Breeding biology of dark-bellied brent geese Branta b. bernicla in Taimyr in 1990 in the absence of arctic foxes and under favourable weather conditions BERNARD SPAANS. MARTIN STOCK, ANDREW ST. JOSEPH, HANS-HEINER BERGMANN and BARWOLT S. EBBINGE Spaans, B.. Stock. M.. St.Joseph. A . . Bergmann. H.-H. & Ebbinge, B. S. 1993: Breeding biology of dark- bellied brent geese Branta b. bernicla in Taimyr in IYYO in the absence of arctic foxes and under favourable weather conditions. Polar Research 12(2). 117-130. In combination with observations in spring staging and wintering grounds in western Europe, a detailed etho-ecological study of nesting dark-bellied brcnt geese Branta b. bernicla in western Taimyr. Krdsnoyarsk, Russia. was made in 1990. Most brent geese arrived on the breeding grounds from 14-19 June and started nesting within a few days. In the study area 264 nests of breeding brent geese were found, mainly on islands but also along small rivers on the mainland. The mean clutch size was 3.0 and 80% of the eggs hatched. Time budget studies showed that incubating females spent on average 138 minutes per 24 hours on feeding. Despite favourable wcathcr conditions and a low density of arctic foxes. only about one-third of the mature birds in the study area bred. I n the autumn an intermediate breeding success of 20% juveniles was recorded in the wintering areas. This was probably due to the relatively poor condition in which the brent geese left their spring staging areas. - v p ~ ~ ~ ~ ~ t . i S T ' ' Bernard Spaans and Barwolf S . Ebbinge. lnsfitufe for Forestry and Nature Research (IBN-DLO). P. 0. Box 23, NL-6700 A A Wageningen, The Netherlands; Martin Stock, Nationalparkamt. Schleswig-Holsteinkche Wattenmeer. A m Hafen 40 A. 0-2253 Tonning, Germany; Andrew St. Joseph, Mellfarm. Pallesbury. Muldon. Essex CM9 8SS. England; Hans-Heiner Bergmann, Fachbereich Biologie Der Unioersitat Osna- briick. Postfach 4469, 0-4500 Osnabriick. Germany. Introduction The factors that determine the large variations in annual breeding success of the dark-bellied brent geese Branta bernicla bernicla are the subject of an ongoing debate (Summers & Underhill 1987, 1990; Ebbinge 1989, 1990). Factors assumed to be the most important are weather conditions in the arctic breeding area (Barry 1962; Boyd 1987), size of lipid and protein reserves upon arrival there (Ebbinge et al. 1982), and nest predation by arctic foxes AIopex lagopus (Roselaar 1979; Summers 1986; Dhondt 1987; Greenwood 1987; Owen 1987; Anthony et al. 1991). Nutrient reserves are stored in the Dutch, German and Danish Wadden Sea in the 5-6 weeks prior to spring migration (Ebbinge et al. 1982; St. Joseph 1982). Weather conditions during migration affect the final condition upon arrival (Ebbinge 1989). The number of arctic foxes and their predation pressure on arctic birds have been hypothesized to depend mainly on the abundance of lemmings Lemmus sibiricus and Dicrostonyx torquatus. In particular in years after a lemming peak, the increased number of foxes are expected t o ter- rorize the nests of arctic breeding birds due to scarcity of lemmings as their main prey. The discussion about the relative importance of each of these factors, however, is hampered by the lack of knowledge about what is really happening in the breeding area of brent geese. The opportunity provided by E. E. Syroech- kovskiy of the Laboratory of Evolutionary Ani- mal Morphology and Ecology of the Russian Academy of Sciences to visit the Taimyr peninsula in 1990 and t o start a detailed study on the breed- ing biology of brent geese gives new dimensions to this discussion. This paper presents the results of our first Tai- myr expedition in 1990 and data on body con- dition and reproductive success gathered in the wintering and spring staging areas in western Europe before and after the same breeding season. These data from the wintering areas, which have been gathered over several years, can now be viewed in perspective with data from Taimyr. 118 B . Spaans et al. Our study on the breeding biology of brent geese in Taimyr focused on the following points: (1) the condition (body-mass) of the geese upon arrival; (2) the importance of body reserves dur- ing the breeding season; (3) the effect of potential predation on behaviour and breeding success of the geese; (4) the effect of lemming density on predation; and finally (5) how these factors are affected by the weather conditions. We cannot presume to solve the riddle of the cyclic variations in annual breeding output of the brent goose on the basis of one season's work, but we feel that our first results can help to shape future work and therefore merit early consider- ation. Study area Habitat The study was carried out in the coastal area north of the Piassina delta, about 200 km ENE of Dickson, Taimyr, USSR. The base camp was situated along the Lidia Bay, 74"07'N, 86"50'E (Fig. 1). This location was chosen in advance because of the presence of a small archipelago with the promising name 'Bird Islands'. We stayed in the study area from 1 June until 28 July 1990. The coastal mainland consists of low, undu- lating tundra traversed by a number of small rivers of which the Lidia is the most important. The tundra vegetation is dominated by mosses, lichens, sedges and grasses. During snowmelt the rivers are wide, but after the snow disappears the water level declines and a lush vegetation of monocots develops on the banks alongside these rivers. This vegetation is dominated by Dupontia psilosantha, Carex aquatilis, Arctophila fulua, Eriophorum scheuchzeri and Puccinellia s p p . Similar vegetation also occurs on the flat outer Beacon Islands (Fig. 1). In June the vegetation on these islands was dominated by mosses, but later in the season young monocots emerged. Island 1 was rather flat with mainly tundra veg- etation comparable to the mainland. Islands 2, 3 and 4 were rocky with bare patches, some areas of tundra vegetation, and a richer grassy vegetation around colonies of the lesser black-backed gull Larus fuscus taimyrensis (Cramp & Simmons 1983). The large island Farwaternie was covered with tundra vegetation similar to that of the main- land. Fig. 1. Upper: Location of the study area in Taimyr, Russia. Lower: the study area north of the Piassina-delta in Taimyr is indicated by grey. The Bird Islands arc numbered. Weather conditions in 1990 Daily minimum and maximum temperatures were measured at the base-camp. Percentage snow- cover was measured by counting the number of steps with and without snow along a 2200 metre long tundra transect from the seashore to the top of a hill, north of the Lidia Bay. Temperature and snowmelt data from 1977 to 1990 collected by a station along the Piassina-river, ca. 40 km southeast of our camp, were made available t o use by Y. Kokorev. Comparison of the mean maximum temperature over the previous 14 years with the minimum and maximum temperatures measured at our camp (Fig. 2) indicates that June and July 1990 were relatively warm. In our camp, the daily maximum temperature was on average Breeding biology of dark-bellied brent geese 119 25 20 15 9 - - 5 10 3 e E $ 5 0 -5 .lot.,, ,, ,, , ,, ,, , , , , ,, , , , , ,, , , ,, , , , , , , , , , I , , , , , , ,J 1 4 7 10 13 16 19 22 25 28 1 4 7 10 13 16 date in June and July 1990 Fig. 4. Duration of snow cover from 1977 to 1990. The number data from thc Piassine weather-station ca. 41) km southeast of the study area. 2" higher than at the weather station. The progress of snowmelt north of the Lidia Bay is shown in Fig. 3. In early July the area was almost snowfree. Snowmelt-data over the previous 14 years show that 1990 was a year with early snowmelt (Fig. 4). Until 10 June it was possible to reach the Bird Islands from the mainland over the frozen bay. Throughout the first week of July, islands 1 and 2 could still be reached after crossing a narrow channel by boat. From the second week of July, the islands were completely surrounded by water and could only be reached by boat. date in June and July 1990 Fig. 3, Percentage of mainland-tundra covered with snow from 4 June to 4 July. Predators Lemmings and arctic foxes. - To establish the effect of predation on the behaviour and breeding success of the geese, counts were made of lemmings, arctic foxes and other potential pred- ators. A population-dynamic study of lemmings was carried out by Russian biologists. Lemming density was measured by catching the animals both with live-traps and ordinary traps. During six weeks of continuous trapping with hundreds of traps less than twenty lemmings were caught, on average 0.25 lemmings/100 trap nights (Ryhli- kova, 1993). On about 750 man-days spent in the tundra by the participants of o u r camp during June and July, only a few lemmings were observed. Snowy owls Nyctea scandiaca, rough- legged buzzards Buteo lagopus and pomarine skuas Stercorarius pomarinus, all lemming-pred- ators, were seen migrating during the first half of June, but these predators disappeared almost completely from our study area, without making any attempt to breed. This is also an indication of a poor lemming year. Not a single arctic fox was observed and local hunters had not even tried to catch foxes the preceding winter because there were so few. Other potential predators. - Lesser black- backed gulls were common breeding birds in our study area (over 1200 pairs). These gulls were already present on the Bird Islands when we 120 B. Spaans et al. arrived on 1 June. Glaucous gulls Larus hyp- erboreus were rare and only a few nests have been found. Long-tailed and arctic skuas Stercorarius longicaudus and S . parasiticus were also breeding in our study area but in small numbers. Methods In western Europe On the Dutch island Terschelling brent geese were caught with cannon nets on 14 and 16 May 1990 to determine their body-mass just before their departure for the Arctic. Migration dates were recorded by several field workers in the Wadden Sea area in the second half of May (from 1976 to 1990). In autumn 1990 we recorded juv- enile percentage and family-size in many staging sites along the coasts of the North Sea, to arrive at a reliable over-all population mean. In this way we accounted for the unequal distribution of families with juveniles over the wintering area (Lambeck 1990a, 1990b). Special attention was devoted to geese with engraved colour-rings to determine pair-bond and breeding success of birds marked in Taimyr. In Taimyr Brent geese were caught with cannon nets in our study area from 15 t o 19 June 1990 to determine pre-breeding body-masses. We marked captured individuals with individually-coded colour rings, of which the inscriptions can be read by telescope at 300 m distance. We surveyed the entire study area for breeding geese; clutch-size and egg mass were determined in the first week of incubation. On seven different days we observed two breeding pairs on island 2 (called pair A and B) for 24 consecutive hours from a hide. On the mainland along the Lidia River we watched, also from a hide, five incubating females (called numbers 1 to 5) and only 1 male (the male of female 3). This male was chosen because both members of pair 3 were individually recognizable (marked with colour-rings in the Netherlands in May 1989). Along the Lidia River we observed on six days for 24 consecutive hours, but only male 3 was observed on one of these days. Due to the fact that incubation of the Lidia females started 5 days later than on the island, calendar days do not correspond when observations of both areas are plotted against the start of incubation. During these 24-hour periods, activity of the geese was noted every 5 minutes. We also recorded starting- time and duration of feeding-recesses of the females, all interactions with potential predators, all intraspecific interactions and time and number of potential predators present near the nesting geese along the Lidia River. In addition, a scan of the activities of a flock of non-breeders (birds without a nest or territory, living in flocks) along the Lidia River was made every 15 minutes. Non- breeders did not occur on the Bird Islands. Nests with a known number of eggs were observed during hatching to determine the frac- tion of eggs that hatched. The size of each observed family was recorded immediately after hatching. Post-hatching behaviour of brent goose families was recorded to establish the effect of predation of goslings by gulls and skuas. We also counted the non-breeding birds on the islands where they were moulting. Results Body condition during spring staging and breeding success as measured in autumn The breeding success measured in autumn flocks is plotted in Fig. 5 against the mean body-mass of adult female brent geese (caught during spring staging just before departure to the north of Siberia), together with comparable data from the other years (Ebbinge 1989; Ebbinge & Spaans 1992). Compared with other years, 1990 showed a relatively poor condition of the geese before departure and an intermediate breeding success of 20% juveniles in autumn. The mean number of juveniles per successful pair was 2.4 (N = 336). In 1990 the geese left the Wadden Sea early (Fig. 6). Wind conditions during migration in the North Sea area and through the Baltic region were favourable with 57% of days with a tailwind blowing in the period 21-30 May in the Baltic (see Ebbinge 1989 for sources and methods of calculation). Arrival and condition on the breeding grounds We saw the first two brent geese on 10 June. Most migrating geese were observed from 14-19 June (Fig. 7). Between 15 and 19 June, nine brent geese were caught, weighed and individually marked Breeding biology of dark-bellied brent geese 121 Fig. 5 . Spring body-mass of female hrent geese just before departure i n the end of May from the Wadden Sea area and breeding succc~s. expressed as percentage of juveniles. in the subsequent autumn and winter. Years that were UnfdVOUrabk due to (head)wind during spring migration are given between brackets and unfavourablc years due to high inferred fox pre- dation on the breeding grounds are circlcd (data from Ebhinge 1989 and Ebbinge & Spaans 1992). (Table 1). The mean weight of the females was only 40g less than the weight of the females caught in the Wadden Sea (Fig. 5). Although the sample is very small, the small difference in mean body-mass indicates that the costs of migration were (partly) compensated on staging grounds between the Baltic region and Taimyr. C L In In 2 v) U 0 m" c L Q, a 3 c E periods in May and June Fig. 6. Timing of departure from the spring staging area. Num- ber of days with hrent goose mass departure from the Dutch Wadden Sea in 5-day periods from 1976 to 1989. The arrow indicates the mass departure in 1990. Fig. 7. Arrival of brent geese in Taimyr and timing of egg- laying, incubation and hatching. The left graph indicates the number of migrating brent geese per day in the Lidia Bay area, the right graph indicates the number of incubating females on island 1. Distribution of breeding brent geese over the study area A total of 264 nests of brent geese were found in the study area, 252 (= 96%) on the small off- shore Bird Islands and 12 nests on the mainland alongside small rivers (Fig. 8). The mean laying date of the first egg on a small colony along the Lidia River on the mainland was about one week later than the two inner Bird Islands. We d o not have data on nest initiation on the outer islands. A review of the timing of migration, egg-laying, incubation and hatching is given in Fig. 7. Most of the geese started nesting within a few days after arrival. A sketch of the distribution of vegetation types on island 2 with the position of 16 brent goose nests is given in Fig. 9. In contrast to the veg- etation the nests seemed t o be rather evenly dis- persed and the mean density was 2,8 nestslha. The highest density was measured on island 1: 3,9 nestslha. Mean clutch-size was 3.0 2 1.2, (N = 72) and ranged from 1 to 6 eggs per nest. Mean egg-weight was 79.5 2 7.5 g, (n = 94 eggs), range 64 to 98 g, fresh weight i.e. measured at clutch completion ( 2 standard deviation). We could not relocate any of the 11 individually marked geese caught in June as a breeding bird in our study area but they could have been missed. 122 B. Spaans et al. Table 1. Pre-breeding body-mass of 9 brent geese caught between I5 and 19 June. The pair-bond and the net breeding succcss have been determined in the subsequent autumn and winter in western Europe. Codes Male body-mass Female body-mass Observed in w. Europe Number of juv. RPGA X.RPGC RPGK x RPGH RPGF RPGJ RPGN RPGP RPGT Mean body-mass 1510 1490 1380 1590 1484 1610 1380 1400 1560 1530 1496 Breeding behauiour Only the female incubated, leaving the nest regu- larly to feed while the male stayed near the nest. Around the nest, the males defended a territory. We were able to precisely estimate the length of the incubation period for one pair which was 24.5 days (measured from clutch completion to hatch of last gosling). The total time off the nest (recess time) per 24 hour period is plotted for the two island females and the five Lidia females in Fig. 10. When a female left the nest, she carefully covered the eggs with down, and frantically started to feed, usually within her territory, but sometimes on another island. On average, 90% of the time females spent off their nest was used for feeding. While the female was off, the male usually stayed close by to defend the nest against gulls or skuas. All females left their nests more often as the Fig. 8. Number of nesting brent geese in the study area. The black square indicates the location of the base-camp. Island 2 was the main observation island. breeding season progressed, varying from once per 24 hours in the beginning to up to 16 times per 24 hours at the end (Fig. 11). The average recess frequency was 11 per day for the Lidia females and 9 per day for the island females. Although there was some variation in the length of a feeding recess per female during the incu- bation, the increasing number of feeding recesses was the main reason for the increasing total time off the nest per 24 hours. However, there were striking differences between females: the mean duration over all observation days of an individual feeding recess of the Lidia females ranged from 10.9 (female 4) to 15.4 minutes (female 2), with a mean of 12.9 minutes, while this was 16.7 for the island female A and 27.3 minutes for the island female B. The large differences in total time off the nest are thus due to a longer mean recess duration on the island. Average recess frequency for all females was 10.6 per day with a mean recess duration of 15.5 minutes. Males have more time for feeding than the incubating females. The percentage of time spent on feeding during the incubation period of both island males, the male of Lidia female 3 and the non-breeding birds present in the Lidia valley decreased in the course of the time (Fig. 12). Both island males spent more time on feeding (40.4% on average) than the Lidia male 3 (31.0%) but the non-breeding geese spent the least time on feeding. Most of the feeding of the incubation females took place in the afternoon and the least in the early morning (Fig. 13). The males showed an opposite pattern: there was a tendency to feed more during midnight (Fig. 14). Breeding biology of dark-bellied brent geese 123 .G 250 E 1 - t Fig. 9. Map of island 2. Vegetation-types are indicated as well as the IocatirJn o f the brent goose nests. the camp and the observation hide. Pairs A and B were studied in detail. - 250- E 1 a, E 200- - Lh 8 150- ?? x - ._ 2 100- 50 always feeding within her own territory which had to be heavily defended by male A , who engaged in frequent conflicts with t h e neighbouring males (Fig. 15). T h e number of conflicts showed a marked decrease in the week before hatching. IntraspeciJic interactions The breeding area, especially on the island, was divided into territories. Male A had a relatively good territory (100% vegetation cover) while male B had a poor territory (<30% vegetation cover, Fig. 9). Male B in fact hardly defended A lemale 2 0 emale V lemale 4 A h a l e 5 - his territory which was only used for-feeding by himself. His female almbst always left the island Interactions with potential predators a, E 200- - Lh 8 150- ?? x - ._ 2 100- 50 to feed somewhere else. Female A , however, was O n the island the most numerous potential pred- A lemale 2 0 emale V lemale 4 A h a l e 5 - 300 3501 r ' i ' i ' A '~'lb'l;'tb'16'1'8'2b' " days from start of incubation Fig. 10. Daily recess timc during incubation of the 5 Lidia femalcs ( 1 to 5 ) and the two island females ( A and B). Females were non-stop observed during 24-hours periods. Lldla- Irland- 0 female 1 temale A A lemale 2 0 lemale 8 V lemale 3 T female 4 A lemale 5 -- 01 " " " " " " " " " " 0 2 4 6 8 10 12 14 16 18 20 days from start of incubation Fig. 1 1 . Daily number of recesses during incubation. Same fcmales as in Fig. 10. 124 B . Spaans et al. 55t 9 50 - m 2 45- a, m 40- 35- 30- a 2 25- 20 - m 0 0 male A 0 male B A male 3 15 A non-breeders 1 . , , , 1 , , , . 1 , , , . 1 , . 1 1 0 5 10 15 20 25 d a y s from start of incubation Fig. 12. Daily percentage of total time spent on feeding by male brent geese during the incubation period. Male A and B have territories on island 2 and male 3 in the Lidia valley. The non- breeders were also observed in the Lidia valley, on the same dates as male 3. ators were the lesser black-backed gulls. Pom- arine skuas and long-tailed skuas were sometimes seen flying over but did not seem t o pay any attention to the breeding geese. Sometimes the threat of the male was enough to keep a gull at a distance but very often a pursuit-flight followed. Male B had more interactions with the lesser black-backed gulls during the incubation period than male A (Fig. 16) which may be explained by the closer proximity of pair B to a gull-colony. Although there is no general trend in time, there was a striking similarity in the interactions of the two males. Apparently there were days on which r * a, a L time of d a y Fig. 14. Diurnal variation in feeding of males during the incu- bation period. The percentage of total daily feeding-time of the 2 island males A and B is given per 4 hour periods. lesser black-backed gulls assailed the geese and days on which they did not. The only successful gull action that we observed on the island was a gull that managed to remove the down cover of a goose nest while the female was absent, but the male chased the gull away before it could take an egg. On other islands we saw egg-predation by the lesser black-backed gull and the arctic skua but in all cases disturbance of the geese by our presence was probably the prime cause of this predation. O n the mainland no lesser black- backed gulls bred and although they regularly flew over the valley they left the geese undisturbed most of the time. The actions of male 3 were mainly directed to the pair of arctic skuas breeding c * a, L a L 0 22 time of day Fig. 13. Diurnal variation in feeding of females during incubation. In this figure the percentage of total daily recess 140- OmaleA a male B 0 5 4 0 - n d a y s from start of incubation Fig. 15. lntraspecific interactions of males during incubation. The total number of interactions per 24 hours is eiven for the time for all females is given per 4 hour-periods. island males A and B. Breeding biology of dark-bellied brent geese 125 0 male A 0 male B A male 3 Lesser Black-backed Gull in this area (Fig. 16). Like the lesser black-backed gulls on the island, the skuas did not seem to be very successful in taking eggs in the Lidia valley. The number of gulls and skuas present near nests in the Lidia valley decreased during the incu- bation period (Fig. 17). It is likely that other more profitable food, for example from marine sources, became available for these birds in the course of the season. After hatching The two studied island pairs hatched their eggs on average 5 days earlier than the five studied females in the Lidia valley. On island 2 the first goslings hatched on 12 July and the last on 16 July. Within a few days after hatching the families left the islands. They presumably swam to the river mouths on the mainland where the food situation was more attractive. The mean observed brood size (island and river-breeders) was 2.4 ( N = 34) while clutch size was 3.0 ( N = 76), thus 80% of the eggs hatched. Both island pairs studied (A and B) were less successful: only one out of the three eggs hatched in each case. The parents were capable of defending their goslings against the lesser black-backed gulls. Only two times did we witness a gull taking a gosling, both times while the family was swimming in the water under windy conditions. The Lidia breeders stayed in the nesting area after hatching. The banks of a small lake, about 80 0 Black-backed Gull OSkuas 70 - 20 - 10 - d a y s from start of incubation Fig. 17. Presence of avian potential predators at thc Lidia brent geese colony. The number of times per 24 hours one or more lesser black-backed gulls and (arctic) skuas were present is given. 1 km west of the nesting site along the Lidia River became one of their preferred feeding sites. The percentage of eggs hatched was 89% (16 goslings from 18 eggs in nest 1 to 5). We observed the families for more than one week. During the week only one gosling was lost. After a week's time the goslings had grown so much that predation by gulls or skuas did not seem likely. We were able to determine the breeding results of two individually marked breeding pairs soon after hatching: Lidia pair 3 had one gosling (from one egg) and the pair OFYA and OFYF had also 1 gosling (number of eggs unknown). Because we left the area on 28 July, we d o not have data about the gosling survival of the local population studied. The large number of moulting non-breeders on the Beacon Islands (Fig. 1) at the end of July, about 1200, suggested that there were many non-breeding birds in 1990. Assuming that the study area is representative for the entire breeding population, 264 breeding pairs with 2.4 juveniles per pair and 1200 non-breeding birds gives a juvenile percentage of 2.4 x 264 (2.4 x 264) + (2 x 264) + 1200 = 27%. Due to losses during growth and migration, juv- enile percentage in winter area should be less than this value. Observations of marked individuals in the winter area Geese with known pre-breeding weights. - No less 126 B. Spaans et al. than 8 of the 9 geese caught in spring in Taimyr were seen by different observers in the subsequent autumn and winter in western Europe. As juv- eniles stay with their parents during the first win- ter it was possible t o determine the family size of two ringed pairs and one ringed female: the pair RPGC and RPGA was accompanied by five full grown juveniles (Ameland, Netherlands), the pair RPGH and RPGK had no juveniles (south- east England) and the female RPGN was seen with an unringed male and 1 juvenile (Texel, Netherlands) (Table 1). Geese with unknown pre-breeding weight but with known breeding results soon after hatching. - Both pairs of this category were seen again. Lidia pair 3 was seen with one full-grown juvenile (coast of Groningen, Netherland) and pair OFYA and OFYF had one juvenile as well (Schiermon- nikoog, Netherlands). This means that both pairs raised their single gosling successfully because both were seen with only one gosling soon after hatching in Taimyr. Discussion The late arrival of brent geese in Taimyr (Fig. 7), compared with the departure from the spring staging areas (Fig. 6), means that there must be staging areas along the northern coast of Russia. No staging areas are known in the Baltic (Kumari 1979; Lampio 1979). Given the fact that in Taimyr almost all vegetation was covered with snow at the beginning of June, it is likely that the staging areas are situated far more to the southwest where snowmelt starts earlier. Kistchinski & Vronski (1979) suggest that spring migration between the White Sea and Taimyr is a non-stop migration, and the dates on which migration is seen along the north Russian coast correspond with our arrival dates in Taimyr. This means that the White Sea is most likely the intermediate staging area. Given the early departure from the Wadden Sea, the geese must have spent almost three weeks on these staging grounds in 1990. This period is long enough to replenish the body reserves used during the first part of the migration (Ebbinge & Spaans, 1992). The mean female body-mass at arrival in Tai- myr was 1496 g (Table l ) . This body-mass is about 250g heavier than lowest mean Wadden Sea body-mass (first half of April) and about 140g heavier than mean winter body-mass (Ebbinge et al. 1982), so it is clear that the geese at arrival in 1990 did have reserves. However, it is likely that in years when departure weights in the Wadden Sea are higher and weather during migration is comparable, the body condition on arrival in Tai- myr will improve. Since the departure weights were low in 1990 (Fig. 5 ) , 1496 g will probably be a rather poor pre-breeding condition. Brent geese started breeding very soon after arriving in Taimyr (Fig. 7). These observations are in sharp contrast to findings in Raveling (1978), Fox & Madsen (1981), Ely & Raveling (1984) and Gauthier & Choinibre (1990) where species such as the cackling goose Branta can- adensis minima, the two subspecies of the white- fronted goose Anser albifrons and the greater snow goose Anser caerulescens atlantica spend 10 to 20 days on the breeding grounds before egg- laying starts. As with the brent geese in our study, however, Ross geese Anser rossii and lesser snow geese Anser c. caerulescens start to lay eggs within 3-5 days after arrival (Raveling 1978). Most brent geese bred on small off-shore islands but some were breeding on the mainland along rivers. If there had been arctic foxes, the island birds would have had a safer location than the birds on the mainland, although most of the islands would have been easily accessible to foxes over the still frozen sea during the first two weeks of incubation. The developing vegetation in the river valleys and on the outer islands seemed to be much more attractive to geese than the vegetation on the inner islands, where nest density was highest (Fig. 8). The fact that the majority of the geese did not choose to breed in the best feeding-areas may be the result of a difference i n potential predation pressure and/or difference in availability of the nest-sites soon after the arrival of the geese. The melting snow in spring may flood river-valleys and the low-lying outer islands to a large extent in the second half of June. Because of the very short summer season at this latitude (74"N), delayed nesting causes problems at the end of the season when goslings must be able to fly before changing weather conditions make feeding impossible. Due to large territories around the nest where the females feed during incubation, nest density is low (maximum 3.9 nests/ha) compared for instance to the barnacle geese Branta leucopsis on Svalbard where nest density on breeding islands has been reported t o be 84.7 nests/ha on Breeding biology of dark-bellied brent geese 127 sen et al. (1989): four female light-bellied brents deserted their nests 19 days after the start of incubation within 32 hours after a snow storm. Predation of eggs is more likely to occur when the recesses are long (Inglis 1977; Prop et al. 1984; Raveling 1989; Thompson & Raveling 1987). We never observed that a male sat on the nest while the female was feeding as is described for whis- tling swans Cygnus c . columbianus (Hawkins 1986). The preference of the females to feed during mid-day, the warmest part of the day (Fig. 13), may be determined by the fact that the eggs will cool less at that time when the nest is left for longer periods. This diurnal rhythm has also been found in light-bellied brent geese (Madsen et al. 1989) and in other goose species (Prop et al. 1980; Aldrich & Raveling 1983; Thompson & Raveling 1987). The abundant presence of down in the nests of brents clearly is an adaptation to their low nest-attentiveness. Thompson & Raveling (1988) showed that eggs in black brant nests cooled less than those in the nests of the more attentive emperor and cackling goose. Some males spent a considerable amount of time in defending their territory against neigh- bouring males (Fig. 15). As the territory of male B was probably not attractive for the neighbours, male B had less conflicts than male A . The decrease in the number of conflicts between the males in the week before hatching could be the result of an improving food situation at that time; intrusion pressure decreased and defending every square metre of the territory became less impor- tant. Spring was early in 1990 (Fig. 2) and the decreasing time spent on feeding by the males and non-breeders during the incubation time indicates that the food situation was improving during the course of the incubation period (Fig. 12). The difference in male feeding time between the island and the Lidia colony might result from the poorer food situation on the island. Because the males of breeding pairs had to defend and guard their families after hatching, it was necessary that they had reserves for this period. This might explain the difference in feeding time between breeding and non-breeding birds. The average feeding per- centage of non-breeders observed by Madsen et al. (1989) during the incubation period was 53.3%; in the Lidia situation it was only 28.0%. The overall picture of decreasing feeding time, resulting in low feeding percentages (Fig. 12), will average (Prop et al. 1984). Here the geese breed in colonies on islands and fly to the mainland for feeding. Madsen et al. (1989), however, found a breeding density of only 0.5 nests/ha for light- bellied brent geese Branta bernicla hrota on the Tusendyane islands, Svalbard. The time a female spends off her nest will depend on her body reserves and the quality and quantity of the food she is able to gather during the feeding recesses. Concerning the food, the river-valley birds had a better local food supply than the island birds. This might explain part of the differences in recess time (Fig. 10). On the island female A fed in her own territory whereas female B , who had a poorer territory (Fig. 9), in most cases flew to the mainland or other adjacent islands to feed, resulting in longer nest recesses (27 minutes). This illustrates the importance of a territory with sufficient food. Madsen et al. (1989) found for the light-bellied brents on Svalbard an average recess frequency of 6.7 per 24 hours and a mean recess length of 19.2 min. This means that on Svalbard the females spent on average less time off their nest (129 min per day) than the dark-bellied brents in our situation (10.6 recesses of 15.5 rnin per day = 165 min). This difference can be the result of a difference in digestibility of the food. Madsen et al. report that the food of the brent on Svalbard partly consists of mosses. We seldom saw the geese eating mosses in our area. The nest attentiveness of the female black brant Branta bernicla nigricans reported by Thompson & Raveling (1987) is about the same as for the dark-bellied but the feeding time per day is shorter, caused by the low percentage of feeding during recesses (6496, based on 3 recesses). The longest feeding time (239 minutes per day) is reported for the light-bellied brents in Canada (Afton & Paulus 1993). The mean daily feeding time for our island female B however was 238 minutes. In the latter case only one out of the three eggs hatched so 238 minutes might be too long but the example shows that such a long feeding time was also possible in our situation. Brent geese are less attentive to their nests than other goose species (Prop et al. 1984; Afton & Paulus 1993). The females apparently do not have sufficient body reserves to fast during incubation. The dependence on local food during incubation makes brent geese very vulnerable for bad weather conditions during this period (Barry 1962). The effect of not being able to feed suf- ficiently during incubation was observed by Mad- 128 B . Spaans et al. be indicative for a favourable food situation in the course of the season. During spring staging it is not unusual that brent geese spend over 89% of the daytime on feeding (Prop & Deerenberg 1991), this means about 16 hours feeding per day. In order to achieve the same feeding time in the 24 hours daylight in Taimyr, on average 67% of the time should be spent feeding. Because of the absence of arctic foxes, lesser black-backed gulls and skuas were the only poten- tial predators. Madsen et al. (1989) observed a heavy egg-predation by arctic skuas on Svalbard, but in 92% of the cases predation took place during disturbance of the brent goose colony by polar bears Ursus maritimus. Although egg and gosling predation by avian predators was also a constant danger for the geese in Taimyr, they were well able to defend their clutch and young goslings. Therefore predation must be considered as a factor of only minor importance in 1990 in our situation. The observations of the two marked pairs, each with one full-grown juvenile in western Europe while both pairs had only one gosling after hatching in Taimyr, are an indication that gosling surviving was indeed high in 1990. We observed a large number of non-breeders in our study area (1200). These birds started moulting already during the hatching period. Because we never observed complete nest failure, we suspect that most of these birds did not even attempt to breed. This means that 69% of the local population failed to produce eggs. Combining the Taimyr and western Europe data The estimated dark-bellied brent geese popu- lation size in January 1990 was 220,000 birds (St. Joseph 1991). Because virtually no dark-bellied brent nested successfully in 1989, there were no yearlings in the population in 1990. Therefore 100% of the birds of the population were theor- etically capable of breeding. This means that there were 110,OOO potential breeding pairs in the sum- mer of 1990, mortality during spring and spring- migration not taken into account. In autumn 1990 the percentage of juveniles was 20% and the mean number of juveniles per pair was 2.4. This means that in every 100 brents there were 100 - 20 = 80 adults (or 40 pairs) of which 20/2.4 = 8.33 were successful pairs. So the percentage successful parents was 8.3/40 = 20.8%. The total number of pairs that brought one or more juveniles to the wintering-grounds was 0.208 x 110,000 = about 23,000 pairs. There has been discussion about the occurrence of a density-dependent reduction in the pro- portion of successful breeding pairs in good breed- ing seasons (30% juveniles or more) (Ebbinge 1985; Summers & Underhill 1991), but 23,000 successful pairs comprise only a fraction of the predicted number of successful pairs in a good breeding season (according to Summers & Under- hill, without density-dependent reduction, it is 44% of the predicted number, and according to Ebbinge it is about 60% when density-dependent reduction occurs). Thus relatively few pairs man- aged to raise juveniles successfully in 1990. In Taimyr snowmelt was early (Fig. 4), the mean temperature in June and July was high (Fig. 2), arctic foxes were absent, predation by birds (gulls and skuas) was probably of little importance and the food-rich riverbanks could easily hold more breeding pairs. If predation and/or weather- conditions would be the most important factors determining the breeding success (Barry 1962; Roselaar 1979; Summers 1986; Boyd 1987; Dhondt 1987; Greenwood 1987; Owen 1987). 1990 should have been a very good breeding year. The fact that population-breeding success was only moderate indicates that another factor played an important role in 1990. It seems plaus- ible that this was the poor pre-breeding condition of the geese (Fig. 5). Although mean condition is poor, there are always, due to the variation in the body-mass, individuals with a condition better than average (Table 1). If only females in better condition have sufficient reserves to start breed- ing, we can explain the intermediate breeding success on population scale. Under this hypoth- esis the intermediate breeding success is not the result of many breeding pairs that were mod- erately successful but of a relatively small number successful breeding pairs and many pairs that did not lay eggs at all. In that case breeding birds were pairs in relatively good condition. The fact that the local breeding birds in our study seemed to have been successful is in agreement with this explanation. Also, the observations of the marked geese with known pre-breeding weight in western Europe, where the heaviest female (1610 g) had five, an intermediate female (1560 g) had one, and the lightest female (1380 g) had no juveniles, correspond with this hypothesis. Acknowledgements. - We are very grateful to, Y . Y. Syroech- kovskiy, Institute for Evolutionary Animal Morphology and Breeding biology of dark-bellied brent geese 129 Ecology in Moscow. who organized and gave us the opportunity to participate in the international exhibition to the Taimyr Peninsula. N. Vronski. A. Abolitz, G. Boere, P. Prokosch and J . Veen played important roles in the organisation of the expedition and co-operation with the Russians. J . Koenes and his carpenters. Zoological Laboratory of the University of Groningen. prepared a prefabricated observation hut suitable for arctic conditions. We are also grateful to the many people who assisted in collecting data presented in this paper. The catching of geese in spring on Terschelling has been carried out by J . Burgers and G. Miiskens. In Taimyr. V. Grabowski, Y. Kokorev, A. Ivanov-Smolenski and A. Moroz helped with the observations. M. Ryhlikova collected data on the occurrence of lemmings. H. Schmauder was important as an interpreter and A. Grosheva prepared excellent meals. The figures hdVC been prepared by D. Visser. Comments by R . H. Drent. J. Tinbergen and two anonymous referees greatly improved this paper. References Afton. A . D. & Paulus. S. L. 1993: Incubation and Brood Care. Pp 62-108 in Batt. B. D. J . , Afton, A. D . . Anderson. M. G.. Ankncy. C. D.. Johnson, D. H . , Kadlec. J. A . & Krapu, G. L. (eds.): Ecology and Management of Brccding Water- fowl. Univ. of Minnesota Press. Minneapolis. Aldrich. W. & Raveling. D. G. 1983: Effects of experience and body weight on incubation behavior of Canada Geese. Auk 100. 670-679. Anthony. R. M.. Flint. P. L. & Sedinger, J. S. 1991: Arctic fox removal improves nest succcss of Black Brant. Wildl. SOC. Bull. 19. 176184. Barry. T . W. 1962: Effect of late seasons on Atlantic Brant reproduction. J , Wildl. Manage. 26.. 1+26. Boyd, H. 1987: D o June temperatures affect the breeding success of Dark-bellied Brent Geese? Bird Study 34, 11s 159. Cramp, S . & Simmons, K. E. L. (eds) 1983: Handbook of the Birds of Europe. the Middle East and North Africa, Vol 3. Oxford University Press. Oxford. Dhont, A. A . 1987: Cyclcs of lemmings and Brent Geese Branfa 6. bernicla: a comment on the hypothesis of Roselaar and Summers. Bird Study 34. 151-154. Ebbinge. B. S . 1985: Factors determining the population size of Arctic-breeding gcese. wintering in western Europe. Ardea 73, 121-128. Ebbingc. B. S . 1989: A multifactorial explanation for variation in breeding performance of Brent Geese. Ibis 131, 196204. Ebhinge, B. S. 1990: Reply by Barwold S . Ebbinge. fbis 132, 48 1-482. Ebbinge, B. S. & Spaans, B. 1992: The importance of body- reserves accumulated in spring staging areas in the temperate zone for breeding of Dark-bellied Brent Geese Branfa b . bernicla in the high arctic. In: Ebbinge. B. S. 1992: Population limitation in Arctic-Breeding Geese. Ph.D. dissertation. Uni- versity of Groningen. Pp 144-155. Ebbinge, B., St.Joseph, A., Prokosch, P. & Spaans, B. 1982: The importance of spring staging areas for arctic-breeding geese. wintering in western Europe. Aquila 89. 249-258. Ely, C. R. & Raveling, D. G . 1984: Breeding biology of Pacific White-fronted Geese. J. Wildl. Manage. 48(3). 823-837. Fox, A. D. & Madsen, J . 1981: The pre-nesting behaviour of the Greenland White-fronted Goose. Wildfowl 32, 48-54. Gauthier, G. & Choinikre. L. 1990: Energy reserves and repro- ductive performance in Greater Snow Geese. Acta XX Con- gressus internationalis Ornithologici (Supplement). P. 425. Greenwood, J . J. D . 1987: Three-year cycles of lemmings and Arctic geese explained. Nafure 328. 577. Hawkins, L. L. 1986: Nesting behaviour of male and female Whistling Swans and implications of male incubation. Wild- fowl 37, 5-27. Inglis, I . R. 1977: The breeding behaviour of the Pink-footed Goose: behavioral correlates of nesting success. Anim. Behao. 27, 747-764. Kistchinski. A. A. & Vronski, N. V. 1979: Migrations of Brant. In: Migrations of birds of Eastern Europe and Northern Asia, Ciconiiformes-Anseriformes. Publishing house Nauka. Moscow. Pp 188-194. Kumari. E. E. 1979: Status of Branfa bernicla bernicla on passage in Baltic USSR. In Smart, M. (ed): Procecdings of the First Technical Meeting on Western Palearctic Migratory Bird Management, Paris 1977. Slimbridge: International Waterfowl Research Bureau. Lambeck, R. H. D. 1990a: The applicability of age ratio and brood size counts in population dynamic studies of the Brent Goose Branfa 6. bernicla. Ardea 78(3). 414-425. Lambeck. R. H . D. 199ob: Differences in migratory pattern and habitat choice between social classes of the Brent Goose Branfa 6. bernicla. Ardea 78(3). 426440. Lampio, T . 1979: Status of Branfa bernicla bernicla in Finland. In Smart, M. (ed). Proceedingsof the First Technical Meeting on Western Palearctic Migratory Bird Management, Paris 1977. Slimbridge: International Waterfowl Research Bureau. Madsen. J.. Bregnballe, T . & Mehlum, F. 1989: Study of the breeding ecology and behaviour of the Svalhard population of Light-bellied Brent Goose Branta bernicla hrofa. Polar Res. 7, 1-21. Owen. M. 1987: Brent Goose Branla 6 . bernicla breeding and lemmings - a re-examination. Bird Study 34. 147-149. Prop, J . & Deerenberg. C. 1991: Spring staging in Brent Geese Branfa bernicla: feeding constraints and the impact of diet on the accumulation of body reserves. Oecologia 87, 1S28. Prop, J., van Eerden. M. R.. Daan. S.. Drent. R. H.. Tinber- gen, J. M.. St.Joseph. A. M. 1980: Ecology of the Barnacle Goose Branfa leucopsis during the breeding season: pre- liminary results from expeditions t o Spitsbergen in 1977 and 1978. Pp. 50-1 12 in Proc. Norwegian-Netherlands Symp. Svalbard. Arctic Centre. Groningen. Prop, J., van Eerden, M. R . & Drent, R. H. 1984: Reproductive success of Barnacle Goose Branfa leucopsis in relation to food exploitation on the breeding grounds. western Spits- bergen. Nor. Polarinst. Skr. 181, 87-117. Raveling, D. G. 1978: The timing of egg laying by northern geese. Auk 95,294-303. Raveling, D. G. 1989: Nest-predation rates in relation to colony size of Black Brant. J. Wildl. Manage. 53(1), 87-90. Roselaar, C. S. 1979: Fluetuaties in aantallen Krombek- strandlopers Calidris ferruginea. Wateruogels 4 , 202-210. Ryhlikova, M. E. 1993: Lemmings in the West Taimyr Arctic: Population density, distribution of territories and relation with other animals. Bericht iiber die Deutseh-Russischen Biologen-Expeditionen naeh Taimyr in den Jahren 1989.1990 und 1991. WWF-Deutsehland, Husum. St.Joseph, A. K. M. 1982: Review of the status of Branfa bernicla bernicla. In Scott, D. A. & Smart, M. (eds). Pro- ceedings of the Second Technical Meeting on Western Pale- arctic Migratory Bird Management, Paris 1979. Slimbridge: International Waterfowl Research Bureau. 130 B . Spaans et al. St.Joseph, A . K. M. 1991: Status of the Brent Goose Branta bernicla, 198689, Dark-bellied Brent. IWRB Goose Research Group Bulletin. 1 . 4-5. Summers, R. W. 1986: Breeding production of Dark-bellied Brent Geese Branta b . bernicla in relation to lemming cycles. Bird Study 33, 105-108. Summers, R . W. & Underhill, L. G. 1987: Factors related to breeding production of Brent Geese and waders on the Tai- myr Peninsula. Bird Study 34, 161-171. Summers, R. W. & Underhill, L. G . 1990: Multivariate analyses of breeding performance in Dark-bellied Brent Geese Branra b. bernicla; comment. Ibis 132, 477-481. Summers, R. W. & Underhill, L. G . 1991: The growth of the population of Dark-bellied Brent Geese Branra b . bernicla between 1955 and 1988.1. Appl. Ecol. 28, 584-585. Thompson, S . C. & Raveling, D. G . 1987: Incubation behavior of Emperor Geese compared with other geese: interactions of predation. body size. and energetics. Auk l04. 707-716. Thompson, S. C. & Raveling. D . G. 1988: Nest insulation and incubation constancy of arctic geese. Wildfowl3Y, 124-132.