Intra-seasonal variation in foraging behavior among Ad��lie penguins (Pygocelis adeliae) breeding at Cape Hallett, Ross Sea, Antarctica

We investigated intra-seasonal variation in foraging behavior of chick-rearing Adélie penguins, Pygoscelis adeliae, during two consecutive summers at Cape Hallett, northwestern Ross Sea. Although foraging behavior of this species has been extensively studied throughout the broad continental shelf region of the Ross Sea, this is the first study to report foraging behaviors and habitat affiliations among birds occupying continental slope waters. Continental slope habitat supports the greatest abundances of this species throughout its range, but we lack information about how intra-specific competition for prey might affect foraging and at-sea distribution and how these attributes compare with previous Ross Sea studies. Foraging trips increased in both distance and duration as breeding advanced from guard to crèche stage, but foraging dive depth, dive rates, and vertical dive distances travelled per hour decreased. Consistent with previous studies within slope habitats elsewhere in Antarctic waters, Antarctic krill (Euphausia superba) dominated chick meal composition, but fish increased four-fold from guard to crèche stages. Foraging-, focal-, and core areas all doubled during the crèche stage as individuals shifted distribution in a southeasterly direction away from the coast while simultaneously becoming more widely dispersed (i.e., less spatial overlap among individuals). Intra-specific competition for prey among Adélie penguins appears to influence foraging behavior of this species, even in food webs dominated by Antarctic krill.     

Foraging movements of emperor penguins at Pointe Géologie, Antarctica

The foraging distributions of 20 breeding emperor penguins were investigated at Pointe Géologie, Terre Adélie, Antarctica by using satellite telemetry in 2005 and 2006 during early and late winter, as well as during late spring and summer, corresponding to incubation, early chick-brooding, late chick-rearing and the adult pre-moult period, respectively. Dive depth records of three post-egg-laying females, two post-incubating males and four late chick-rearing adults were examined, as well as the horizontal space use by these birds. Foraging ranges of chick-provisioning penguins extended over the Antarctic shelf and were constricted by winter pack-ice. During spring ice break-up, the foraging ranges rarely exceeded the shelf slope, although seawater access was apparently almost unlimited. Winter females appeared constrained in their access to open water but used fissures in the sea ice and expanded their prey search effort by expanding the horizontal search component underwater. Birds in spring however, showed higher area-restricted-search than did birds in winter. Despite different seasonal foraging strategies, chick-rearing penguins exploited similar areas as indicated by both a high ‘Area-Restricted-Search Index’ and high ‘Catch Per Unit Effort’. During pre-moult trips, emperor penguins ranged much farther offshore than breeding birds, which argues for particularly profitable oceanic feeding areas which can be exploited when the time constraints imposed by having to return to a central place to provision the chick no longer apply.     

On a wing and a prayer: the foraging ecology of breeding Cape cormorants

The Cape cormorant Phalacrocorax capensis is unusual among cormorants in using aerial searching to locate patchily distributed pelagic schooling fish. It feeds up to 80 km offshore, often roosts at sea during the day and retains more air in its plumage and is more buoyant than most other cormorants. Despite these adaptations to its pelagic lifestyle, little is known of its foraging ecology. We measured the activity budget and diving ecology of breeding Cape cormorants. All foraging took place during the day, with 3.6 ± 1.3 foraging trips per day, each lasting 85 ± 60 min and comprising 61 ± 53 dives. Dives lasted 21.2 ± 13.9 s (maximum 70 s), attaining an average depth of 10.2 ± 6.7 m (maximum 34 m), but variability in dive depth both within and between foraging trips was considerable. The within-bout variation in dive depth was greater when making shallow dives, suggesting that pelagic prey were targeted mainly when diving to <10 m. Diving ecology and total foraging time were similar to other cormorants, but the time spent flying (122 ± 51 min day⁻¹, 14% of daylight) was greater and more variable than other species. Searching flights lasted up to 1 h, and birds made numerous short flights during foraging bouts, presumably following fast-moving schools of pelagic prey. Compared with the other main seabird predators of pelagic fish in the Benguela region, Cape gannets Morus capensis and African penguins Spheniscus demersus , Cape cormorants made shorter, more frequent foraging trips. Their foraging range while feeding small chicks was 7 ± 6 km (maximum 40 km), similar to penguins (10–20 km), but less than gannets (50–200 km). Successful breeding by large colonies depends on the reliable occurrence of pelagic fish schools within this foraging range.     

Foraging ecology of Gentoo Penguins Pygoscelis papua at Macquarie Island during the period of chick care

The diet, diving behaviour, swimming velocity and foraging range of Gentoo Penguins Pygoscelis papua were studied at Macquarie Island during the breeding season in the 1993–1994 austral summer. Gentoo Penguins are considered to be inshore feeders, and at Macquarie Island the diet and estimated foraging ranges supported this. The diet consisted of 91.6% fish and 8.3% squid, by mass. The dominant prey taxa were the fish Gymnoscopelus sp. and Paranotothenia magellanica. A mixture of pelagic and benthic prey was consumed, with a greater proportion of benthic species occurring later in the season. The penguins exhibited a strong diurnal pattern in their diving behaviour. Deep diving (≥30 m) began near sunrise (03.00 h) and finished close to sunset (21.00 h). Diving at night was less common and very shallow (<10 m). Early in the breeding season, dive profiles indicated that birds were probably following vertically migrating pelagic prey through the water column and were foraging in waters over 100 m deep. Later in the season, more uniform, shallower depths were used, suggesting an increase in benthic foraging activity. These changes in dive pattern and depth were consistent with the habitat preferences of prey species found in the diet. Gentoo Penguins swam at 1.04 m per s and had a maximum potential foraging range of about 26 km for single-day trips. They tended to forage within 14 km of the colony, with a mean range of 5.4 km. This range encompassed the deep ocean habitat to the west and east of the island and a shallow area to the north.     

Satellite tracking of light-mantled sooty albatrosses

Five light-mantled sooty albatrosses (Phoebetria palpebrata) breeding at Macquarie Island were tracked with miniaturised satellite transmitters during foraging trips of the incubation period. Birds moved rapidly to specific sectors of the Southern Ocean, where they spent several days foraging before returning to their nests. These specific sectors were at an average distance of 1516 km from Macquarie Island and located in pelagic Antarctic waters, mostly along the Antarctic continent. The maximum foraging range was in average 1721 km and the total distance covered by two birds for which there were complete tracks was 6463 and 6975 km. This study confirms previous suggestions that light-mantled sooty albatrosses are able to forage in the waters of the high Antarctic while breeding in the sub-Antarctic. The implications of the extreme separation of feeding zones from nesting grounds, in terms of conservation and life-history strategies, are discussed.     

What is out there: diversity in feeding of gentoo penguins (Pygoscelis papua) around the Falkland Islands (Southwest Atlantic)

During austral spring 2000, the diet of Gentoo penguins, Pygoscelis papua, was studied and compared to prey availability, assessed by trawl and plankton surveys, in waters adjacent to a number of colonies along the coast of West Falkland. There was good agreement in size and abundance of key prey species in the penguin diet and from the plankton and trawl surveys. The diet of birds breeding adjacent to shallow sheltered waters was dominated by demersal species, in particular crustaceans, Munida spp. whilst birds breeding adjacent to deeper exposed waters generally preyed upon pelagic species of fish, Falkland herring Sprattus fuegensis and squid, Moroteuthis ingens. Gentoo penguins generally foraged at depths of up to 30 m, relatively close to the coast, and illustrated a certain degree of selection of prey species where possible, thus providing evidence that they are not fully opportunistic predators.     

Diet and prey consumption of Antarctic petrels Thalassoica antarctica at Svarthamaren, Dronning Maud Land, and at sea outside the colony

The diet of the Antarctic petrel Thalassoica antarctica was studied during two seasons at Svarthamaren, an inland colony in Dronning Maud Land, Antarctica, and in the pack ice off the coast of Svarthamaren. The most important food (wet mass) at Svarthamaren was crustaceans (67%), fish (29%) and squid (5%); however, individuals collected in the pack ice took mostly fish (87%). The prey composition and lengths of prey are comparable to what has been documented in other studies on this species. Estimates of food consumption by birds breeding at Svarthamaren (ca. 250,000 pairs) suggest that approximately 6500 tonnes of crustaceans, 2800 tonnes of fish and 435 tonnes of squid are consumed during the breeding season. The annual consumptions of these birds are estimated to be 34,100 tonnes of crustaceans, 14,700 tonnes of fish, and 2300 tonnes of squid. Satellite telemetry data indicate that Antarctic petrels from Svarthamaren may fly more than 3000 km during one foraging trip, and thus may cover a huge ocean area to obtain their prey. Received: 1 September 1997 / Accepted: 3 February 1998     

Species‐specific foraging strategies and segregation mechanisms of sympatric Antarctic fulmarine petrels throughout the annual cycle

Determining the year‐round distribution and behaviour of birds is necessary for a better understanding of their ecology and foraging strategies. Petrels form an important component of the high‐latitude seabird assemblages in terms of species and individuals. The distribution and foraging ecology of three sympatric fulmarine petrels (Southern Fulmar Fulmarus glacialoides, Cape Petrel Daption capense and Snow Petrel Pagodroma nivea) were studied at Adélie Land, East Antarctica, by combining information from miniaturized saltwater immersion geolocators and stable isotopes from feathers. During the breeding season at a large spatial scale (c. 200 km), the three species overlapped in their foraging areas located in the vicinity of the colonies but were segregated by their diet and trophic level, as indicated by the different chick δ¹⁵N values that increased in the order Cape Petrel < Southern Fulmar < Snow Petrel. During the non‐breeding season, the three fulmarines showed species‐specific migration strategies along a wide latitudinal gradient. Snow Petrels largely remained in ice‐associated Antarctic waters, Southern Fulmars targeted primarily the sub‐Antarctic zone and Cape Petrels migrated further north. Overall, birds spent less time in flight during the non‐breeding period than during the breeding season, with the highest percentage of time spent sitting on the water occurring during the breeding season and at the beginning of the non‐breeding period before migration. This activity pattern, together with the δ¹³C values of most feathers, strongly suggests that moult of the three fulmarine petrels occurred at that time in the very productive high Antarctic waters, where birds fed on a combination of crustaceans and fish. The study highlights different segregating mechanisms that allow the coexistence of closely related species, specifically, prey partitioning during the breeding season and spatial segregation at sea during the non‐breeding season.     

Activity Time Budget during Foraging Trips of Emperor Penguins

We developed an automated method using depth and one axis of body acceleration data recorded by animal-borne data loggers to identify activities of penguins over long-term deployments. Using this technique, we evaluated the activity time budget of emperor penguins (n = 10) both in water and on sea ice during foraging trips in chick-rearing season. During the foraging trips, emperor penguins alternated dive bouts (4.8±4.5 h) and rest periods on sea ice (2.5±2.3 h). After recorder deployment and release near the colony, the birds spent 17.9±8.4% of their time traveling until they reached the ice edge. Once at the ice edge, they stayed there more than 4 hours before the first dive. After the first dive, the mean proportions of time spent on the ice and in water were 30.8±7.4% and 69.2±7.4%, respectively. When in the water, they spent 67.9±3.1% of time making dives deeper than 5 m. Dive activity had no typical diurnal pattern for individual birds. While in the water between dives, the birds had short resting periods (1.2±1.7 min) and periods of swimming at depths shallower than 5 m (0.25±0.38 min). When the birds were on the ice, they primarily used time for resting (90.3±4.1% of time) and spent only 9.7±4.1% of time traveling. Thus, it appears that, during foraging trips at sea, emperor penguins traveled during dives >5 m depth, and that sea ice was primarily used for resting. Sea ice probably provides refuge from natural predators such as leopard seals. We also suggest that 24 hours of sunlight and the cycling of dive bouts with short rest periods on sea ice allow emperor penguins to dive continuously throughout the day during foraging trips to sea.     

Divergent responses of Pygoscelis penguins reveal a common environmental driver

The responses of predators to environmental variability in the Antarctic Peninsula region have exhibited divergent patterns owing to variation in the geographic settings of colonies and predator life-history strategies. Five breeding colonies of Pygoscelis penguins from King George Island and Livingston Island, South Shetland Islands, Antarctica, were examined to (1) compare the responses of sympatric congeners to recent changes in their Antarctic ecosystem and (2) assess underlying causes for such responses. We used linear regression and correlation analyses to compare indices of abundance, recruitment, and summer breeding performance of the Adélie (P. adeliae), gentoo (P. papua), and chinstrap penguins (P. antarctica). Breeding colonies of Adélie and chinstrap penguins have declined by roughly 50% since the mid-1970s, and recruitment indices of Adélie penguins have declined by roughly 80%, but no such patterns are evident for gentoo penguins. Fledging success, however, has remained stable at all breeding colonies. The different trends in abundance and recruitment indices for each species, despite generally similar indices of summer performance, suggest that winter conditions contribute to the divergent responses among the penguins. In particular, strong correlations between indices of penguin and krill recruitment suggest that penguins in the South Shetland Islands may live under an increasingly krill-limited system that has disproportionate effects on the survival of juvenile birds.     

Finding food in the open ocean: foraging strategies in Humboldt penguins

Penguins are excellent “model” organisms allowing us to study the behaviour of marine homeotherms at sea. Penguins regularly return to their breeding colonies, enabling biologists to equip them with remote sensing devices such as physiological or behavioural data-loggers, radio- or satellite transmitters. Foraging trips at sea can last from days to weeks and after return of the birds to their breeding sites, the devices can easily be removed for analysis of on-board stored data, yielding a wealth of information. Investigation of penguin behaviour at sea becomes particularly revealing when other sources of information can be matched to the data set, such as satellite data on wind, temperature, ice cover, and chlorophyll-a concentrations.

Penguins and other marine homeotherms are true inhabitants of the high seas. Depending on the season, the marine behaviour varies: during reproduction, penguins are central-place foragers, and must return regularly to their nest to feed their chicks. During the remainder of the year, there are no constraints and the birds travel large distances at sea.

Breeding Humboldt penguins react to climatic change by varying their daily foraging range and dive duration. Similar to other representatives of the family Spheniscidae, Humboldt penguins avoid food shortages by migrating into more productive marine areas. Navigational clues such as daylength, sea surface temperature, local wind direction and olfaction might provide important aids in finding patchily distributed prey in the open ocean. DMS, a chemical compound produced by decaying algae, seems to be a further clue that indirectly points the way to feeding areas.     

Diving pattern and performance in relation to foraging ecology in the gentoo penguin, Pygoscelis papua

We present data on diving pattern and performance (dive depth, duration, frequency and organization during the foraging trip) in gentoo penguins Pygoscelis papua , obtained using time-depth recorders ( n = 9 birds, 99 foraging trips). These data are used to estimate various parameters of foraging activity, e.g. foraging range, prey capture rates, and are compared in relation to breeding chronology. Foraging trip duration was 6 h and 10 h, and trip frequency 1.0/day and 0.96/day, during the brooding and creche periods, respectively. Birds spent on average 52%of each foraging trip diving. Dive depth and duration were highly bimodal: shallow dives (< 21 m) averaged 4 m and 0.23 min, and deep dives (> 30 m) 80 m and 2.5 min, respectively. Birds spent on average 71%and 25%of total diving time in deep and shallow dives, respectively. For deep dives, dive duration exceeded the subsequent surface interval, but shallow dives were followed by surface intervals 2–3 times dive duration. We suggest that most shallow dives are searching/exploratory dives and most deep dives are feeding dives. Deep dives showed clear diel patterns averaging 40 m at dawn and dusk and 80–90 m at midday. Estimated foraging ranges were 2.3 km and 4.1 km during the brood and creche period, respectively. Foraging trip duration increased by 4 h between the brood and creche periods but total time spent in deep dives (i.e. time spent feeding) was the same (3 h). Of 99 foraging trips, 56%consisted of only one dive bout and 44%of 2–4 bouts delimited by extended surface intervals > 10 min. We suggest that this pattern of diving activity reflects variation in spatial distribution of prey rather than the effect of physiological constraints on diving ability.     

Satellite tracking of Adélie penguins

Summary Female Adélie penguins (Pygoscelis adeliae) that take too long on their first post-laying foraging trip are a major cause of breeding failure, but in the ice-filled waters of Antarctica, determining where they go and why they are away so long has proved difficult. Here we describe the first successful attempt to track penguins at sea using satellite telemetry. Four females foraged in different locations, dispelling the notion of a common feeding ground. They moved up to 272 km from the rookery and covered from 551 to 1,121 km on their trips, swimming at minimum average speeds around 1.2 m/s. The birds were most likely to be in the water between 0630 and 1430 when light intensity, important for a visual predator, was greatest. Carrying the transmitters reduced rates of fat deposition (weight gain), increasing the duration of foraging trips of females, and suggested that they may forage until their fat depots reach a minimum threshold level. This has two implications: (i) durations of these postlaying foraging trips could potentially be used as an indicator of krill abundance (Euphausia sp), the almost exclusive food of Adélie penguins during this period, and (ii) any reduction in krill stocks caused by harvesting could increase foraging trip durations with a concomitant increase mi breeding failures.     

Gene flow on the ice: genetic differentiation among Adélie penguin colonies around Antarctica

Each summer Adélie penguins breed in large disjunct colonies on ice-free areas around the Antarctic continent. Comprising > 10 million birds, this species represents a dominant feature of the Antarctic ecosystem. The patchy distribution within a large geographical range, natal philopatry and a probable history of refugia, suggest that this species is likely to exhibit significant genetic differentiation within and among colonies. We present data from seven microsatellite DNA loci for 442 individuals from 13 locations around the Antarctic continent. With the exception of one locus, there was no significant genic or genotypic heterogeneity across populations. Pairwise FST values were low with no value > 0.02. When all colonies were compared in a single analysis, the overall FST value was 0.0007. Moreover, assignment tests were relatively ineffective at correctly placing individuals into their respective collection sites. These data reveal a lack of genetic differentiation between Adélie penguin colonies around the Antarctic continent, despite substantial levels of genetic variation. We consider this homogeneity in terms of the dispersal of individuals among colonies and the size of breeding groups and discuss our results in terms of the glacial history of Antarctica.     

Satellite tracking of male rockhopper penguins Eudyptes chrysocome during the incubation period at the Falkland Islands

The foraging patterns of ten male rockhopper penguins Eudyptes chrysocome from the Falkland Islands were recorded during the incubation period by using satellite telemetry. Irrespective of study site and year, two different foraging areas could be identified. Three foraging trips were directed towards the slope of the Patagonian Shelf c. 140 km to the northeast of the breeding colony and these trips had a duration of 11–15 days. The seven other rockhopper penguins travelled c. 400 km towards the edge of the Falkland Islands' waters; all these foraging trips followed an anti-clockwise direction and lasted 16–27 days. The calculated mean daily travelling speed, based on the time spent underwater and the distance covered between two positions, was significantly higher in birds travelling to the edge of the Falkland Islands' waters compared with those foraging at the shelf slope (4.4±1.6 km/h vs. 3.4±2.0 km/h, respectively). The consistent foraging patterns exhibited during the long trips may be linked to the Falklands current, allowing the penguins to reach remote areas while reducing their energy expenditure. Potential interactions between commercial fisheries and hydrocarbon exploration are discussed.     

Group Size in Foraging African Penguins (Spheniscus demersus)

Diving synchrony was examined for varying group sizes of African penguins (Spheniscus demersus) travelling to their foraging grounds from their breeding islands. Groups of fewer than 12 birds always dived synchronously, whereas groups of more than 17 birds always dived asynchronously. Since travelling penguins do not dive deeply, large groups of birds can remain together irrespective of diving synchronization. Observations from boats showed that foraging penguins rarely occurred in groups of more than 17 birds. We suggest that groups of penguins that do not have synchronized dives cannot forage effectively, because foraging penguins dive deeply.     

Planktonic microbial assemblages and the potential effects of metazooplankton predation on the food web of lakes from the maritime Antarctica and sub-Antarctic islands

Antarctica is the continent with the harshest climate on the Earth. Antarctic lakes, however, usually presents liquid water, at least during part of the year or below the ice cover, especially those from the sub-Antarctic islands and the maritime Antarctic region where climatic conditions are less extreme. Planktonic communities in these lakes are mostly dominated by microorganisms, including bacteria and phototrophic and heterotrophic protists, and by metazooplankton, usually represented by rotifers and calanoid copepods, the latter mainly from the genus Boeckella. Here I report and discuss on studies performed during the last decade that show that there is a potential for top–down control of the structure of the planktonic microbial food web in sub-Antarctic and maritime Antarctic lakes. In some of the studied lakes, the effect of copepod grazing on protozoa, either ciliates or flagellates, depending on size of both the predator and the prey, could promote cascade effects that would be transmitted to the bacterioplankton assemblage.     

Open-ocean barriers to dispersal: a test case with the Antarctic Polar Front and the ribbon worm Parborlasia corrugatus (Nemertea: Lineidae)

Open-ocean environments provide few obvious barriers to the dispersal of marine organisms. Major currents and/or environmental gradients potentially impede gene flow. One system hypothesized to form an open-ocean dispersal barrier is the Antarctic Polar Front, an area characterized by marked temperature change, deep water, and the high-flow Antarctic Circumpolar current. Despite these potential isolating factors, several invertebrate species occur in both regions, including the broadcast-spawning nemertean worm Parborlasia corrugatus. To empirically test for the presence of an open-ocean dispersal barrier, we sampled P. corrugatus and other nemerteans from southern South America, Antarctica, and the sub-Antarctic islands. Diversity was assessed by analyzing mitochondrial 16S rRNA and cytochrome c oxidase subunit I sequence data with Bayesian inference and tcs haplotype network analysis. Appropriate neutrality tests were also employed. Although our results indicate a single well-mixed lineage in Antarctica and the sub-Antarctic, no evidence for recent gene flow was detected between this population and South American P. corrugatus. Thus, even though P. corrugatus can disperse over large geographical distances, physical oceanographic barriers (i.e. Antarctic Polar Front and Antarctic Circumpolar Current) between continents have likely restricted dispersal over evolutionary time. Genetic distances and haplotype network analysis between South American and Antarctic/sub-Antarctic P. corrugatus suggest that these two populations are possibly two cryptic species.     

The winter distribution of chinstrap penguins from two breeding sites in the South Shetland Islands of Antarctica

Satellite telemetry was used to determine the winter movements of eight chinstrap penguins from two breeding colonies in the South Shetland Islands, Antarctica during the 2000 and 2004 austral winters. Chinstrap penguins foraged largely inshore, on the shelf, north of the South Shetland Islands during 2000, but foraged mainly offshore in pelagic waters in 2004. Analyses of foraging trip durations during the chick-rearing periods that preceded the 2000 and 2004 winters suggest that prey were more abundant inshore during the summer of 2000 than in 2004. Oceanographic data further revealed the presence of a strong shelf-slope front in 2000 that was absent in 2004. In addition, two of the six chinstrap penguins from the colony in Admiralty Bay migrated from the South Shetland Islands region to the vicinity of the South Orkney and South Sandwich Islands, distances of 800 and 1,300 km, respectively. We postulate that the differences in winter migratory behavior among chinstrap penguins from this colony may reflect individual ties to different ancestral epicenters of chinstrap populations; one older and local in the South Shetland Islands and one relatively recent, arising from the emigration of chinstrap penguins that occurred during the expansion of this species in the mid-1900s.     

Long-term population trends in seven Antarctic seabirds at Pointe Géologie (Terre Adélie) Human impact compared with environmental change

Antarctic seabird populations have been much studied over the last decades as bioindicators of the nature of variability in the Southern Ocean marine ecosystem, and most attention has been focused on the role of food supply and the extent of sea ice. In addition, the rapid spread of tourism and the activities of researchers since the early 1960s have raised questions related to their real and potential impact on bird populations. Our data sets start in 1952 for several species of Antarctic seabirds and this study documents the trends over a 14-year period (1985–1999) in seven species breeding on Pointe Géologie archipelago (Terre Adélie, Antarctica). This is the first study where the direct impact of destruction of breeding sites (for building of an airstrip) is examined and where such long-term populations trends have been assessed in such a number of Antarctic species at one site. Trends from 1985 show that for the whole archipelago and when excluding islands destroyed, Adélie penguins and south polar skuas were the only species to show a significant increase (>3.5% annual change). The others species showed opposite trends, three increasing slightly (southern fulmars +0.4%, cape petrels +2.3%, snow petrels +0.9%) and two decreasing (emperor penguin −0.9%, southern giant petrel −3.9%). Three species particularly affected by the destruction of their breeding habitat (Adélie penguin, cape petrel, snow petrel) showed the capability to restore their populations. The availability of food and nesting sites is discussed in relation to environmental change. Species feeding on krill (Adélie penguins and cape petrels) increased more than other species; however, decrease of ice cover can increase availability of nesting sites. The importance of long-term studies is shown when assessing the role of human activities in Antarctica compared to larger-scale changes. Accepted: 18 September 2000