Abstract Adult Dunlin migrate in numbers in the Sound area, S. Sweden, from late June till mid-August, after that (pentades 47 - 59) there is a low, but constant presence of migrant adults among juveniles. The latest, migrating adults, arriving with complete or almost complete remiges in October have grey medians and belong to the subspecies alpina.
From July onwards adult Dunlin have low weights, indicating small fat deposits; juveniles have increasing weights from late September till early November. Juvenile mean bill and wing lengths increase from August to early September, in September values decrease significantly, finally they increase and reach a maximum in late October and early November. A cohort of juveniles puts on fat from late October and leaves the area, no later than 15 November. This cohort is thought to belong to the subspecies centralis, its goal area could be the E. Mediterranean, but there are no recoveries, neither from winter nor from later migration seasons. From mid-November, when the wintering population is established, all mean values (bill, wing, weight) decrease. The bill length distributions of males and females may be skewed, this is suggested by a material of morphologically sexed adults from July/August. If skewness is a general feature, it makes the separation method suggested by Griffiths 1968, 1970 inapplicable. The extension of white on inner primaries, ranked by a scale 0 - 3, increases throughout the autumn and abruptly decreases by mid-November, this is thought to reflect a continuous flow of populations and finally a shift from eastern to western provenience. In general, there is a continuous flow of juveniles bound for distant goals in the area throughout the autumn, only those groups of juveniles which are close to some sort of in-between targets (staging area, wintering grounds) lag behind. Birds with alleged sakhalina biometry occur in the Baltic area throughout the year. Where alpina birds predominate, the large-sized specimens must be natural extremes occurring in large samples, in October/November they are mostly centralis birds. The "sakhalina problem" of the past may be a pseudoproblem, possibly caused by long Poisson-like "tails" and an all-pervading skewness of the biometrical distributions (bill length, possibly wing length and body size to some extent as well) of male and female Dunlin.
A fundamental temporal division between age classes of Dunlin Calidris alpina migrating through the Baltic area was demonstrated more than a century ago by Kolthoff 1896, in recent years Brenning 1987 and Gromadzka 1986, 1989 have added new shades to the original insight. The Baltic is not completely abandoned in favour of milder wintering grounds, however; Dunlin are known to have wintered successfully in the straits connecting the Baltic with Kattegat and in the German part of the Baltic for at least three decades now. In addition there are staging birds in Germany (Waddensea and Baltic), Denmark (Waddensea, Kattegat, Belt Sea) and S. Sweden from March onwards. To "Wintering and spring staging Dunlin...". These different events and movements each have their own, specific time-tables, together resulting in a highly intertwined pattern, that at times makes Dunlin migration in the Sound area seem almost chaotic. There is no month without Dunlin migratory movements in S. Scania, and the populations involved fan out over 180º: from Orkney and the Hebrides to Egypt and maybe even the Persian Gulf (Brenning 1989, Gromadzka 1989, Rösner 1997, Serra et al. 1998).
At Ottenby Bird Observatory in the South Baltic, adults make up on average 85 % of the annual catch (Pettersson 1994). These birds are caught mainly at the cape proper between July and mid-August, while birds feeding in areas north of the cape (therefore not caught) later in the season were thought to be almost exclusively juveniles; from September Dunlin migration in the daytime is quite unimportant at this site (Edelstam 1972). According to Martin-Löf 1958 c. 1 % of all birds caught between 26.8 and 21.9.57 were adults (the suspicion of faulty age identification lies close at hand, the correct figure is 7 - 8 % by this time of year), and Edelstam 1972 more generally remarks that "minor waves (of adults) have been noted in September and, very occasionally, in October". Beyond these brief mentions there seems to be no published material on late adults from this important Baltic site. In the Bay of Gdansk and the Bay of Wismar (Isle of Langenwerder) 10 - 15 % of the annual catch of adults are caught in September (Brenning 1987, Gromadzka 1986), however, and adult migration goes on at least till the second pentade of October at Langenwerder. The Swedish check-list explicitely states: A late wave of migrants, in some years occurring in late September and October, is thought to consist of juveniles from western Siberia (SOF. 1978, 1990).
The alleged occurrence in W Europe of Dunlin belonging to the East Siberian subspecies sakhalina presents another problem of long standing (Nørrevang 1955, Holgersen 1963, Meltofte 1993). Dunlin from the sakhalina area moult on breeding grounds and migrate south and south-east to East Asia (Greenwood 1983). Brenning 1987 discusses the possible occurrence of East Siberian Dunlin in the Baltic area at some length, assuming that birds with wing lengths > 125 mm and bills > 36.5 mm should be assigned to sakhalina, and concludes (in translation): Granted that this subspecies is really involved, it is by no means a rare straggler to the Baltic area, but a regular visitor occurring in small numbers. With the differing moult/migration schedule in mind, Stiefel & Scheufler 1989 put a question mark at the occurrence of sakhalina in Western Europe, but they too admit, that morphological evidence (in their view: wing lengths over 125 mm, bills longer than 36 mm) is difficult to explain without considering East Siberian populations. These limits: wing in excess of 125 mm, bill longer than 36 - 36.5 mm, have been traditional criteria for sakhalina; they occur in e.g. Greenwood 1979, 1986. Genetical studies in recent years (e.g. Wenink 1994) have revealed no evidence for Beringian haplotype in W. Europe, and there is a problem to this approach: it overrides the biometrical and morphological definition of "subspecies" without so far daring to replace it with a new, genetically based definition. (And the possible gain is dubious). Still, the problem of overlap between subspecies and haplotype and the relevance of morphology/biometry for determination of well-defined populations remains, therefore the morphology and biometry of long-billed and long-winged birds - the once-presumed sakhalina - in the Sound area will be discussed at some length here.
2.1.1. General phenology. Roos 1984 lists 614 occasions of adult Dunlin ringed or retrapped at Skanör; 47 (= 7.5 %) of these up till and including 15.7. Extreme dates are 3 July and 20 August, median 29 July. This period could be termed the summer moult migration in the Baltic; the main catching effort at Ottenby, Öland also coincides with this passage. As a rule the first small flocks of migrating Dunlin in the Sound area are seen by midsummer.
Between 1991 and 2002 1,148 adults (605 3c+, 333 2c, 56 2c+ before 1 September, 154 2c+ from 1 September onwards) and 3,207 1c birds were caught on the Falsterbo peninsula between 16 July and 20 November. There were no catches before mid-July because of ongoing breeding of protected species at catching sites, apart from this restriction the catching effort has been much the same throughout the period. Out of the 1148 adults, 154, i. e. 13.4 % of the catch from the migration period were taken after September 1st. This estimate is on the low side, due to problems with water-levels there was practically no late catch in 2001 and 2002. If the swift female migration in late June and early July is considered as well, my estimate is that every tenth adult Dunlin, migrating by way of the Sound area, passes between September 1st and October 15th - 25th, and this estimate should hold for at least Ottenby and Langenwerder as well. The presence of this contingent is obvious from Brenning 1987, it remains unconsidered by Holmgren et al. 1993b and is ignored by Dierschke 1996. The distribution of 1c and 2c+ birds from July to November is shown in Fig. 1.
2.1.2. Adults: bill length distributions, differences between sexes. Between 13 July and 31 August 1995 - 2002 biometrical data (wing length, bill from feathers and weight) were collected from 804 adult (284 2c, 42 2c+, 478 3c+) Dunlin. From 1999 onwards 182 birds of these were sexed as males, 144 as females according to plumage characters given by Ferns & Green 1979, Stiefel & Scheufler 1989 - this means that measurements were not considered in sexing. This kind of approach is difficult with the extremely worn feathers by this time of year - but half of all birds investigated were left unsexed; sexing was attempted only in distinct cases. Mean values for different categories are given in Table I, the overall bill-length distribution and the distributions of morphologically sexed birds in Figs. 2 and 3. Note, that the coefficient of variation in bills (6 - 7) is three times the CV in wings (appr. 2), there must be a premium for variation in bill-lengths in Dunlin. The peaks in Fig. 2 are artefacts; there is a tendency to note analogue measurements "31.9" as "32.0", the only way to avoid this is to use digital callipers. The male distribution of Fig. 3 has skewness 0.60 (strongly skewed to the right), kurtosis 0.02 (normal), the female distribution skewness -0.25 (skewed to the left), kurtosis -0.31 (which means platykurtic: fewer items at center and tail than the normal curve). The surplus of males by this time of year of course has influenced the character of the overall distribution of (Fig 2); with sex ratio 50:50 it would have been clearly bimodal, but the tendency for bimodality is still there. This material suggests, that treatment of bill length materials according to the algorithm suggested by Griffiths 1968, 1970, and used again by Pienkowski & Dick 1975, OAG Münster 1976 and Zajac 1980 may be based on false assumptions of normality and should not be pursued any further. All bill length materials presented without separation in this paper will be treated anew together with other materials in "Bill length distributions in Dunlin".
|Category||Mean wing length|
(mm) ± 1 s.e.; 1 s.d.
|Mean bill length|
(mm) ± 1 s.e.; 1 s.d.
|Mean weight (g)|
± 1 s.e.; 1 s.d.
|All (n=804)||118.3 ± 0.1; 3.3||32.5 ± 0.1; 2.5||45.4 ± 0.2; 4.5|
|2c (n=284)||117.5 ± 0.2; 3.1||32.6 ± 0.1; 2.5||45.6 ± 0.3; 4.7|
|3c+ (n=478)||118.7 ± 0.1; 3.2||32.4 ± 0.1; 2.5||45.2 ± 0.2; 4.4|
|Males (n=182)||117.6 ± 0.2; 2.7||31.9 ± 0.2; 2.0||44.5 ± 0.3; 4.6|
|Females (n=144)||120.3 ± 0.2; 2.9||34.5 ± 0.2; 2.3||47.6 ± 0.4; 4.3|
2.1.3. Adults: regression WEIGHT on BILL. Summer migrants in the Sound area carry surprisingly little fat and many arrive in the early morning hours, quite likely after a night's extended flight, with (almost fat-free) weights 37 - 39 g. Whatever fat load they may have had from the outset, they have now emptied their reserves almost entirely. Note the repudiation by Mascher & Marcström 1976 of fat-free weights calculated by Mascher 1966 and Fuchs 1973; at least the value of Fuchs (37 g) holds good! Dierschke 1996 considers Helgoland juveniles below 35 g as lacking fat reserves. (Fat-free weights from the Waddensea in May are probably higher, due to muscle growth, and females must have larger fat-free body mass than males; c40 g with bill 38 mm against c35 g with bill 28 mm, cf. Fig. 4). The regression WEIGHT on BILL in July/August adults is shown in Fig. 4. This is the standard regression of W Palearctic Dunlin approaching the end of their migration journey, it is put here as a basic reference to be compared with the regression WEIGHT on BILL of juveniles in October/November (section 2.2.3 of this paper).
2.1.4. Adults: biometry and phenology of birds with "adult buff" coverts. There are measurements from 47 migrating adults with "adult buff" median coverts (Gromadzka 1986, 1989) between 1995 and 2002; mean values for standard measurements are given in Table II:
|Category||Mean wing length|
(mm) ± 1 s.e.; 1 s.d.
|Mean bill length|
(mm) ± 1 s.e.; 1 s.d.
|Mean weight (g)|
± 1 s.e.; 1 s.d.
|All (n=47)||117.0 ± 0.4; 2.6||31.9 ± 0.4; 2.5||45.3 ± 0.7; 4.4|
The time distribution of 53 adult Dunlin with "adult buff" coverts is shown in Fig. 5, the median date is 10 August. Some 5 % of all adults caught on the Falsterbo peninsula in July and August are distinguished by this character, and there are single birds throughout September as well.
2.2.1. Juveniles: do bills grow in autumn? OAG Münster 1976 lists seven cases outside the breeding area, where juvenile bills are said to have been 3 - 4 mm shorter than adult bills, but this list is based on a misreading of at least Pienkowski and Dick 1975. Soikkeli 1966 states that bill and wings are full-grown when juvenile Dunlins are appr. 40 days old, and according to Cramp et al. 1983 juvenile bills are equal to adult bills c. 2 months after fledging.
Mean values of bill lengths per half-month in 1,789 1c birds, measured between 1 August and 31 October, are shown in Fig. 6. The value of juveniles caught in the first half of August is significantly lower than the remaining values from autumn, provenience (Scandinavian alpina, even a few schinzii) and continued growth may both lie behind, a juvenile Dunlin is little more than one month old in early August, and the possibility that there is still some minute growth of the bill going on, can not be ruled out. From late August the mean value of all samples has stabilized between 32.5 and 33.0 mm, and there is no positive regression.
2.2.2. Juveniles versus adults: general autumn biometry. From a wintering population in the Camargue Fuchs (1973) reported no change in wing-lengths and little difference between ages from October onwards. On the other hand minute weight differences of 1 - 2 g between age classes were reported by Mascher 1966, Pienkowski et al. 1979, Johnson 1985 in Europe, and 2 - 2.5 g in Africa (Pienkowski & Dick 1975). In Figs. 7, 8 and 9 adult and first-year measurements of maximum chord wing, bill from feathers and weight are brought together for comparison. The intervals are half-months from August till November, next a full month (December) and finally two months (January - February) pooled.
Adult wings are significantly shorter than juvenile wings in the second half of September and in October (S II: F = 1.12, z = 2.99; O I: F = 1.35, z = 3.08; O II F = 1.32, z = 4.06; p < 0.01 in all three cases), the difference must be caused by a surplus of males among migrating adults by this time of year. Some adults have primaries 8, 9 and 10 still growing, and a few adult measurements have been excluded for this reason (in spite of this, there may still be individuals with a few mm wing growth remaining, unnoticed by the ringer). All 1c / 2c+ bill lengths except those from the first half of August lie within 1 s. e. of each other; if necessary, adult and juvenile bill lengths can be pooled in autumn and winter.
The obvious fattening from late October onwards is a complication, here all possible differences between age-classes and populations are overthrown. Adult mean weights increase till November, still they remain c5 g lower than adult weights from the Waddensea between July and September (Boere & Smit 1981, Dierschke 1996, Johnson 1985, Rösner 1997). Juvenile weights are on average higher than adult weights till the end of October and lower from November onwards. When the wintering population is established mean weights decrease, probably reflecting an exchange in the "basic material": from a numerically strong cohort of heavy birds with fat-free weights appr. 45 g in October and early November to light birds with fat-free weights 40 g or less after November 15th or 20th (for this shift also see Figs. 14 and 16). Adults are on average slightly heavier than 1c / 2c birds throughout the winter, but all values lie within 1 s.e.. The dip from S I to O I (ages pooled, wing SI - SII: F = 1.27, z = 3.29, bill: F = 1.02, z = 0.82; wing SII - OI: F = 1.01, z = 3.10, bill: F = 1.05, z = 3.89; wing OI - OII: F = 1.04, z = 3.55, bill: F = 1.00, z = 1.94. p < 0.01 in wings and in S II - O I bills) affects wing lengths and bill lengths alike, a fact that exonerates measuring technique from suspicion; the most likely reason is a quick turnover of a sequence of populations with minute biometrical differences throughout the autumn. This is confirmed by numerous observations of nocturnal departure and dawn arrival. Still, the fact that measurements continue to decline after new-year demands an explanation, this matter will be brought up in "Bill-length distributions in Dunlin".
2.2.3. Juveniles: fattening birds from October. In the second half of September, the regression coefficient (weight on bill) of juvenile Dunlin from the Falsterbo peninsula is even lower than that of summer adults shown in fig. 4 (0.60, p < 0.001; n = 435), but from October onwards it starts to rise: Oct I: 0.92 (p < 0.001; n = 474), Oct II: 1.19 (p < 0.001; n = 409), Nov I: 1.27 (p < 0.001; n = 202). This sequence illustrates a process of fattening, which is followed by departure for a distant goal-area; in the second half of November the regression coefficient is again back to normal: 0.76 (p < 0.01, n = 115). The sudden correction back to normality comes before any exterior strain (frost, strong winds, ice, disappearance of prey) has had a chance to affect the birds, and as a matter of fact the fat birds are gone (and never met with again: there are no recoveries indisputably related to this group). It should be emphasized here: all birds do not put on fat at the same time, and many birds are not putting on fat at all (since they are not leaving the area). Furthermore: among the fattening birds, some are only going to W. Europe and they do not need large deposits for that journey. But a fairly large contingent adds on average 15 g of fat (relationship in Piersma & van Brederode 1990) in late October and early November; with at most 60 km/h air speed (Lane & Jessop 1985, Zwarts et al. 1990) this implies a non-stop migration flight of at least 1700 km (McNeil & Cadieux 1972, this formula uses actual fat load, also see Davidson 1984a). My guess is: the heavy birds are due for the E. Mediterranean, the Red Sea or the Persian Gulf as intermediate goals before turning up as staging birds in the Black Sea in spring. If they fly with the northwesterly tailwinds so common in many late autumns, the task will be an easy one and ground speeds exceeding 100 km/h quite possible. The weights of these Dunlin stand out in Fig. 10 when compared with summer adult weights in Fig. 4; the task is to sever as much of them as possible for statistical investigation.
When fat loads increase in a Dunlin cohort, linear regression of weight on wing or weight on bill gradually gives a poorer approximation of actual distributions; the process is slightly non-linear, possibly because there is additional, non-linear muscle growth as well (Lindström & Piersma 1993). This can be seen in Fig. 10, values to the right tend to bend up and create a boat-shaped distribution. In the optimal case the severing should be done with such a curve as point of departure, but here I have confined myself to linear severing, adapted by means of trial and error. Starting value was the overall regression coefficient 1.27 from Nov I, but this is not sufficient, the (non-linear) rise of the upper part of the diagram is steeper. With slope 1.4 the peripheral, non-"clustered" part of the diagram can be severed in a satisfactory way. The resulting material is shown in Fig. 11, it contains 155 birds, 80 % between 21 October and 18 November, median 5 November, mean weight 65.8 ± 0.6 g, s. d. 7.0 g. Note the tendency for bimodality, and the obvious skewness of at least a discernible female component; with sample sizes 100 - 200 Dunlin materials never will be more regular than this one. The material of Fig. 11 will be dealt with further in "Bill-length distributions in Dunlin".
3. The "sakhalina problem"
3.1. The "sakhalina problem" in the south Baltic: biometry. The bill length limit for safe determination of "sakhalina" birds suggested by Brenning 1987 and Stiefel & Scheufler 1989 seems to be on the low side; 10 % of all Dunlin measured on the Falsterbo peninsula after 1 September have bill-lengths > 36.5 mm (own material), at Bottsand/Bay of Kiel even 10 % of all birds measured between 16.7 and 3.9.95. (H. Behmann in litt.). In order to get a more even match, limit values 126.0 (maximum chord; 3.8 % of the material) and 37.0 mm (4.5 % of the material) will be used here. Data for birds equaling or exceeding these limits have been brought together in Figs. 12 and 13. The bill measurements come from 2 winter / spring birds and 118 autumn birds, median of the latter 4 October, the wing measurements from 2 winter / spring birds and 100 autumn birds, median of the latter 13 October. (Note that all birds in Figs. 12 and 13 exceed the upper limit for sakhalina wing lengths attributed to Greenwood 1979 in Cramp & Simmons 1983, there is an irony in this extreme, unchallenged abundance. Part of the discrepancy must be caused by shrinkage of museum specimens (cf. Engelmoor et al 1983), however, and many older measurements were taken with natural wing curve.
From these diagrams it is evident, that there is practically no positive correlation between bill and wing when long bills or long wings are treated in isolation. Bill length in particular is connected with sex more than provenience in W Palearctic Dunlin populations; any long-billed birds of eastern origin - if they had occurred, whether males or females - would be swamped by the natural variation of alpina and - in particular - centralis populations. It should be added, that long-billed birds may be caught at all times of the year. In contrast, long-winged birds tend to occur late in the season; if the limit is moved down to 125.5 mm (n = 135) the median falls on 17 October, two weeks later than the median for long-billed birds. The increase of body mass, noted in late October and early November, is likely to contain a subspecific component, reflected in longer wing but not in longer bill.
Between 1991 and 1999 249 Dunlin weighing 60 g or more have been caught on the Falsterbo peninsula; 32 in winter, 2 in spring, 26 in Aug - Sept and 189 in Oct - Nov. Table III gives statistics for the heavy October-November birds, the division is half-months:
span; mean ± 1 s.e.
from 126 mm
span; mean ± 1 s.e.
from 37.0 mm
span; mean ± 1 s.e.
|1 - 15 Oct||30||117-126; 121.1±0.5||2||30.0-38.0; 34.3±0.4||2||60-71; 63.6±0.6|
|16 - 31 Oct||61||115-127.5; 122.4±0.4||6||28.6-38.0; 34.3±0.3||8||60-78; 63.6±0.5|
|1 - 15 Nov||70||111.5-130; 121.0±0.5||5||28.0-39.5; 33.5±0.3||4||60-83.5; 69.3±0.7|
|16 - 30 Nov||28||110-123; 119.1±0.6||0||28.2-39.7; 33.9±0.5||2||60-76; 67.3±0.9|
Some 10 % of the heavy late October and early November birds have exceeded the provisional sakhalina limits mentioned earlier. Still, there is only a very weak correlation between body mass and wing or bill length in these Dunlin, and one reason for this is obvious: the birds do not normally weigh 60 - 70 g, but rather 45 g (fat-free weight, i.e. they are more robust than July alpina migrants), and birds will be caught at all stages of the fattening process: lean, half-fat, fat, and so there will be only poor correlations between weight and size parameters. At any rate the proposed limits will discern only extreme females and no males in a true sakhalina cohort from centralis or alpina birds; according to Greenwood 1979 even wing length only marginally separates sakhalina from other races.
3.2. The "sakhalina problem" in the south Baltic: morphology. In search of supplementary criteria it is natural to turn to the extent of white on outer webs of inner primaries; white reaching rachis on the inner primaries was noted in Hungary by Horvath & Keve 1956 and discussed as a possible sakhalina criterion by H. F. Witherby (Witherby et al. 1958). Engelmoor and Roselaar 1998 set out to use this variable in a larger context, but were led astray by the irregular Anglo-Saxon notation (P4 termed "P7") of Greenwood 1986. On the Falsterbo peninsula the extent of white on inner primaries has been ranked according to a 4-grade scoring scale, developed by W. Meissner (in litt.) and used by the workers at Reda mouth, Poland (also see pictures under "The Meissner scale" in the menu).
|Extension of white||Score|
|distance between white and rachis|
longer than 1/4 of width of outer vane
|white colour doesn't touch rachis:|
but a distance between white and rachis is
shorter than 1/4 of width of outer vane
|white colour touches rachis:|
width of line of contact very short,
shorter than width of rachis
|white colour touches rachis:|
width of line of contact longer
than width of rachis
|age: score||n||wing mean ± 1 s.e, s.d||bill mean ± 1 s.e.; s.d.||"adult buff" medians|
n; % of total
|1c: 0||1117||119.9 ± 0.1; 3.3||32.6 ± 0.1; 2.7||-|
|1c: 1||265||120.5 ± 0.2; 3.1||32.8 ± 0.2; 2.6||-|
|1c: 2||158||121.1 ± 0.3; 3.4||33.8 ± 0.2; 2.6||-|
|1c: 3||133||120.9 ± 0.3; 3.0||33.4 ± 0.2; 2.7||-|
|2c+: 0||434||118.4 ± 0.2; 3.5||32.4 ± 0.1; 2.6||6; 1.4|
|2c+: 1||113||118.2 ± 0.3; 3.3||32.2 ± 0.2; 2.5||5; 4.4|
|2c+: 2||95||118.9 ± 0.3; 3.1||32.5 ± 0.3; 2.5||6; 6.3|
|2c+: 3||111||119.0 ± 0.3; 3.0||33.2 ± 0.2; 2.4||11; 9.9|
In contrast the regular and significant decline of rating 0 (or increase of ratings 1 - 3) in juveniles is striking. Here the corresponding events described under 2.2. should be kept in mind: staging birds with large fat deposits depart no later than 15 November and adults belonging to the wintering population join juveniles by the same time. The final departure situation in November, with a mix of staging and wintering birds, is characterized by a 50+ % ratio of ranking 0, ten days later there will be at least 70 % of this ranking in the "pure" wintering population. This difference will be detected with 90 % probability at the 5 %-level with a sample sixe of at least 110 (Sokal & Rohlf 1998 (not active)). Colour-ringed adults belonging to the wintering population have been observed as migrants on 3 September (one, Westerhever, Germany) and 12 September (two; Skanör), this decade also has an average 70+ % share of rating 0. There are seven intervals between decade 2 and decade 9; if we assume that they represent a continuous migration of juveniles from 20°E to 70 or 80° E, with the point of gravity starting at 20° E and ending with the staging birds as the easternmost elements, the wintering population should belong between appr. 2/7 and 3/7 of the total span; this rough estimate locates the wintering population somewhere around 40° E, near the mouth of the White Sea. On the other hand: if there is an inversion with birds of extreme eastern origin migrating already in September (wing and bill length maxima of Figs. 7 and 8), the estimate is upset. But why then the continuous development of the ranking score? I believe, there is more consistency in colour rankings than in biometry, they are subject to less environmental induction (cf. the discussion of Wenink 1994), but I still do not know how to prove my point.
Summing up: There must be birds with sakhalina biometry in alpina or centralis populations. Published biometry is faulty, the assumption of normality is wrong, at least in the case of bill-length distributions: the long-billed and most probably the long-winged birds as well must be part of the natural variation of "non-sakhalina" populations. White reaching rachis on inner primaries occurs among migrants from July to October; 10 % of adults with ranking 3 have had "adult buff" coverts (a similar observation lay behind Meissner's creation of this scale, the same observation recurs in Engelmoor and Roselaar 1998), but the Sound material still is too small on this point to allow significant conclusions. Long wings, long bills, white on inner primaries in alpina or centralis birds make true sakhalina impossible to determine in the hand, if any occur in Western Europe. In my opinion records of "sakhalina" birds, based on measurements and / or morphology, should not be admitted to European check-lists. The easternmost limit of the recruitment area of Dunlin so far known to occur in W. Europe is marked by the two recoveries of birds on Taimyr Peninsula at 79° 50' E listed by Tomkovich, Lappo & Syroechkovski 2000. And with these recoveries the centralis area is barely entered.
4.1. Adult phenology. In the past the catching effort has been broken off prematurely at practically all Baltic dunlin-catching stations. Meissner (in litt.) sums up on behalf of Poland: "We stop catching on the turn of September and October, when Dunlins are still present. That is why I have no morphological data about late migrants (October-November). (...) For sure adults are still present in small numbers in October or November flocks. Some of them have many feathers from breeding plumage." From Brenning 1987 it is clear, that there was notable, continuous catch of adults at Langenwerder, Germany at least till 3 October, and still a few in the four pentades between October 3rd and 23rd (1976-1985). In the past Langenwerder has the best Baltic material from the end of the migration period. The summer catching effort at Ottenby is made up of 85 % adults and fades away by mid-August (Pettersson 1994), and there has always been uncertainty surrounding the age structure of late migration (e.g. Edelstam 1972) - probably made worse by incorrect age identification - where Ottenby material has been involved (i.e. in most of the Swedish writing on the subject).
In the Sound area a wave (or two separate waves, one in July, one in August, cf Fig. 17) of Siberian birds with "adult buff" coverts thrusts itself into the regular alpina migration in summer, the overall "adult buff" median falling on August 10th. These birds come from, or east of, Gydan longitude (Gromadzka 1989). From September onwards there have been few adults with "adult buff" coverts so far, but the material is small, the early September catch has never been really successful. Two "adult buffs" on 13.9 and 16.9.02 call for caution in this question, however, and the ranking dip of Fig. 17 does the same. Furthermore, the catching effort in the Sound area has included the last wave of migrating adults, arriving with juveniles in October. These birds arrive with fresh remiges or remiges in the last stage of growth, and practically all of them have grey medians. This is a wave of alpina birds, by all likelihood originating from somewhere around the Yamal longitude, where most birds still have grey medians (Gromadzka 1989). There is one October recovery supporting this view. The wintering adults returning from 15 November are of even more distinct alpina type and on average smaller than the October migrants. The essence of this is: from mid-August, throughout September and well into October every juvenile flock arriving at the Sound area contains adults, and adults are present in every departing flock (whether for the Mediterranean or the Waddensea) throughout the same period; this observation effectively refutes Dierschke's (Dierschke 1996) hypothesis of "differential migration". And I do not believe, that adults are ballast in these flocks, they know the way.
4.2. Juveniles: phenology, biometry. The key figures to juvenile migration are Figs. 14 and 16. They should be viewed in connection with Figs. 7, 8, 9 and 10, showing developments in juvenile biometry. Wing-lengths and bill-lengths are linked, they increase and decrease simultaneously throughout the autumn, with a notable dip in the second half of September and the first half of October. Their standard deviations are much the same, which means that coefficients of variation for bill-lengths are 3 - 4 times higher than CV:s for wing-lengths. A similar pattern of increase and decrease was recorded by Fuchs 1973 on the Mediterranean flyway, and Brenning's 1987 quickly disappearing August juveniles should be kept in mind. There may be some degree of inversion here; juveniles from distant, easterly populations arriving (and resting) before more westerly populations, so the alleged early passage of Siberian juveniles in the Baltic (Goede, Nieboer & Zegers 1990) is not entirely without evidence. Still, the biometrical development in late October and early November shows that there is a continuous filling up of easterly, long-winged populations of juveniles in late autumn, and this second wave of long-winged juveniles has quite another average ranking of inner primaries, here we have undisputed Siberians. I do not exclude the possibility, that late autumn juveniles in the Sound area congregate from all directions, even from the Waddensea and Kattegat, and maybe visit the Sound area twice in a single autumn. This in turn means, that the extreme October weights could have been obtained in the Waddensea. Still, the Sound weight peak in November by far exceeds the simultaneous mean weights at e.g the Wash: some 53 g in juveniles and 55 g in adults (Pienkowski et al. 1979), it is not part of the same pattern.
Again, one single point will be made: the migration of juvenile Dunlin in the Baltic area is coherent and consistent, the same pattern repeating itself in consecutive years, although the relative weight of different groups may alter. Early juveniles (from early or mid-August) are very average alpina with ranking "0" on inner primaries, but there will always be a handful of deviating juveniles accompanying adults with "adult buff" medians by this time as well. An easterly population, slightly more long-winged than "standard" alpina, zooms in on its staging area, rather early, with a bill-length and wing-length maximum by mid-September, but with continuous additions; when they have settled a process of fattening begins. (Or they accumulate fat elsewhere and congregate gradually at Skanör). Their biometry brings the biometrical differences noted by Martin-Löf 1958 at Ottenby in September to mind, they are of centralis type, barely accompanied by adults - this is the most independent, but also the most well-developed cohort of juveniles. By the same time, or even a little later, the first juveniles belonging to the wintering population converge on the same area (Fig. 16, also see Fig. 2 of "Wintering and spring staging..."). Wintering juveniles have no program for fattening or departure, no distant goal, and when the staging birds have departed, mean wing-lengths, bill-lengths and weights decrease abruptly. Again there is a target for my conclusion: Rösner 1990, 1997. Given the extent of preparation, adjustment, readiness on part of the juvenile birds in the Sound area, I think Rösner's hypothesis of "trial and error" progress is shattered, instead we should look for similar preparations in every single cohort of juveniles in W. Europe by this time of year.
4.3. The "sakhalina problem", general provenience of juveniles. Throughout this period, from August till November, there will be long-winged and long-billed birds - Dunlin with alleged "sakhalina" biometry (Figs. 12 and 13). The early ones have no characteristic morphology, no extra tendency for "adult buff" coverts or white on inner primaries, the extreme late ones may have, but their particular context is a staging presumed centralis population and they belong there. When these birds have left there will again be long-billed and long-winged birds with grey medians and mainly primary ranking "0", particularly at Barsebäck, where females may be in majority. My assessment is: long-winged and long-billed birds are part of the natural variation when sample-sizes exceed, say, 250, they are extremes from all populations regularly visiting the Baltic area - and mainly extremes on the female side, nota bene. As a rule they are juveniles, bills longer than 38 mm are rare in adults (cf. Figs. 3, 4), but so far their samples are smaller than samples of juveniles. Extremes on the adult side occur in the bill-length material from the Waddensea in spring, however (Rösner 1997), where sample sizes exceed 1,000.
The "Meissner ranking scale" (Meissner in litt., Horvath & Keve 1956, Witherby 1958, Engelmoor & Roselaar 1998) offers a new tool for morphological characterization, but its gradation may not be quite to the point; rankings 1 - 3 give coherent values only when pooled (Fig. 16), and again there are too many "sakhalina" suspects (10 %) according to old thumb-rules. I suggest that "0" be restricted to dark on outer vane > 50 % of width (the normal schinzii pattern on inner primaries), "1" 10 - 50 % of width, "2" 0 - 10 % of width and < 10 mm long on no more than two primaries, "3" > 10 mm long on two primaries or white reaching rachis on three or four primaries (often in connection with "adult buff" medians and "droplet" primary coverts); a scale of this kind might offer a useful "fingerprint" of Dunlin populations. It is reasonable to assume that the mean latitude of origin for migrating juveniles shifts eastwards as the season rolls on, and that birds from the western parts of the breeding distribution are the first to visit the Sound area (first decade of Fig. 16 possibly excepted; cf. remarks on early juveniles in Brenning 1989). The increasing mean values of wing- and bill-lengths from Aug I - Sep I (Figs. 7 and 8) speak in favour of this assumption. I still do not understand what causes the dip in Sep II and Oct I - one possibility is that the biometrical cline from Scandinavia to Gydan peninsula is not linear, but has some sort of dip in between - a "morphological divergence of evolutionary closely related populations" in the sense of Wenink 1994 - another that extreme northern populations (e.g. Novaja Zemlja) make their entry here.
By the end of October the eastern element is very prominent, many birds are long-winged and heavy, there is a high ratio of inner primary ranking 1 - 3, and median coverts (worn juvenile as well as new 1W plumage) look slightly aberrant, with a tinge of "adult buff". At the same time short-billed and short-winged juveniles begin to mix with the flocks; I suggest that their presence in numbers causes the divergence of rankings "0" and pooled "1" - "3" in decade 10 (3 - 12 Nov) in Fig. 14, and the significant jump in "0" ratio from decade 9 to decade 10 in Fig. 16. Assuming an overall linear decrease of ranking "0" from W to E, the staging birds causing the 50+ % ranking "0" in decade 9 should originate from the easternmost so far known breeding areas of W European migrants: Gydan peninsula or Taimyr (Tomkovich et al. 2000), while the wintering birds with 70+ % "0" ranking could come from somewhere around the mouth of the White Sea (30 - 40 % of the distance between 20° E and 70 or 80° E). Rankings of breeding birds from different areas are needed to justify the whole approach - in this case rankings on museum skins would be quite sufficient.