zyxwvutsrqpo zyxwvutsrqp lbis (2001) 143, 63-71 zyxwv zyx Significance of the White Sea as a stopover for Bewick's Swans Cygnus columbianus bewickii in spring BART A. NOLET'', VALERY A. ANDREEV2, PREBEN CLAUSEN3, MARTIN J. M. POOT4& ERIK G.J. WESSEL' Netherlands lnstitute of Ecology, Centre for Limnology; Postbox 1299, NL-3600 BG Maarssen, The Netherlands 2Committee of Environment Protection and Natural Resources of Arkhangelsk Region, Troitsky Avenue 49, Arkhangelsk, Russia 3National Environmental Research Institute, Department of Coastal Zone Ecology; Grenivej 12, DK-8410 Ronde, Denmark 4Zoological Laboratory; University of Groningen, Postbox 14, NL-9750 AA Haren, The Netherlands zyx zyxwv zyxwv We searched for a major stopover site of Bewicks Swans Cygnus columbianus bewickii in the White Sea following the suggestion that one should exist on the stretch between Estonia and the breeding grounds (1750 km). We discovered 733 Swans in Dvina Bay during a late aerial survey in spring 1993. Subsequently, ground-based research was carried out in May 1994, 1995 and 1996 in the Dry Sea, a tidal, shallow bay with fresh t o brackish water just north of the Dvina Delta. The total number of passing Bewicks Swans was estimated at 10974 (1994), 9593 (1995) and 17972 (1996) (32-60% of the flyway population). Estimated peak numbers staging were 1500-2000 (9 May 1994), 4937 (1 7 May 1995) and 4457 (24 May 1996) (> 5-16% of the flyway population). The Swans foraged almost exclusively on submerged water plants apart from some supplemental feeding on emerged food plants around high tide. Stoneworts Chara spp. were an important food in the late spring of 1996, because they grew in places where bog streams quickly melted the ice. At this latitude (65"N) food alternatives to the submerged macrophytes are rare in spring, but we cannot rule out the possibility that the Swans forage on grass rhizomes on inundated pastures. One bird tracked by satellite staged 15 days in Dvina Bay, of which four days were spent in the Dry Sea, in accordance with other indications that the Dry Sea is part of a larger stopover site within Dvina Bay. Recent evidence shows that the Swans largely slup the White Sea during autumn migration. However, in spring the birds probably need this stopover to be able to carry reserves to the breeding grounds. At present, the preservation of the submerged vegetation in Dvina Bay seems t o be crucial t o the conservation of this Bewick's Swan population. Tundra Swans Cygnus columbianus have a circumpolar distribution. The nominate columbianus breeds on the tundra of North America (> 150 000 birds). The swans on the tundra of Europe and Asia belong to the subspecies bewickii; those breeding in the Lena Delta and further to the east winter in Japan, Korea and China (about 30000 birds), whereas the majority (also about 30 000 birds) of more westerly Bewick's Swans winter in northwest Europe (Beekman 1997, Rees et al. 1997). A recent survey of important stopover sites of waterfowl indicated no main site for Bewicks Swans between Estonia and arctic Russia, a stretch of about *Corresponding author. Email: nolet@cl.nioo.knaw.nl 02001 British Ornithologists' Union 1750 km (Scott 81 Rose 1996). In theory, Bewicks Swans should be able to cover this distance in one nonstop flight but, based on the timing of the migration, Beekman et al. (1991) suggested that a further major stopover site should exist. Rees (1991) reported that nine marked Bewick's Swans had been observed between Estonia and the breeding grounds, but a major stopover site was not identified. Beekman et al. (1991) thought that a stopover was located somewhere in Karelia and/or in the White Sea, based on the presumed preference of migrating Bewick's Swan for tubers of Fennel Pondweed Potamogeton pectinatus and the conceivable lack of alternative food sources (terrestrial or agricultural plants) above 60"N in the migratory period (see also Beekman et al. 1996). zyx zy 64 zyxwvutsrqpo zyxwvutsrqp zyxwvutsrqp B.A. Nolet et al In our search for the missing link, we first performed an aerial survey of the White Sea in spring 1993. We subsequently visited the most promising site in the springs of 1994, 1995 and 1996. Our aim was twofold. First, we wanted to assess the numerical importance of this spring stopover site for Bewick's Swans. Secondly, we studied the food choice in relation to ice break-up in order to obtain data on food preference and the Swans' dependence on submerged water plants. STUDY AREA In the White Sea, submerged vegetation is present in Ukhta and Unskaya Bay on the Onega peninsula (Zostera spp.; Beekman et ul. 1996) as well as in the southwestern part of Dvina Bay (Fig. 1). Here waterplants are found in the delta o f the Severnaya (Northern) Dvina river and in the nearby Dry Sea east of the island Mud'yug. The latter was our main study area. The Dry Sea (64"51'N, 40"17'E) is a tidal area with an amplitude varying between 0.3 and 1.2 m. We divided the area into south, central and northern parts (Fig. 1, inset). There are clear vegetation zones with (from high to low) Reed Phrugmites australis, Bulrush Scirpus lacustris or stoneworts Chara spp., Fennel Pondweed Potumogeton pectinatus and, in the brackish waters, Eelgrass Zosteru marina. O n at least one site, Fennel Pondweed is replaced by the hybrid of Fennel Pondweed and Fineleaf Pondweed Potumogeton filiformis, i.e. Potamogeton x suecicus. As a result of the discharge of the Northern Dvina and numerous small streams in spring after ice break-up, the water in the Dry Sea is practically fresh (0-8 mM salt) a t that time of the year, except at the entrance and in the middle of the bay in the north part where the water was braclush (50-60 mM salt). Salt concentrations were calculated from the measured conductivity and water temperature on 3 June 1996 using linear regression equations derived from Tables 3.1 and 3.2 in Golterman (1970) METHODS Aerial surveys The aerial surveys were conducted on 3 1 May, 1 and 2 June 1993, and 7 June 1996 with AN2 and L410 planes, respectively In 1993, the whole coastline of the southeastern part of the White Sea from Kem to and including MeLen Bay was covered, whereas in 1996 only the stretch between Dvina Bay and Mezen Bay was counted (Fig. 1) Flight altitude was 50-1 00 m and @ 2001 British Ornithologists' flight speed about 80 km/h. The trajectory was 200-500 m away from the coast, but in 1996 an inland route was taken over the tundra between Dvina Bay and Mezen Bay (Fig. 1). A pair of observers was situated on each side o f the plane. The number of birds counted, and the time of observation, were noted. The observations were checked for double counts and mapped. zyxwvut Fieldwork We counted the number of passing swans from a 10 m high tower on the weather station at the southern tip of the island Mud'yug using 7-10x binoculars and 2 0 - 6 0 ~telescopes. Bewick's Swans were distinguished from Whooper Swans Cygnus cygnus (less than I Yo of the swans). Counts were conducted every day between 05:OO and 22:OO h from 7 May to 5 June 1994 (393 h), 4-17 May 1995 (125 h) and 1-27 May 1996 (260 h). We assumed that no swans migrated over sea during the dark night hours (see Pennycuick et al. 1999). In 1994 continuous counts were made during the day, whereas in 1995 and 1996 two hours of observation were followed by a one-hour pause and we used a linear interpolation to estimate the numbers passing during these observation gaps. The swans were foraging in flocks around low tide. These were counted occasionally in 1994 and almost daily in 1995 and 1996. Flocks were mapped by taking two bearings from the shore, using landmarks as reference points, or by measuring the angle (corrected for the local declination of +14") to this flock from observation posts using a range finder (Leica Geovid 7 x 42 BDA). The map location of these observation posts was in turn determined using a Garmin GPS 40. At the end of May and the beginning of June xve mapped the submerged vegetation along transects, using the rangefinder or a compass and a 100-m long rope for navigation. The borders of the beds in the north part of the Dry Sea were also surveyed from a small boat during high tide, again using compass bearings. The species composition and remaining plant biomass were determined by taking samples at regular intervals (10-100 m, depending on the size o f the bed). In Eelgrass beds, we placed a square steel frame (0.1 m2) a t the bottom (total sampled area: 8.0 m' in 1995 and 4.4 m? in 1996). After taking care that ramets growing inside or outside the frame ended up a t the appropriate site of the frame during placement, Lve cut a square of turf along the inside of the frame. In stonewort and pondweed beds we took sediment cores 25-40 cm deep and 10-14 cm in diameter (total sampled area: 3.5 mx in 1995 and 3.2 m* in 1996). The zyxw zyxwvutsrqp Union, Ibis, 143, 63-71 Staging areas for swans in the White Sea zy z 65 zyxwvutsrqpo zyxwvutsrqpon Figure 1. The White Sea with the trajectories of the aerial surveys conducted between 31 May and 2 June 1993 (- - -) and on 7 June 1996 (. . .). Inset shows the Dry Sea (Guba Sukhoye More). Light grey indicates shallows that fall dry during extreme low tides. The locations of the major swan feeding sites are indicated (P, Fennel Pondweed Potamogeton pectinatus except P8 where it is Potamogeton x suecicus; C, stoneworts Chara spp.; Z, Eelgrass Zostera marina). m, Location of the observation post on the weather station. cores were then sieved through a .I-mm sieve (in the north part of the Dry Sea) or a 3-mm sieve (in the south and central part of the Dry Sea). Samples (subsamples for pondweed tubers) were dried to constant weight a t 70°C and biomass density given as dry weight. Pondweed tuber densities were logtransformed (to achieve normality; thus, zero values were excluded) and differences between years (1995 and 1996) and beds (P4, P5 and P6) were tested with an analysis of variance after checlung for normality (Kolmogorov-Smirnov test) and homogeneity of variances (Levene's test). We recorded the behaviour of the swans by scan sampling (Altmann 1974), distinguishing foraging (i.e. paddling, head-dipping or up-ending, see Rees et al. 1997) from non-foraging behaviour. Scans were made zy zyxw @ 2001 British Ornithologists' Union, Ibis, 143, 63-71 66 zyxwvutsrqpo zyxwvutsrqpo B.A. Nolet et al. a t the pondweed beds P4, P5 and P6 and the Eelgrass bed Z1 (Fig. 1, inset) between 18 and 24 May 1995. We made additional scans a t the pondweed beds between 17 and 28 May 1996. The distance from the observer to the swans was 250-1500 m at the pondweed beds and 3000 m a t the Eelgrass bed. Swans were scanned once every 15 or 30 minutes using a 2 0 - 6 0 ~ telescope. When more than 100 swans were present, we usually first counted the swans and subsequently scanned 100 of them evenly distributed over the flocks. We calculated the average for each quarter of an hour, weighting the scans according to the number of swans present (pondweed: 167 scans in total with on average 165 swans, range 7-718; Eelgrass: 86 scans in total with on average 214 swans, range 100-600). Table 1. Ice break-up in the three years of study 1994 ~ ~~ ~~~~~ Ice break-up at Northern Dvina river First openings in Dry Sea Dry Sea largely ice-free -~ 1995 ~~ ~ 1996 ~ c 1 May 23 April 8 May 9 Maya 11 May c 1 May c 4 May 9 May 24 May aOpenings may have been earlier in the north part. zyxwvu Chemical analysis Eelgrass leaves collected in 1995 and 1996 were dried and later analysed for Kjeldahl nitrogen, cell wall components (neutral detergent fibre NDF, acid detergent fibre ADF and lignin; Van Soest &Wine 1967), and fat (ether extraction) and ash content. Protein content was calculated as 6.25 x Kjeldahl nitrogen, hemicellulose content as the difference between NDF and ADF, cellulose content as the difference between ADF and lignin, and soluble carbohydrate content as the part remaining after subtraction of protein, fat and NDF content. Pondweed tubers collected from different beds in 1995 were analysed only for ash and ADF content. RESULTS that the ice disappeared first in the south part and last in the north part. However, small openings in the ice formed as early as 9 May 1996 above the stonewort bed C1 in the north part where small bog streams entered the Dry Sea. The first openings above pondweed beds developed on 14 May in the south (Pl) and central part (P3). The small Eelgrass bed 2 2 in the central part was ice-free as early as 15 May, but no part of the large Eelgrass bed Z1 in the north part was exposed until 18 May. Finally, the pondweed bed P9 and the adjacent stonewort bed C4 in the north part became only ice-free on 20 May. Swan passage In May 1994, 10 974 Bewick’s Swans were counted flying in a northerly or easterly direction along the island of Mud’yug; 786 individuals flew in the opposite direction. In May 1995 and 1996, respectively, we counted 6944 and 14 1 12 Bewick’s Swans landing or flying into and 156 and 1071 flying out of the Dry Sea. Correcting for the observation gaps, we estimated the total number of northerly or easterly flying Bewick’s Swans at 10974 in 1994, 9593 in 1995 and 17972 in 1996. This represents 37%, 320/0 and 60Y0, respectively, of the flyway population, estimated at 30 000 birds. For 1994 and 1995 these are minimum estimates, since migration had already started when the counts started in 1994 and continued after we stopped counting in 1995. In 1994, peak migration occurred on 10 May. The majority of swans passing the observation post at the weather station between 7 and 9 May flew over the largely ice-covered Dry Sea and continued their journey northeastwards. Despite the large difference in the onset of spring in 1995 and 1996, the majority of Bewick’s Swans passed through in the second decade of May of 1995 and 1996 with peak numbers on 17 and 18 May, respectively. Most of the swans passing on 14 May 1996 entered an ice-covered Dry Sea, and flew on into the northeast. zyxwvutsrqp zyx Aerial surveys In 1993, concentrations of Bewick’s Swans were only seen in Dvina Bay on 1 June: 118 in the Dvina Delta and 61 5 (including one bird with a blue neck-collar) in the Dry Sea (Fig. 2). The birds were partly resting along the water edge, partly floating on the water. In 1996, the aerial survey was conducted after the swans had left the Dry Sea. O n 7 June, we counted only very small numbers of Bewick’s Swans north of the Dry Sea: 76 on small tundra lakes between Dvina Bay and Mezen Bay, and another 23 along the east coast of Mezen Bay. Ice break-up Spring was relatively early in 1995, late in 1996 and intermediate in 1994 (Table 1). In 1996, we observed zyxwvutsrqpon @ 2001 British Ornithologists’ Union, Ibis. 143, 63-71 Staging areas for swans in the White Sea zy zy 67 zyx Figure 2. Black ellipses depict the concentrations of Bewick’s Swans in Dvina Bay during the aerial survey of 1 June 1993:flock sizes are: (1) 118,(2) 295 and (3)320 swans. Letters indicate the locations of Bewicks Swan 1 1 OU tracked by the ARGOS satellite system during its stopover of two-weeks in Dvina Bay in spring 1996.Subsequent locations are labelled in alphabetical order. Accuracy of location represented as black (150-350 m), grey (35&1000m) or white (> l km). Star indicates position determined by visual observation 4.5 h before satellite location v. Swan staging and foraging zyxwvuts zyxw In 1994, between 1500 and 2000 Bewick’s Swans were staging in a large opening in the ice in the Dry Sea on 9 May. Up to 15 May, about 900 Bewick’s Swans were present, mainly feeding on pondweed tubers. In 1995, the maximum number of staging Bewick’s Swans in the south and central part of the Dry Sea was counted on 15 May with 3865 birds foraging on pondweed (Fig. 3). Unfortunately, the north-part of the Dry Sea was not counted until 18 May, when we observed 1325 Bewick’s Swans feeding on Eelgrass there (Fig. 3). Under the assumption that this number was the maximum number of Bewicks Swans in the zyxw @ 2001 British Ornithologists’ Union, Ibis, 143, 63-71 68 zyxwvutsrqpo zyxwvutsr B.A. Noletet al north part in 1995, we estimated by linear interpolation that the maximum total number of staging Bewick's Swans in the Dry Sea in 1995 was 4937 on 17 May. In 1995, the estimated total number of swan-days in the Dry Sea was 52 089, of which a t least 9.9% were spent on Eelgrass. In 1996, the number of staging Bewick's Swans in the south and central part was less than half the number in the previous year. A maximum of 1610 birds was counted on 19 May. Again, these swans were mainly feeding on pondweed tubers (Fig. 3). However, on 16 May, during the early stages of ice break-up, most swans were foraging on the small Eelgrass bed ( Z 2 ) . Many movements were observed between birds feeding on ZZ and birds exploring the pondweed beds P4, P5 and P6, which had become available again in the course of the day after ice, formed during the night, had melted away. In contrast, the numbers in the north part were much greater in 1996 than in 1995 with 3600 swans staging on 24 May. Here they started feeding on stonewort bulbils, which became available first. The swans switched to the pondweed bed P9 as soon as it was exposed on 20 May, but it was difficult to judge which proportion of the flock was feeding on the adjacent stonewort bed C4. The maximum number in the Dry Sea as a whole (4457 staging birds) was reached on 24 May, i.e. one week later than in 1995. The total number of swan-days was 32 910 in 1996, 8.8%of which were spent on Eelgrass. The submerged water plants were only available to the Bewick's Swans at low tide, and their behaviour was strongly related to the tidal cycle. In both 1995 and 1996, the swans in the central part were feeding on rhizomes of Reed and (later in the 1995 season) emerging leaves of Bulrush around high tide (Fig. 4). During one cycle, a swan actively foraged for, on average, 4.0 h on Fennel Pondweed and another 0.6 h on Reed and Bulrush; with a tidal cycle of 12 h 15 min, this corresponds to 7.9 h and 1.2 h per day, respectively. Eelgrass was available for longer periods than Fennel Pondweed (Fig. 4), namely 6.5 h/cycle, equivalent to 12.8 h/day. Our preliminary chemical analyses suggest that Eelgrass leaves had a relatively low soluble carbohydrate content, b u t high protein content compared with pondweed tubers (Table 2). zyxwvut zyxwvutsr z 0on ice 2000-7 north $ 0 ._ i mL + 0 & a $ L Significance of Dvina Bay as a spring stopover site Our migratory counts indicate that between 32% and aReed = stoneworts Eelgrass zy 4000, north 2000 1 south zyxwvutsrqp zyxwvutsrqpo zyxwvut zyxwvutsr a 4000- central 3 (I) pondweed DISCUSSION 3000- 2000- 1000- II 0 2000 1 south 1000 - 7 1000 - ? 0 5 10 15 20 May 1995 25 30 35 0 5 10 15 20 25 30 35 May 1996 Figure 3. Number and diet of Bewick's Swans staging in the three parts of the Dry Sea in May 1995 and 1996. The swans feeding on pondweed in the north part were partly foraging on an adjacent stonewort bed. The birds counted on 8 May 1995 in the north part were roosting during high tide. @ 2001 British Ornithologists' Union. /b/s. 143, 63-71 Staging areas for swans in the White Sea 0Reed/Bulrush pondweed -6 -4 0 2 -4 -2 0 2 1994 and 1996 showed that peak numbers of staging Bewicks Swans in the Dry Sea were reached more than two weeks later than in Estonia. In 1994, numbers peaked sharply in Estonia and in the Dry Sea on 24 April (Luigujije et al. 1996) and 9 May, respectively. In 1996, numbers remained high in Estonia between 27 April and 7 May', and peaked in the Dry Sea on 24 May. The swans might have spent the intervening time in the Gulf of Finland (Rees 1991), but there are indications that a large part stages elsewhere in Dvina Bay. The most direct evidence comes from a swan (1 lOU) tracked by satellite in spring 1996. Beekman and Laubek (1997) report that it flew from Estonia to Dvina Bay in only three days. A detailed analysis revealed that this swan used exactly the same areas that swans had been observed in during the aerial survey three years earlier. It spent three days in the south part of the Dry Sea, subsequently staged in the Dvina Delta for 11 days and then moved to the north part of the Dry Sea for a short stop prior to its flight to the Pechora Bay (Fig. 2). Of its two-week staging period in Dvina Bay, this satellite bird thus spent only a few days in the Dry Sea, and this seems to be true for the majority of the swans. Division of the total number of swan-days by the number of passing swans yields an average staging period of five days in 1995 and two days in 1996. Since we observed that some of the swans did not land in the Dry Sea, or only landed briefly in order to drink, the staging period of individual swans that did forage must have been longer. However, observations of neckcollared birds confirm the short staging period in the Dry Sea (Nolet & Drent 1998). The peak numbers of Bewick's Swans present in the Dry Sea are close to 5000 birds. This justifies a classification as a major stopover site sensu Scott and Rose zyxwvuts 4 " -6 69 zyxwv zy 0other -2 z zy 4 6 Time to low tide (h) Figure 4. The behaviour of Bewicks Swans in relation to the tide. (a) Scans in the central part of the Dry Sea in May 1995 and 1996 (1-1 1 scans per bar; the black bar indicates no observation). Non-foraging behaviour is split into resting and other behaviour. (b) Scans in the north part in May 1995 (1-6 scans per bar). zyxwvutsrqpon zyxw zyxwvut 60% of the West-European flyway population of Bewick's Swans passes through Dvina Bay in May. Simultaneous counts a t Mud'yug and in Estonia in Table 2. Chemical composition of main food items of Bewick's Swans. Chemical composition (% organic dry matter) - Protein Fat Soluble carbohydrates NDF ADF Lignin Hemicellulose Cellulose Eelgrass leaves ( n = 2) Pondweed tubers (Van Eerden et a/. 1997) 28.6 3.4 2.1 1.1 26.3 i 2.7 43.0 * 1.8 31.9* 1.6 3.6 1.9 11.1 i 3.4 28.3 * 0.4 14.8 5.0a 36.9 43.2 6.7b * * - 36.5 aAssumed. bWe measured 6.1 & 1.8 in pondweed tubers ( n = 6) collected in the Dry Sea. @ 2001 British Ornithologists' Union, Ibis, 143, 63-71 70 zyxwvutsrqp zyxwvutsrqp zyxwvutsrq 6.A. Nolet et al ( I 996) (> 10 %I of the flyway population). However, as discussed above, the Dry Sea is probably only part of a spring stopover. Peak numbers around and within Dvina Bay as a whole are likely to be closer to 10 000 birds. The utilization of the bays on the Onega peninsula and Dvina Delta by Bewick’s Swans urgently needs confirmation by aerial counts that are conducted earlier and more frequently in the migratory season than our own survey. during autumn stopover in the Veluwemeer, The Netherlands (Van der Winden et al. 1997). Unfortunately, data on the nutritional value of stonewort bulbils are lacking. Secondly, at a time when the wintering population of swans was much smaller than it is now, and prior to the decline of aquatic vegetation in western Europe, Bewick’s Swans were also foraging on both pondweed and Eelgrass (Brouwer & Tinbergen 1939, Sparck 1958). Whether Bewicks Swans prefer pondweed over Eelgrass remains uncertain. We found that the swans can exploit Eelgrass considerably longer (12.8 h/day) than pondweed tubers (7.9 h/day plus 1.2 h/day on Reed rhizomes and Bulrush leaves). The swans can extend the foraging period on Eelgrass by eating the leaves during rising and falling tides. This advantage may at least partly offset any disadvantage due to a lower soluble carbohydrate content of Eelgrass leaves. The leaves, which start growing while under the ice (McRoy 1969), may also be attractive because they contain about twice as much protein as pondweed tubers. In absolute numbers, swan usage of Eelgrass was most pronounced in 1995. In that year Eelgrass was more abundant than in 1996 (1 9.5 g/m* + 22.1 sd versus 10.9 17.3, PZ = 44 in both years, t,,- = 2.02, P < O.O5), even after exploitation by swans, maybe because of less ice-scouring damage and an earlier onset of the growth of Eelgrass in the early spring of 1995. The average pondweed tuber densities varied between 5.5 and 20.1 g/mz. The three pondweed beds that were sampled in both 1995 and 1996 after the swans had left, did not differ in tuber density between years (F,,,?, = 1.49, ns; corrected for differences between beds), despite the large difference in swan exploitation (22 958 versus 6123 swan-days on pondweed tubers in the central part in 1995 and 1996, respectively). This indicates that, during the staging period, more pondweed tubers must also have been present in 1995 than in 1996. If the swans depleted the pondweed tubers during an autumn stopover in the Dry Sea, no tubers would have been available during the spring since the tuber stock would not grow in the meantime but would decrease due to winter mortality and respiration of the tubers (Beekman et al. 1991). Local fishermen told us that the number of swans in the Dry Sea is much smaller in autumn than in spring. Recent results with satellite tracking show that, during the autumn migration, Bewick’s Swans skip the White Sea or only stop there for two days’. The use of Dvina Bay as a stopover in spring only can be esplained by the advantage conferred on the swans of carrying body reserves to the zyxwvuts Diet The Bewick’s Swans staging in the Dry Sea foraged almost exclusively on submerged water plants. These become available to the swans as soon as the ice breaks. We recorded only supplemental feeding on emerged food plants (rhizomes of Reed and, in the early spring of 1995, sprouting leaves of Bulrush) around high tide. During their stopover in Estonia, the swans also mainly refuel on submerged vegetation (pondweed and stoneworts), but additional feeding on cultivated plants has occurred since 1978 (Rootsmae 1990 in Luigujoe et al. 1996). The preponderance of agricultural plants in the diet of the swans fluctuates from year to year, both in autumn and spring, strongly depending on the weather (Luiguj6e et al. 1996). In our case, swans were never seen foraging on terrestrial food plants. Agriculture around Arkhangel’sk is limited mainly to the pastures along the Northern Dvina and in its delta. These pastures might be used by swans when flooded in spring, since grass rhizomes have been found to be important spring food for swans a t lower latitudes (Rees et al. 1997). However, the data at hand point to an increase in the importance of submerged vegetation as food for refuelling swans with latitude (cf. Beekman et al. 1996). The opportunistic food choice among submerged plants we observed may be due to the strong increase of the population in the last decade (Beekman 19973, some birds being forced to feed on less preferred water plants than Fennel Pondweed. We think that this is only part of the explanation for the two following reasons. First, the swans exploited the stoneworts only in the late spring of 1996 when they had no good food alternatives due to ice-cover. The swans apparently included pondweed tubers in their diet as soon as the nearby pondweed bed became ice-free, although not all stonewort beds had been depleted, as indicated by the high biomass remaining at C1 (40.6 g/m’ 69.7 sd, n = 24) compared with C3 (3.2 f 4.9, n = 16) and C4 (4.0 3.6, n = 16). Bewick’s Swans also showed a preference for pondweed tubers over stonewort bulbils zyxwvutsrqp zyxw zyxwvut z * * 02001 British Ornithologists’ Unlon, ibis, * 143, 63-71 zy zy zyxwvutsrqpo zyxw Staging areas for swans in the White Sea breeding grounds (Cooke et al. 1995). Arrival on the breeding grounds is about ten days after the peak in staging numbers in the D r y Sea (cf. Rees et al. 1997). We conclude t h a t Dvina Bay contains a major spring stopover site for Bewicks Swans. W h i c h other shallows in Dvina Bay besides the D r y Sea are important, and to w h a t extent flooded pastures along the Northern Dvina river are utilized b y swans, remains to be investigated further. At the moment, however, the preservation of the submerged vegetation in Dvina Bay seems to be crucial to the conservation of this Bewicks Swan population. We greatly acknowledge the Meteorological Office, and in particular Pjotr A. and Marij D. Malygin of the Mud'yug weather station for their hospitality. We also thank Sergej V. Dokunihin and Yurij V. Asutchenko for their valuable logistic support. We thank Ebbe Bagebjerg, Alexander S. Cherenkov, Anne Dekinga, Rudi Drent, Piet Duiven, Martin Epe, Frederik Haas, Anita Koolhaas, Harry Korthals, Oscar Langevoord, Roef Mulder, Jeroen Nienhuis, Vladimir A. Semashko and Eugeny E. Syroechkovslu Jr for their good company and hard work in the field. Jan Beekman kindly placed his spring satellite trackmg data at our disposal, and Leho Luiguj6e gave us access to his report on the counts of swans in Estonia in 1996. Peter Mikkelsen digitized the maps. Jan Beekman, Kees Carnphuysen, Marcel Klaassen and Eileen Rees commented on drafts of &s paper. The study was funded by the Netherlands Organization for Scientific Research (NWO grant 047-002-008 to B.A.N.), the Danish Natural Science Research Council (grant 950342 1 to P.C.), the Beijerink Popping Fonds and the Swedish Natural Science Research Council (grant to h e Lindstrom, Lund University). This is publication number 2610 of the Netherlands Institute of Ecology. 71 resource of Potamogeton pectinatus during autumn in the Netherlands. Wildfowl Suppl. 1: 238-248. Beekman, J.H., van Eerden, M.R., Mineyev, Y.N., Luigujk, L. & den Hollander, H.J. 1996. LANDSAT satellite images for detection of submerged macrophytes: in search of potential stop-over feeding sites for Bewick's Swans (Cygnus columbianus bewickif) along their migratory route. Game Wildl. 13: 421450. Brouwer, G.A. & Tinbergen, L. 1939. Feeding-habits and some of the factors influencing the distribution of Swans, Cygnus b. bewickii Yarr., hibernating in the Zuiderzee. Limosa 12: 1-18 (in Dutch with English summary). Cooke, F., Rockwell, R.F. & Lank, D.B. 1995. The Snow Geese of La Perouse Bag Natural Selection in the Wild. Oxford: Oxford University Press. Golterman, H.L. (ed.) 1970. Methods for Chemical Analysis of Fresh Waters. IBP Handbook No. 8, 2nd edn. Oxford and Edinburgh: Blackwell Scientific Publications. LuigujBe, L., Kuresoo, A., Keskpaik, J., Ader, A. & Leito, A. 1996. Migration and staging of the Bewick's Swan (Cygnus columbianus bewickif) in Estonia. Game Wildl. 13: 451461. McRoy, C.P. 1969. Eelgrass under Arctic winter ice. Nature 224: 818-819. Nolet, B.A. & Drent, R.H. 1998. Bewick's Swans refuelling on pondweed tubers in the Dvina Bay (White Sea) during their spring migration: first come, first served. J. Avian Biol. 29: 574-584. Pennycuick, C.J., Bradbury, T.A.M., Einarsson, O., & Owen, M. 1999. Response to weather and light conditions of migrating Whooper Swans Cygnus cygnus and flying height profiles, observed with the Argos satellite system. lbis 141: 434-443. Rees, E.C. 1991. Distribution within the USSR of Bewick's Swans Cygnus columbianus bewickii marked in Britain. Wildfowl Suppl. 1: 209-21 3. Rees, E.C., Bowler, J.M. & Beekman, J.H. 1997. Bewick's Swan and Whistling Swan. BWP Update 1: 63-74. Scott, D.A. & Rose, P.M. 1996. Atlas of Anatidae populations in Africa and Western Eurasia. Publ. 41. Wageningen: Wetlands International, Wageningen. Sparck, R. 1958. An investigation of the food of swans and ducks in Denmark. Dan. Rev. Game Biol. 3: 45-47. Van der Winden, J., Poot, M.J.M., Van den Berg, M.S., Boudewijn, T.J. & Dirksen, S. 1997. Stoneworts: food for large numbers of waterbirds in Lake Veluwemeer. Levende Nat. 98: 3 4 4 2 (in Dutch with English summary). Van Eerden, M.R., Beekman, J.H., Smit, M. & Oosterbeek, K. 1997. Patch use by Bewick's Swans Cygnus columbianus bewickii feeding upon Sago Pondweed fotamogeton pectinatus in shallow lakes in The Netherlands: variation in exploitation threshold caused by social, environmental and time dependent factors. In Van Eerden, M.R. (ed.) Patchwork: Patch Use, Habitat Exploitation and Carrying Capacity for Water Birds in Dutch Freshwater Wetlands, Van Zee tot Land 65: 187-214. Lelystad: Rijkswaterstaat Directie Idsselmeergebied. Van Soest, P.J. &Wine, R.H. 1967. Use of detergents in the analysis of fibrous feeds. IV. Determination of plant cell wall constituents. J. Assoc. Off. Anal. Chem. 50: 5Cb55. zyxwvutsrq ENDNOTES 1L. Luiguj6e & A. Kuresoo, unpublished report, Institute of Zoology and Botany, Estonian Academy of Sciences, Tartu. 2B.A. Nolet, J.H. Beekman, M. Klaassen, B. Laubek, 0. Langevoord & E.G.J. Wessel, unpublished report, Netherlands Institute of Ecology. REFERENCES Altmann, J. 1974. Observational study of behavior: sampling methods. Behaviour49: 227-267. Beekman, J.H. 1997. International censuses of the north-west European Bewick's Swan population, January 1990 and 1995. Swan Specialist Group Newsl. 6: 7-9. Beekman, J.H. & Laubek, B. 1997. Satellite tracking Bewick's Swans on spring migration. Swan Specialist Group Newsl. 6: 1 6 - 1 8. Beekman, J.H., van Eerden, M.R. & Dirksen, S. 1991. Bewick's Swans Cygnus columbianus bewickii utilising the changing zyxwvuts Received 7 June 1999; revision accepted 17 January 2000 @ 2001 British Ornithologists' Union, Ibis, 143, 63-71