus leucurus) in western North America. Seasonal altitudinal movements occur even on relatively low mountains, such as the Great Dividing Range in southeast Australia, where several montane species appear in lowland towns and farms in winter.

Comparisons between hemispheres

Migration of landbirds from their breeding areas is a much more obvious phenomenon in the northern hemisphere than in the southern. This is partly because land covers three times the area in the northern hemisphere as in the southern hemisphere, and the difference is most marked at high latitudes (we can ignore Antarctica because it holds no landbirds). In North America, Greenland and Eurasia, some landbird habitat extends north of 80°N, but in the southern hemisphere, South America reaches only to 55°S, Africa to 35°S, Australia to 43°S, and New Zealand to 47°S. The net result is that latitudes 30–80°N hold 15 times more land than do latitudes 30–80°S, and it is at these latitudes that winters are coldest, and migration is most developed (for discussion of area effects in South America see Chesser 1994, and in Australia see Chan 2001). The greater latitudinal spread of land in the northern hemisphere results not only in more marked migration, but also in generally longer journeys than are undertaken by southern hemisphere breeders, which are closer to the equator.

Another factor is temperature, which has a steeper downward gradient north of the equator than south of it. For example, in the New World the mean midwinter (January) temperature at the Tropic of Cancer is about 13°C, while at 50°N it is –15°C, a 28°C difference. In contrast, the mean midwinter (July) at the Tropic of Capricorn is 16°C, while at 50°S it is 0°C, a 16°C difference. Similar hemisphere differences are apparent in much of the Old World too. This difference in steepness of temperature gradient between hemispheres may explain why greater proportions of species leave from temperate latitudes in the northern hemisphere than in the southern. For example, about 29% of species leave completely for the winter from Morocco, but only 6% of species leave from equivalent latitudes around the Cape in South Africa (Harrison et al. 1997, Newton & Dale 1996a). It is presumably largely for both these land-related and temperature-related reasons that landbird migration is much more marked in the northern than in the southern hemisphere.

Regarding length of journeys, many landbird species that breed in the northern hemisphere migrate south of the tropics, yet no landbird species that breed in the southern hemisphere move north of the tropics. This may be because the much larger landbird populations of the northern hemisphere have to travel far to the south to find sufficient accommodation, while the smaller landbird populations of the southern hemisphere find sufficient accommodation by travelling relatively shorter distances to the north. This difference is another likely consequence of the imbalance in available land areas between the two hemispheres.

In extent of seasonal migration, pelagic seabirds provide a revealing contrast with landbirds. Associated with the reduced land areas in the southern hemisphere, the sea areas are correspondingly larger than in the northern hemisphere. Linked with these greater sea areas and larger numbers of scattered island breeding sites, pelagic seabirds breed much more numerously in the southern hemisphere than in the northern, both in terms of species and of individuals. Correspondingly, a greater proportion of southern than of northern hemisphere breeding seabird species make long migrations. Five (11%) of 47 species that breed north of the tropics extend to south of the tropics in the northern winter, whereas 14 (23%) of 61 species that breed south of the tropics extend to north of the tropics in the austral winter (calculated from maps in Harrison 1983). Some seabird species that breed in the southern hemisphere, such as Wilson’s Storm-petrel (Oceanites oceanicus) and Sooty Shearwater (Puffinus griseus), are amongst the commonest seabirds in the North Atlantic and North Pacific in the northern summer (austral winter).The implication is again that the sheer numbers of individual birds, in relation to the habitat available, influence the distances moved and areas occupied outside the breeding season.

Species that migrate between the northern and the southern hemispheres gain the advantage of ‘summer’ conditions in both their breeding and non-breeding homes. The question then arises why the same individuals do not breed twice in one year, in both northern and southern quarters. One reason in many species is that individuals moult while in their non-breeding areas, a process that takes several weeks or months and could not be undertaken at the same time as breeding (the two processes being mutually exclusive in most birds). Another reason is that many migratory species do not remain for long in the same area in winter, but periodically move to other areas in response to changes in food supplies. This exploitation of temporary abundances is one way in which migrants in the southern hemisphere could avoid competing with the local birds which, breeding at that time, are tied to fixed nesting areas. Neither explanation applies to all trans-equatorial migratory species, however, and there are still some that are sedentary while they are in both breeding and non-breeding areas, and would seem able to breed in both, six months apart, but do not.

Sex and age differences

In many bird populations, sex and age differences occur in the proportions of individuals that migrate, in the timing of outward and return movements, and in the distances travelled. The latter give rise to geographical gradients in sex and age ratios from high to low latitudes across the wintering range. In many species, adult birds winter, on average, nearer to the breeding areas than first-year birds, and males, on average, nearer than females. Winter sex ratios have been studied in detail in the Common Pochard (Aythya ferina) in which the proportion of females in local flocks increases from north to south across the European wintering range (Carbone & Owen 1995).

Sex differences in migration within species can often be linked to the different roles of the sexes in breeding, and both sex and age differences partly to dominance and competition within populations, duration and timing of moult, and perhaps also to other (as yet unknown) factors. Thus, in many bird species, males are responsible for establishing nesting territories, and arrive in breeding areas before females. In addition, in species in which only one partner looks after the young, the other leaves breeding areas earlier (males in ducks, males in some shorebirds, females in other shorebirds). In Curlew Sandpipers (Calidris ferruginea), the males play no part in parental care, so depart early, leaving the females to raise the young. In Spotted Redshanks (Tringa erythropus), the opposite occurs, in that the females depart earlier, leaving the males to rear the young. In yet other waders, both partners help to the same stage with parental care, and the two sexes migrate at about the same time, as in the Northern Lapwing (Vanellus vanellus) and Black-tailed Godwit (Limosa limosa). Typically, in all these species, adults leave breeding areas earlier than young of the year.

Variants on a migratory theme

Several variants occur on the return two-way movement between regular breeding and wintering areas, associated with the ecology of the species concerned.

Moult migration

Moult migration is a regular movement which many waterbirds and others perform each summer, travelling long distances from their nesting areas to assemble at traditional sites that offer food and safety (Jehl 1990). Here they pass the flightless period, replacing all their large wing feathers simultaneously within the space of a few weeks. Geese migrate to moult mainly in latitudes higher than their nesting areas, but ducks migrate in any directions from their nesting areas. After moulting, depending largely on location, the birds may return to their nesting areas, move on to their wintering areas, or stay in their moulting areas all winter, returning to nesting areas in spring.

One of the best known moult migrations concerns Common Shelducks (Tadorna tadorna), in which individuals from large parts of Europe gather each summer on the vast tidal mudflats of the Grosser Knechtsand in the German Wadden Sea, where they feed on the abundant mud snail (Hydrobia ulvae). The birds converge on this site from all directions, travelling up to several hundred kilometres from their nesting areas. Their numbers peak at more than 200,000 individuals. Yearlings and young adults arrive first, followed by failed breeders, then successful adults which leave their well-grown young behind (Patterson 1982). After moult, the birds drift back to their breeding sites over a period of weeks or months or move on to wintering areas, depending on whether their particular breeding areas are habitable in winter. In some other duck species, individuals have been found by ringing to travel distances of more than 2000 km from breeding to moulting sites, and some of the biggest concentrations hold tens of thousands or hundreds of thousands of birds. In general, sea-ducks perform longer journeys than freshwater ducks. For example, King Eiders (Somateria spectabilis) from most of eastern Canada travel to coastal areas off mid-western Greenland to moult, travelling up to 2500 km, and forming a concentration of more than 100,000 birds (Salomonsen 1968). Moult migrations are also known among grebes, auks and other waterbirds.

Movements in the breeding season

Some multi-brooded bird species nest in more than one locality each year, migrating from one site to another during the course of the breeding season. Typically, individuals raise a brood in one locality, then migrate several hundred kilometres to another locality, and raise another brood. Such ‘itinerant breeding’ has been recorded in various seed-eaters, including the Red-billed Quelea (Quelea quelea) of the African savannas. Throughout the dry season, Red-billed Queleas subsist on dry grass seeds picked off the ground, but when the rains break, this seed suddenly germinates, thus removing the food supply. The birds then fly over the approaching rain front to areas where rain fell about two months earlier and new grass seed has already formed. This new seed, together with insects associated with growing vegetation, enable the birds to breed. After raising their young, the birds move en masse, stopping again in an area where rain has recently fallen, and raise another brood (Jaeger et al. 1986, Ward 1971). Conditions suitable for rearing young do not last long in any one place, and despite a short breeding season of only five weeks, Red-billed Queleas cannot raise successive broods at the same place. Only by remaining within the slowly shifting zone of seeding grasses are individual Red-billed Queleas able to raise more than one brood per year, each time moving some hundreds of kilometres.

Itinerant breeding has also been suspected among Common Quails (Coturnix coturnix), as they breed first in North Africa and then in the northern Mediterranean region, and possibly a third time in temperate Europe, as spring begins progressively later with distance northward (Cramp 1980). It has also been suspected among Common Redpolls (Carduelis flammea) in northern Europe, as they breed first in the boreal forests as conifer seeds are shed from the opening cones, and then further north on the strength of birch and grass seeds (Peiponen 1967).

Movements in the non-breeding season

Many birds make long movements within the non-breeding season, usually to escape food shortages. Some such movements occur in response to depletion of fruit and seed crops, while others occur in response to cold and snow. For example, following the first heavy snow-fall in temperate latitudes, thousands of birds can be seen moving southwards, particularly ground-feeding species, such as plovers, larks, pipits and finches. Such escape movements vary from year to year, depending on the severity of the winter, and in years without serious snow, no such movements occur.

Nomadism

Apparent nomadic movements occur in species which exploit sporadic habitats and food supplies, concentrating temporarily wherever conditions permit. They are shown by some boreal finches which move around to exploit sporadic tree-seed crops, by certain owls which exploit the cyclic peaks of rodent populations, and by many desert bird species which move around in relation to irregular rainfall which generates their food supplies (Dean 2005, Newton 2008). Such species seem to show little year-to-year consistency in their movement patterns, but shift from one area to another, residing for a time in whichever parts of their range food is plentiful at the time.

Every few years, however, at times of widespread food-shortage, such species may move longer distances, and appear well outside their usual range. In the central Australian desert, the populations of many species increase greatly as a result of good breeding in occasional wet years, and then many individuals move outwards to the more humid peripheral districts in the following dry years (Nix 1976). Many birds probably die before they can return, but individuals of some species spend long periods in coastal localities and occasionally move inland to breed in vast numbers when rain creates suitable conditions. For example, Banded Stilts (Cladorhynchus leucocephalus) can live for years as non-breeders on scattered briny coastal lagoons. But within days of rain falling inland, they concentrate in tens of thousands on newly formed shallow lakes, feeding on the freshly-hatched swarms of brine shrimps (Burbidge & Fuller 1982, Robinson & Minton 1989). They breed while conditions last, making repeated nesting attempts, while the young form huge crèches. Water evaporates rapidly, however, and the land soon resorts to its usual parched state. The birds return to the coast, if necessary leaving the last eggs and young to die. Years may pass before they can breed again, but not necessarily in the same sites. A similar outward movement of desert species at times of widespread drought is illustrated in Eurasia by the westward movements of Rose-coloured Starlings (Sturnus roseus), Pallas’s Sandgrouse (Syrrhaptes paradoxus) and others from the steppes into western Europe.

One of the puzzles of nomadic birds, which seem to move in any direction from one breeding area to another, is how they find their way to suitable areas. One possibility is that they search at random, gradually settling wherever conditions are good, but this seems inefficient, and desert birds, in particular, may have other means of detecting areas where rain has fallen, responding perhaps to atmospheric pressure changes, cloud formations or changing odours. As yet, though, it remains a mystery how birds find widely separated patches of temporarily favourable habitat in the vastness of the desert.

Migratory flight

The reason that long-distance movements are so well developed in birds is that most species can fly. One of the main advantages of flight is its speed, which is much faster than the alternatives of walking, running or swimming. Flight requires more energy per unit time, but because of the greater distance covered, it is also the cheapest mode of transport overall. One type of flight, by soaring–gliding, is cheaper still, but is practised mainly by larger species, such as albatrosses, which can travel the Southern Ocean with little more energy expenditure than sitting still. Long-distance flight also allows birds to cross hostile areas that would otherwise act as barriers to their movements. Nevertheless, while most birds migrate by flying, penguins and some other seabirds migrate by swimming, and some landbirds by walking for part or all of their journeys.

The flight speeds of birds have often been measured using a car or airplane travelling alongside, or by using radar to track the movements of particular flocks or individuals (Bruderer & Boldt 2001). Measures taken from a vehicle or airplane cannot be corrected for wind effects, and are often of doubtful accuracy, so are of limited value. Radar measures can be obtained specifically for birds on migration and can be corrected to allow for wind speed, but do not always provide a reliable identification of species. Other values used for comparative purposes are the theoretical flight speeds calculated from aerodynamic principles on the basis of body mass, wingspan and wing area (Pennycuick 1969). The main findings to emerge from all these various sources of information are that, while individual birds can vary their flight speeds according to circumstance, larger birds generally fly faster than small ones, although body and wing structure also have a major influence.

Birds are assumed to migrate, whenever possible, either at their ‘minimum power speed’ (which minimises energy cost per unit time) or at their ‘maximum range speed’ (which minimises energy cost per unit distance). Both speeds are somewhat slower than the maximum speed of which the bird is capable. On average, 10 g birds have a theoretical maximum range speed in flapping flight of around 22 km per hour, 20 g birds of around 32 km per hour, 100 g birds of around 55 km per hour, 1000 g birds of around 85 km per hour and 10 kg birds of around 90 km per hour (Pennycuick 1969). As a rule of thumb, the theoretical maximum range speed of birds roughly doubles for every 100-fold increase in body mass up to around 15–20 kg, the approximate weight limit for flying birds. These are mean theoretical values, however, and species of similar weight would be expected to vary somewhat in their actual flight speeds, according to body and wing shape and other features, which differ from one type of bird to another. Hummingbirds, pigeons, ducks and auks fly faster than expected from their body weight, while terns, harriers and owls fly slower.

In steady flapping flight, a bird must generate the forces which support its weight against gravity and which at the same time provide the forward thrust necessary to overcome the friction and other forces that make up drag. The power for both the lift and forward thrust is supplied by the breast muscles, while directional control is provided mainly by the tail. Given sufficient fuel reserves, some birds that migrate by continuous flapping flight, such as passerines and waders, can travel for hours or days on end. They can cross water or other hostile terrain, and can fly by night as well as by day. In moving between their breeding and wintering places, therefore, such birds often travel directly, taking the shortest routes. As populations, they migrate mostly on a broad front, but concentrate to some extent through mountain passes or along coasts or other ‘leading-lines’ that deviate little from their main direction. Because flapping flight is expensive, however, such species must normally lay down substantial body reserves, especially for travelling over large areas of sea or other inhospit­able substrate where they cannot feed. Sustained flapping also produces heat, which may enable birds to fly at high latitudes and altitudes without having to burn extra fuel to keep warm. In hot conditions, however, heat production can result in the need for evaporative cooling (panting), which increases water loss and dehydration risk.

The situation differs somewhat in birds that migrate mainly by soaring–gliding flight, notably the broad-winged raptors, storks and pelicans, which gain most of the energy they need for flight from the ambient atmosphere. Typically, these birds make use of updrafts created either by wind striking a cliff or slope being directed upwards, or by ‘thermals’, which are columns of rising air caused by uneven heating of the ground. Typically, soaring birds circle upwards in a thermal, then glide with loss of height to the next thermal and rise again; they repeat this process again and again along the migration route, and over long distances in ideal conditions they seldom need to flap their wings (Figure 4). Because the lift comes largely from rising air currents and the forward motion partly from gravity, this still-wing flight mode requires much less internally-generated energy than continuous flapping. Many soaring species use a mix of gliding and powered flight, with intermediate costs, but seek to maximise the contribution from gliding, resorting increasingly to flapping as thermal conditions deteriorate. Flapping during a glide (‘power gliding’) can provide additional lift and speed, but at a cost in fuel use. The extreme soaring species thus depend for their movements mainly on a source of energy external to their own bodies, and unlike flapping birds, soaring species can continually correct for the effects of cross-winds without wasting energy.

Soaring–gliding flight has other consequences. Because of their dependence on updrafts, soaring landbird species must migrate mainly over land, favouring routes where appropriate conditions develop. They are also constrained to travel by day when the sun heats the land surface, creating rising air currents. Their migration typically reaches its peak, and moves most rapidly, in the middle part of each day when thermal activity is greatest. It is then that the birds achieve the greatest heights, and can make the longest and fastest glides across country (Spaar 1997, Spaar & Bruderer 1996). Soaring landbirds also tend to concentrate along narrow land bridges (such as Panama), or at narrow sea crossings (such as Gibraltar or the Bosphorus), and thereby avoid spending long periods over water where thermal soaring is seldom feasible. In this way, soaring landbirds often take long roundabout routes between breeding and wintering areas in order to make as much of their journey as possible over land, and minimise their use of expensive flapping flight. Despite the greater distances, their total energy consumption is thereby much reduced. Moreover, because their travel routes are determined by geography and topography, they tend to take the same traditional narrow-front ‘corridor’ routes year after year.

Soaring–gliding flight is not confined to landbirds. Many seabirds make use of up-currents formed either as the wind is deflected off waves (equivalent to slope soaring), or as a wave of ‘swell’ displaces air upwards as it moves over the sea surface. Some seabirds also use ‘dynamic soaring’, which depends partly on wind speed being slowed by the sea surface, an effect which is lessened with height up to about 16 m. The bird first climbs into the wind, then makes a high leeward turn, gaining distance by gliding with the wind whilst losing height. After making a low turn in the trough of a wave, it starts the cycle again. A bird could also make use of discontinuities in wind flow near the sea surface, as it flies first behind a wave crest and then emerges for a time into the unobstructed wind. At this moment, the bird tilts its body so that the temporary gust strikes its ventral surface, providing lift, enabling further onward gliding flight (Pennycuick 2002). Over most oceanic areas, soaring seabirds are normally constrained to fly low over the sea surface, where conditions are most favourable.

Migration mode

Some birds may benefit by migrating as rapidly as possible, thus minimising the time spent on travel (the ‘time minimisation model’ of Alerstam & Lindström 1990). Such birds would then gain the longest possible time on their breeding, wintering or moulting sites, but would require large fuel stores to permit long, non-stop flights. Other birds may have food available throughout their migration route, so that they can stop and feed almost anywhere. Because heavy fat loads mean greater transport costs, one way to save energy is to keep fuel loads small and fly only short distances at a time, refuelling as necessary (the ‘energy minimisation’ model). Moreover, because extra weight reduces flight performance (notably climb rate and agility), minimising fuel loads can also reduce predation risks (the ‘predation minimisation’ model). The second and third options may thus be combined as the ‘load-minimising’ strategy. The particular migration mode adopted by individuals in any population might be a compromise between any of these different options, depending partly on the type of terrain over which birds travel, the distribution of potential feeding sites, and the risks of predation. Moreover, any ‘ideal strategy’ that the bird might have is likely often to be compromised by external conditions, such as adverse weather or poor food supplies.

Difficult journeys

Landbirds that migrate over oceans provide some of the most extreme examples of endurance flight. They travel without opportunity to feed, drink or rest, over vast stretches of open water, where they cannot stop, as birds do overland, when the weather turns against them. Yet millions of landbirds regularly cross the Mediterranean Sea and the Gulf of Mexico at their widest points (about 1200 km), and smaller numbers regularly cross longer stretches, such as the western Atlantic between northeastern North America and northeastern South America (2400–3700 km), or the northern Pacific between Alaska and Hawaii and other central Pacific Islands (5000–6000 km). To judge from their normal flight speeds, landbirds would take more than 100 hours of non-stop flight in still air to accomplish the longer of their overwater journeys, but by taking advantage of favourable winds, they can shorten their flight times, sometimes by as much as one half. Participants include many passerines and shorebirds, but also waterfowl which, unlike the others, can (and do) rest on the sea if need be. However, the most impressive of all overwater migrations by a landbird is undertaken by Bar-tailed Godwits (Limosa lapponica) from eastern Siberia and Alaska, which in autumn apparently accomplish an astonishing 175-hour non-stop 10,400 km flight to New Zealand (Gill et al. 2005). These figures are minima, and recent satellite-based radio-tracking of godwits has shown that both non-stop flight times and distances can be even longer.

Some overland journeys are also difficult. Long desert crossings are made by the many species (including passerines) that travel between Eurasia and tropical Africa (Moreau 1961). Most west European species cross at least 1500 km of the Sahara Desert immediately after crossing the Mediterranean Sea. In autumn some species may make this Mediterranean–Saharan flight without a break, a total journey of 1500–2700 km, depending on the route taken. Other individuals on the same journey fly by night but stop and rest in the shade by day, but without feeding or drinking (Biebach 1990). Yet other birds from further east cross the central Asian deserts, and then another 1700 km of southern Arabia and its bordering gulfs, before reaching East Africa. In Australia, some shorebirds cross the central desert in moving between southeast and northwest coasts, a journey of more than 2000 km.

Yet other birds cross high mountain ranges, including the Himalayas and Tibetan plateau. One such species is the Bar-headed Goose (Anser indicus) which in the pro­cess can rise to more than 8 km above sea level, where the air is thin and very cold. Other species cross extensive areas of pack-ice that lie in spring between Siberia and Alaska or between Norway and Svalbard. A few species cross 2000 km of the 2-km-high Greenland ice cap on journeys between northeastern Canada and western Europe.

Seabirds migrating entirely over the sea would seem to have plenty of opportunity to pick up food en route. But this is not always the case. Many species breeding at high latitudes migrate over the equator, and tropical seas are notoriously poor in food. In any case, foods such as surface fish tend to be concentrated in particular nutrient-rich places, which may be few and far between. Evidence is accumulating that, like some landbirds, some seabirds refuel at widely-separated staging areas before continuing migration. For example, after breeding in western Europe, Black Terns (Chlidonias niger) assemble at one major feeding area, the IJsselmeer on the Dutch coast. Here they increase in body mass by 25–30% within 2–3 weeks, which would then enable a non-stop flight of more than 3600 km to West Africa. The birds ascend in the evening to more than 500 m and start migration at night. Although Black Terns are seen at localities en route, no important stopover site is known between the IJsselmeer and West Africa (van der Winden 2002). In Namibia, a similar increase in the body mass of Black Terns was noticed prior to spring migration.

Duration and speed of migration

It is not so much the speed of flight that determines the speed of migration, but the time spent on stopovers where fuel reserves are replenished. Migration speed is a function of both flight times and stopover times. From knowledge of the average fuelling rates and flight costs of small passerines, it has been calculated that, for every hour spent in flight, about seven hours would need to be spent on the ground, feeding and resting (Hedenström & Alerstam 1998). Such a ratio has been borne out by field observations. In larger birds which refuel at a slower rate, the proportion of stopover time is two to three times greater than in passerines (Newton 2008).

Actual migration speeds have been measured mainly by use of ring recoveries of birds identified at different points on their migration routes, or in recent years from the satellite tracking of radio-marked individuals or other electronic means. Such studies have revealed that the times spent on migration by different bird populations are enormously variable, depending partly on features of the birds themselves (such as body size, wing shape and flight speed), but largely on the distances travelled and the conditions encountered en route. At one extreme, some birds can complete their migration in less than one day (such as a radio-tagged Bald Eagle (Haliaeetus leucocephalus) that flew 435 km between its wintering site in Michigan and its nesting place in Ontario, Grubb et al. 1994). At the other extreme, some landbirds take at least four months to reach their distant winter quarters, and at least two months to return, so that more than half of every year is spent on migration. Long journey times are also shown by some marine species, including shearwaters and petrels, which have a fixed base only during the breeding season, and are effectively on migration for the rest of the year, pausing to feed at places where food is available en route. Whatever the advantages in migrating as quickly as possible, external conditions provide severe constraints, notably the rate at which food can be obtained and converted into body reserves to fuel the flights, and also the weather at the time, which could speed or slow the journeys. Rain or snow, cold or ice, or unfavourable winds can delay migration for days or weeks at a time.

We should not be misled by birds that make their whole over-sea migration in one flight. In favourable winds, wild geese have been found to fly 5000 km over water within 60 hours, giving mean speeds of around 2000 km per 24-hour day, or around 80 km per hour. Bar-tailed Godwits travelling from Alaska to New Zealand covered more than 10,400 km non-stop in seven days, around 1500 km per day, or 60 km per hour (Gill et al. 2005). However, these estimates exclude the initial fuelling period, which could extend to several weeks, considering the large fuel reserves necessary to power such long over-water flights. Strictly speaking, fuelling times are part of migration, and should be added to flight times. These record speeds would then not seem so unusual.

Weather effects on migration

Weather has obvious effects on bird migration. It influences the times when birds can travel, the energy costs and risks of the journey, and the visibility of any celestial or ground-based cues that birds might use for navigation. Yet assessing the effects of weather on the volume of bird migration is not straightforward. This is partly because migration depends less on the prevailing weather than on the intrinsic migratory state of the birds themselves. The numbers of individuals migrating on particular days depend not just on the prevailing weather, but on the weather over preceding days, the date in the season, and the proportion ready to leave at the time. Towards the end of the migration season few birds may be left to migrate, however good the weather. The association between the volume of migration on particular days and the prevailing weather is therefore not constant, which complicates analysis. In addition, because species differ in body size, flight mode and other aspects, they are affected by adverse weather to different extents, some species being able to travel in conditions that would ground others (Alerstam 1978, Lack 1960).

Another problem is that different weather factors tend to be associated with one another, with some occurring under cyclonic and others under anticyclonic conditions (Lack 1960, Richardson, W.J. 1990). Even with the help of multivariate statistics, it is often hard to tell which factors are critical to migration and which are coincidental. Almost certainly, migrants do not react to the general weather situation as such, but to one or more components of it, such as wind and rain. Nevertheless, for the human observer, the synoptic weather situation of fronts and pressure systems gives a good indication of how much migration is likely to occur at different places on particular days.

Visual records from ground-based observers armed only with binoculars may miss birds flying too high to be seen by day, and provide little or no information on nocturnal migration. Yet radar has revealed that most migration of birds that fly by flapping flight occurs above binocular range. In fact, the proportion of birds flying within sight, and the proportion that come to ground, tend to be greatest in conditions that are unfavourable for flight (Lack 1960). Thus, migrants tend to fly low in opposing rather than following winds, and to settle whenever they encounter strong opposing winds, mist or rain, or reach coastlines or islands. The observer equipped only with binoculars might conclude that these were the very conditions that favoured migration, a once firmly held view but the opposite to reality. Birds also tend to fly low along coasts in these conditions, reluctant to strike out over water. It is therefore important to distinguish the influence of weather in promoting migration from its influence in making migration conspicuous (Alerstam 1978, Lack 1960).

The advent of radar greatly clarified the situation, because it enabled migrants to be detected at almost all heights (missing only those below the radar horizon), day and night, and in all weathers. From radar-based studies, consensus has now emerged that, within the appropriate seasons, migration is favoured by fine anticyclonic conditions with favourable tailwinds, and also by rising temperatures in spring and by falling temperatures in autumn. In effect, at both seasons the birds prefer to migrate under clear skies with following or light winds. Clear skies assist navigation, especially at night, by making celestial cues more visible, while following winds reduce the time and energy spent on the journey, and the risk of being blown off course. In contrast, birds seldom take off to migrate in strong opposing winds, dense cloud, mist and rain. Opposing winds make progress difficult or impossible, cloud hampers navigation, while mist or rain can soak many kinds of birds and force them down.

While temperature is important to migration, the basis of this relationship is less clear. In spring, warmth occurs in association with other conditions favourable to flight, as does cold in autumn. But temperature may have direct effects through influencing the energy balance of the birds, and more importantly through influencing their food supplies, because all vegetative growth, insect activity, and ice melt are temperature-dependent. It is therefore of obvious advantage for migrants to adjust their migratory schedules to year-to-year variations in temperature, and they are clearly deterred in spring by extreme cold and snow.

If migrating birds meet low cloud and unfavourable wind en route, they may be forced low and, if over land, they can settle and wait for conditions to improve. Over the sea, as radar has revealed, landbird migrants that enter cloud or mist banks often become disorientated, milling in all directions and gradually drifting downwind, or actively flying downwind which gives a good chance of reaching clearer weather (Lack 1960, Richardson, W.J. 1990). If cloud persists, migrants over the sea are sometimes attracted in large numbers to lighted ships or oilrigs (Bourne 1979). Although heavy overcast appears inimical for migration, some birds seem to maintain more or less straight courses with complete cloud cover. Below the cloud they can see the ground, and above it the sun or stars (although flying above the cloud seems infrequent).

Importance of wind

In general, wind speeds are stronger at mid-day than at night or early morning, and increase from the ground, where friction slows the flow, up to several thousand metres. In addition, the air mass in which birds migrate is continually changing in speed and direction, and birds must continually adjust their behaviour if they are to migrate to a predetermined destination in the most energy-efficient way. That birds respond to wind is shown by the frequent observations that: (1) they tend to depart only in fav­ourable (following) winds; (2) they often select flight altitudes where winds are fav­ourable; and (3) they compensate for wind drift, at least to some extent, providing that they can see the ground below.

Wind assistance is like food or fat reserves: it is a resource that fuels migration. If a bird with a given flying speed was blessed with a tailwind of the same direction and speed, it could in theory fly twice as far on the same fuel and in the same time, or the same distance on half the fuel and in half the time. Yet a bird flying into a headwind of the same speed could make little or no progress, however great its fuel reserve. In practice, it seems from radar studies that birds flying overland do not behave in quite this way, but fly slower than usual with a tailwind, and faster than usual against a headwind (Alerstam & Gudmundsson 1999, Hedenström et al. 2005, Liechti 2006). It is as though they conserve energy when conditions allow, and expend more than usual when necessary. The net effect is that birds make slower progress than expected in a tailwind, and faster progress than expected in a headwind, but further study is needed to find how widespread this behaviour is. In any case, following winds reduce the energy cost of migration, which is important to many migrants.

If the wind blows from the side, the bird can in theory correct for this by adjusting its heading so as to remain on track with respect to the ground (Figure 5). Birds do not then progress in the direction they are heading, but at some angle to it, which is closer to the intended track. The greater the crosswind component for a given flight speed, the greater this compensating angle must be (aircraft pilots refer to the angle between heading and track as the drift angle, alpha). This angle can be reduced if the bird flies faster. The point at which a bird is no longer able to compensate for lateral drift is thus a function of wind speed and direction, as well as the maximum flight speed that the bird itself is able to maintain (called the threshold for drift, Evans 1966).

Radar observations have been used to explore not only the extent to which migrating birds can compensate for drift by crosswinds, but also how much the prevailing circumstances (wind strength, visibility, day or night, over land or over sea, and cloud cover) affect the response. Patterns of compensation, full lateral drift, partial drift and downwind drift have all been observed, depending on conditions. Birds can presumably ‘know’ they are being drifted off course only when they have reference to some stable feature, such as the ground below; for drift would have to be substantial before it could be detected from celestial or magnetic cues. These considerations may explain why compensation seems usual in low-altitude (<1 km) diurnal migrants, but is much less frequent or complete at high altitudes, and why it is more frequent in light winds (up to 5 m per second) than in stronger ones (Bruderer & Jenni 1990). In any case, it is rarely complete in nocturnal migrants and over large water bodies (Richardson, W.J. 1990). Drift by crosswinds is, of course, one of the commonest ways in which migrants turn up as vagrants in places off their normal routes.

With respect to wind, soaring birds provide a partial exception to the usual patterns. Updraughts strongly reduce the energy cost of migration for such birds, which often fly in side-winds or light opposing winds if updrafts are present. Thermals develop in calm or light wind conditions, but not in strong winds, which suppresses soaring bird migration in some regions, regardless of wind direction. Soaring birds also show no particular tendency to migrate on cold days in autumn, as do many other birds, probably because the necessary thermals develop best on warm days (Alerstam 1978, Kerlinger 1989).

Flight altitudes

As revealed mainly by radar studies, the altitude of migratory flight is related to prevailing atmospheric conditions, especially wind speed and direction, but also cloud thickness and height, topography and other factors, as well as features of the birds themselves. Over low land, most migration takes place within 1.5 km of the ground, with decreasing numbers of birds at higher altitudes up to 3 km or more (Bruderer 1999). When birds need to cross high mountains or find favourable airstreams, they sometimes fly much higher, occasionally reaching more than 7 km.

Several species migrate regularly over the Himalayas, and as mentioned above, at least one species, the Bar-headed Goose, has been recorded at more than 8 km above sea level. Some birds can thus achieve feats of high altitude performance unmatched by other animals, and can do so without needing time to acclimatise. Clearly, they must be pre-equipped physiologically to avoid altitude sickness (hypoxia). Not all birds have such flexibility in their flight altitudes and, whatever the advantages of high-altitude flight, some species (such as swans) seem confined to migrate at low elevations, presumably for physiological or energy-based reasons (one published high altitude estimate for swans probably being erroneous).

Soaring birds are limited in flight altitude by the height reached by thermals, which is greatest around noon over land, and in most conditions seldom exceeds 1.5 km. During their migrations, soaring birds are continually rising and falling, as they climb in successive thermals and lose height between them (Leshem & Yom-Tov 1996b, Spaar & Bruderer 1996). Unlike many other birds, therefore, soaring species cannot maintain constant altitude over long distances.

Diurnal and nocturnal migration

As indicated already, some birds migrate mainly by day and others mainly by night. Nocturnal species such as owls and nightjars, or optional diurnal–nocturnal species such as shorebirds, might be expected to migrate under cover of darkness. Surprisingly, however, many normally diurnal species also prefer to travel at night. To judge from their eye structure, diurnal birds may have no better vision at night than do humans, but this would still enable them to fly safely through the open skies, and recognise star patterns and landscape features that might help them find their way.

Apart from soaring landbirds, which depend on daytime thermals, it is not immediately obvious why particular species migrate at one time rather than another. Among passerines: crows, finches, pipits, larks, wagtails, tits, swallows and others migrate primarily by day; while warblers, flycatchers, thrushes, chats and others migrate primarily by night. Among non-passerines: pigeons, raptors, cranes, herons and egrets migrate by day; while cuckoos, shorebirds, rails, and grebes migrate mainly by night. Comparing different families, there is no obvious and consistent connection between migration times and difficulty of journey, habitat, diet or other aspects of ecology. Among closely-related families, some striking differences occur, as in the passerines just mentioned, and also among waders, in which plovers (Charadriidae) migrate more by day than sandpipers (Scolopacidae). Although most species within a family seem consistent in their migratory behaviour, occasional revealing exceptions occur, with the tendency to nocturnal migration increasing with migration distance (Dorka 1966). For example, most species of Emberiza buntings in western Europe migrate by day over short distances, but the Ortolan Bunting (E. hortulana) migrates by night over long distances, being the only species that winters in Africa south of the Sahara. Similarly, most pigeons migrate by day over short distances within Europe, but the European Turtle-dove (Streptopelia turtur) migrates partly by night over long distances to Africa. In addition, Tree Pipits (Anthus trivialis) are more nocturnal than Meadow Pipits (A. pratensis), Willow Warblers (Phylloscopus trochilus) than Common Chiffchaffs (P. collybita), and Common Redstarts (Phoenicurus phoenicurus) than Black Redstarts (P. ochrurus). In all these species pairs, the first mentioned migrates further than the second. Nevertheless, there are still some puzzling exceptions: for example, European Robins (Erithaca rubecula) and Common Firecrests (Regulus ignicapilla) are short-distance migrants, but still travel mainly at night.

The division between day and night migrants is most obvious from take-off times, with diurnal migrants leaving mainly in the morning and nocturnal ones mainly in the evening. However, whether day or night, landbirds of both groups must continue flying if they find themselves over water, as must waders and waterfowl over dry land.

The main supposed advantages of nocturnal migration are that: (1) more time is left for feeding during the day, the only time that most birds can feed, so the entire journey can be accomplished more quickly; (2) temperatures are lower at night than in the day which could help to prevent over-heating and dehydration in warm regions; (3) humidities are usually higher at night and early morning, which could further reduce dehydration risk; (4) energy demands are lower, because it costs less to fly in cooler denser night air than in warmer daytime air; (5) wind speeds are generally lower at night, thus reducing the effects of headwinds or crosswinds, and vertical turbulence is less, further reducing the total energy costs of flight; (6) the use of stars for navigating is possible; and (7) the likelihood of predation during flight is much reduced. The main threat to flying migrants is from falcons or eagles during the daytime (plus gulls over water), but a wide range of other raptors take migrants when they are on the ground. Owls do not normally fly high enough to encounter migrants and in any case seldom catch prey on the wing. Given these advantages, it is surprising that any birds (apart from thermal-dependent soaring species) migrate by day.

Control of migration

In considering the timing of migration, it is helpful to separate the ultimate from the proximate causal factors. The ultimate factors, notably seasonal changes in food supplies, are those that act through the survival and reproduction of individuals, favouring those genotypes which migrate in spring in time to take advantage of optimal conditions for breeding, but leave in autumn before their continual survival in breeding areas would become precarious. The proximate factors are those, such as changing daylengths, which act to trigger migration at appropriate dates each year. The birds may thus be said to migrate in spring because they can then take advantage of a developing food supply in breeding areas (the ultimate cause) or because they are stimulated to do so by increasing daylengths and other prevailing conditions (the proximate cause).

As every bird-watcher can see for himself, many long-distance bird migrants arrive at their nesting or wintering places every year at around the same dates. This implies the existence in the birds of precise timing mechanisms that, in response to external stimuli, trigger migration at about the same dates each year and maintain it for long enough to allow the bird to cover the distance required. The relatively small differences in timing that occur from year to year are mainly associated with variations in prevailing weather or food supplies.

In the scientific literature dealing with the timing of bird migration, German words are often used, even in texts written in English or other languages. This custom persists from a time when Germany led the world in migration research, mainly in the first few decades of the 20th century, when many of the basic concepts and paradigms were laid down. Anyone who reads widely in this subject area can thus expect to come across words like Zugdisposition (readiness to migrate), Zugunruhe (migratory restlessness), and Zeitgeber (time keeper).

The migratory state

Migratory condition (Zugdisposition) in birds is marked by a sudden weight increase, due largely to the deposition of additional fat which serves as the main fuel during the journey. In captive birds, it is also marked by the development of so-called ‘migratory restlessness’ (Zugunruhe). This is a distinct form of behaviour in which birds hop and flutter round their cages, and undertake long periods of ‘wing-whirring’ in a perched position (perhaps equivalent to migration in a cage). Typical diurnal migrants show this behaviour by day, and nocturnal ones by night. It gives a useful indication of migratory condition in captive birds because it can be quantified automatically. From studies of body weight and restlessness in caged birds, much has been learned about the proximate control of migration. Under natural daylengths, caged birds from migratory populations develop fat reserves and migratory restlessness at appropriate dates in autumn and spring, at about the same times as their wild counterparts (Berthold 1996, Gwinner 1986).

Endogenous rhythms

While the ultimate extrinsic factor controlling the annual cycles of birds is the seasonality of the environment, the primary intrinsic proximate factor is apparently an endogenous rhythm within the bird. This self-sustaining rhythm tends to ensure that the major processes of migration, breeding and moult occur in the correct sequence each year, and at roughly the right times. The evidence for the existence of an internal rhythm has come largely from studies on captive birds kept for up to several years under rather specific but constant daylengths (Berthold 1996, Berthold & Terrill 1991, Gwinner 1968, 1986, 1996, Gwinner & Helm 2003). Such birds have no clue from the outside world as to what the date might be. Yet they usually moult and reach breeding and migratory condition in the correct sequence, and at roughly appropriate intervals, with corresponding cycles in body weights, gonad sizes and hormone levels. This finding is taken to imply the existence of some underlying ‘endogenous’ controlling system. However, in conditions of constant daylength, the cycles do not stick strictly to a year, but tend to drift, getting either shorter (rarely) or longer, hence the term ‘circannual’ cycles, which typically last 9–13 months.

The existence of internal circannual rhythms, underlying the natural yearly cycles, and persisting for at least two cycles, has now been shown experimentally in more than 20 different bird species, including resident and migratory, temperate and tropical, passerines and non-passerines, as well as in other animals and plants (Berthold & Terrill 1991, Gwinner 1981, 1986, 1996). Circannual rhythms evidently underlie the control of seasonal activities in a great variety of organisms. Among birds, circannual cycles may be reflected in gonad condition alone, in moult alone, in migratory condition alone, or in any combination of these activities. The cycles can thus be viewed as consisting of separate but integrated components involving the different activities (Wingfield 2005). Which of these components are expressed in captive birds depends largely on the constant photoperiod to which the birds are exposed, and perhaps also on the time of year (= internal physiological state of the birds) when the experiment starts.

Spontaneous endogenous rhythms are most apparent in long-distance migrants, which are normally exposed to varying photoperiodic regimes on migration, and in which the need for some form of endogenous control is greatest. They are also apparent in some resident species of equatorial regions, where daylengths are constant year-round. For example, equatorial Common Stonechats (Saxicola torquata axillaris) kept caged in Germany under constant 11.8 hours of light alternating with 11.2 hours of dark went through up to 12 reproductive-moult cycles in a 10-year period (Gwinner 1996). However, in temperate zone residents and short-distance migrants caged in constant conditions, the cycles tend to continue for less long, and are more variable among individuals; they seldom proceed for more than one year, and the different events tend to become increasingly out of phase with one another. They are thus less rigid and persistent, as found in some Blackcap (Sylvia atricapilla) populations (Berthold et al. 1972) and in Common Stonechats (S. t. rubicola) from Europe, as opposed to Africa (Gwinner 1996).

The role of daylength

Under natural conditions, the endogenous cycles of many birds are kept in phase by seasonal daylength changes, and in experimental conditions particular events can be advanced or retarded by appropriate use of an electric light (Farner & Follett 1966, Lofts & Murton 1968). For example, if migratory birds of some species are exposed in late winter to photoperiods longer than natural days, their gonads begin to grow earlier than usual, and they show migratory and reproductive behaviour prematurely (King 1972, Lofts et al. 1963, Rowan 1925, Wolfson 1959). Similarly, if they are exposed in late summer to shorter days than usual, they moult and reach migratory condition earlier than usual.

The importance of daylength as a time-keeper (Zeitgeber) derives from its reliability. Its seasonal changes are consistent between years, making it the most obvious environmental feature that, at most latitudes, gives a reliable cue to date. The synchronisation of the internal annual cycle to photoperiod has been shown most convincingly in experiments in which birds were exposed to seasonal photoperiodic cycles with periods deviating from 12 months (e.g. six month cycles). As a rule, the birds’ biological rhythms then conformed to the altered photoperiodic regime (Gwinner 1986, 1990b). For example, when the normal annual cycle of daylength was shortened to six months without altering its amplitude, Garden Warblers (Sylvia borin) went through four instead of two moult periods within one calendar year, two instead of one gonad cycle, and four instead of two periods of migratory restlessness (Berthold 1996). The same occurred in Sardinian Warblers (Sylvia melanocephala), in which the usual one annual moult occurred twice within one calendar year (six months apart). It also occurred in Common Stonechats which underwent two gonad and moult cycles in one calendar year (Gwinner & Helm 2003). More remarkably, Dark-eyed Juncos (Junco hyemalis), which were exposed to four periods of short (9-hour) days and five periods of long (20-hour) days in one year, showed in this time five periods of gonadal activity, five of fat deposition and two of moult (Wolfson 1954).

Time and distance programmes

In general, the longer the distance between breeding and wintering areas, the greater is the duration and intensity of migratory restlessness (Zugunruhe) shown by caged birds. Different Sylvia warblers that breed in Europe migrate average distances varying from a few hundred to nearly six thousand kilometres, and show corresponding average periods of restlessness varying from less than twenty to more than 1000 hours (Figure 6). This implies that periods of restlessness in these different populations are endogenously (genetically) determined. In addition, the amount of fat accumulated by captive birds at migration times is related to the types of journeys they make in the wild. Birds that migrate by long flights, as from Europe to sub-Saharan Africa, typically accumulate more fat in captivity than do birds that migrate short distances within Europe (Berthold 1973, 1984). This contrast is evident in comparisons between related species, such as Willow Warbler and Common Chiffchaff (Gwinner 1972), and between different populations of the same species which migrate different distances, as in the Blackcap (Berthold & Querner 1981). Hybrids between short-distance and long-distance populations of the Blackcap show intermediate patterns of restlessness and fattening (Berthold & Querner 1981).

Directions

The timing and duration of migratory restlessness, and patterns of fattening, are not the only features under endogenous control, as the same applies to directions. Birds from populations that take different directions in the wild show the same directional preferences when tested in captivity in special ‘orientation cages’. These circular test cages have a wire top, so the bird can see the sky, and some system of automatic recording which indicates the directional preference adopted by the bird during periods of restlessness.

Blackcaps in western Europe mostly migrate southwest in autumn, while those from eastern Europe migrate southeast, the two types separated by a distinct ‘migratory divide’. Birds from the two types, reared together in captivity and then tested in orientation cages, showed appropriate directional preferences, either southwest or southeast, depending on their region of origin (Helbig 1991). Moreover, hybrids between the two types showed intermediate directional preferences, more or less due south. These and other findings provide strong evidence that directional preferences are inherent, and endogenously controlled.

Some wild migrants in spring retrace their path from the previous autumn, but others take different routes at the two seasons (so-called loop migrants). When tested for directional preferences in orientation cages, hand-reared Garden Warblers kept in constant (12Light:12Dark) conditions changed their mean heading from southwest to southeast part way through their autumn migration period. This corresponded with a change they would normally make part way through their journey between central Europe and Africa (Gwinner & Wiltschko 1978). They made no such change in spring, when they return by a more direct northerly route, requiring no change in direction during the journey. This gave another indication that directional preferences were endogenously controlled, and that spring directions were not simply the reverse of autumn ones.

Integration of time–distance and direction programmes

The combination of inherent time–distance programmes and directional preferences could explain how juvenile birds migrating on their own can reach wintering areas unknown to them but specific to their population. On this basis, naïve autumn migrants would fly in the right direction for an appropriate period, and would not need to experience the particular conditions of their wintering areas before they stopped migrating. After an appropriate time, caged birds from both European and North American breeding areas lost their autumn restlessness and fat reserves, even though they had moved no further than the confines of their cages. Moreover, when captive juveniles were experimentally transported to their species-specific wintering areas, or even beyond their normal wintering range, the migratory activity they showed in cages persisted as long as that of individuals kept in the breeding area (for experiments on young Garden Warblers, Lesser Whitethroats (Sylvia curruca) and European Pied Flycatchers (Ficedula hypoleuca) see Gwinner 1971, Rabøl 1993). Similarly, juvenile Common Starlings and other species that were trapped on migration, flown by airplane and released immediately in the usual wintering areas for their population, or at some other locality off the normal route, resumed migration. Ring recoveries revealed that these transported juveniles moved in the same direction and covered about the same additional distance that they would have travelled had they not been displaced (Figure 7; Perdeck 1964, 1967). These experiments again suggested control of timing and direction by an endogenous programme, rather than by location. This type of migration is often called ‘clock-and-compass’ or ‘vector’ migration.

The role of experience

On current thinking, then, the urge to migrate in autumn and spring is genetically controlled. It is reflected in an autonomous rhythm of physiology and behaviour, which is kept on schedule by daylength changes (Berthold 1996, Gwinner 1972, 1986). This inherent system controls both fattening and migratory restlessness, as well as the general direction and time-course of migration. However, most of the experimental work which gave rise to these ideas was based on naïve juvenile birds that had no previous knowledge of the wintering range of their population. The situation differs somewhat in experienced birds migrating to a known site. This was first shown in experiments with Common Starlings and others, in which adults birds displaced off their normal route did not stick to their inherent direction, like the juveniles described above, but corrected for their displacement, changed direction and headed back to their previous wintering areas (Figure 7). These birds revealed more than one-directional orientation; they had a map sense, as they knew where they ought to be, and headed towards their goal (Perdeck 1958). On return migration to breeding areas, both juveniles and adults proved able to return to the region of their birth, but in this case both age-groups had previous experience of the area (Perdeck 1958).

Later experiments on Dark-eyed Juncos showed that individuals that were held captive on their wintering areas all summer did not show the usual autumn restlessness and fattening. Many of those that were released in autumn in wintering areas stayed nearby, often within the same home range they occupied the previous winter (Ketterson & Nolan 1986). They acted as though they knew where they ought to be at that time of year, and remained there without migrating. However, they disappeared from the area in the following spring and some reappeared in the next autumn, so they were assumed to have spent the summer as normal on more northern breeding areas.

Comparable findings emerged from an experiment in which migratory Indigo Buntings (Passerina cyanea) were held in their breeding areas through the winter, and released there in spring (Ketterson & Nolan 1990, Sniegowski et al. 1988). Seven out of 20 males released in spring in their nesting territories remained, while eight out of 20 that were transported and released 1000 km to the south returned to their nesting localities. The remaining birds were unaccounted for, but it seemed that when migrants were exposed in spring to their previous nesting place, they did not migrate. These findings gave a further indication that the inherent template of migration, so important to the first migration of juveniles, could be modified by experience.

Other factors influencing the timing of migration

Within the migration season, migratory flights may be influenced by prevailing weather, as discussed earlier, and also by food supply. The role of feeding conditions in affecting autumn fattening rates, body condition and departure dates has been established in the field for a number of species ranging from the Sedge Warbler (Acrocephalus schoenobaenus), studied by Bibby & Green (1981), to the Greylag Goose (Anser anser), studied by van Eerden et al. (1991).

When they reach an appropriate weight, migrants normally leave immediately if weather permits. This would be expected because, once acquired, fat reserves are dangerous and expensive to maintain, conferring increased vulnerability to predation. For example, when captive Blackcaps were exposed to simulated predator attacks, individuals carrying a fuel load equivalent to 60% of lean body mass (the maximum recorded in this species) were calculated to suffer reduction of 32% in angle of ascent and 17% in velocity, compared to lean Blackcaps (Kullberg et al. 1996). This degree of difference could put fat birds at substantially greater risk. Research on captive birds, or on wild ones caught during migration, has confirmed that the motivation to proceed with migration is related to fat reserves, as is the strength of directional preference. In general, individuals with the greatest body reserves at any one time showed the greatest migratory restlessness and the strongest directional preferences (Bairlein 1992, Berthold 1996, Dolnik & Blyumental 1967, Sandberg et al. 2002, Yong & Moore 1993).

Autumn migration

Migrants normally leave their breeding areas when conditions deteriorate, but before their continued survival there would become precarious. From any one area, however, departure dates also differ widely between species, between late summer and late autumn, depending largely on their type of food and when it becomes scarce. For example, species that depend on insects from fresh leaves, such as warblers, normally leave before those that depend on fruit or seeds, such as thrushes and finches. More­over, in association with the earlier onset of winter at high latitudes, many species withdraw from high-latitude parts of their breeding range first, and from lower latitude parts later. The assumption is that, in order to prepare for autumn migration, populations have evolved responses to different daylength regimes, appropriate to the latitude at which they breed.

Obligate and facultative modes

In considering the proximate control of autumn migration, a useful distinction can be drawn between obligate migrants (formerly called ‘instinct’ or ‘calendar’ migrants), and facultative migrants (formerly called ‘weather’ migrants). In obligate migrants, all main aspects are viewed as under firm internal (genetic) control, mediated by daylength changes, which gives a high degree of annual consistency in the timing, directions and distances of movements. For the most part, each individual behaves in the same way year after year, migrating at similar dates and for similar distances. Obligate migrants often leave their breeding areas well before food supplies collapse, and while they still have ample opportunity to accumulate body reserves for the journey. They tend to migrate long distances, often to the tropics or beyond. Examples include swallows and warblers.

In contrast, facultative migration is viewed as a direct response to prevailing conditions, especially food supplies, and the same individual may migrate in some years but not in others. Within a population, the proportions of individuals that leave the breeding range, the dates they leave and the distances they travel, can vary greatly from year to year, as can the rate of progress on migration, all depending on conditions at the time (e.g. Moore et al. 2003, Newton 2006, Svärdson 1957, Terrill 1990). In consequence, facultative migrants have been seen on migration at almost any date in the non-breeding season (at least into January in the northern hemisphere), and their winter distributions can vary greatly from year to year. Extreme examples include irruptive seed-eaters, such as Common Redpoll, Eurasian Siskin (Carduelis spinus) and Bohemian Waxwing (Bombicilla garrulus). Although in such facultative migrants, the timing and distance of autumn movements may vary with individual circumstances, other aspects must presumably be under firmer genetic control, notably the directional preferences and the tendency to return at appropriate dates in spring.

Compared with obligate migrants, facultative migrants tend to migrate shorter distances, although many exceptions occur. The two types of migrants thus have different distribution patterns in mid winter. Whereas obligate migrants are concentrated in a distinct wintering area, usually at long distance from the breeding area, facultative migrants are typically found over the whole migration route from breeding to wintering areas, usually tailing off with increasing distance from breeding area, but with marked annual variations.

In general, it seems that obligate migration occurs in populations whose food supplies in breeding areas are predictably absent in winter, whereas facultative migration occurs in populations whose food supplies in breeding areas vary greatly from one winter to another, according to weather or other variables. The distinction between the two types is thought to reflect the degree to which individual behaviour is sensitive to prevailing external conditions, and hence varies from year to year. However, obligate and facultative migrants are best regarded, not as distinct categories, but as opposite ends of a continuum, with predominantly internal control (= rigidity) at one end and predominantly external control (= flexibility) at the other.

Another reason for not drawing a sharp distinction between the two categories is that many birds seem to change from obligate to facultative mode during the course of their journeys, as the endogenous drive to migrate wanes with time and distance, and the stimulus to continue becomes more directly dependent on prevailing local conditions (Helms 1963, Terrill 1990, Terrill & Ohmart 1984). Theoretically, the initial obligate phase of any journey might take the migrant across regions where the probability of overwinter survival is practically zero: where any individuals that attempted to winter there in the past were eliminated by natural selection. As migration continues into more benign areas, and survival probability increases, the bird switches to a facultative mode, in which it benefits by responding to local conditions, stopping where food is abundant. The obligate phase would therefore be expected to be undertaken much more rapidly, on average, than the facultative phase, which involves longer and more variable stops. Such a two-phase migration, with obligate and facultative stages, would also ensure that, in any particular year, the bird migrated no further than necessary. In some species only the tail end of the migration may be facultative, in others the entire journey. Most irruptive migrants are near the latter end of the spectrum.

Arctic-nesting geese provide circumstantial evidence for a two-phase migration, in which the first part is obligatory and the second part facultative. Geese need to leave the arctic every year before survival there becomes impossible, and they tend to depart en masse on about the same dates every year. But once they reach suitable wintering areas, their movements become much more variable in timing and extent, depending on local food availability. They appear to change from a primarily endogenous migratory phase (obligate migration) to a stage when the stimulus for further migration is primarily environmental. In effect, as birds travel south in autumn, the drive to continue becomes increasingly dependent on food and other local conditions.

Role of dominance in facultative migrants

Because birds compete for food, and vary in dominance or feeding efficiency, some individuals could survive in conditions where others would die unless they moved out. In facultative migrants, the subordinate sex and age groups typically migrate in greater proportions, at earlier dates, or extend further from the breeding areas, than the dominants. Thus, in many bird species, adult females are more migratory than adult males, juveniles more than adults, and late-hatched young more than early-hatched ones (Gauthreaux 1982, Smith & Nilsson 1987). Such differences have led to the notion that, in facultative migrants, competition (or its effect on body condition) is involved as a proximate mechanism stimulating migration in those individuals least able to survive in local conditions (Gauthreaux 1982).

Timing of autumn migration

In obligate migrants, where all individuals leave the breeding range each autumn, the dates of migration are in general fairly consistent from year to year. This is apparent not only in the dates that birds leave their breeding areas, but also in the dates they pass particular places on their migration routes and arrive in wintering areas. For example, among raptors migrating through Israel en route to Africa, the timing and duration of passage varied greatly between species, but within species the autumn passage dates were remarkably similar between years (as were spring dates). Over nine years, the confidence intervals of the mean autumn dates ranged between 1.5 and 3.4 days (versus 2.1–5.5 days for spring dates), depending on species (Leshem & Yom-Tov 1996a).

In facultative migrants the situation differs. Such populations generally moult after breeding and, in contrast to obligate migrants, individuals do not necessarily depart immediately after finishing moult. If conditions are favourable, they may linger longer in the breeding or stopover areas, leaving only when food supplies dwindle or are shut off by snow and ice. Mean autumn migration dates may therefore vary greatly from year to year, depending on local food supplies, as may rates of travel. This situation is exemplified by most short-distance migrants, especially irruptive seed-eaters, and by waterfowl and others affected by frost. For example, the peak date for passage of Eurasian Siskins through Falsterbo Bird Observatory in south Sweden during 1949–1988 varied from 15 August (in 1988) to 17 November (in 1958), the last date the station was manned that year (Roos 1991). At another site, Eurasian Siskins passed in largest numbers, and at the earliest dates, in years when birch seeds (the main autumn food) were scarce (Svärdson 1957). Likewise, the date on which the last Whooper Swans (Cygnus cygnus) left Lake Chuna on the Kola Peninsula each year during 1931–1999 varied between about 20 September and 9 November, depending on when the lake froze over (Gilyazov & Sparks 2002). These observations illustrate the point that some migrants depart only when deteriorating conditions encourage them to leave. Together with variable conditions on migration routes, this facultative response gives wide variation in the dates that particular species arrive in their various wintering areas, the more distant of which may be reached only in occasional years.

Split migrations

Although autumn migration is usually considered as a single event, consisting of alternating periods of flight and fattening, some species break their autumn journeys for periods of several weeks, much longer than is needed for refuelling. Some even moult during this break in migration. This behaviour is shown by many Eurasian migrants to Africa, such as the Marsh Warbler (Acrocephalus palustris) and Garden Warbler, which remain in the northern tropics for several weeks, and only later move on to the southern tropics (Jones 1995). In this way, they get the best from both regions, remaining in the northern tropics until conditions deteriorate, and arriving in the southern tropics at the optimal time, after fresh rains have promoted vegetation growth. Interrupted migrations are also evident in various irruptive and other facultative migrants that break their journeys to exploit food supplies they encounter en route, and travel much further from their breeding areas in some years than in others (Newton 2006, 2008, Svärdson 1957). One consequence of split migration is that the outward journey (including the break) takes more than four months in some species, whereas the return journey in spring can take only 1–2 months.

Spring migration

Birds normally leave their wintering areas so as to reach their nesting areas in time to breed at the most favourable season. Many migrants winter so far from their breeding areas that they could not judge conditions there from their position in wintering areas. They can only leave their wintering areas at a time that natural selection has decreed is appropriate, using an internal timer combined with local conditions as a cue. Among species wintering in the temperate zone, the main known environmental stimulus for spring migration (perhaps superimposed on an endogenous rhythm) is increasing daylength which promotes extra feeding, fattening and migratory restlessness at appropriate dates for the population concerned (King 1972, Lofts et al. 1963, Rowan 1925). In species that undergo a spring moult, this is also initiated by increasing daylengths, as is gonad growth, each of these processes occurring in appropriate overlapping sequence through the season. The role of daylength has been shown repeatedly in experiments on captive birds, in which longer-than-natural photoperiods advance all spring-occurring processes, whether gonad growth, pre-nuptial moult or migration.

As in autumn, however, the effects of daylength may be modified by both temperatures and food supplies. Every bird watcher outside the tropics can see that summer migrants arrive later in cold springs than warm ones, and that migrants can be held up for many days during a cold snap. This situation has been duplicated experimentally. For example, captive White-crowned Sparrows (Zonotrichia leucophrys) exposed to air temperatures increasing from 5°C to 26°C advanced the time of migratory restlessness compared with control birds (Eyster 1954, Lewis & Farner 1973). However, it is hard to tell whether temperature acts directly on the birds, or through its effect on their energy needs and the food available for fuelling.

Different populations of a species wintering in the same area

Different species wintering in the same area, and hence subject to the same daylength regime, may start their migrations at different dates, weeks or sometimes months apart, depending on the distance they have to travel and the dates their breeding areas become fit for occupation (for shorebirds see Piersma et al. 1990). Such differences are also found in different races (or populations) of the same species wintering in the same area, as shown for White-crowned Sparrows in California (Blanchard 1941) and for Yellow Wagtails (Motacilla flava) in tropical Africa (Curry-Lindahl 1958, 1963). Although exposed to the same winter conditions, the members of the various races differ in the dates at which their gonads develop, and at which they accumulate fat and depart for breeding areas (Blanchard 1941, Curry-Lindahl 1963, Fry et al. 1972). In general, in these northern hemisphere species, races that breed furthest south are first to leave their wintering areas, and those that breed furthest north are last to leave. Thus, in the Yellow Wagtail, the first to leave in spring is the southern race M. f. feldegg, then M. f. lutea, followed by M. f. flava and M. f. flavissima, and finally M. f. thunbergi. These various races arrive in their breeding areas in the same sequence, spanning the period March–June, from south to north. Inherent differences in endogenous rhythms could explain such population differences, as could inherent differences in the threshold daylengths required to trigger departure. Either mechanism could account for how birds of different races can leave their shared wintering area in appropriate sequence and reach their respective breeding areas (at different latitudes) at appropriate dates. However, the Yellow Wagtails are the more remarkable because several races winter together on the equator, where daylength is constant year round. In these birds, endogenous control seems essential, with different races responding differently, according to where they breed, and setting their ‘internal clocks’ before they reach the equator, so as to leave at appropriate dates some weeks or months later.

Return migration from variable wintering areas

In many bird species, the migrants from particular breeding areas can winter over a wide span of latitude and daylength regimes. For example, Eurasian Siskins breeding in the northern boreal forest of western Europe may winter anywhere between mid Sweden and Morocco, a latitudinal span of about 30 degrees, and the same individ­uals may winter at widely separated places in different years (Newton 2006). The general pattern in such species is that return migration begins earliest from the most distant (most southern) parts of the wintering range, and latest from the most northern parts. This sequential withdrawal from lower to higher latitudes can be spread over many weeks. It has been examined in detail in White-crowned Sparrows wintering in western North America. In these birds, the date of onset of pre-migratory fattening varied linearly with latitude (and hence with solstical daylength), averaging 3.3 days later for each degree of latitude northward. The start of withdrawal was thus spread over seven weeks from the 14 degrees of latitude involved (King & Mewaldt 1981). Mean rates of fattening were the same in all areas, regardless of latitude.

Deferred return to breeding areas

In some long-lived bird species with deferred maturity, individual migrants leave their natal areas towards the end of their first summer, and do not return in the next spring, but only in a later one, when they are two or more years old. Other individuals may return part way towards breeding areas, or may visit breeding areas only for a short time each year, leaving wintering areas later and returning earlier than breeding adults. They perform both migrations in less hurry than the breeding adults, and in more favourable conditions. Such patterns are shown by various raptors, seabirds, shorebirds and others in which individuals do not breed until they are several years old.

Most of the first-year shorebirds that stay in ‘wintering areas’ show no sign of pre-migratory fat deposition or spring moult into breeding plumage, but remain light in weight and in well-worn winter plumage until the next ‘post-breeding’ moult in late summer into new winter plumage. In other individuals, ‘pre-breeding’ moult and fattening are much delayed, sometimes into July, too late for the birds to breed that year (McNeil et al. 1994). Lack of both weight gain and pre-breeding moult was apparent among juvenile Curlew Sandpipers in South Africa, among Ruddy Turnstones (Arenaria interpres) in Scotland, and among Western Sandpipers (Calidris mauri) in Panama, while adults wintering in the same places began moult, accumulated body fat and left in spring in the usual manner (Elliott et al. 1976, Metcalfe & Furness 1984, O’Hara et al. 2002).

Although the birds that stay year-round in wintering areas do not always undergo the pre-breeding moult into summer plumage, their late summer post-breeding moult can occur up to several weeks earlier than in adults returning from breeding areas. Ruddy Turnstones over-summering in England moulted seven weeks earlier than adults returning from their arctic nesting grounds (Branson et al. 1979), and Western Sandpipers over-summering in Panama moulted 3–4 weeks earlier than returning adults (O’Hara et al. 2002). They provide an example of birds moulting at a more genial time of year when not constrained by breeding to a less favourable time later in the year. More study is needed before we can hope to understand the mechanisms that control the occurrence and timing of migration in such species during their early years of life.

Recent changes in migration patterns

Mainly in association with climate warming, many birds have changed their migratory behaviour in recent decades, arriving earlier in spring on their breeding areas, and often also departing later in summer or autumn (Newton 2008). In some once wholly-migratory species, some individuals are now remaining in breeding areas year-round, while other individuals are migrating less far than before. In general, such changes are most obvious in regions with the greatest climate change, and are more marked in short-distance than in long-distance migrants. Other species have extended their breeding areas into higher latitudes, where they have become more migratory. Not all these changes are necessarily linked with climate warming, however, as some major changes in food-supplies were human induced. An example is the increased feeding of garden birds in winter, which has enabled some formerly migratory species to remain at higher latitudes in winter, almost entirely dependent on these handouts. The important point is that migratory behaviour has proved flexible, changing rapidly in response to alt­ered conditions. Some of these behavioural changes may have a genetic basis, resulting from the action of natural selection, whereas others may represent facultative responses to environmental changes. Whatever the basis, it is this facility for rapid modification of migratory behaviour which presumably enabled birds to respond to the great glacial changes of the past, and will enable them to adapt to future changes.

Migratory fuelling

At times of migration, as mentioned above, many birds put on extra body fat and other reserves for use as fuel during the journey. Typically, they divide their migration into periods of flight, during which reserves are depleted, and stopovers, when reserves can be replenished by feeding. Species that travel over favourable terrain tend to migrate in short flights, each lasting up to several hours, broken by periods of rest and foraging, when they can replace the relatively small amounts of fuel used on each flight. Given suitable weather, migratory flight can, in theory, occur for part of every day until the journey is completed. However, birds that migrate over seas and other large inhospitable areas have to sustain much longer fasts during flights of up to several days. These flights are preceded by days or weeks of feeding when much larger body reserves are accumulated. Passerines and shorebirds typically take 1–3 weeks to accumulate the fuel reserves necessary for such long journeys, and before departure some may have doubled their normal weights.

Energy needs and body composition

Per unit of weight, fat provides much more energy than any other storable biochemical fuel available. The use of 1 g of fat will yield around 9.2 kilocalories (or 38 kilo­joules) of energy, compared with only about 1.3 kcal (5.3 kJ) from 1 g of protein or 1.0 kcal (4.0 kJ) from 1 g of carbohydrate (Table 1). Weight for weight, therefore, fat contains 7–9 times more energy than alternative fuels, providing the maximum energy storage for the minimum weight gain. Fat is an even more efficient fuel than high-octane vehicle fuel, and has the additional advantage that its oxidation yields an equal weight of water, thus contributing to another of the bird’s needs during long-distance flight. Moreover, unlike any other potential bio-fuel, fat can be stored without water or protein, and can also be digested efficiently with less loss of heat and no effect on body glucose. The main recognised drawback of fat is that its metabolism requires the breakdown of small amounts of protein to provide enzymes for the chemical processes involved (the citric acid cycle).

Fat is laid down as adipose tissue in various parts of the bird’s body, especially under the skin, and in well-defined deposits within the wishbone (tracheal pit) and around the gut. Just before departure, the subcutaneous fat layer in some long-distance passerine migrants covers most of the body, only the central part of the breast muscle remaining uncovered.

Pre-migratory weight increase involves not only the deposition of fat, but also of body protein. Fuel should therefore be regarded as a combination of the two, but not necessarily in consistent proportions. In the most extreme species, protein contents increase prior to migration by less than two-fold, whereas fat contents may increase by more than 10-fold. Nevertheless, the protein and fat levels usually increase in step with one another during migratory fuelling, so appear to be closely correlated. Some species, such as Sandhill Crane (Grus canadensis), add protein and fat at migration times in the approximate ratio of 1:10 (Krapu et al. 1985), whereas other migratory birds lay down roughly equal weights of protein and fat in the ratio of 1:1 (see later). This difference may result simply from differences in the diets of different species, or in their metabolism, but it may also represent an adaptation to the different types of journey they undertake or their needs after arrival.

Some species alter the ratio of stored protein to fat between seasons. For example, the 40–50 g mass gain by 200 g Eurasian Golden Plovers (Pluvialis apricaria) during autumn stopovers consists almost entirely of fat, but a similar mass gain in spring consists chiefly of protein tissue (mainly muscle). This difference may be because Eur­asian Golden Plovers face energy deficits on autumn migration and in winter when they eat mainly protein-rich earthworms, but in spring they risk protein deficits, when after arrival in arctic breeding areas they eat mainly berries but must soon produce eggs (Piersma & Jukema 2002). Another indication that birds can adjust their body reserves to oncoming needs derives from King Penguins (Aptenodytes patagonicus), which double their body mass before long fasts on land, but with reserves consisting of about 14% protein before incubation, and 29% protein before feather moult (Cherel 1995).

Compared to fat and protein, carbohydrate is of relatively minor importance as an energy source for migratory birds. It is present in the form of glycogen in the liver and muscle tissue. The highest glycogen values reported from birds amount to no more than about 3.0% of liver mass and about 0.5% of total body mass (Blem 1990).

The bodies of resident bird species, or of migrants outside the migration seasons, typically contain fuel amounting to 3–5% of their lean body mass. Some migrants apparently travel with reserves no greater than this. However, most regular passerine migrants depart with fuel loads amounting to 10–30% of their usual body mass, and those making especially long flights accumulate fuel loads between 40 and 70% of their usual mass, approaching 100% in a few species (Blem 1990, Fry et al. 1970, Lindström 1991, Moreau & Dolp 1970). Similarly, some shorebirds attain very large fuel loads, as high as 50–90% of lean body mass, but with a maximum of around 100% in those embarking on the longest non-stop flights. They may then lose up to half their body mass during their flights over the next few days.

In addition to fuel deposition, preparation for migration in many birds involves enlargement of the breast muscles, heart and blood vessels, and shrinkage of other organs less important in migratory flight (see below). It also involves the activation of enzyme systems for the storage and rapid mobilisation of fat, an increase in the erythrocyte (haematocrit) content of the blood to enhance oxygen transport during long flights (Jenni-Eiermann & Jenni 1991), and the modification of various aspects of behaviour, including the diurnal rhythm of activity to permit nocturnal flights in some otherwise diurnal species. All these changes vary between species, according to the types of journeys they undertake.

Mechanisms of fuel deposition

To deposit body fuel, a bird must increase its rate of food intake. This may be achieved in various ways, such as feeding more rapidly or for longer than usual each day, or by selecting from potential foods the most calorific and easily digestible items. This is shown by some warblers and other insectivores, which switch to mainly fruit at migration times, the sugars therein being easily converted to fats. A bird might also increase the size of its digestive tract, so as to increase the throughput and processing of food, as again shown in many species of passerines. Some waterfowl and shorebirds can feed both by day and by night, and can thus achieve higher rates of food intake than other birds, and correspondingly higher rates of fuel deposition. But the most obvious way in which a diurnal bird can conserve feeding time is to migrate at night. While this may not increase feeding time over what is usually available, it at least prevents potential feeding time being reduced by flight time.

Not only are birds able rapidly to store and metabolise large amounts of fat, they also undergo many other physiological changes, affecting skeletal muscles and various internal organs (Battley et al. 2000, Piersma 1998). Before departure on long journeys, exercise organs (pectoral muscle and heart) tend to enlarge and nutritional organs (stomach, intestine and liver) tend to shrink. This makes sense on long flights where weight reduction is at a premium. Prior reductions in nutritional organs appear most pronounced in populations about to over-fly oceans that offer few or no opportunities for emergency landings, let alone feeding. In other species, the digestive tract is apparently reduced during the flight itself, rather than beforehand, contributing to the fuel and water needs of the migrant on its journey. Digestive organs are rebuilt after arrival at a staging or wintering site. Hence, a bird refuelling for long-distance migration is not like a plane landing, refuelling and taking off again. Unlike the plane, the bodies of long-distance migrant birds have to be partly reconstructed at each stopover and modified again before take-off (Piersma 1998). This is much less true of species that migrate by short flights.

In some waterfowl, notably geese, reserves accumulated before arrival in breeding areas, and especially at the last stopover site, help in egg formation and survival through incubation (Drent et al. 2003, Ebbinge & Spaans1995, Newton 1977). Such geese often arrive before vegetation growth has begun in their breeding areas, when little food is available. They are described as ‘capital breeders’, because they reproduce largely on the strength of existing body reserves, and contrast with ‘income breeders’ which breed on the strength of food eaten at the time. In different goose populations, weight gains of 25–53% have been recorded before the birds set off on spring migration (McLandress & Raveling 1981). This is a lot for birds of this size. Females accumulate more weight than males, in association with the needs of egg production and incubation (in which males do not participate, except in nest guarding).

At least some species of high arctic geese, such as the Lesser Snow Goose (Anser c. caerulescens), Ross’s Goose (A. rossii) and Brent Goose (Branta bernicla), can start egg-laying 2–5 days after arriving in breeding areas, before plant growth has begun. They seem to rely entirely on body reserves. Others feed and regain some weight after arrival, but still depend partly on body reserves accumulated further south (Bromley & Jarvis 1993). In Lesser Snow Geese, the relationship between body reserves and reproductive output was studied in females shot at various stages of breeding in the Northwest Territories of Canada (Ankney & MacInnes 1978). The potential clutch size of pre-laying females was found from the number of large vascularised follicles in the ovary, and it emerged that females with larger body reserves had, on average, larger potential clutches. In other females collected after laying, body reserves had been partly used, but the mean weights of remaining reserves from females that laid clutches of different sizes were not then significantly different. Apparently, clutch size in Lesser Snow Geese was determined by the size of nutrient reserves: females with the largest body reserves produced the most eggs. Breeding females used most of their remaining fat and protein reserves during incubation (85% and 24% respectively). Late in incubation when females had depleted their body reserves, some left their nests to feed, while others were found dead on their nests from starvation. Hence, to reproduce successfully in this area, female Lesser Snow Geese had to accumulate beforehand enough reserves to support the last stage of migration, egg production and maintenance during the four weeks of incubation. Only after hatch were females able to feed intensively again, and build up body condition for the return migration to wintering areas.

Navigation

One of the most amazing aspects of migration is how birds find their way over long distances, often through unknown terrain. Many species are capable of migrating year after year between exactly the same breeding and wintering places, located up to thousands of kilometres apart. Some pelagic seabirds wander widely over the oceans, yet each year return unerringly to their own particular nesting islands. Great Shearwaters (Puffinus gravis), for example, nest on the isolated Tristan da Cunha Islands, lying at 40°S in the South Atlantic and more than 2000 km from Africa, the nearest continent. In the non-breeding season these birds migrate northward in their millions, ranging over large parts of the North Atlantic. But they return each year with pinpoint accuracy to their tiny breeding islands, which are spread over only 45 km of ocean. Individuals occupy the same nest burrows from year to year, often lying within a metre of those of other individuals. These and other seabirds that migrate long over-water distances to small oceanic islands must surely be among the greatest of animal navigators. But how do birds achieve these remarkable feats of orientation and navigation over such huge distances?

It is not just a question of finding the way. Birds must know where in their journeys they need to do particular things, such as change direction or accumulate extra body reserves in preparation for a long non-stop flight. The fact that they can respond appropriately at specific places on their route again implies that they possess some geographical sense —an ability to detect and respond in an appropriate manner to conditions at particular locations.

The most obvious way in which birds and other animals could find their way around on a day-to-day basis is by use of landmarks. But such features are useful only in familiar areas. When moving over longer distances into unknown terrain, a reliable geographical reference system is needed by which to navigate. At least two types of factors can provide this reference —celestial and geomagnetic— and both are used by birds as directional aids (for reviews see Able 1980, Åkesson 2003, Emlen 1975, Wiltschko, R. & Wiltschko 1995, Wiltschko, W. & Wiltschko 2003). In migratory birds, compasses based on the sun, various sunset cues, stars, and magnetic information have been studied in detail, but a prior requirement for using any compass is that the bird should ‘know’ beforehand —either by inherent preference or experience— in what direction it needs to head.

Moreover, one feature of celestial cues, such as the sun and stars, is that they appear to change in position through each 24-hour cycle, as the earth spins on its axis. In the northern hemisphere, the sun lies in the south and moves during the day from east to west, and at night the stars rotate anticlockwise around the geographical north. In the southern hemisphere the sun lies in the north and moves from east to west, while the stars rotate clockwise around the geographical south. In using the sun and related factors in direction finding, therefore, birds in both hemispheres must allow for time of day. The same is not necessarily true for star patterns if they are used solely to indicate geographical north or south, determined by the centre of rotation of the night sky. Time-keeping depends on the internal clock, kept to time by the regular day–night cycle of light and dark.

The sun compass

The height of the sun’s arc in the sky varies with latitude and season, but it is always symmetrical with respect to true north or south. Its highest point in the sky at mid-day indicates due south in the northern hemisphere and due north in the southern hemisphere. The use of the sun as a compass by birds has been known for more than 50 years. Under a sunny sky, Common Starlings kept in circular wire cages during the migration period oriented in the same direction as free-living birds. They varied the angle they took to the sun according to the time of day. If the sky became overcast, their directional preference disappeared. When their view of the sun’s direction was changed using mirrors, the birds oriented at the same angle to the apparent sun as they would to the real sun (Kramer 1952, 1957). These experiments confirmed that starlings made use of a sun compass, which gave accurate information only if used in association with an internal clock, allowing adjustment of directional preference as the sun moved across the sky. Additional experiments in which Common Starlings were kept under artificial cycles of day and night of the same duration as natural time, but out of phase, revealed clearly that, in order to orientate, the birds used both the sun’s position on the azimuth (direction from the observer) and the time of day. Given a simulated stationary sun, a caged migrant orientated at different angles to it according to the time of day. The use of a sun-azimuth compass has now been confirmed experimentally in several species, and may be commonly used by diurnal migrants.

If homing pigeons and other species known to use the sun in orientation are transported to the southern hemisphere, they orientate themselves incorrectly, interpreting the sun as if they were in the northern hemisphere, indicating south rather than north. Regular trans-equatorial migrants must presumably be able to make the necessary adjustment, but how they do so is still unknown.

Skylight polarisation patterns

The ability of birds to detect sky polarisation patterns, which change with respect to the sun’s position (being particularly striking around the time of sunset), has also been demonstrated by experiment (Able 1993). Several species of normally nocturnal migrants have been shown to respond to manipulations of polarised skylight, especially around the time of sunset. The birds were tested individually outdoors in otherwise normal conditions in cages covered by sheet polaroids (Able 1982, 1989, Helbig & Wiltschko 1989, Moore & Phillips 1988, Phillips & Moore 1992). In each case, the birds changed orientation as predicted by alterations in the alignment of the polaroids. The experimental birds were clearly responding to polarised light as such, rather than to other sunset features.

The star compass

Providing that enough of the night sky is visible, nocturnal migrants proved able to use the stars as a guide. When tested in orientation cages, they could orientate correctly on clear starry nights, but became inactive or disoriented under overcast skies. They also became confused if star patterns were varied experimentally in a planetarium (Emlen 1967a, 1967b, Sauer 1957). When Indigo Buntings were tested under a natural starry sky during autumn migration, they preferred southerly directions. They maintained this southerly preference under an artificial star pattern imitating the natural sky in a planetarium. But when the artificial star pattern was changed by 180°, the birds changed their directional preference to the north. Under a static night sky, no obvious migratory restlessness occurred. The development of a star compass evidently involved learning, with celestial rotation as a directional reference, and captive Indigo Buntings without early experience of the night sky failed to orientate correctly in a planetarium (Emlen 1967b, 1975). Similar results were later obtained with Garden Warblers (Wiltschko et al. 1987).

In experimental conditions, some birds learnt to respond to a simplified and reduced star pattern, with some constellations blocked out, so long as this pattern rotated about a single conspicuous star. Nestling Indigo Buntings raised under an artificial sky with the star Betelgeuse (in the constellation Orion) as the point of rotation treated Betelgeuse as the Pole Star when subsequently tested. Detecting the rotation of the night sky probably takes considerable time —it could not be determined at a glance. Not surprisingly, therefore, the birds did not depend on the axis of rotation per se, but rather learned star patterns that indicated the axis, and thereafter they relied on those patterns. As with the sun compass, however, the ability and tendency to acquire this knowledge was apparently innate.

The use of a star compass has now been demonstrated experimentally in at least six different bird species, and may be general in nocturnal migrants. If the birds use the rotating star pattern only to define the position of the poles, then no correction for time of day is necessary. They may, however, gain further information from star patterns: as long-distance migrants proceed on their journeys over several weeks, stars that were once visible disappear below the horizon behind them, while others appear above the horizon in front, another indication that birds are unlikely to rely throughout on particular star patterns.

Celestial cues and time shifts

In using celestial cues, long east–west migrations present greater navigational problems than north–south flights because they involve time shifts, as the birds pass through successive time zones. If long-distance migrants using celestial cues to navigate did not allow for time shifts when travelling east–west (or west–east), they would make ever greater directional errors, and veer progressively further off course. The problem created by time shift is greatest at the highest latitudes, where the longitude lines are closest together, requiring more rapid adjustment. Hence, a second presumed function of an internal clock is to measure the changes in timing of sunrise and sunset, as the bird flies long distances west or east. High-latitude east–west or west–east flights are fairly common, being performed every year, for example, by the many waterfowl and seabirds that migrate along the northern coasts of Eurasia and North America to reach the Atlantic or Pacific Oceans on either side.

The magnetic compass

The second major system of bird orientation makes use of the earth’s magnetic field, which has both horizontal and vertical components. Imagine the earth as a hugely powerful magnet, whose north magnetic pole is situated fairly close to the geographic North Pole, and whose south magnetic pole is similarly close to the geographic South Pole. Running through the atmosphere between the two magnetic poles are invisible longitudinal lines of magnetic force, which circle the globe rather like the segments of an orange. At the equator, the magnetic force lines run horizontal to the earth’s surface, but toward higher latitudes they dip more and more strongly into the earth, becoming vertically downward at the magnetic poles. The inclination of the field lines thus varies within each hemisphere according to latitude. Hence, for any creature that can measure the inclination of the force lines, the earth’s magnetic field can give a cue to latitude and direction (toward the equator or pole) within each hemisphere. However, it cannot give a reliable cue to longitude, so it cannot provide a firm basis for ‘bi-coordinate navigation’ (although this may be possible in some regions because of irregularities in the field). In contrast to celestial cues, however, the magnetic field can give consistent information in all weather conditions, both day and night, and unlike the sun compass, it needs no correction for time of day.

As revealed by radar studies, nocturnal migrants can occasionally orientate correctly even under completely overcast conditions, as can caged birds with no view of the sky. Caged birds lost this ability when isolated from both the sky and the earth’s magnetic field behind metal-reinforced walls. Moreover, when the magnetic field experienced by caged European Robins was rotated using a powerful electromagnetic coil so that, for example, magnetic north was shifted to the east while the field’s total intensity and inclination, as well as other potential directional cues, were kept unchanged, the birds altered their orientation accordingly (Wiltschko 1968). This crucial experiment showed conclusively that birds can respond appropriately to the earth’s magnetic field. Since then, the use of a magnetic compass has been demonstrated experimentally in about 20 bird species, and its use may be widespread, but mainly in association with other information. Although the method of its perception is still unclear, birds are sensitive to both inclination and intensity, but apparently not to polarity (Wiltschko & Wiltschko 1972).

Use of multiple orientation cues

Experiments thus indicate that birds can use a number of different compasses for orientation during long-distance migration, based on information from the sun and related pattern of skylight polarisation, from star patterns and from the earth’s magnetic field. Birds can learn and modify their use of all these compasses during their early lives or later. These different cues would normally give the same directional message. However, they might vary in reliability between regions, seasons and local conditions, so that for instance the sun compass cannot be used if the sky is totally overcast, the star compass might not be visible during the round-the-clock daylight in high-latitude summers, and a magnetic compass based on the angle of inclination is of little use around the geomagnetic poles and the geomagnetic equator (Åkesson et al. 2001). In practice, individual birds probably use information from more than one compass mechanism, with emphasis on whatever cues are most reliable in the conditions prevailing, switching from one type of cue to another during the course of a journey, depending on location, weather and light values (Muheim et al. 2006, Wiltschko, Weindler & Wiltschko 1998).

Which of several potential compasses is most important to a bird at a given time has been investigated in so-called ‘cue-conflict’ experiments. These involve presenting a bird with two (or more) orientation cues at once, manipulating one of them while leaving others unchanged, and monitoring the response of the bird (Able 1993). A typical experiment might involve placing an orientation cage surrounded by electric coils outdoors under a clear night sky. In this situation, the bird would have access to two known orientation cues, the stars and the magnetic field. The coils can be used to shift the direction of magnetic north so that magnetic compass directions differ from star-based ones. If, as compared to control birds tested in an unaltered magnetic field, the birds experiencing the cue-conflict changed direction in line with the magnetic field shift, one would conclude that in this situation magnetic information took precedence over stellar information.

Just such an experiment was conducted on three species of Sylvia warblers and European Robins captured on migration (Wiltschko & Wiltschko 1975a, 1975b). When the directions of stellar and magnetic north were at variance, the birds seemed to orient preferentially with respect to magnetic cues. The warblers changed direction during the first test in the conflict situation, but the robins did not shift until tested in cue-conflict for several consecutive nights (a finding later replicated on the same species elsewhere, Bingman 1987). Robins apparently needed longer to work out the changed relationship between stellar and magnetic cues.

Cue-conflict experiments have proved especially useful in assessing the cues used around sunset, a time when many birds set off on migratory flights. Orientation based on visual cues between the time of sunset and the appearance of the first stars could be based on the sun itself (e.g. the azimuth of sunset) or on patterns of polarised skylight as mentioned above. Both could provide the same reliable directional information. However, in cue-conflict experiments involving several different species, polarised light appeared to be the predominant influence on directional preference when placed in conflict with the sun’s position or magnetic directions (Able 1993).

So far, we have considered only compass directions, but experienced birds can practice bi-coordinate navigation to re-find places they have previously visited. Such navigation could be provided by any two non-parallel gradients, and in theory the different gradients could be provided by different types of cues, for example one coordinate being based on a celestial cue and another on a magnetic cue. Whereas latitude can be fixed by both celestial and magnetic cues, longitude seems much more difficult to determine. Displacement experiments with White-crowned Sparrows in arctic North America indicated that a combination of geomagnetic and celestial information might be used to define longitude, but the precise mechanism remains unclear (Åkesson et al. 2005).

Another way in which different types of cues are not necessarily independent of one another is that a bird may calibrate one compass cue against another. Thus, while the rotation of the earth relative to the sky provides a stable reference for defining geographic north and south, changing geomagnetic declination (waviness in the force lines) renders the earth’s magnetic field less reliable in this respect. Accordingly, young birds were found able to use celestial information to calibrate a migratory orientation response to the earth’s magnetic field (Weindler et al. 1996). The combined experience of the night sky and the natural geomagnetic field seemed crucial for songbirds at high latitudes to find the appropriate migration direction to a population-specific wintering area (Weindler et al. 1996). Experienced adult Savannah Sparrows (Passerculus sandwichensis) also used celestial cues to recalibrate their migratory orientation to an experimentally shifted magnetic field (Able & Able 1995). Subsequent experiments revealed that Savannah Sparrows used polarised light cues from the region of sky near the horizon to calibrate the magnetic compass at both sunrise and sunset (Muheim et al. 2006).

In another study, Catharus thrushes caught on migration were exposed to an experimentally deflected magnetic field during twilight, and then released and radio-tracked on their subsequent night flights (Cochran et al. 2004). Their tracks indicated that the thrushes recalibrated their magnetic compass in relation to twilight cues, and then relied on their (miscalibrated) magnetic compass for their nocturnal flight, apparently ignoring stellar cues. The experimental birds changed to normal orientation again on succeeding nights, apparently having recalibrated their magnetic compass (correctly) back to north. Daily recalibration of the magnetic compass could explain how birds cope with changes in magnetic declination during the route, as well as various local magnetic anomalies; it could also explain how birds operating with a magnetic inclination compass can cross the equator (where the force lines are horizontal) without becoming disoriented.

Response to specific areas

Experiments using an artificial magnetic field implied that some birds use magnetic information to indicate regions where they must stop migrating, change direction or accumulate large fat reserves before crossing a barrier. For example, young migratory European Pied Flycatchers from western Europe showed a distinct change in compass heading when exposed in captivity to values of the magnetic field normally encountered in southern Europe, where the normal migratory route shifts from southwest to south. However, the altered magnetic field was followed by the shift in orientation only when applied at the appropriate time during the migratory period (Beck & Wiltschko 1988). In European Pied Flycatchers, therefore, the magnetic conditions of the location where the change is to occur and the time programme evidently interact to produce an appropriate response in an appropriate region.

In another experiment, some juvenile Thrush Nightingales (Luscinia luscinia) were caught in autumn in Sweden and exposed there to the geomagnetic conditions they would normally experience in northern Egypt. These birds promptly accumulated high fat levels appropriate to the subsequent desert crossing. They contrasted with control birds, exposed to local geomagnetic conditions, which accumulated much smaller fat reserves typical for south Sweden (Kullberg et al. 2003). However, birds trapped late in the onset period of autumn migration accumulated a high fat load irrespective of magnetic treatment. It seemed that the relative importance of endogenous and environmental factors in individual birds was affected by time of season, as well as by geographical location.

These experiments suggest that inexperienced birds on their first migration can detect and make use of the geomagnetic field, at least to indicate when major changes are needed during their journeys. The implication is that such birds have an inborn response to external geographic cues (especially geomagnetic cues) that are characteristic of certain latitudes or regions, and that they can use particular conditions to trigger a change in direction or fattening regime (Beck & Wiltschko 1988, Fransson et al. 2001). At least two mechanisms seem to be involved. The first is an endogenous time programme which switches particular activities on and off at appropriate times in the migration cycle (see above). The second is a response to particular latitudes or more specific regions, at least partly through regional magnetic or other conditions, which can similarly trigger appropriate changes in migratory behaviour. How much these separate mechanisms act independently or in conjunction with one another is an open question, but different species would not necessarily be expected to respond to particular experiments in the same way.

Birds that have experienced a wintering area may use the magnetic conditions there to halt migration on subsequent journeys. Adult Tasmanian Silvereyes (Zosterops lateralis) were tested near the mid-point of their south–north migration in southeast Australia. Birds exposed in captivity to artificially generated magnetic field values of inclination and intensity normally experienced near the start of their migration, oriented correctly toward north-northeast. In contrast, birds exposed to magnetic field values that they would experience near the end of their migration ceased to show any significant directional preference; in this respect, they acted as though they had arrived in wintering areas (Fischer et al. 2003). In contrast, no effects of changing the artificial magnetic field were noted in inexperienced young birds caught prior to their first migration, and which had therefore never visited the wintering area. The implication is that birds can learn the magnetic conditions of areas important to them, and use these conditions to indicate when to stop migrating.

Rhumbline and great circle routes

An unresolved question in migration research is whether long-distance migrants travel on straight ‘rhumbline’ routes (also called loxodromes) or on great circle routes (also called orthodromes). Navigationally, rhumbline routes are the most straightforward, because the bird could set off in the appropriate direction and maintain the same compass heading throughout its journey. If the route ran directly north–south, it would also be the shortest route between two points on the earth’s surface and would not involve a time-shift. However, if the journey had an easterly or westerly component, so that it involved crossing lines of longitude (as most routes do), a constant heading would still be the simplest but not the shortest route. The great circle route covers the shortest distance between two longitudinally separated points on the globe, but requires continual change in heading during the journey. Great circle routes are thus more demanding in their navigational needs. They can be accomplished by aeroplanes, with sophisticated navigation equipment, but whether by birds remains uncertain. Moreover, on any journey that involves longitudinal displacement, whether on rhumbline or great circle routes, the bird is also subject to time-shifts, as mentioned above. These time-shift problems are greatest at high latitudes where the longitude lines are closest together, but it is also at high latitudes where the distance savings on great circle routes are greatest.

The tracking of individual birds on their journeys has shown that many take a straight, constant-direction rhumbline route, even when they would save much time and distance by taking a great circle route. This held, for example, in Brent Geese migrating between the Wadden Sea and the Taimyr Peninsula in Siberia. However, it was not clear whether these geese took a rhumbline route because they were not capable of navigating the shorter (over-water) route, or in order to stay near the coast with its feeding areas. The shortest (great circle) route was about 4300 km, compared to the rhumbline of about 4700 km. Although the birds that were tracked kept closer to a rhumbline than a great circle, they also made continual minor deviations, bringing their average flight distance to 5000 km, at least 700 km (16%) further than the shortest possible route (Green et al. 2002). Similarly, Brent Geese travelling from Iceland to the Queen Elizabeth Islands in northeast Canada tended to migrate along fairly straight rhumbline routes to their breeding areas, and Red Knots (Calidris canutus) performed likewise (Gudmundsson et al. 1991). Again these routes took the birds mostly over land, where they could come to ground in inclement weather. Many other birds that have been studied have followed rhumbline routes.

In contrast, evidence that any birds take great circle routes (other than north–south) is as yet rather slender. Use of radar on the coast of northern Siberia revealed the occurrence of an east-northeast post-breeding migration, indicating direct flights between Siberia and North America, 1800–3000 km across the Arctic Ocean (Alerstam & Gudmundsson 1999). If the migrants gradually changed their orientation to the right during these flights, they would travel towards Alaska and neighbouring parts of Canada along the shortest possible great circle route to South America. The commonest species involved were the Pectoral Sandpiper (Calidris melanotos) and Red Phalarope (Phalaropus fulicaria), which winter on and near South American coastlines, respectively. This is one piece of evidence indicating that some long-distance migrants might travel along approximate great circle routes, but it will remain inconclusive until birds have been followed along more of the route. Ring recoveries from these or other candidate species are also insufficient to confirm travel by great circle routes. Only when not influenced by topographic features, important feeding sites or weather patterns, would migrants be expected to follow either straightforward rhumbline or great circle routes, and this situation may be quite rare.

As a further point, among the known orientation mechanisms, only a sun compass with no compensation for changes in local time could lead birds along a track similar to a great circle route. If birds did compensate for the time shift, using the sun compass they would follow a route similar to a rhumbline route. Hence, either type of route would appear possible using the sun as a compass. However, routes based on use of the stars or the earth’s magnetic field would invariably run closer to a rhumbline than a great circle route. One way in which birds could follow an approximate great circle route would be to divide the journey into stages with one or more appropriately positioned stopover sites, flying straight from one to another, but making a directional change at each one. Many landbirds are known to take roundabout routes in order to avoid long water crossings or high mountains, or to make use of re-fuelling sites that are off the most direct route. The journey is thus divided into successive legs with different main orientations. For example, most of the migrants that travel in autumn from western Europe to West Africa travel southwestward into southern Iberia, and then take a more southerly course into Africa. If they continued heading southwestward, they would end up far over the Atlantic Ocean.

Overall, straight rhumbline routes based on constant compass headings appear more likely in many bird populations, and are consistent with the routes frequently recorded by ringing and radio-tracking. They also fit the experimental evidence (based mainly on passerines) of a genetically fixed directional preference that steers inexperienced juveniles towards their wintering areas (although they may change directions at specific points on their journeys). Clearly, more research is needed on the precise routes taken by long-distance migrants before their navigation systems can be more thoroughly assessed.

Route-finding

In conclusion, the navigational tools available to migrating birds include: (a) a celestial compass based on sun, skylight polarisation and star patterns; (b) a magnetic compass based on the earth’s magnetic field; (c) an internal clock, recording diurnal (circadian) and longer-term time changes; and (d) an inherited mean migratory direction and time programme, which together ensure that the bird flies in an appropriate direction for an appropriate time. Some, if not all, birds also have a map sense used for homing to a previously experienced place. Moreover, the fact that birds can re-find places they have already visited implies a good spatial memory.

Various other navigational methods may also be used by migrating birds, including the use of ultra-sound and odours. Tube-nosed seabirds, in particular, have a well developed olfaction sense which may be used for locating food and nest burrows, but may also be used in longer-distance navigation, although this idea is conjectural only.

Equipped with these navigation aids, a bird could use at least four different route-finding strategies:

1.   In guiding or ‘follow-the-leader’, some birds might complete their migration by following others which know the way, thereby learning the route. Providing the leaders were experienced, they could pass on knowledge of travel routes to younger individuals by cultural transmission, assuming the youngsters could record and memorise their journey in some way. This strategy is used by swans, geese and cranes, in which young migrate with their parents. Any birds using this method would benefit from a back-up mechanism (such as clock-and-compass) in case they were left to migrate on their own. This method cannot be used by many other bird species, however, because the young migrate independently of the adults, and sometimes at a somewhat different time of year. One example is the Common Cuckoo (Cuculus canorus) in which the adults leave their breeding areas a month before their latest young, reared by various host species, have even left the nest.

2.   In clock-and-compass (vector) navigation, birds aim to head in a constant migratory direction (which may change once or more times during a journey) for an innately determined amount of time controlled by an internal clock. By this mechanism birds could reach previously unknown but appropriate wintering areas. Theoretically, birds of all ages could use this orientation strategy, which has been demonstrated experimentally in young passerines and others. On this mechanism alone, birds are unable to determine their position and are therefore unable to correct for wind drift, directional mistakes, over-flight, or experimental displacement. Also, adults would be unlikely to return to the same precise localities in successive years, which many are known to do.

3.   In bi-coordinate navigation, birds can sense at least two global coordinates forming a reliable grid through which they can determine their geographical position. Bi-coordinate navigation could provide continual positional feedback, enabling birds to correct for drift or directional mistakes. Theoretically, birds of all ages could use this strategy, but experimental evidence from several species suggests that it is used primarily by experienced birds returning to a known area.

4.   In piloting, a migration route is retraced by using a sequence of learnt landmarks. This method would require birds to build a landmark-based map during a previous migratory journey which is retraced during each subsequent migration. Such landmarks could be visual, auditory, magnetic or olfactory. This method could not be used by inexperienced migrants on a first-time journey, but could help birds returning to a known area.

While we still have much to learn about this fascinating subject, migratory birds clearly have a number of orientation and navigation mechanisms available to them, and are not restricted to just one. They also have an inherent ability to learn to make use of the more important navigational cues, re-assessing them and if necessary cross-checking them at points along the route. By these various means, they can find their way each year between widely-separated localities, to the continuing amazement of mankind.

Birds as colonisers

Like other terrestrial organisms, land-birds that have the opportunity readily expand their geographical ranges across land areas, occupying all suitable habitats. A spectacular example is the Eurasian Collared-dove (Streptopelia decaocto) which in the mid-twentieth century colonised much of Europe from breeding areas further east. It is now a common village bird in areas where a century ago it was totally unknown. In addition, however, powers of sustained flight enable land-birds to cross extensive sea areas, and thereby to colonise distant lands in a way that most other animals cannot. As a result, land-birds of one sort or another are found on almost every oceanic island in the world, where other animal groups are poorly represented or altogether lacking.

Every bird-watcher delights in seeing occasional individuals of certain bird species, which normally live far away but periodically turn up as rarities. There can be no doubt that individual birds continually reach localities hundreds of kilometres beyond their normal range boundaries. Bird vagrancy is a familiar phenomenon, especially in well-watched places such as the British Isles, where more than half the species recorded in the last 200 years are classed as vagrants. Some appear in small numbers every year, but others only singly at intervals of many years.

In both Europe and elsewhere, vagrancy is especially evident on off-shore islands. Although the variety of species is somewhat restricted, and most vagrants are drawn from migratory populations, bird movement is clearly not the major obstacle to range extension. The main problem is the difficulty in establishment. Migrants blown off-course are usually programmed to re-fatten and move on, reducing the chance that they will stay in a new area. Unless they are adapted to the journey, birds that have crossed an ocean are likely to arrive so exhausted and low on body reserves that they have poor survival prospects, even in favourable environments; and even if individ­uals recover from the journey, remain and survive long enough to breed, their low numbers may render them vulnerable for several years.

Despite the difficulties, some remarkable trans-oceanic range extensions have occurred among birds in recent times. One remarkable example is the Cattle Egret (Bubulcus ibis) which around 1880 is thought to have crossed the Atlantic from Africa unaided, reaching Surinam in South America. From there it gradually spread to occupy grassland habitats through much of the New World, including Caribbean Islands. In the opposite direction, the same species spread from Asia via New Guinea to Western Australia, and then on to New Zealand. It has doubtlessly been helped by forest clearance, and the proliferation of cattle ranching, as it naturally associates with large grazing mammals to feed on the insects they disturb. Moreover, in the relatively short time since it arrived in these new areas, the Cattle Egret has established new migrations, including one between New Zealand and Australia.

Since the occupation of New Zealand by European people, and the resulting habitat transformation, several bird species have crossed the 1600 km of sea and colonised the country from Australia (Baker 1991, Bell 1991). Besides the Cattle Egret which arrived around 1958, new colonists include: the Silvereye (Zosterops l. lateralis), since 1855; Grey Teal (Anas gibberifrons gracilis), re-established since 1916; Welcome Swallow (Hirundo neoxena), 1920s; Spur-winged Lapwing (Vanellus spinosus), since about 1932; Masked Lapwing (Vanellus miles), since 1940; White-faced Heron (Egretta novaehollandiae), since 1940; Royal Spoonbill (Platalea regia), since 1950, still rare; Common Coot (Fulica atra), since 1954; and Black-fronted Dotterel (Elseyornis melanops), since 1954. Such species may have reached New Zealand in earlier times, but did not persist, possibly because suitable habitat was lacking then. They contrast with many other animals which have reached New Zealand only with more direct human help.

Concluding remarks

From time immemorial, the seasonal movements of birds have fascinated people. They have continually raised questions about where particular species come from, where they go, how they time their journeys, and how they find their way. After more than a century of scientific research, we have come a considerable way towards answering these questions in broad terms, at least for northern hemisphere species. But substantial gaps in our understanding still remain, notably on important aspects of navigation, fuel accumulation and use on journeys, and the control of migratory timing. In addition, for obvious reasons, we still have only skimpy knowledge of the movement patterns of pelagic birds which spend most of their lives on the open sea. As time goes by, research is likely to extend to a wider range of species, and to other parts of the globe, while developments in technology will make more things possible. In the foreseeable future, smaller and more sophisticated radio-transmitters are likely to become available, enabling us to follow much smaller birds, such as swallows, day by day on their journeys, recording the heights at which they fly, the precise routes they take, and when and where they stop to rest and feed. No doubt future discoveries, like previous ones, will generate excitement and awe in scientists and lay people alike, as we continue to be amazed at what migrating birds can do.

 

 

Ian Newton

Bibliography

The full references to the citations included herein can be found towards the back of the volume, in the General List of References.