In the Öresund area, S. Sweden, gap distributions of moulting Dunlin are clumped to the zero end (small gaps), and primary growth has been negligible in birds caught twice within a few days. Dunlin must suspend primary growth before migrating and at best grow incomplete primaries on "low jets" during rests, while fresh or seemingly growing primaries have been shed and grown on or near breeding-grounds. In spite of suspension or reduced growth, the schedule of early moult is arranged so as not to claim the metabolic capacity of the bird the way it does later on; here lies a clear adaptation to the sequence "migration first - moult next". Most of the birds initiating moult before migration have grey median coverts and are of general "western" alpina type, in addition two moulting birds of schinzii type are listed. This in turn means, that moult start on breeding-grounds is a widespread strategy in Dunlin populations breeding west of the Ural mountains as well.
43 % of all adult Dunlin caught in the Öresund area had started to exchange primaries, 2c birds to a significantly higher extent than 3c+ birds, 2c birds also shedding P1 and P2 simultaneously to a higher degree than 3c+ birds. Moulting birds, with or without gaps, had much the same mean weights as non-moulting birds, no consistent adaptations could be discerned here, but all 3c+ categories weighed slightly less than the corresponding 2c categories, non-moulting 3c+ significantly less. The average individual duration of exchange of remiges was estimated at 61 days from the regression "Time on percentage primary mass grown", hidden in the seemingly unbroken growth may be a suspension pause lasting for a few days and up to a week. Birds with "adult buff" median coverts start to moult at least five and possibly ten days ahead of birds with grey medians. The last birds with all remiges old (grey medians) and the first birds with all remiges fresh ("adult buff" medians) were noted at the same time: by the end of August. Western (and maybe at the same time: northern) populations may be more ready for a drawn-out wait for favourable breeding conditions than the easterly populations with "adult buff" coverts. From late August many migrant Dunlin fly with larger gaps than July/August birds: at most 1.6 - 2.3 primaries and up to 8.0 rectrices. Some of these birds are exhausted and have low weights on arrival; again the adaptive value of low weights can be doubted, such birds are vulnerable. From mid-October the exchange of P9, P10, inner secondaries and scapulars is arrested; birds with old outer primaries have been noted in the wintering population. 108 adults, 17 with "adult buff" medians or "droplet" primary coverts, 91 with grey medians, arrived with fresh remiges from late August and into October; these birds must have moulted completely on or near breeding-grounds. This capacity is by no means exclusively connected with "adult buff" medians, but since many birds from Yamal peninsula have grey medians, some of the October birds with fresh remiges and grey medians may still be Siberian.
All in all: moult study is a tricky business, and the literature is full of partly permanented errors, caused by lack of understanding among authors as well as among journal editors and reviewers. In this paper I will present moult data showing that the growth rate of feathers during migration is negligible, at least when migrants reach the Öresund area (and most probably throughout the south Baltic as well), consequently the major part of the recorded feather exchange must have taken place on or near breeding-grounds and will be resumed again when the migration journey is over. Because of the general level of insight into moult problems, I will be particularly meticulous, offering a lot of full data (moult protocols) quoted in the Stresemann manner (Stresemann & Stresemann 1966). Furthermore the problem of lacking linearity in conventional moult scores (based on feather length) from waders will be addressed, departing from the preliminary amendments of Summers et al. 1983 and Underhill & Joubert 1995, but here some work still remains to be done as far as Dunlins are concerned.
2.1. Transformation from length to weight score. In a Dunlin the weight ratio between primaries, secondaries and rectrices is approximately 4 : 2 : 1. The three outermost hold almost half of the total mass of primaries (Underhill & Joubert 1995), primaries 1 - 5 only 30 %. The central rectrix is shed with P7 or P8, and rectrices grow mainly with P8 or P9. In normal cases the first secondaries are shed when P7 is in pin, sometimes with P6, and in late moulters in September secondaries may be growing already with P4. The overall schedule for exchange of remiges is shown in Fig. 1, this line-up clearly shows how the demands on metabolism from growing feathers reach a maximum when P8 and P9 are growing; in the early phase of moult the metabolic resources are not fully claimed. This is the best bid for a genetically coded (but still dynamically modified) adaptation to the particular migration/moult pattern in Dunlin and a strong argument for converting ordinary moult score data based on relative length of primaries (e.g. Ginn & Melville 1983) into at least "percentage primary mass grown", in the ideal case "percentage remige mass grown".
In the ten-year-period 1995-2004 moult data were collected from 682 migrant Dunlin with growing or suspended primaries in the Öresund area. These data have been transformed from "moult score" to "percentage primary mass grown", the conversion greatly facilitated by the fact that feathers were scored in tenths of full length; where outer primaries are involved this gives better dissolution than the traditional score. The resulting overall regressions "TIME on SCORE" for both materials are shown in Fig. 2; the improved linearity (and increased explanation power) of the converted material is evident here. This conversion should be continued to include secondaries and tail-feathers as well, but the present material does not meet this wish, there are too many omitted secondary and rectrix scores. The general effect of weight transformation is to suppress the relative influence of lightweight primaries 1 - 6 (only 40 % of overall remige mass) on regressions.
2.2. Remige moult, suspension, gaps; early migrants. Table I gives median dates and conventional mean moult scores for relevant categories in late summer: birds in their second calendar year (2c), birds in at least their third calendar year (3c+) and birds with "adult buff" medians or "droplet" primary coverts. (Gromadzka 1986, 1989, about adult buff coverts) Birds with moult scores 0 (moult not initiated) and 50 (fresh) have not been included.
|Category||n||median||mean moult score ± 1 s.e.; s.d.|
|2c, non-moulting||128||24 July||-|
|3c+, non-moulting||641||28 July||-|
|2c, moulting||289||3 Aug||16.6 ± 0.5; 8.9|
|3c+, moulting||225||4 Aug||15.0 ± 0.6; 8.9|
|"abc" and/or "dc", ages|
|86||4 Aug||23.8 ± 0.9; 8.4|
Moult depends on age; 2c birds initiate moult before migration to a much higher extent than 3c+ birds do (cf. Fig. 1 and text of "Migratory progress..."). The average percentage primary mass grown for 2c and 3c+ birds in July and August is shown in Fig. 5, the overall material is presented with regressions TIME on PERCENTAGE PRIMARY MASS GROWN in Fig. 6 (birds with "adult buff" medians and "droplet" primary coverts) and Fig. 7 (birds with grey medians). Note that standard deviations increase in spite of increasing sample sizes in Fig. 5; the material becomes less and less homogeneous as the season proceeds.
Some birds complete growing primaries on or near breeding-grounds, then suspend the moult process and migrate, others migrate with one or (in rare cases) two or three incomplete primaries. Even schinzii birds follow this pattern, but I do not know their provenience, moulting schinzii may come from other Baltic countries than Sweden. The following moult-cards are not the only ones possibly attributable to schinzii birds, they are quoted because the subspecies determination has the highest degree of certainty in these cases. The second bird was taken at a breeding site, Foteviken, together with a. o. two not very well-developed schinzii yearlings. (Note that many alpina migrants occurring on migration in the Baltic before 15 July have belly-patches with separation along the midline; in some areas in Lappland visited by me, breeding males have been very schinzii-like!):
|Category||n||Mean weight ± 1 s.e., 1 s.d.||Mean gap ± 1 s.e., 1 s.d.|
|2c, no moult||111||46.4 ± 0.5 g, 4.8 g|
|2c, suspension; no gap||48||45.5 ± 0.7 g, 5.0 g|
|2c, suspension; gap||196||46.0 ± 0.3 g, 4.7 g||0.49 ± 0.03, 0.38|
|3c+, no moult||511||44.9 ± 0.2 g, 4.3 g|
|3c+, suspension; no gap||43||45.0 ± 0.7 g, 4.5 g|
|3c+, suspension; gap||147||45.5 ± 0.4 g, 4.5 g||0.44 ± 0.03, 0.38|
The distribution of gaps in 2c birds between July and September is shown in Fig. 8, the distribution of gaps in 3c+ birds in Fig. 9, the time distribution of all gaps in Fig. 10. It is advisable to compare the gap distributions in Figs. 8 and 9 with Greenfinch gaps in October - December (Fig. 4) and Fig. 10 with GAP versus TIME in Greenfinches (Fig. 2, Fig. 3)! The overall gap distribution consists of superimposed distributions, one starting with gap 0, the next with gap 1.0, these component distributions have a general Poisson character (mean value = variance), but as a matter of fact they are even more accentuated (variance > mean value): clumped to the zero end. The meaning of a clumped distribution in this particular case - and of deviations from it - will be adressed in the discussion, but the central feature will be emphasized once again: in the Öresund area scores of gaps in individual remiges in migrant Dunlin are not randomly and evenly distributed, but clustered to the zero end of distributions.
3c+ birds may shed P1 (n = 40) or P1 - P2 (n = 53) and suspend at full length or migrate with a gap; in 2c birds it is much more common to grow both P1 and P2 more or less simultaneously (n = 71) than to shed P1 only (n = 25) (G-test of independence; adjusted G = 6.01, p < 0.05; material 1995 - 2004). During summer months Dunlin arrive at the Falsterbo peninsula with rather small gaps: 60 out of 205 3c+ and 59 out of 274 2c with moult initiated had suspended "perfectly", showing no gap at all (adjusted G by G-test of independence = 3.72; the null hypothesis that complete suspension is independent of age can not be rejected at the 0.05 % level. Before 2002 there was a highly significant difference, in 2003 and 2004 this pattern was upset).
The 2c pattern from migration was shared by a brooding 2c female at Snannaviejgiera 68° 07' N, 19° 30' E), 870 m a.s.l., Swedish Lapland: it had P1 0.5, P2 shed but not showing on 3 July 1997. In the delta of Rautasjaure, 560 m a.s.l., but only a few km from the former site, two Dunlin pairs (or 2 + 2 birds) were observed on 9 July 1997. There had been downpour, snow and frost in the area on 8 July, the fact that obvious males escorted obvious females 300 m below the breeding altitude indicated that both pairs had lost their broods. In one pair one bird - as far as I could see the female, but the age could not be determined - had shed one or two inner primaries, by all likelihood two, or I would not have noticed it in the field.
2.3. Remige moult: growth rate. My own controls of adults from the Öresund area indicate very limited feather growth during migration flight, and there is little evidence (retraps) of birds shedding remiges on migration in July or early August, although it has occurred that a remige was shed when a bird was handled (cf. LIST 4). Eight representative cards including retraps are given below, all birds were caught after midnight, the retraps in 2006 coincided with a heat wave where a flock of some 50 Dunlin stayed for at least a week. The general impression is: summer migrants are in full control of the moult process, there may be substantial weight increases on stopovers, still the growth of remiges is not accelerated. When remiges shed on breeding-grounds are full-grown or just suspended, feather growth is effectively shut off as long as the transport phase of the annual cycle goes on. Late moulters may arrive in poor condition, they are likely to lose weight for a day or two if wind flats are unavailable, and if these birds show some feather growth, it is minute and restricted to secondaries and tail, the most crucial areas in flight:
There have been few exceptions from the rule, but one night in 2002 five birds collectively broke the "Öresund pattern" of small gaps by appearing with short "growing" innermost primaries. In the past, single birds have broken the pattern in this way, but I never had five together before. From the state of the feathers they could have been on breeding-grounds the day before, but at least one had not; by all likelihood it had flown at least 3,000 kilometers before showing up at Skanör. One guess is as good as the other: primaries in pin were shed the day before, after the birds had arrived at Skanör, or they were shed on breeding-grounds, and growth had been suspended throughout the migration flight. Gender may be involved as well, and season; 16.7.02 saw my largest early catch so far: thirty Dunlin.
In addition the rare cases from preceding years with primaries shed and no new growing primary are added for comparison, the list is finished by one late case of "perfect" suspension. (I repeat here: the new feather has reached the apparent point of attachment of the old one when the latter is shed (Watson 1977), so when the new one is not visible, it must lie just underneath the skin surface). In no. 9 primaries 1 - 5 were practically completed before P6 was shed, in no. 11 primaries 1 - 3 completed before P4 and 5 were shed simultaneously. No. 8 is of particular interest; here one possibility is that P1 was shed on breeding-grounds, and that there was a conflict between the growth and the suspension mechanisms; one feather grew at some advantageous time, one was suspended. Nos. 1 - 3 may belong to the same category. The same applies to practically all of these examples; one primary may have been shed on breeding-grounds and its growth suspended when the bird departed, the suspension of growth lasting throughout the migration journey:
Summing up the above: on average 56.9 ± 1.3 % of all adult Dunlin migrating by way of the Öresund area from c15 July onwards had old remiges. In July and August 21.5 ± 2.5 (2c mean value) - 29.3 ± 3.2 % (3c+ mean value) of the birds with initiated moult migrated with no gap at all, in most other cases there was one growing primary and probably minute or zero growth. A handful of cases where birds lacked one or several primaries are listed, their share of all birds with initiated moult is order of magnitude 1 % - these birds could have shed their primaries on breeding-grounds and suspended because of imminent departure, or they could have shed when reaching the Baltic, there is no way of knowing in the individual case. Single birds have shed primaries while being investigated.
2.4. Remige moult, suspension, gaps; late migrants. The capacity of migrating Dunlin is by no means exhausted with these moderate gaps, however, larger gaps occur from late August onwards, when P6 - P8 or P9 - P10 and a lot of secondaries and tail-feathers may be growing simultaneously. The outermost secondary (S1) is shed with P6 or P7, the tail grows with primaries 8 - 10, alula is always linked to P10. The obvious cause is the forced coincidence of moult "agenda" for different feather tracts, illustrated in Fig. 1, resulting in a sudden extension of gaps (cf. Fig. 10). In order to demonstrate the qualitative leap taking place here, a few moult-cards, all from the Falsterbo peninsula, are given in extenso:
With primary gaps amounting to 1.6 - 2.3 middle-sized primaries, 1.8 - 3.6 secondaries and 2.8 - 8.0 rectrices the relatively well-managed summer moult migration passes into something else in September: a race towards moulting-grounds, where the cost to migrants is reflected in exhaustion and in a few cases low (fat-free) weights at arrival. Weakened, newly arrived moulting males with weights 36 - 39 g is a recurrent feature in the Öresund area in autumn; ragged wing / tail and migratory effort do not comply well. Under such conditions, if predators are present, the moulting birds are the first ones to be predated; they are slow starters when the flock flies up. There is one very illuminating recovery, disclosing the origin of one such moulting bird; in Fig. 7 there are only three birds from later dates:
2.5. Migrants with fresh remiges. In the overall material there are 97 individuals (6.5 % of the total material) migrating with completely moulted remiges in the interval 25 August - 26 October, median 26 September. There are two waves of them, by all likelihood from two different areas; they have moulted remiges on breeding-grounds but have intact belly-patches as often as not. These birds form the "roofs" of diagrams 6 and 7, their temporal distribution is shown in Fig. 11.
2.6. Non-moulted extremes, arrested moult, body-feathers. Note, how the last birds with old remiges (27.8, 29.8, 30.8; all 3c+ with grey medians) coincide with the first birds with fresh remiges (2 x 29.8, 1.9; 2c grey, 2c+ abc, 2c abc) if Figs. 6 and Figs. 7 are overlapped. The unmoulted birds were investigated indoors, under a fluorescent lamp, to exclude the possibility that they had moulted a few primaries and then suspended. In the moult diagram shown in Ginn & Melville 1983 there are no adults with all remiges old in late August, but in Boere 1976 there are a few birds with old remiges from late August and one from c8 September (I suggest: just arrived). At the other end of the scale there are cases, where it is obvious that all remiges were not going to be exchanged, and cases of arrested moult from late October. Scapulars are often left unmoulted (being ornamental and shielding but not particularly functional in flight); if this is the case they are among the first feathers to be shed in the prenuptial moult in April - May, often one feather and one side at a time. Some males do not moult winter scapulars at all (1 - 2 of list below; the first one had started moulting inner primaries on 6.8, the second had not), maybe their short bills make them sufficiently attractive to females (Cf. "Wintering and spring staging Dunlin..." and "Bill length distributions..."), or the preceding ice-winter in N. Europe had made prenuptial moult particularly difficult in the spring of 1996.
A few more birds with arrested remige moult and old scapulars were noted in winter in the early nineties, but there are no moult-cards. As a rule all tail-feathers and outer secondaries are growing by September 1st, R1 and S1 often completed by that time, but no. 11 in List 4 under 2.3. shows, that moult may be at least suspended and possibly arrested even here. In general the remiges have precedence, while scapulars and body-feathers often lag behind, and secondaries and rectrices grow more than primaries, at least in terms of length:
The faded belly-patches still stand out in a few adults by mid-October, Wlodek Meissner (in litt.) mentions them from Poland and Klaus Hein (in litt.) from Fehmarn/Germany in October/November, but by the turn of October the few adults still left in the Öresund area are beginning to complete their winter-plumage. As a rule the October Dunlin with belly-patches (growing outer primaries and secondaries, old scapulars) are transmigrants; they stay for a day or two and are gone. Their presence catches the eye in an environment of 2,000 immaculate juveniles, and their departure never remains unnoticed. Next, when they occur in the Waddensea (or the Mediterranean), they will hardly be recognized as fresh arrivals, there is no way of telling this among tens of thousands of moulting adults, here everything remains status quo to the observer. This fact is noted and regretted already by Boere 1976.
3.1. Adult time-table on breeding-grounds, representativity of materials from other areas.
Time of departure: Adult Dunlin in the last stages of moult or in fresh plumage continue to migrate by way of the south Baltic area throughout the autumn. Obviously performing a complete moult in the breeding area, these late migrants may have hatched their young no sooner than 15 July and deserted them no sooner than 10 August (this is Knot time-table in Taimyr, cf. Tomkovich & Soloviev 1996), inner primaries shed at the latest by the same time. In August there is a rich supply of insect food even in high arctic areas, little competition (Holmes 1966 - a precondition is that other competing species have already migrated) and the option to remain in the breeding-area and moult there seems to be worth considering for the slender-built Dunlin. Lagging behind may involve some degree of scarcity as autumn advances, however, and that may be the reason why some moulting males arrive with fat-free or almost fat-free weights at the Öresund area as late as early October.
Goede, Nieboer & Zegers 1990 arrived at much the same time-table for adults, but they rush the juveniles: After the breeding season the adults of the European as well as the Siberian population leave for W. Europe late July/early August.(...) The Siberian juveniles leave the breeding area shortly after the departure of the adults. Ten years earlier and with fewer possible references, Hardy & Minton 1980 present a more conservative picture: Adult Dunlins leave their breeding grounds before the juveniles and migrate about a month ahead of the latter. I will not choose between these options, because neither view is entirely to the point; at any rate I am convinced that some (order of magnitude: 10 %) adults and quite a few juveniles will lag behind rather long; maybe in pockets of favourable microclimate. Some of the adults arriving at the Öresund area in October, accompanied by juveniles, are exhausted after a long stage, they are not merely arriving from Polish or German resting-areas.
Moult start of birds with "adult buff" and grey medians: In July, August and early September moulting migrants in the Baltic area may exhibit "adult buff" median coverts (Gromadzka 1986, 1989). These birds originate from some West Siberian population that starts moulting earlier than their European conspecifics; in Poland they had scores 40 - 50 when birds without "adult buff" had score 30 or less. I have added birds with "droplet" primary coverts to this category, since they are early moulters as well, and since "droplets" and "adult buff" often occur in one and the same bird. The regression of Fig. 6 indicates faster growth in such birds than in birds with grey medians, but this is probably incorrect, caused by a lack of material from the early moult phase. By all likelihood "adult buff" birds shed their first primaries at least a week ahead of more western birds with grey medians (regressions suggest five days), and grow remiges at much the same rate (the "adult buff" regression would be less steep with more early material).
From September onwards practically all migrating adults in the Öresund area will have grey medians, typical of "standard" alpina, but grey coverts occur as far to the east as Yamal Peninsula (Gromadzka 1989) as well. 91 birds belonging to this latter category have been caught on migration in the Öresund area in September and October (see under 2.2.) with completely new remiges, others have had growing outer primaries and inner secondaries (Fig. 7). Here is an indication, that the harshness of September weather at northern latitudes in Russia is exaggerated by authors; in some autumns regular heatwaves are likely to invite birds to lag behind - and the wintering population of the south Baltic area amply demonstrates, that Dunlin can weather periods of frost without flinching.
Non-moulting birds: Finally: 57 % of all adult Dunlin migrating by way of the Öresund area have had old remiges; they must arrive at the North Sea, only 500 kms away, in much the same state. A corresponding amount of unmoulted birds has never been noted in the Waddensea, however (e.g. Boere 1976), and according to Ginn & Melville 1983 a mere 2 % of all alpina arrive on the Wash in suspended primary moult, which is another puzzling statement. This fact demonstrates the confusion and general inaccessibility of the Waddensea area, good and representative catches may be possible only at spring tides. However, Boere 1976 points out that in all Waddensea samples there will be Dunlin with only one or two primaries in concurrent growth, commenting on this: "which is a normal situation in species in which arrested (suspended) moult is common".
Two points are made here: there is some uncertainty concerning the final departure of birds from breeding-grounds on the northern tundra; the August/September phenology of these areas is not very well documented. By all likelihood there is an extended departure; from late June till early October, and all combinations of moult and migration occur in all populations. And the "arrival phenology" of the Waddensea may be equally extended; from Baltic and Öresund catches (Brenning 1987, Meissner 1998, my own material) it is evident, that adults - with suspended moult or fresh remiges - not continuing by the Mediterranean Flyway (e.g. Brenning 1989, Gromadzka 1989), "fill up" at the Waddensea till well into October.
3.2. Energetic costs; moult suspended on migration, hastened in September.
Moult a lavish energy expenditure on migration: 25 % of a bird's dry weight are accounted for by feathers, and in passerines the overall metabolic rate may increase by as much as 1/4 during moult (Koch & De Bont 1944, Wallgren 1954). King 1981 estimated that moult seizes 16 % of the metabolized energy in a House Sparrow Passer domesticus, on the other hand Lindström et al. (1993) showed that the efficiency of feather production is a mere c3 % in captive Bluethroats Luscinia svecica and Redpolls Carduelis flammea; the main expenditure is believed to consist of costs for maintaining tissues necessary for feather production. This is a crucial cost for a bird expecting to fly 3 - 4,000 km in the shortest time possible, in particular the low efficiency speaks against any investment in feather growth during migration. I believe that the transport of energy to follicles with growing feathers is simply switched off in migrating Dunlin. There are 60 or 62 remiges to be considered; if e.g. 50 % of P3 or P4 are grown "on migration" (or stopovers), this means that some 3.5 % of the total remige mass is replaced during this phase. [P3 + P4 = 12.5 % of total primary mass according to Underhill & Joubert 1995); 15 secondaries + 5 rectrices together weigh c75 % of ten primaries]. September migrants are less favoured, with more remiges (tail, secondaries) involved in moult and primaries still growing (cf. Fig. 1); the increased overall gap makes flight more costly, hence their poor weight condition in a couple of cases. A second disadvantage is the fact that the outer primaries have more substance than inner ones; P8-P10 make up almost 50 % of the total primary mass in Dunlin, P1-3 only 15 %. (But the protecting primary covert grows simultaneously; PC1 weighs 2/3 of P1). On the other hand, late moulters seem to increase their metabolic effort; in the Wash Johnson & Minton 1980 found the growth-rate of remiges to be lower in the early phase than in the late. This holds true of the September bird controlled twice and shown in Fig. 8; its estimated moult duration is 57.5 days against diagram average 61.
Adaptive weight decrease? In the summer months all categories of migrating Dunlin, moulting and non-moulting, in the Öresund area have had much the same mean weights: 45 - 46.5 g, but it could be noted that all categories of 3c+ birds weighed less than the corresponding 2c category, the difference significant in non-moulters (Table II). (In addition the data of this table indicate that it may be slightly more costly to grow feathers to full length than just to proceed with an existing, open gap. The weight difference between moulters with and without gaps is not significant, but at least this observation offers no support for the findings of Holmgren et al. 1993b). Ten moulting birds from the Waddensea studied by Goede & Nieboer 1983 weighed substantially more: 49 - 58 g, still these authors concluded, that the reduced flight-efficiency during moult requires that Dunlin keep their weights low. If the Dutch birds weren't females, it seems reasonable to assume that the higher weights of Waddensea moulters were caused by extra tissue, supporting feather growth. On the other hand, when migrating males with growing tail and growing outer primaries weigh less than 40 g in September, this tends to adaptation ad nauseam; in reality they may simply have overstrained their resources, leaving themselves vulnerable to predation.
Cost reduction: closing the gaps before migration: In addition the refusal by Dunlin to commit themselves to moult on migration can be judged from the fact that most "growing" primaries are close to completion and do not grow in the Öresund area; fresh primaries have been replaced, growing primaries shed on breeding-grounds or in some staging area close to the breeding-grounds. The remark that the "intensity of moult was reduced" in Holmgren et al. 1993a lacks quantitative confirmation; but in Holmgren et al. 1993b the slowdown is evident from the remark, that the growth stage (ranking) never changed between the first and the last measuring of single feathers in Dunlin held captive for 2 days. The fact, that the "gap function" (Figs. 8 and 9) is clumped to the left (against the vertical axis) should be seen against this background: the bird is close to completion of a particular primary when occurring on migration in the Öresund area, the rate of feather growth has been substantially slowed down, and no new remiges will be shed on migration. Some ten cases of full gaps (primary shed, the new one not showing) in the Öresund area complicate the picture; either they were shed on breeding-grounds and completely suspended or shed the moment birds were beginning to "feel" the proximity to the Waddensea area.
3.3. Moult start as a characteristic of populations. Individual degree of freedom.
Given these facts, the low efficiency and the high cost, as well as the obvious slowdown of growth, I find it most likely that the option "moult start in the breeding-area + suspension" - and no appreciable "moult on migration" - embraces the whole breeding area of the subspecies alpina as well, it is not an exclusive characteristic of populations from east of the Urals as once claimed by Greenwood 1983. (Gromadzka 1989 adopted this point of view, and, in a surprisingly absolute way, Engelmoor & Roselaar 1998). Furthermore: all stages of moult are met with in the Baltic area, not only scores 0 - 30 (40 if "adult buff" are included) as suggested by the material in Gromadzka 1986 and Holmgren et al. 1993b. (In the weight study by Meissner 1998 the full extent of moult in the Baltic area is clear; 370 adults at Rewa + Jastarnia, more than 4 % of all adults caught, had fresh remiges in eight years of study, and some 10 % flew with almost complete wing). Fig. 11 should be seen in this context, demonstrating that conditions in Poland and S. Sweden are much the same; 6.5 % of the material from Falsterbo peninsula migrated with fresh remiges, and again some 10 % with almost fresh wing (cf. upper sections of Figs. 6 and 7). Furthermore the passage of these birds extends over two months, with two obvious peaks, it must include the whole of the eastern alpina area as well as what marginal passage there is of "centralis" in late September and October. Alpina birds (without any "centralis" taint) are not only capable of starting on breeding-grounds, they are capable of performing a complete exchange of feathers there. With a provocative wording: Every single Dunlin occurring at Baltic sites with fresh or "growing" remiges has grown these on breeding-grounds, birds with fresh remiges have grown 98 - 99 % there; the option "feather growth on migration" is of such little importance to the overall exchange of remiges, that it may be ignored.
When it comes to assessing moult start in Arctic areas the date of visit to a breeding site is of the utmost importance; a period of a few days must be targeted, after that a bird with shed remiges has abandoned its breeding territory. Ferns 1978 found no moulting Dunlin in NE Greenland, where he had expected to find them - but as far as I know he was wrong there, probably influenced by North American findings (Holmes 1966; Alaskan adults don't migrate in order to moult, they stay together with juveniles on breeding-grounds and moult there, the same way some Siberian and European Dunlin obviously do). At any rate he concludes, that he may have caught them too close to hatching date, thereby correctly identifying the problem. In this paper I report of a Lappland bird moulting before the hatching of her eggs, another five can be credited to Rösner 1997, and Gromadzka 1989 had two moulting birds out of eleven investigated museum specimens from European Russia. But as a matter of fact I am convinced that each single one of the 10 - 40 % 3c+ and 40 - 95 % (dependent on time interval) 2c C. a. alpina caught in the Öresund area with moult score in excess of zero have shed and to a grest extent grown their feathers on breeding-grounds. And values from the Öresund area (Fig. 3) are fully compatible with values from the Gdansk area (for comparison see Fig. 2 in Gromadzka 1986), there is no major change over the 375 kms separating the two areas - c1/10 of an average migration flight between Yamal and the Waddensea. Particularly among early migrants there are many birds that abandoned broods or had them predated, such birds can not be caught and in the past they were no easy targets for collectors, since they were no longer tied to a territory; museum materials won't help us on this point.
In addition there may be differences in this respect between areas; the division of labour between sexes must be an important precondition for early moult among Golden Plovers Pluvialis apricaria in Swedish Lapland - and probably among Dunlin as well. There is seldom more than one bird present in a breeding territory at the 800 - 900 m level, the other bird feeding only a few kilometers away, but at 400 - 500 m altitude, where the food supply is much richer. The bird from Snannaviejgiera mentioned above bred in a bog where the permafrost level lay one decimeter below the mud surface, the water temperature was +4° C, insect prey very scarce in the water. And it was on its own, two were never seen. On the other hand the Rautasjaure delta, 300 meters lower and only a few kilometers away, is one of the richest habitats in this part of Lapland. Where the landscape is flatter and less elevated the possibility of alternating between levels does not occur. In Russia Kola Peninsula, Novaja Zemlja and Taimyr have mountain ranges, but I do not know if Dunlin are breeding at high altitudes here. At any rate, I believe that local, rather "patchy" sources with high food abundance are likely to favour moult start on breeding-grounds throughout the breeding area of the subspecies alpina. And I do not believe, that a bird is tied to some sort of hereditary strategy (Rösner 1997 also gives some emphasis to this standpoint); it may moult on breeding-grounds one year and on moulting-grounds the other, this must have been the case for the (unmoulted) bird ringed on the Falsterbo peninsula in July one year and retrapped at Langenwerder three years later in October; in between it had been retrapped as a staging bird in the Black Sea area. A bird of eastern origin, still it had options. There is an individual degree of freedom when it comes to moult; the focus should not be exclusively on populations.