Most of the experimental research discussed above was concerned with the modification and further development of migratory behaviour, and not with its inception. Most research also involved only one species, although some findings were confirmed on other species. It implies that, given sufficient underlying genetic variation and strong enough selection pressure, some wild birds could change their migratory habits within only a few generations in response to environmental change (for more examples see Chapter 21). Moreover, the intermediate behaviour of hybrids suggests control by several genes, and not just one (for any behaviour controlled by a single gene would be expected to show an on-off pattern, with no scope for intermediates). This does not mean that all populations could adapt rapidly to environmental change. Insufficient genetic variation in a trait can prevent adaptation occurring altogether, whatever the selection pressure. Once the available genetic variation within the population had run out, time would then be needed for mutation to create yet more variation on which selection could act. Whether a population will respond to a new selection regime also depends on whether the trait under selection co-varies with other traits. If selection to change one trait beneficially brings concurrent detrimental changes in other traits, the resulting behaviour is likely to be a compromise. In terms of the length of journey, there are also limits to how much of each year a bird could spend on migration, and still have time to reproduce and moult, and how much body fuel it could carry for non-stop flights. These aspects act as constraints on evolutionary change, at least in the short term.
Artificial selection for higher and lower levels of migratory activity among captive Blackcaps resulted in changes in subsequent generations not only in the frequency of individuals showing high levels of activity, but also in the amount of migratory activity they showed. The correlation between these two traits was very strong, suggesting that both traits were controlled by the same genes: in other words, genetically, both incidence and amount were aspects of the one trait. This finding has important implications for the evolution of migration: in particular, that obligate migrants and non-migrants could be present in all bird populations, although the frequency of one or other may be very low in populations classed as completely sedentary or completely migratory. As a consequence, all bird populations could be considered as partially migratory, differing only in degree of migratory behaviour.
High correlations between migratory traits (date of onset, incidence, and duration of migratory activity) indicate that such traits are expressed as a syndrome: that is, different traits do not occur in isolation, but as a suite of connected features. This means that if selection changes one trait, others will change at the same time. If, for instance, the survival of sedentary individuals increases due to milder winters in the breeding areas, the frequency of non-migrants in the population will rise as a direct response to selection. At the same time, migratory individuals in that population would be expected gradually to delay their autumn departure from the breeding area and to shorten their migration distance as correlated responses (Pulido & Berthold 1998). This fits the facts that partial migrants often travel only short distances and, compared with other migrants, usually leave late and return early to their breeding areas. It also implies that the difference between facultative and obligate migration might be a question of degree, rather than type, involving genetical influence over a moveable threshold. Migratory syndromes may also include other features, such as morphology and physiology but, as far as I am aware, these aspects have not yet been investigated in this context (for further discussion of migratory syndromes see Dingle 2005).
On the basis mainly of the experimental results described above, Berthold (1999) proposed what he called a 'comprehensive theory' of migration control. The theory centres on the concept of obligate partial migration, in which individuals vary in their inherent migratory activity. This situation is clearly widespread (and possibly universal) in birds, enabling complete migratory or complete resident behaviour to be achieved or lost, depending on selection. That occasional migrants exist in basically sedentary populations has long been evident, as exemplified in recent analyses of British ring recoveries (Wernham et al. 2002). No clear-cut difference separated resident and migratory species, but different species showed a continuum of variation in the proportion of individuals that had made long directed moves. This accounts for many familiar aspects of bird movements: for why occasional individuals of normally migratory species (such as Barn Swallow Hirundo rustica) can be seen in breeding areas in winter, long after other individuals of their species have left; and why, even in the most sedentary of species, occasional ringed individuals are recovered in autumn or winter at long distances, all in a restricted direction south of their natal areas. In the Reed Bunting Emberiza schoeniclus and Stock Dove Columba oenas, for example, these long-distance birds provided less than 1% of all recoveries of British-ringed birds.
What is clear is that, once obligate partial migration has evolved, the whole range of behaviour from strictly resident to strictly migratory can develop in short-lived birds in much less than a human lifetime. If the propensity for resident or migratory behaviour exists in a population as a gradient rather than as a dichotomy (as research implies), selection to change from resident to migratory (or vice versa) could begin anywhere on the behavioural gradient, and is not necessarily dependent on the prior presence of a few migratory individuals in an otherwise resident population (or vice versa).
This theory of in-built flexibility contrasts with an earlier suggestion that migration might have evolved independently several times in birds by convergent evolution. If this were so, it could be controlled by different mechanisms in different types of birds. Because most of the relevant work so far has involved passerines, this earlier theory of multiple control mechanisms cannot yet be considered as having been invalidated, however unlikely it may seem. Moreover, all the known examples of variable or changing migratory behaviour, whether from wild or captive birds, refer to species that live in seasonal environments, where migration is the norm. Whether birds that have spent their entire evolutionary history in the relatively stable environment of lowland tropical rainforest could develop latitudinal migration so rapidly is much less certain, and provides an obvious opening for further research. Examples of Neotropical families or subfamilies which contain no known migratory species include the Piprinae, Pipromorphinae, Dendrocolaptinae, Formicariidae, Rhinocryptidae, Conophagidae and Thamnophilidae.
Another gap in our understanding concerns the extent to which differences in migratory characteristics (such as distance and direction) are reflected in the overall genetic differentiation between populations. The few studies undertaken so far suggest that differences in migration behaviour between populations occupying contiguous breeding ranges generally do not correlate with strong overall genetic differentiation, as reflected in microsatellite or mitochondrial DNA (for Blackcap Sylvia atricapilla, see Helbig 1994, Pérez-Tris et al. 2004; for Willow Warbler Phylloscopus trochilus see Bensch et al. 1999; for Prairie Warbler Dendroica discolor see Buerckle 1999; for Great Bustard Otis tarda see Pitra et al. 2000). Rather, changes in their migratory behaviour seem to result from selection on relatively few loci (Helbig 2003). This conclusion agrees with the findings that few genes may be involved in the expression of migratory traits (Helbig 1994, 1996), and that strong correlations exist between these traits (Pulido & Berthold 2003). It also agrees with the finding that some evolutionary changes in migratory behaviour can happen rapidly (within a few generations), and that such adaptations are population-specific rather than species-specific.
This discussion leads to the question of how many different aspects of migratory behaviour (or traits) there might be on which selection could act (see also Chapter 1). We have one apparent gradient in behaviour, as reflected perhaps in Zugunruhe, which determines the timing and distance of migration (from residents to short-distance and long-distance migrants). Associated with this is another gradient in patterns and levels of fuel deposition in preparation for different types of journeys. We have a third apparent gradient between mainly endogenous control (regular longdistance migrants) to mainly external control (facultative irruptive migrants). We have a fourth apparent gradient in breadth of directional preference (from no preference to strong and narrow preference in a particular direction). Fifth, we have an apparent binary response on whether the movement is one-way or return (but even this might be regarded as a graded response from no return, through partial return to full return, Chapter 15). These are the behavioural gradients on which selection can act. The discovery of correlations between different aspects of migration may tempt us to reduce these different aspects of variation to as small a number as possible. But given the enormous diversity of movement patterns found among birds, pooling aspects of behaviour in this way (from correlations in a small number of similar species) brings the risk of masking other kinds of variation. Each additional aspect of modification allows an extra 'degree of freedom', and hence finer adjustment of behaviour to environments and species needs. Acknowledgement of all aspects of variation may be necessary to account for the wide range of movement patterns found among modern birds.
Was this article helpful?