Risk-prone or risk-averse? Dunlin Calidris alpina migrating with and without moult-gaps in the Baltic area

by Christer Persson

Christer Persson, Ljungsätersvägen 43, S-236 41
Höllviken, Sweden. E-mail cp.hollviken'at'swipnet.se

Dunlins Calidris alpina are assumed to migrate in short steps in the Baltic area in late summer and autumn, but little is known about the length of bouts within Russia, before the birds reach the Baltic. For a long time they were believed to "moult" on migration, but Persson 2003 produced evidence strongly indicating that effective feather growth is halted during the transport phase. On the Falsterbo peninsula, S. Sweden, there is a fall of migrating Dunlin from midnight onwards, and there follow a few hours of foraging before the birds depart again, the nocturnal stage including both migration and foraging. It is possible that the time schedule is aligned so that the main arrival of migrants coincides with a nocturnal activity peak of their invertebrate prey.
The body weight regressions of Dunlin are influenced by both moult state (suspended or no moult) and age. Birds with initiated moult showed a significant weight loss as long as there was ongoing migration; this loss was halted in 2c birds by 04h, while at least some 3c+ birds seemed to press on. With weights transformed to 118 mm wing and 00h status, both 2c and 3c+ with old remiges weighed on average 45 - 46 grams [-20 to +30 % deviation], while birds with initiated moult weighed on average 1.5 g more. The extra weight is assumed to compensate for inferior performance/higher costs to birds with worn remiges and moult-gaps at some level: flight, regulation or metabolism.
Gaps amount to the same average ratio from primary 1 to primary 10, given this extent they seem to weigh lightly with flight. If the larger extension and larger mass of outer primaries is considered, distal gaps represent a larger absolute measurement and larger mass than proximal gaps. The regression WEIGHT on GAP SIZE is poor in all categories, the only significant influence from gaps being the GAP POSITION in 3c+ birds, here weights increase as it moves outwards. It is suggested that birds fan proximate feathers across gaps and that weight changes are not connected with aerodynamic performance, but rather prepare for the growth of energy- and nutrient-consuming outer primaries.
The large weight variances within a wave of migrating Dunlins (4 s.d. = 40 % of mean value) in the Baltic area in late summer are considered to be characteristic of a particular mix of risk-averse and risk-prone strategy that avoids predation by moving and foraging in darkness, at the same time taking chances with food availability in a stochastic and extremely patchy environment. This strategy may be valid all the way from the Arctic breeding areas for birds moving on a (hypothetical) southerly route; on such a route much migration and foraging might proceed unnoticed and out of reach of diurnal predators in the nighttime. Birds migrating in short bouts late in the season after performing part or most of their postnuptial moult on breeding-grounds will experience more protective darkness on all routes.

Straight text version for printing: text version.

Contents

1. Introduction
2. Results
2.1. General progress
2.2. Diurnal timetable
2.3. Descriptive statistics, multiple regressions, weight transformations
2.4. Transformed weights versus time
3. Discussion
3.1. Are Dunlin most active in the daytime or the nighttime?
3.2. Implications of large weight variances
3.3. Weight development of birds with initiated moult and birds with wing-gaps
3.4. End of the journey: is the material representative of the whole of the postbreeding migration?

1. Introduction

Weighing of birds in different life situations is the standard method to obtain basic information about changes in their body masses. Weights are unspecific, however, pooling contributions from different sources like water, fat, muscle tissue, food ingested; at best they reveal that some sort of change is taking place. Inference, involving both additional investigation and statistical analysis, must follow in order to make the original, raw measurements useful in some respect.

In most cases inference takes place within some sort of theoretical framework. This could lead to e.g. assumptions about optimising "trade-offs" between basic options, one of them involving mass, the choice resulting in some sort of "fitness" to be redeemed at a crucial point of time in an individual's life history. From the very start the approach of the theory hinted at here was hampered by one particular shortcoming: it lacked an intuitive feeling or a sense for dynamic interaction and dynamics at large, in most cases tending to a static either-or (choice, selection between one of two major options). In addition the gesture that was used to mark biology as a theoretical science (expressed mainly in terms "selected for" or "adapted to") has from the very outset, by its mere nature, been hard to falsify, often a sullen assertion rather than a scientific statement. The scientific dilemma is aggravated by an intricate system of slightly deficient and misleading cross-quotation, where the quantity of Darwinist tautology easily outnumbers what there is of quality in the shadow of Darwin. Here lies the reason why "outsiders", with a background in physics and mathematics but little experience of biology, have tended to repudiate Darwinist inference as sheer stupidity; it violated and violates their basic instinct for dynamic action and interaction. In general I share this scepticism towards Darwinist inference; many actions that are considered coded for and in some sense relying on "instruction" from the genome are, when it comes to it, inspired by individual experience (and to some extent inspired by inexperienced individuals using their senses) and transferred to aggregates by means of dynamic action and interaction. The underlying dynamics as such (its basic message being: change, e.g. "dissolve the flock") in turn has no remembered prehistory, it doesn't depend on coded instruction, and it leaves no imprinted memory - at least there is no unambiguous evidence for that. In the end such repudiation also involves suspicion that data are manipulated or ignored in order to suit the simplifications. Such suspicion shouldn't be taken lightly; much Darwinist "inference" can be upheld only by ignoring or suppressing part of the empirics available, this is the rule rather than the exception. I don't believe in the metaphoric "trade-offs" of an optimising œconomia naturæ (Swaddle et al. 1999 spoil a promising dynamic approach to moult gaps by introducing a "trade-off between height gain and speed gain"); what we see are the dynamic effects of aggregates, where the presence of experienced individuals (experienced and instructed bodies, instructed organs) creates a bias, that will pull or tilt the overall dynamics towards some sort of purpose or desired result. Still, the perspective of optimising economy may have some primitive-didactic merit, much like the old Ptolemaian cosmology, it shouldn't be discarded lightly, before we know what we have instead.
/My general theoretical point of departure is the four volumes "Towards a Theoretical Biology", ed. C. H. Waddington, 1968, 1969, 1970, 1972 (Edinburgh Un. Press), and to some extent the writings of the theoretical physicist Robert M. May and his "school", e.g. "Theoretical Ecology" 1976 - which i read as "Not Yet Theoretical Ecology"; the development of biology towards theoretical and scientific status is much hindered by today's Darwinist hegemony, right now i can discern no positive development at all. [8.10.06; CP]/

When there was some overflow from physics or mathematics to biology the situation tended to improve, here lies some hope for biology. The introduction of a subversive concept like "risk-sensitive foraging" (e.g. Real & Caraco 1986, Caraco & Lima 1987, McNamara & Houston 1992) and different manifestations of "dynamic modelling" or programming (Mangel & Clark 1986, 1988, McNamara & Houston 1986, Houston & McNamara 1988, McNamara et al. 1991, Houston et al. 1988, Bednekoff & Houston 1994) are examples; by such approaches the analysis of typical situations has been greatly diversified and gradually improved - maybe even to the satisfaction of non-committed outsiders. (On the whole there would be fewer problems with all kinds of biological modeling if the warning by Mangel & Clark 1986 was heeded:...any behavioral model which is simple enough to be operational is necessar(-il)y too simple to be biologically realistic). New problems are created when systems theory is confronted with Darwinist concepts, however; the variance of features connected with a risk-prone strategy may be disturbing to a theory based on evolution theory; how is it to handle large variance in a feature considered to be crucial to survival - maybe even large variances in two or three parameters simultaneously? The problem cannot be solved by postponing the redemption of fitness to some Arctic blizzard (this is done), or by using fitness to pay off installments, that leaves us with a concept of fitness reminding of the old Nordic 'huldra' (wood-spirit): she had a hollowed-out back. And how are we to separate risk-prone strategies from "neutral" behaviour in situations where choices don't pay off? The problem is as old as Darwinism; the originator touches it in "The Origin of Species": Variations neither useful nor injurious would not be affected by natural selection, and would be left a fluctuating element. There must be a lot of situations where the possible impact of selective pressures can be ignored, where a wide variety of strategies is actually adopted, and where each single strategy is viable, but still not selected for. In such cases large variances, when they occur, are connected with "neutral" situations with little (not decisive) selective pressure on individuals.

A selection dichotomy of this kind is met with in wader migration to and from Arctic areas; there is an indisputable selective pressure on northbound migrants, there may be some additional pressure on breeders, while the return journey tends to proceed fairly undisputed. O'Reilly & Wingfield 1995 summarised the situation: same distance, different strategies. The reversed summer/autumn migration in some way proceeds "downhill", compared to the risky and demanding "uphill" migration of late spring. Still, there have been many attempts to introduce trade-offs and orthodox Darwinist vocabulary to the post-breeding migration as well, and new ones pop up all the time; migrating waders are said to avoid expected presence of predators (Lank et al. 2003; N. America, avoidance at a lower level is suggested by Dierschke 1998), different fixed adaptations in moult and feeding have been suggested for south-/westbound Dunlin in the Baltic (Holmgren et al. 1993b). Such attempts tend to be a little alienating to informed readers, in general only faulty data and square concepts can uphold the preference of selected and coded "adaptations" on autumn migration to the more or less ad hoc "solutions" of information-processing dynamics. In NW Europe in late summer and autumn any segment of the passage of a (not overly specialised) wader species with sex dimorphism and several distinct age-classes is likely to offer: no clean-cut, unequivocal advantage, no clean-cut "fitness" options and no clean-cut, one-sided adaptations. Dunlin most probably can skip (long rests), hop and jump (fast long-distance movements), on their journey from Arctic breeding-grounds to W European wintering-grounds, all according to particular conditions, this assumption questioning the much too single-track exposition of Piersma 1987. (Piersma in turn is quoted by Skagen et al. 1999. Here I wonder if possible North American inland resting sites for Dunlin were monitored in the nighttime; else the conclusion of the paper, as far as Dunlins are concerned, is worthless). And another example: the migration of Charadriiformes in the Baltic area merges with one of the densest concentrations known of migrating hawks and falcons, and as far as I can see the birds are able to handle this challenge with a wide spectrum of adaptative behaviour while still sharing the migration "territory" with raptors, an observation that should lead to some scrutiny of the suggestions in Lank et al. 2003.

In my opinion the migration of adult Dunlin from the NW Palearctic area by way of the Baltic offers an example of such apparently "non-optimising" progression; it extends in time from midsummer till mid-October, and the variances in weights and moult state of all discernable categories are substantial (Persson 2003a, 2003b, Persson 2004), for fast and retarded moult see e.g. pictures 12, 28 and 46 of the "Dunlin Website" picture collection. In the present paper, biometry and moult records collected from 1211 adult migrating Dunlin on the Falsterbo peninsula in the period 1995 - 2005 are investigated for differences between the appointed categories: 2y and 3y+, males and females, birds with and without moult gaps. Methods and catching circumstances are described in Studies of migrating Dunlin Calidris alpina in the Sound area, S. Sweden: Introduction, three advantages of the material should be pointed out here: sexing, instant handling and detailed recording of moult. Note that all "moult" is considered suspended, for details see Persson 2003a. For an extensive introduction to windflat ecology in the Baltic see Dierschke 1996, Dierschke & Helbig 1999, Dierschke et al. 1999a, Dierschke et al. 1999b. The basic stochastic nature of prey availability in this environment should be emphasized; feeding conditions are never reliable and foraging will follow the pattern concluded by Houston et al. 1993, gains and expenditures are stochastic, some individuals will do better than average, some will lag behind. During windy periods, when wind flats are inundated, migrating sandpipers simply pass by the Öresund area, and probably many other, similar areas as well, without resting; that way much of July may be "lost" to migrant Dunlin and Dunlin ringers. (Strong winds between W and N also tend to link the migration towards the South Baltic area, producing the most spectacular falls from wader migration, however). I could add here: good performance in a wind flat area is connected with experience, the presence of one or a few experienced birds in a flock may save the others much effort.


2. Results

2.1. General progress of migration

The time-schedule of arrival, rest and continued migration of Dunlin visiting the Falsterbo peninsula in late summer and autumn can be summarised as follows:


2.2. Diurnal timetable

The last of the daylight is fading when the nets are set up: by 23h in late June, as early as 21 h in late August. Water levels will determine the overall number of resting birds in all catching areas, but a few Calidris are present at most sites under all circumstances; when water levels are high they forage at the edges of inundated meadow or on floating seaweed banks. With ongoing migration there will be a pronounced fall of Dunlin from midnight onwards; the distributions of four categories of birds caught are shown in Fig. 1, the cumulative percentage distribution between 00h and 08h in Fig. 2. 50 % of the total catch has been taken already by 02.30, with some allowance for handling and some allowance for lower catching rates after dawn, it's fair to say that the migration culminates by that time. On average birds that have initiated their moult seem to be a little quicker to halt and feed, but any difference in presence between the categories has disappeared before the culmination.


2.3. Descriptive statistics, multiple regressions, weight transformations

Since the distributions in time are almost identical there is some point in departing from the overall statistics:


Table I. Biometry of moulting (m) and non-moulting (nm) adult Dunlin, aged 2c (2nd calendar year) and 3c+ (at least 3rd calendar year).


CategoryMean 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.
2c nm (n=148)118.2 ± 0.3; 2.9 (n=131)33.3 ± 0.2; 2.6 (n=131)46.3 ± 0.43; 4.9 (n=131)
2c m (n=372)117.6 ± 0.2; 3.1 (n=332)32.7 ± 0.1; 2.5 (n=332)45.7 ± 0.27; 4.8 (n=319)
3c+ nm (n=710)118.9 ± 0.1; 3.2 (n=581)32.6 ± 0.1; 2.6 (n=580)44.9 ± 0.18; 4.3 (n=574)
3c+ m (n=253)118.7 ± 0.2; 3.5 (n=222)32.8 ± 0.2; 2.4 (n=221)45.4 ± 0.30; 4.4 (n=218)



If uncorrected weights are considered, 3c+ birds with suspended moult weigh on average more than 3c+ birds with old primaries, while the difference is the inverse in 2c birds. Neither difference is significant at the 5-%-level, however (2c: F=1.04, z=1.44; 3c+: F=1.05, z=1.19). Moulting and non-moulting birds could have the same mean weights. The regressions Weight versus Time and the development of mean weights per hour are shown in Fig. 3 and Fig. 4.

In order to eliminate the effects of body-size (measured by wing-length) and time, multiple regressions were calculated for all four categories (equations 1 - 4 below). Since the procedure ultimately aims at regressing weights against wing-gaps in birds having initiated their moult, the time distribution of gaps was checked first. The median gaps in 2c birds are appr. = 0.3 primaries, 3c+ birds have median gaps appr. = 0.2 primaries, mean values of gaps between 00h and 05h are shown in Fig. 5. In most cases there is no significant difference, birds at 03h and 04h lie 1 + 1 s.e. from each other.

(1) 2c nm (n=130): WEIGHT = - 58.4 + 0.883 WING + 0.080 HOUR; Radj2 = 25,8 %, p < 0.001 for Y intercept and wing constant, time constant n.s. ANOVA: F = 23.4.

(2) 2c m (n=311): WEIGHT = - 31.3 + 0.664 WING - 0.388 HOUR; Radj2 = 19,4 %, p < 0.001 for Y intercept and wing constant, p < 0.005 for time constant. ANOVA: F = 38.4

(3) 3c+ nm (n=562): WEIGHT = - 24.7 + 0.591 WING - 0.185 HOUR; Radj2 = 19,4 %, p < 0.001 for Y intercept and wing constant, time constant n.s. (p=0.072). ANOVA: F = 68.7.

(4) 3c+ m (n=208): WEIGHT = - 11.0 + 0.487 WING - 0.514 HOUR; Radj2 = 16.6 %, Y intercept n.s., p < 0.001 for wing and time constants. ANOVA: F = 26.0.

When data is subdivided into distinct categories, it turns out that each category has, or is likely to have its own characteristic pattern. Rather little (15 - 25 %) of the variation in weight is explained by variations in body size, and only moulting birds have a significant weight decline with time. Table II shows weights reduced to 118 mm wing and 00h values by means of the above regressions. With this transformation moulting birds weigh significantly more than non-moulting ones (2c: F=1.04, z=2.84; 3c+: F=1.08, z=4.75):


Table II. Mean weights of 2c and 3c+ Dunlin, reduced to wing-length 118 mm and midnight state with regressions (1) - (4). m = moulting, nm = non-moulting.


CategoryMean weight (g) ± 1 s.e.; 1 s.d.
2c nm (n=130)45.8 ± 0.4; 4.2
2c m (n=311)47.1 ± 0.2; 4.3
3c+ nm (n=562)45.0 ± 0.2; 3.8
3c+ m (n=208)46.5 ± 0.3; 4.0


With contributions from size and time eliminated, transformed weights can be tested for "among" and "within" variances with single classification ANOVA. There is a significant but small added variance sA2 = 1.14 among groups (F=20.4), which means that the percent of variation within groups is 93.4 %, among groups only 6.6 %. If moulting and non-moulting birds resp. are pooled, the among variance sA2 = 1.43 (F=53.7), which means that the percent of variation within these two groups is 91.9 %, among them 8.1 %.

dfSSMSF
Among groups3983.2327.7220.36
Within groups120719425.816.09
Total121020409.0


Sexes were determined from plumage characteristics according to the recommendations by Ferns & Green 1979; in addition I have developed and used a few marks of my own during the course of the field-work. Nuptial plumages are beginning to get abraded by the end of July and some birds are intermediate in both measurements and colouring; only birds without any question-marks have been assigned to either sex. The overall material is presented in Table III, mainly for orientation. In the long run it seems desirable that the weights of each sex be transformed to some sex-specific average wing and 00h values separately, but the size of the material is not adequate for this yet.


Table III. Descriptive statistics for sexes in 2c and 3c+ Dunlin, moulting and non-moulting. Weighing is discontinued if wind gusts rock the weighing bag (biometry collected in the open!), hence the variation in n.


Age, category, sexnMedian dateMedian hourMean wing (mm) ± 1 s.e.; 1 s.d.Mean bill (mm) ± 1 s.e.; 1 s.d.Mean weight (g) ± 1 s.e.; 1 s.d.Mean gap ± 1 s.e.; 1 s.d.
2c moult male134-1521 Aug03h116.8 ± 0.2; 2.631.9 ± 0.2; 2.144.6 ± 0.4; 4.40.40 ± 0.03; 0.42
2c moult female83-9428 July03h119.7 ± 0.3; 2.934.8 ± 0.2; 2.147.8 ± 0.5; 4.80.36 ± 0.04; 0.37
3c+ moult male81-995 Aug03h117.6 ± 0.3; 2.931.6 ± 0.2; 1.943.8 ± 0.5; 4.60.29 ± 0.04; 0.36
3c+ moult female58-6430 July03h120.4 ± 0.5; 3.734.5 ± 0.3; 2.146.6 ± 0.5; 3.80.23 ± 0.04; 0.33
2c nonmoult male56-6425 July03h117.1 ± 0.4; 2.831.8 ± 0.3; 2.144.2 ± 0.6; 4.7
2c nonmoult female47-5422 July03h119.8 ± 0.3; 2.234.9 ± 0.3; 2.049.2 ± 0.5; 3.7
3c+ nonmoult male258-31028 July03h117.6 ± 0.2; 2.831.3 ± 0.1; 2.043.4 ± 0.3; 4.0
3c+ nonmoult female162-20424 July03h120.7 ± 0.2; 2.734.8 ± 0.2; 2.147.1 ± 0.3; 3.9

The span of the median dates of eight discernible categories is fourteen days, birds with initiated moult migrate on average one week later than birds with old remiges. Females have longer wings, higher weights and smaller gaps than males. The wings of 2c males are on average 0.5 - 0.8 mm shorter than the wings of 3c+ males, the wings of 2c females 0.6 - 0.7 mm shorter than the wings of 3c+ females. This will subtract appr. one mm from a wingspan of order of magnitude 300 mm, barely affecting the aspect ratio of the wing. The negative influence of worn remiges in flight, if there is one, should not be sought here, but rather at the ragged edges of individual feathers. The tail often is of much poorer quality than the primaries (in extreme 2c cases only the rectrix shafts remain in July), while there is little visible wear in secondaries.


2.4. Transformed weights versus moult gap extensions

The linear regressions TRANSFORMED WING versus GAP SIZE have coefficients of determination almost 0 and neither is significant, the regression of 3c+ birds more distinctly linear than that of 2c birds (Fig. 6). On the other hand the quadratic regression TRANSFORMED WING versus GAP SIZE (Fig. 7) at least indicates that weights are not randomly distributed over gaps in 2c birds. Figs. 8 and 9, linear and quadratic regressions resp., detail the within variation of the materials, note the dip of 3c+ birds at gap 1.0 and of 2c birds at gaps 1.1 and 1.2 (often a newly shed primary with the preceding primary almost full-grown). As a matter of fact the lowest mean weights of the overall materials occur here. There are other agreements as well; the jerks between gap size 0 and 1.0 are much the same in both diagrams, and it seems as if the development from gap size 1.0 or 1.1 onwards merits a regression analysis of its own (preferably with a larger material).

In section 2.3 we saw that there is an extra variance component when non-moulting birds and birds with initiated moult are separated, and since the gap size in itself seems to be of such little importance, the role of the position of the gap should be investigated. Gaps are plotted versus position in wing in Figs. 10 and 11, note that growth of any one of the ten primaries may be suspended during migration, which means that Dunlin will migrate from the northern parts of Russia and Siberia in any stage of moult (or with fresh plumage). Covariation analysis and multiple regression again show that gap size doesn't really contribute to weights, therefore it is disregarded and only the one-parameter regressions TRANSFORMED WEIGHT (118, 00h) on GAP POSITION calculated:
(5) 2c m (n=306): TRANSF WEIGHT = 46.3 + 0.201 GAP POS; Radj2 = 0,4 %, p < 0.001 for Y intercept, p = 0.14 for gap constant (n.s.) ANOVA: F = 2.19, p = 0.140

(6) 3c+ m (n=208): TRANSF WEIGHT = 45.1 + 0.417 GAP POS; Radj2 = 3,4 %, p < 0.001 for Y intercept, p = 0.004 for gap constant. ANOVA: F = 8.34, p = 0.004
There is good significance in the 3c+ case, but weights increase as the gap moves outwards, weights also increasing in the 2c case, but here the correlation is weaker and non-significant.


3. Discussion

3.1. Are Dunlin most active in the daytime or the nighttime?

The adaptive value of nocturnal foraging cannot be doubted, although a rhythm biased towards the nighttime may be double-edged (Andrews 1992, Stienen & Brenninkmeijer 1997). Dunlin are mainly tactile feeders, with a highly flexible joint near the sensitive bill-tip and highly developed random searching routines (see Ehlert 1964, Mouritsen & Jensen 1992); they don't need eyesight to locate their prey. Even in the daytime they seemingly close their eyes when probing the substrate with slightly opened bill. An extra advantage is presented by the fact that some invertebrate prey has an activity peak in the nighttime (Dugan 1981), the period when most Dunlin descend from the migration level is likely to coincide with that peak. (Nocturnal feeding is mentioned from Mellum (tidal flats) in Ehlert 1964, but it was not studied specifically, although it is suggested that Dunlin don't sleep at all in the nighttime; Dierschke 1996 worked in both the North Sea and the Baltic, but his concern, too, is mainly with diurnal feeding. Particular emphasis on nocturnal feeding in Dunlin is found in Mouritsen 1992, Mouritsen & Jensen 1992, Mouritsen 1994, Hötker 1995, Shepherd et al. 2003, in addition there is a recent study of night-feeding Redshank Tringa totanus by Burton & Armitage 2005; Goss-Custard 1969 stated that Redshanks fed in the nighttime only in winter). The traditionalism of foraging Dunlins - particularly important in the nighttime, when the feeding substrate may be difficult to scan for worm casts - is the topic of Luis & Goss-Custard 2005; the numerous retraps of migrating Dunlin between seasons at all Dunlin-catching sites in the Baltic area (cf. Roos 1984) belong in this context; migrants re-visit wellknown sites if they are available. In general I think nocturnal feeding has one distinct advantage over diurnal to Dunlin; no safety precautions have to be taken, the birds are more confident than in the daytime and seem to focus their attention exclusively on one task, insofar foraging "optimally". If the prey densities differ between night and day as well, nocturnal foraging may yield better than diurnal, considering the time and energy invested. On the other hand: the lower midnight temperatures are likely to neutralize this advantage, and in the end Dunlin probably have to balance the advantages and disadvantages of nocturnal and diurnal feeding against each other. Again: no clear-cut, one-sided adaptations.

3.2. Implications of large weight variances

The migration of Arctic waders by way of the Baltic area in late summer and autumn has not attracted at all the same amount of theoretical attention as the spring migration in different regions (e.g. Piersma & Jukema 1990, Gudmundsen et al. 1992, Farmer & Wiens 1999, Clark & Butler 1999); on the whole the Baltic has provided rather straightforward empirical reports. One explanation for this lies at hand: there is something improvised and ad hoc about the general migration progress in this rather patchy area (with as much coastline as the whole of the Atlantic-exposed NW Europe) that is unlikely to inspire e.g. the local protagonists of "optimal migration" (Alerstam & Lindström 1990, Klaassen & Lindström 1996, Hedenström & Alerstam 1997). As a consequence there has been no concerted effort to elucidate the distances covered in single flights in autumn, in particular the stages over inland or Arctic Ocean Russia. Meissner 1998 does little more than reflect this situation when remarking: Low amount of accumulated fat suggests that this species (the Dunlin; my comm.) migrates along the southern Baltic in small steps, similarly to the rest of Europe. (As a matter of fact there are a handful of recoveries illustrating extremely swift long-distance movements by Dunlin within the Baltic area). The same problem is touched upon by Tomkovich & Soloviev 1996 when they remark that Knots Calidris canutus seem to leave Taimyr at much the same time as they appear in W Europe. In his analysis of observation material from Blåvand, Jutland, Netterstrøm 1964 suggests that the larger part of Knot migration from Siberia as a rule passes south of Scandinavia, and that substantial migration at the Norwegian and Danish west coasts occurs only as a "northwest displacement" in anticyclonic weather; the explanation for variations in Dunlin numbers probably should be sought along the same line.

In the present material birds range from fat-free weights 36-38 g to just over 60, and in particular the uncorrected distributions are extended and flat - still much of the cohort arriving after midnight seems to rest only a couple of hours, replenishing and leaving in the same mixed state. There is no evidence whatsoever that lean birds stay longer in order to replenish; in general I think that birds resting for longer periods at Baltic sites were healing minor injuries or fitness losses incurred when being caught in walk-in traps on peak days; the resting-times reported from Poland, Ottenby and Langenwerder are artefacts. The regressions (1) - (4) belong in this context; birds with initiated moult seem to spend more energy per hour and their weight distributions are a little skewed to the low side. The higher midnight weights may simply be compensation for inferior performance at some level: flight, regulation, metabolism. Taking this one step further: each single category that can be discerned in the field seems to have its own, characteristic balance of muscle, fat and water (with sexes separated there would most likely be eight distinct regressions) and in the end probably its own, specific metabolic pattern as well, an observation that brings to mind e.g. the assumed different metabolic rates of small and large Redshanks in Davidson & Evans 1982, the same assumption recurring in Kvist & Lindström 2003, and the rapid internal changes in refueling Knots reported by Piersma et al. 1999; patterns are not even stable, but transient, connected with particular situations. The important point here is that the mass of a bird is determined not only by overall body size, but by age and moult state as well. Not even the discovery that large fat-loads can be more quickly deposited and more efficiently converted to energy than hitherto assumed (Kvist et al. 2001, Kvist & Lindström 2003) removes from the agenda all problems surrounding wader "transport"; in addition there may be some sort of hierarchy in migrating flocks, at least in autumn, with heavy and fit birds "breaking the air" (using the advantages of foraging effectivity, large fat-loads and effective conversion) and the weaker categories more or less coasting along. All Arctic habitats do not offer favourable conditions for fattening at all times in summer, and it is interesting to note how e.g. Pitelka 1959 considers the possibility that breeding grounds should be surrendered by idle birds in order to create space for the growing young. If such birds leave under all circumstances, lean or fat, they will experience a distinct, hitherto not adequately considered advantage in migratory aggregates.

These findings seem to call for some sort of arbitration between the original, rather straightforward and totalitarian concepts of "optimisation", gaining some ground within migration theory after the opening paper by Alerstam & Lindström 1990, and less straightforward approaches, like the one of "risk-sensitive foraging", based on a non-linear relationship between energy reserves and assumed fitness (e.g. Real & Caraco 1986, Caraco & Lima 1987, Bednekoff & Houston 1994, McNamara, Merad & Houston 1991). I am not adducing these sources because I think their approach is entirely to the point, but because I think they stand one step closer to true, dynamic reasoning and modeling. In the case presented here some birds will arrive at moulting grounds in the Waddensea with low weights, in Darwinist vocabulary e.g. "lowering their risks of being predated at arrival". But is some gain of this kind at the core of the whole strategy, and should foraging and general progress of the birds involved be called risk-prone, risk-averse or simply risk-ignoring? (The large variances reveal that simple optimisation isn't at work, but variances in themselves don't reveal anything about underlying causality). With nocturnal migration and nocturnal foraging we clearly see a case of straightforward and effective risk-aversion, the predation risk reduced almost to zero (cf. Mouritsen 1992), but the variances of body weights remain large, because the members of the flock or the migration wave stick together. This indicates some sort of division of labour within flocks, and possibly a separation of heavy and lightweight individuals after arrival at the Waddensea. In the end an element of risk-averse behaviour (nocturnal migration and foraging) may be balanced by a risk-prone element, connected with stochastic food availability in an extremely patchy area (some individuals at some stage migrating with almost fat-free weights), creating a sort of neutral overall progress when all risks are considered - although large variations between years in the net outcome of risk-sensitive strategies are likely. Note that an Arctic Ocean and even a White Sea route will offer no darkness and higher predation risks in July, while a hypothetical southerly route ending in Lithuania, Poland or North Germany (or rather: the northern flank of a cross-continent transport heading inland for the German and Dutch parts of the Waddensea) will experience much more protection from darkness. Birds performing part of their postnuptial moult on breeding-grounds on average migrate later than birds with old remiges (Table III), one purpose of this strategy could be to prolong the time span available for migration and foraging in darkness; the bright Nordic night is over by Aug 7th. (This observation must be highly relevant to the case of Dunlin subspecies moulting partially or completely in Alaska, see Holmes 1966, Holmes 1971). Departing from e.g. Rybinsk on the Volga by 19h and interrupting the migration flight temporarily at 05h, maybe with several intermediate halts, a Dunlin flock would cover at least ten longitudinal degrees and gain almost an hour of predation-free flight in darkness. On the other hand, foraging should be more risky than migration flight, and many migrant Dunlin seem to spend the better part of the night foraging - indicating that they migrate when they have fuel stores (in the daytime) or when there are no suitable feeding-grounds, while nocturnal flight is interrupted for some time as soon as they spot promising feeding-grounds. The time-schedule of migration could even be aligned to arrive at good feeding-grounds in the middle of the night.


3.3 Weight development of birds with initiated moult and birds with wing-gaps

Gaps amount to much the same ratio from inner primaries to outer primaries (Figs. 10 and 11), this means a gradual increase in absolute measurement, and if the mass increase from inner to outer primaries is considered, gaps involving primaries 7 - 10 represent a far larger mass than gaps involving primaries 1 - 6. The untransformed weight materials, with half a dozen of the extreme points excluded, will give zero regressions when regressed against gap sizes. Were the analysis to halt there, we would conclude that there is no significant effect of wing gap sizes manifesting itself as e.g. low weights (decreasing flight efficiency) or high weights (extra fuel, muscle growth) over most of the material. The extreme points to the right are an exception, but they are too few to be conclusive; they represent the "failures" in a material exceeding 1000 birds. When weights are transformed to 118 mm wing/00h status with regressions (2) and (4), distinct patterns emerge, however; moulting birds have added on average c1.5 g to an average level 6 - 8 grams above the fat-free level (Table II). Could there be anything artificial or unallowed about these transformations? The constants of the multiple regressions pertaining to moulting birds are significant, the regressions in themselves are significant, and the temporal regressions are evident from Figs. 3 and 4 as well). There is some small, mainly non-significant development in gaps between 00h and 05h (for averages, see Fig. 5), it is likely to be either stochastic noise or caused by the passage of birds from two different departure areas. In the latter case it cannot be ruled out that corrections from e.g. 06h back to 00h give slightly faulty values because there is a leapfrog migration Lithuania/Poland - Bornholm - S Scanian coast, with birds descending and foraging e.g. every two hours or whenever there is a chance. There is a tendency for weight increase in 2c birds from 04h onwards (cf. Figs. 3 and 4), but not in 3c+ birds. Such an error would affect late morning values most, and the possibility that the elevated extreme right-side values of Figs. 8 and 9 are to some extent artefacts cannot be excluded.

A rather recent interest in effects of tail and wing asymmetry (e.g. Thomas 1993) has gradually switched to investigations of moult-gaps in a few easily manipulated species, e.g Starling: Swaddle et al. 1996, Swaddle & Witter 1997, Swaddle et al. 1999 and Tree Sparrow: Lind 2001, Lind & Jakobsson 2001 and Lind et al. 2004. The difficulties met with when working up rather anonymous field materials is amply illustrated by e.g. Holmgren et al. 1993b, this paper barely contains one correct statement - the main reason must be a lack of biological common sense, but the square data-collecting routines of Ottenby Bird Observatory probably are partly to blame; very little can be done with the crude grid of the Euring moult score. Modeling meets with problems as well: Hedenström & Sunada 1999 developed a primitive and barely worthwhile model for moult-gaps in general, predicting that proximal wing-gaps would be more detrimental to aerodynamic performance than gaps among outer primaries. If the positive, significant regression (6) for 3c+ Dunlin is considered, their suggestion raises the question about the response to effects of moult gaps: is it passive, merely reflecting some shortcoming, or active, constantly regulating both wing structure, flight performance and muscle volume? The early Swaddle Swaddle et al. 1996, Swaddle & Witter 1997 clearly is in favour of regulation, and I myself likewise expect changes in muscle tissue, fat weight, wing structure to be regulatory, in some agreement with the demands of control theory: too straightforward corrections will result in "jumpy" performance, a correction that both derivates and integrates the rate of change in some load function - a "flyball governor" - is most likely to achieve its ends. The lesson from the material presented here, however, is that gaps up to the extent of one primary can be handled with little or no correcting effort by migrating Dunlin, the real problems don't set in until the gap from two more or less absent primaries is to be compensated for (Figs. 8 and 9, with the above reservations). When compared with Fig. 8, Fig. 9 seems to illustrate an immature, undeveloped regulation; a dip when there is an extra load, followed by "overreaction" when some crude measure is taken. It may simply be that there is an element of learning in flight regulation, the birds have to get experienced in the handling of their own resources. (It's much like young Luke Skywalker waving his light sabre for the first time; he can't hit his target). Why then do transformed weights increase when the gap moves outwards, and why doesn't gaps up to the extent of one full primary seem to be of much importance? Firstly, I believe that the bird wing can and will always act as a fan, covering gaps with proximate feathers, and that this is what actually happens in both moulting and non-moulting birds. The wader wing is artfully tiled in a half-folded state, the tiles supporting each other in flight, but it's probably impossible to compensate fully for the absence of two full primaries in this way. Secondly, considering the increasing mass of outer primaries, the most likely explanation for increasing weights when the gap moves outwards is simply that the bird stores increasing amounts of nutrients and energy as well as building supportive tissue for feather growth; aerodynamic performance is not at the heart of the weight changes.


3.4 End of the journey: is the material representative of the whole of the postbreeding migration?

Dunlin caught on the Falsterbo peninsula are less than 500 km from the northern parts of the Waddensea and there are scores of excellent resting and foraging sites along their route, at least under favourable weather conditions. Do the migrants change behaviour in some crucial respect when coming that close to their proximate (or ultimate) goal, or does the journey proceed much the same way in a backward reconstruction: nocturnal migration/nocturnal foraging in Poland, Lithuania, Latvia, Estonia, Finland, Russia...? The whole of this NW European area was formed under the influence of the last glaciation; streams, lakes and ponds are ubiquitous, May and June tend to be dry months when fields and meadows dry up, but the central summer-month July is likely to offer much rain, insects and worms abound in every wet corner. Meissner 1998 seems to claim that migrating Dunlins routinely halt when arriving at the river mouths of the Bay of Gdansk, but at the same time he refers to birds "leaving Gulf of Gdansk after a short stay". How short? The waves of migration never last more than one or two days, but both Gdansk/Rewa (Poland; Gromadzka 1986, Meissner 1998) and Langenwerder (Germany; Brenning 1987) report birds staying for a week or more, "average stay" of adults at Langenwerder 2.9 days. Is it possible that we primarily see effects of the walk-in traps here? In general I think that the fieldwork routines of fifty years of pioneering wader study in the Baltic area should be reflected upon; the trade-off between the desire to catch large amounts of birds and the need to safeguard their fitness for the continued journey hasn't always been optimal. Waits of half a day may be acceptable to stationary birds on tidal feeding-grounds, but not on migration. Dierschke 1996 is more restrictive on this point, flatly stating that most Dunlin visit the Baltic only for a couple of hours or at most for a day, he calls Baltic wind flats "emergency landing areas". Very little is said in all sources about the ratio between resting birds and transmigrants, the actual turnover, and in particular very little is known or said about the rate of nocturnal migration and foraging. I forward as a hypothesis that part of the "unseen" and obviously very rapid wader migration from Arctic areas in European Russia and Siberia is nocturnal, in that way avoiding falcon and hawk predation, and that in particular eastern birds progress more or less nonstop in a "leapfrog" fashion on an inland route, determined by the specific landscape (for a theoretical view, see e.g. Farmer & Parent 1997, Farmer & Wiens 1998), utilising every foraging possibility in the increasingly stochastic environment as they approach the Waddensea. The same swift, short-stage progress was concluded for Knots (Calidris canutus) migrating by way of Ottenby by Helseth et al. 2005, and the southerly inland route (probably including C. ferruginea and some C. alpina as well) is suggested by e.g. Netterstrøm 1964. The pattern observed in the Öresund area might well be valid all the way from the Arctic breeding-grounds to the Waddensea, but most likely there will be differences between years, determined by the overall summer weather (Atlantic regime, with low passages, rain and headwind, wind flats inundated - Continental regime with highs, drought and tailwind, wind flats of huge extension).




  • To "Studies of migrating Dunlin Calidris alpina in the Öresund area, S. Sweden: Introduction"
  • To "Migrating Dunlin Calidris alpina in the Baltic area: the moult issue"
  • To "Phenology and biometry of Dunlin Calidris alpina migrating by way of the Öresund area, S. Sweden"
  • To "Wintering and spring staging Dunlin Calidris alpina in the south Baltic area"
  • To "Migratory progress of juvenile and adult Dunlin Calidris alpina from two perspectives: the Baltic and the Waddensea"
  • To "Bill-length distributions in Dunlin Calidris alpina"
  • To the bill length account
  • About "adult buff" coverts
  • To the Meissner scale
  • To the literature list A - J
  • To the literature list K - Z
  • Back to start page


    First published 24.2.06, last corrected 11.2.08, language corrections 6.3.06.