s-Jones 1991). Individual feathers or groups of feathers can of course be replaced at any time if they are accidentally lost, but this does not entail starting a full moult cycle.In all birds, feathers become worn and require replacement if the plumage is to function efficiently in the mechanical demands of flight, and in providing insulation for the body. The process of moulting the feathers follows a particular sequence, so that the normal functions are disrupted as little as possible, but the form which it takes will depend upon the lifestyle of the species concerned.
Growing new feathers involves protein synthesis, which leads to increased energy demands. Moult is thus a potentially stressful process in physiological terms, and would not be expected to occur at the same time as other activities with high energy demands such as breeding or migration. Wing moult will also affect the bird’s flying ability, so that it should not occur during migration or other times when efficient flight is at a premium. General moult of the body plumage will make insulation less effective, and thus should not take place during the coldest months; in fact the bird would be better off with new plumage at times when the demands of thermoregulation are highest. These principles provide a starting point for reviewing our knowledge of moult in the starlings.
Recording moult
The recording system which has become established amongst bird-ringers focuses primarily on wing moult. Each feather is assigned a score on a 0–5 scale, with old feathers scoring 0 and new feathers 5. Since the moult is normally symmetrical on both wings, the resulting ‘moult score’ provides an indication of how far advanced the moult is, and by a graphical comparison with date, the time when moult starts and is completed in the population, and its duration in individual birds, can be estimated (Ginn and Melville 1983).
The moult score method is a quick and consistent means of recording wing and tail moult on both live birds or museum skins. For other feather regions, anything more than a crude assessment of the presence and extent of moult is extremely time-consuming. Some mathematicians have long been unhappy with the moult score as a basis for calculations, particularly since many authors claim better estimates of the duration of moult by regressing date against moult score, which is the reverse of the intuitively correct procedure - the date is surely the ‘fixed’ variable in this case! This has led to new models, based on the percentage feather mass grown (Underhill and Zucchini 1988). For birds, the total mass of the plumage is quite small, and less than half of it is made up by the wing and tail feathers: for instance in an African Pied Starling, with a total body mass (unplucked) of 95 g, the plumage constituted 7.8 g, with the contribution of the primary wing feathers being 1.3 g, the secondaries and tertials 0.8 g and the tail feathers 0.7 g (AC, unpub. data). Once these figures are known for a particular species, the moult score can be used to calculate the mass of primary wing feathers grown, and a much more accurate estimate of the duration of moult and indirectly of the extra energy requirements is possible. So far this method has been applied to one starling species only, the Common Starling in South Africa (Cooper and Underhill 1991).
The sequence and rate of moult
With very few exceptions, passerine species replace the primary feathers descendantly, from the innermost primary outwards to the wingtip. The secondaries are replaced from the first feather, adjacent to the first primary in the standard numbering system, inwards. Secondary moult generally starts when primary moult is about half complete. The three innermost wing feathers, the tertials, are renewed before secondary moult is completed, in a variable sequence. The tail feathers are often replaced from the innermost pair of feathers outwards, but tail moult is much less symmetrical than wing moult, and irregular sequences are common. Body moult commonly accompanies tail moult and wing moult, and for many species the period of primary wing moult spans the moult in all other body regions (Ginn and Melville 1983). Starlings have 19 flight feathers (remiges) on each wing; ten primaries, of which the outermost one is much reduced, six secondaries, and three tertials; and 12 tail feathers (rectrices).
Interrupted wing moult may occur, in which case birds will be found with some new feathers adjacent to old feathers, while there is no active moult. If moult is later resumed at the next feather which is due for replacement, the condition is termed ‘suspended moult’, whereas after ‘arrested moult’ a new moult cycle starts from the beginning, and the old feathers which have been retained are replaced in the new moult sequence (Ginn and Melville 1983). The condition in which more than one moult cycle is in progress in the same wing is termed ‘Staffelmauser’ (Stresemann and Stresemann 1966).
In several African starlings, interrupted wing moult was noted, but the only species in which it seems to be a common occurrence is Wattled Starling, which shows suspended wing moult (Craig 1983a, 1996). Suspended wing moult has been recognised as a regular part of the cycle in some long-distance migrants (Stresemann and Stresemann 1966) and Traylor (1971) found that interrupted moult occurred in some migrating Amethyst Starlings. Naik and Naik (1969) found a few cases of interrupted secondary moult in the migratory Rose-coloured Starling during the non-breeding season. In the Wattled Starling it also seems likely that suspension of moult coincides with periods when the birds are especially likely to be flying long distances, rather than periods when breeding occurs (Craig 1996). Local food shortages could be the trigger both for emigration and for moult suspension. Evans (1986) reported cases of interrupted moult in Common Starlings, in both migratory and resident populations, though it seemed to be most common in sedentary island birds from localities such as the Faroe and Shetland islands. He suggested that for some of these birds interrupted moult might occur for energetic reasons, during periods of temporary food shortage. However, Meijer (1991) demonstrated that under experimental conditions of restricted access to food, male Common Starlings started moulting later, but females did not, while there was no effect on the duration of moult in either group.
The rate at which moult is completed will clearly depend both on the growth rate of individual feathers, and on the number of growing feathers at any one time. This in turn will determine the effect which moult has on flying ability and energetics. The only passerines in which wing moult regularly occurs so rapidly that the birds may be temporarily flightless are the dippers Cinclus cinclus and C. mexicanus (Stresemann and Stresemann 1966). In the Common Starling, there is evidence that northern populations have a more rapid moult before migration than those in other areas (Lundberg and Eriksson 1984), and birds which start moult later, such as late breeders, also moult more rapidly (Meijer 1991). Based on the number of growing wing feathers, the rate of moult in starling species occurring in the same area may differ significantly (Craig 1983a).
The energetics of moult
The traditional view, based primarily on experience of birds in the north temperate region, was that the energy demands of moult are incompatible with breeding. However, research on tropical birds soon showed that moult-breeding overlap did occur in a significant number of species (e.g. Foster 1975). Detailed studies of the actual energy requirements during moult are now available for a few species of passerines, and these have shown that the energy costs of moult greatly exceed the mere costs of feather production, and require changes in diet to increase the intake of certain amino acids, as well as a higher overall food intake. This suggests that at high latitudes, with an increased daylength and thus feeding time, moult can be completed more rapidly so that moulting in these regions could be advantageous (Murphy and King 1991).
The starlings are predominantly a tropical group. Northern populations of the Common Starling and other species migrate to regions with a milder climate during the winter, so that the role of plumage in thermoregulation is less critical than for species which overwinter at high latitudes. The large communal roosts of the Common Starling do seem to be located at sites which provide a warmer environment to the birds using them (Yom-Tov et al. 1977). There has been very little physiological work done on any species other than the Common Starling, but it is probable that for most starling species the energy demands of moult are not the primary factor determining its timing in the annual cycle.
Age and plumage changes
When they first leave the nest, young birds often have a distinctive plumage, and only later come to resemble the adults, once they have undergone a moult. In species which take several years to mature, the young may have a series of immature plumage stages. There are also birds which change their plumage and appearance in the course of a year, alternating between a dull eclipse plumage and a brightly-coloured breeding or nuptial plumage, as is found in many members of the African weaver family (Craig 1983a).
The only starling species which shows seasonal changes in plumage related to breeding is Wattled Starling, but here the changes involve the loss of feathers which are then replaced in the course of the next moult, so that an ‘extra’ moult is not required. There also appear to be no cases in which plumage changes continue beyond the first year, although other features such as iris coloration may take more than one year to reach the adult condition, as in African Pied Starling (Craig 1983c).
The post-juvenile moult may be complete, which means that the wing, tail and body feathers are replaced, or partial, in which case it usually involves only the body plumage. Both forms are found within the starlings, and there appears to be a clear division based on systematic relationships. In the genus Sturnus, Common Starling (Bährmann 1964), Spotless Starling (Peris 1988), White-cheeked Starling (Kuroda 1963) and Rose-coloured Starling (Naik and Naik 1969) all have a complete post-juvenile moult, and so do Common Myna and Bank Myna in the genus Acridotheres (Naik and Naik 1969). Sometimes a few juvenile secondary feathers may be retained in Common Starling (Scott 1965). Under experimental conditions, juvenile Common Starlings kept on short days (10 hours light, 14 hours dark) did not undergo a post-juvenile wing moult, and showed only irregular moult of some body plumage (Williams et al. 1989). In Rose-coloured Starling, some juveniles show delayed moult (Roberts 1982, van den Berg 1982), and perhaps interrupted wing and tail moult (Roberts 1982, Herroelen 1987). Peris (1988) noted that the moult of young Spotless Starlings from second broods started much later than in the adults and young from first broods, so that these may represent late-moulting birds which are unable to complete wing moult.
Wattled Starling, which is closely related to the Common and Asian starlings, has a complete post-juvenile moult, whereas other African species apparently have a partial post-juvenile moult, in which the body plumage is replaced and sometimes some secondaries and rectrices. However, Wilkinson (1983) found evidence of a complete post-juvenile moult in Chestnut-bellied Starling.
The timing of moult
For most passerines there is a single complete moult, which takes place once a year. Seven species are known to show two complete moults within a year; these include both migrant and resident species, but none are starlings (Prs-Jones 1991). Individual feathers or groups of feathers can of course be replaced at any time if they are accidentally lost, but this does not entail starting a full moult cycle.
Often the complete annual moult follows directly after the breeding season, after which the birds may leave the area. This is particularly obvious in migratory species, and is clearly shown by Scandinavian populations of Common Starling (Lundberg and Eriksson 1984). In this case the moult is rapid, and in the wings several feathers will be growing simultaneously. However, in some migratory species the moult takes place in the non-breeding area, so that the post-breeding migration is undertaken with old plumage, while the birds will have fresh plumage on the return to the breeding grounds. This is evidently the case in Rose-coloured Starling (Naik and Naik 1969, van den Berg 1982), and probably also in Spot-winged Starling (Marien 1951). The timing of breeding and migration need not determine the timing of moult, as is shown in Amethyst Starling, an intra-African migrant in which breeding and migration times differ regionally, whereas the timing of wing moult appears to be consistent between the different populations. Thus some Amethyst Starlings breed, moult, then migrate; others breed, migrate, then moult; while some moult, and then migrate to the breeding area. This species is also unusual in that wing and tail moult are completed well before moult of the body plumage (Traylor 1971).
The Wattled Starling of Africa follows a nomadic life style. There is some indication that the main period of moult differs in eastern and southern populations, but the picture is confused by the frequent occurrence of suspended moult, which means that birds at very different stages of moult may be present in the same flock (Craig 1996).
Many non-migratory starlings show a regular post-breeding moult, and this seems to apply to most Asian species and many African species. There may be striking differences between resident birds in the same habitat. Thus Cape Glossy Starling has a fairly rapid moult directly after breeding, and there is no evidence of moult-breeding overlap in individual birds, whereas at the same localities in South Africa, African Pied Starlings start moult while breeding, and the moult proceeds more slowly with fewer wing feathers growing at the same time (Craig 1983b). Further north, both Burchell’s Glossy Starling and Meves’ Long-tailed Starling complete their moult before breeding (Brooke 1967a, 1967b, 1968). In East Africa, Dittami (1987) found sympatric populations of Rüppell’s Long-tailed Glossy Starling, in which moult and breeding often overlapped, and Greater Blue-eared Glossy Starling, which moulted after breeding was completed. In West Africa, Chestnut-bellied Starling has two breeding seasons within the year, and moult starts toward the end of the first, overlapping with breeding in the second period (Wilkinson 1983). Specimens examined in the present study showed that moult-breeding overlap is likely in several other African species, such as Fischer’s Starling.
If wing moult is very slow, extending over much of the year, it will inevitably occur at the same time as breeding. Red-billed Oxpecker has a very protracted wing moult lasting more than 300 days, even though this species inhabits a seasonal savanna environment and has a clearly defined breeding season (Stutterheim 1980c).
The control of moult
From the above account it appears that moult in starling species shows some correlation with systematic relationships, suggesting that inherited characteristics play a role. In passerine birds in general it appears that the form of the post-juvenile moult is genetically determined, and may be consistent at least at the generic level (Stresemann and Stresemann 1966), while the sequence in which wing feathers are replaced follows a limited number of alternative patterns, which may be common to many species.
What about the timing of wing moult? Breeding has often been regarded as the central event in the annual cycle, but moult could be the fixed point in the annual cycle, while breeding varies regionally in response to local conditions. This would imply a large hereditary component in the control of the timing of wing moult. It is also known that in many bird species, periodic events such as moult, breeding and migration are under the control of circannual rhythms, inborn patterns whose timing is set by particular environmental stimuli. Experimental studies of Common Starling have shown that manipulation of the circannual rhythm can induce several cycles of moult within a single year (Gwinner 1977). The moult season of Amethyst Starling discussed above, and the differing moult times of two populations of Red-winged Starling, which are concordant with the patterns of morphological variation (Craig 1988a), strongly suggest that the timing of moult is often under direct genetic control.
The role of hormones in the control of moult is much less clear. The Common Starling moults after breeding, and the male hormone testosterone will inhibit moult in male birds. Sustained high testosterone levels lead to a delay in the start of wing moult, which then commences not with the first primary but further along the wing, so that only the outer primary feathers are replaced (Schleussner et al. 1985, Schleussner 1990).
Later experiments applying testosterone during the moult led to interruption, which was usually resumed at the point where it had stopped, but in some cases a completely new cycle started. Simulating decreasing daylength generally increased the rate of moult (Dawson 1994). Testosterone treatment also delayed moult in male Wattled Starlings, although wing moult was not the focus of the investigation (Hamilton 1959). Hormone treatments which mimic precocious sexual development will also induce earlier development of adult plumage in many species. Testosterone seems to produce particularly strong effects, whereas the role of female hormones and of hormones which may stimulate, rather than inhibit, moult is not clear. Schleussner et al. (1985) suggested that in male Common Starlings moult is retarded by testosterone and stimulated by thyroxine produced in the thyroid gland, so that moult would be regulated by a push-pull action of these two hormones, within the limits of the ‘moulting window’, which is a set period in the annual cycle during which moult may occur. Their study showed that when moult started very late, it ended before all wing feathers had been replaced, apparently terminated by some pre-set deadline. In starlings which moult while breeding, either the response to hormones such as testosterone is different, or hormonal levels may have dropped below a certain critical threshold before moult starts. In both African Pied Starling and Red-winged Starling, wing moult occurs while the birds are attending second broods (Craig 1983b), and it is possible that hormone levels in the blood are significantly lower during this phase of the breeding season. Dittami (1987) found that in Greater Blue-eared Glossy Starlings moult did not occur during the period of peak levels of testosterone and luteinizing hormone in males, or luteinizing hormone and 17-ß-estradiol in females. However, there was no cyclic variation in the levels of these hormones in Rüppell’s Long-tailed Starling, and moult apparently occurred in all birds at the same period, despite great individual variations in blood hormone levels.
The moult of the Mimidae
If the mockingbirds and thrashers are the starlings’ closest relatives as suggested by Sibley and Ahlquist (1984), how similar are they in respect of moult? The ten North American species all have 10 primaries with the outer feather much reduced, 9 secondaries and tertials, and 12 rectrices. None of them has a pre-breeding moult, and the post-juvenile moult is usually partial except in two species, which sometimes have a complete post-juvenile moult. For five of the ten species, age-related changes in iris coloration have been reported (Pyle et al. 1987). Moult-breeding overlap has been recorded in Northern Mockingbird Mimus polyglottos (Zaias and Breitwisch 1990). There are no data on the moult of the other 20 Central and South American species. So although there are no striking differences from the moult pattern of the starlings, the known moult characteristics of the mockingbirds are also shared by other passerine families.
Bathing and anting
Starlings are keen bathers, and all the African species which have been observed in the wild bathe regularly and often daily fly several kilometres to visit water holes. Dust bathing has not been reported, however, except in oxpeckers. Starlings also sun themselves regularly, on the ground or perched in trees, adopting specific postures for this activity (Simmons 1989).
The behaviour known as anting, in which birds either passively ‘bathe’ in ants, allowing the insects to swarm over their plumage, or actively hold individual ants in the bill and pass them over the feathers, is much less common. Anting has been studied in detail in Common Starling (Querengässer 1973), and it has been observed in three African and three Asian species in the wild. There are also records of five African and 11 Asian starling species anting in captivity (Poulsen 1956, Simmons 1961, 1966). Only the active form of anting has been observed in starlings. The ants used in the process are apparently usually species which release large quantities of formic acid when molested, and this may reduce the number of ectoparasites on the plumage, although direct evidence is lacking. There are also reports of ‘anting’ with millipedes which produce defensive secretions (Clunie 1976), and even of starlings using other strong-smelling materials such as mothballs (Clark et al. 1990). Such behaviour may be associated with moult and feather care, though we lack a fully satisfactory explanation for its occurrence (Craig in press c).