Introduction

Kongsfjorden, a glacial fjord on the northwest coast of Spitsbergen in the Svalbard archipelago, is strongly influenced by both Arctic and Atlantic water (AW) masses (Svendsen et al. 2002), and to a lesser extent glacial discharge (Maclachlan et al. 2007). The relative dominance of these water masses changes seasonally and inter-annually such that the shelf waters and adjacent fjords switch from a state of Arctic dominance to Atlantic dominance and back (Cottier et al. 2005; Svendsen et al. 2002). The fjord opens to the West Spitsbergen Shelf (WSS) through a common mouth it shares with the adjacent Krossfjorden (Fig. 1). A submarine glacial trough, Kongsfjordrenna, acts as a deep-water connection between the outer basin of Kongsfjorden and the shelf, so there is no topographic barrier to exchange between the shelf and the fjords (Svendsen et al. 2002). Therefore advection of water masses plays a significant role in determining the physical and biological characteristics of the fjord (Hop et al. 2002). Unlike other fjords in the Nordic Seas region, the zooplankton fauna in Kongsfjorden is relatively rich and comprised of co-occurring boreal and Arctic species, which respond to variations in the distribution and dynamics of the West Spitsbergen water masses (Kwaśniewski et al. 2003; Willis et al. 2006). Consequently, the composition of the zooplankton community in Kongsfjorden is a function of hydrographic processes and the origin of the water masses. Thus, the fjord is well suited to studying the effects of climate change on water mass distribution, sea ice conditions and ecosystem function in Arctic shelf ecosystems (Hop et al. 2006).

Fig. 1
figure 1

a The Svalbard archipelago with the Kongsfjorden-Krossfjorden system on the northwest coast of Spitsbergen highlighted by the box. The two major currents flowing north along the West Spitsbergen Shelf are the cold coastal current originating as the East Spitsbergen Current (ESC) and the West Spitsbergen Current (WSC) which carries warmer Atlantic water (AW). b The location of the mooring (cross symbol) in Kongsfjorden. The dark grey arrow in (b) shows the topographically steered currents (ArW and AW) originating from the shelf and circulating round the fjord. The mooring is located on the northern side of this circulation

Full size image

The WSS and adjacent coastal waters are influenced primarily by water masses of Arctic origin (Saloranta and Svendsen 2001), although there are often intrusions of warm AW crossing the Arctic Front from the West Spitsbergen Current onto the WSS and into the fjords. It is believed that the major intrusion of AW into the fjords occur primarily in the summer months, with a density front at the fjord entrance preventing winter intrusions (Cottier et al. 2005). However, during the winter of 2006, there was extensive advection of AW across the WSS and into the fjords of West Spitsbergen, which raised water temperatures by several degrees (Cottier et al. 2007).

The close relationship between water mass advection and changes in zooplankton community structure in Kongsfjorden during the summer months has been demonstrated using sediment traps (Willis et al. 2006). Although sequential sediment traps are not specifically designed to sample zooplankton they have been used to monitor zooplankton communities in regions where it is not practicable to use standard sampling methods (Hargrave et al. 1989, Forbes et al. 1992; Willis et al. 2006). With the exception of a study on Pseudocalanus and Oithona (Lischka and Hagen 2005), little is known about the winter zooplankton community in Kongsfjorden, so here we present a unique data set collected by a sediment trap in the outer basin of the fjord from September 2005 to May 2006. Crucially, it captures a distinct switch in the zooplankton community during the exceptionally warm and strongly advective conditions during the winter of 2006.

Methods

A sequential sediment trap (McLane Parflux 78H-21 with 21 cup carousel and 0.5 m2 aperture) was moored at a depth of 115 m on the north side of the outer basin of Kongsfjorden at 79° 01.2N 11° 46.4E (Fig. 1; bottom depth ∼210 m). The sampling period was from 16 September 2005 to 17 May 2006, with a sampling frequency of one month per bottle from 16 September to 15 March, and seven days thereafter. Swimmers from the seven day samples (16 March to 19 April and 20 April to 17 May) were pooled for comparison with the monthly trap samples. Prior to deployment, sediment trap sample bottles were filled with filtered seawater containing NaCl to provide a density discontinuity relative to ambient seawater, and 2% formalin buffered with sodium borate to preserve the deposited material. Temperature (T) and salinity (S) were recorded during the sediment trap deployment period with Seabird 37 Microcats positioned at depths of 33 and 103 m. The microcats were accurate to 0.01°C (T) and better than 0.01 (S).

The trap samples were passed through a 300 μm mesh sieve to remove the larger swimmers. Any smaller species that may have been present in the samples (such as Microcalanus, Oithona and juvenile Limacina helicina) may not have been routinely retained on the sieve. All animals retained on the sieve were intact and showed no signs of decomposition, suggesting that they entered the trap actively and were subsequently killed by the preservative. All zooplankton retained on the sieve were sorted and identified. Developmental stages of Calanus were identified to species based on morphology and prosome lengths of individual copepodid stages according to Kwaśniewski et al. (2003). Calanus prosome lengths were measured under a stereomicroscope equipped with a calibrated ocular micrometer. Remaining copepods were identified to the lowest possible taxon. Zooplankton species are presented as numbers per trap sample, with sampling intervals of one month, starting on the 16th day of each month.

Results

Copepod abundance was highest in September, with C. glacialis (CV), C. hyperboreus (CIV), C. finmarchicus (CV) and Paraeuchaeta norvegica (CV) dominating the trap sample (Fig. 2). Other species present in low numbers in September included the amphipods Melita palmata, Themisto abyssorum and T. libellula, the chaetognaths Sagitta elegans and Eukrohnia hamata, and the decapod Eualus gaimardii (Fig. 3). Small numbers of M. palmata were also present in the October–December trap samples, while the small boreal copepod Bradyidius similis peaked in abundance in December (Fig. 2). Overall species abundance was lowest in the October trap sample with only small numbers of M. palmata, C. glacialis, Metridia longa and P. norvegica present. The appendicularian Oikopleura cf. vanhoeffeni peaked in abundance in November. Later life stages (CV, CVI male and female) of the boreo-Arctic copepod M. longa dominated the samples from November to January, with a progressive increase in abundance over this period (Fig. 2). Peaks in abundance of later stages of C. finmarchicus, C. glacialis and the boreal P. norvegica occurred in February and March, corresponding with the influx of AW into the fjord (Fig. 4). A small peak in abundance of C. hyperboreus was also obvious in February. Abundance of C. finmarchicus and C. glacialis remained high in April, whilst P. norvegica was absent from the trap sample. The ostracod Discoconchoecia elegans peaked in abundance in January and February (Fig. 3). Other species occasionally present in the trap samples, albeit in very low numbers, included Heterorhabdus norvegicus, Gaetanus tenuispinus, Pseudocalanus spp., Microcalanus spp., Thysanoessa spp., Tomopteris helgolandicus, siphonophore fragments, and various polychaetes.

Fig. 2
figure 2

Abundance of copepod developmental stages in the sediment trap deployed at a depth of 115 m in Kongsfjorden between September 2005 and May 2006. Trap sampling periods are one month, beginning on the 16th day of each month. Developmental stages are copepodid stages CIII to CVI males (M) and females (F). The arrows in the top figures indicate the start of the AW influx. Note: y-axis scales are different

Full size image
Fig. 3
figure 3

Abundance of zooplankton swimmers in the sediment trap deployed at a depth of 115 m in Kongsfjorden between September 2005 and May 2006. Trap sampling periods are 1 month, beginning on the 16th day of each month. The arrows in the top figures indicate the start of the AW influx. Note: y-axis scales are different

Full size image
Fig. 4
figure 4

Time series of temperature and salinity recorded at the a upper (33 m) and b lower (103 m) microcats. A 3.5-day box filter has been applied to the data. Vertical broken lines indicate the sampling periods for each sediment trap bottle starting on 16 September 2005 and ending on 17 May 2006. The heavy broken line indicates the start of the AW influx and transition period in zooplankton species composition

Full size image

The hydrographic conditions at the location of the mooring were captured by the microcats (33 and 103 m) and the temperature and salinity time series are shown in Fig. 4. From September through to mid-January, the temperature at both microcats decreased from approximately 3 to −1°C, with small temperature fluctuations in early January at the lower microcat. Salinity at the upper microcat (33 m) decreased throughout September and then remained relatively stable at ∼34 from early November through to early February (Fig. 4a). At the lower microcat (103 m), salinity was initially around 34.8 reducing to 34.2 through December (Fig. 4b). From mid-January, as indicated by the heavy broken line in Fig. 4, both temperature and salinity increased at the lower microcat. A similar increase, beginning late in February, was also observed in the upper water column. At the end of February and in early March, both upper and lower microcats recorded water of ∼3°C and salinity 34.9, indicating the presence of AW in the fjord (Swift 1986). This was followed by a period of intense cooling of the entire water column during March and April through increased surface heat flux.

Discussion

There is an increasing body of information detailing changes in the physical environment of the Arctic in response to climate warming, whereas, there is little information available on associated changes in the biotic communities. The deployment of a sediment trap in Kongsfjorden has provided us with the first winter time series of the larger zooplankton species composition in the fjord. Furthermore, the winter of 2006 was unusual across the WSS in that there was extensive transport of AW across the shelf starting at the end of January. This inflow raised water temperatures by more than 3°C, reduced the extent and thickness of the sea ice cover, and transported zooplankton species onto the WSS and into Kongsfjorden, and presumably into other West Spitsbergen fjords.

The basis of the winter cross-shelf advection was an unusual pressure system over northern Europe bringing sustained southerly winds during December and January followed by a wind reversal and a sustained period of northerly winds. These meteorological conditions generated sustained up-welling at the shelf edge, leading to advection of AW across the WSS (Cottier et al. 2007). The effect was further enhanced by increases in northward transport of AW (Walczowski and Piechura 2007). The underlying cause of the meteorological conditions is beyond the scope of this paper; however, we do not attribute it to changes in the North Atlantic Oscillation index as the winter value for 2005–2006 does not show any strong anomaly.

Prior to the AW influx, the zooplankton sediment trap community was dominated by the boreo-Arctic copepod M. longa, with peaks of O. vanhoeffeni and D. elegans in November and January, respectively. There is no historical information on the community composition of the larger zooplankton species in Kongsfjorden during winter. However, the species composition in the trap samples was comparable to that in four sediment traps moored at similar depths on the shelf break of the Canadian Beaufort Sea during the winter of 1987–1988 (Forbes et al. 1992). As in the Beaufort Sea traps, calanoid copepods, and in particular M. longa females, dominated the Kongsfjorden trap samples. Unlike Calanus, M. longa remains active in the upper water column during the autumn and winter (Grønvik and Hopkins 1984; Båmstedt and Ervik 1984; Båmstedt et al. 1985), and its abundance progressively increased in the trap samples from October through to January, but then declined considerably following the AW influx in February. In two of the Beaufort Sea traps an overall trend of increasing M. longa abundance was apparent from November through to March (Forbes et al. 1992). Likewise, in Balsfjorden, Norway, M. longa increased in abundance throughout autumn and early winter, peaking in December (Barthel 1995).

From mid January 2006, there was a significant shift in species composition when the fjord’s resident zooplankton community was augmented by expatriate species transported into the fjord from the WSS in the AW influx. Conspicuous peaks in abundance of C. finmarchicus, C. glacialis, C. hyperboreus and P. norvegica were associated with this influx. Given the timing of the AW influx in February, we must consider the possibility that the abundance peaks observed in the trap samples were unrelated to it, and instead reflect increasing copepod migratory activity towards the end of the Arctic winter, and seasonal migrations. However we suggest that for the AW indicator species in particular, this is unlikely. The existence of regular synchronised diel vertical migration (DVM) throughout most of the winter months has been identified in data from an ADCP (300 kHz) deployed during winter in Kongsfjorden at the same location as the sediment trap (not during the sampling period reported here), although the species responsible are unknown. This would suggest that the February abundance peaks are unlikely to represent the onset of synchronised DVM following its cessation during the Arctic winter. Instead, given that DVM behaviour occurs through winter, we would expect the abundances of the migrating species to remain relatively stable in the sediment trap samples throughout the deployment period, which was obviously not the case.

The peak of P. norvegica in the trap samples was clearly associated with the AW influx as this copepod resides in deep water during the winter (Baliño and Aksnes 1993; Skarra and Kaartvedt 2003), below the sediment trap sampling depth. Further south in Oslofjord, P. norvegica females were reported to be at their deepest (100–200 m) in February and March by Skarra and Kaartvedt (2003), with little evidence of DVM. We do not know exactly when the Calanus species begin their seasonal ascent to the surface following diapause in Kongsfjorden. It is likely that C. glacialis ascends earlier in the season than C. finmarchicus as it reproduces well before the spring bloom. In Lurefjord (western Norway), 80% of the immature C. glacialis females were located in the top 100 m of the water column by the end of February (Niehoff and Hirche 2005), and Arnkværn et al. (2005) estimated that reproduction started in late February/early March in Billefjorden, Svalbard. Thus, the increased abundance of C. glacialis in the February sediment trap sample is likely to reflect the combined influences of seasonal migration and shelf water advection.

Reproduction occurs later in C. finmarchicus and tends to coincide with the spring phytoplankton bloom (Diel and Tande 1992). Low-rate pre-bloom spawning by C. finmarchicus has been observed in AW in the Barents and Norwegian Seas (Melle and Skjoldal 1998; Hirche et al. 2001), and may also occur in Billefjorden (Arnkværn et al. 2005), where females peaked in abundance at the end of March. Thus, we suggest that the C. finmarchicus in the February sediment trap sample are likely to be individuals from the WSS transported into the fjord in the AW influx. The small peak of C. hyperboreus in February is harder to explain, but it was probably also due to the advection of shelf waters into the fjord. This oceanic species over-winters at depths greater than 1,000 m in the West Spitsbergen Current (WSC) south of Svalbard (Hirche 1997), where it does not begin the spring ascent until April. The high abundance of Calanus in the September sediment trap sample may be related to their seasonal descent in late summer.

Species associations in Kongsfjorden respond to variations in the distribution and dynamics of the regional water masses, with the proportions of each varying seasonally and inter-annually depending on the timing and volume of Atlantic and Arctic Water intrusions (Kwaśniewski et al. 2003; Willis et al. 2006). Copepod biomass in the fjord is generally augmented during influxes of water masses from the WSS in late spring and summer. Consequently, it is unlikely that the spring phytoplankton bloom in April–May is heavily grazed under normal conditions in the fjord. Changing patterns of atmospheric circulation may enhance coastal up-welling and advection of water masses from the shelf edge into the west Spitsbergen fjords throughout the year. The effect would be a change in the zooplankton species composition and biomass on the shelf and in the fjords, with ensuing impacts on the marine pelagic foodweb and the vertical flux of biogenic matter.

For example, it has been proposed that the lipid rich Arctic Calanus species will decrease in abundance in an AW dominated system, with a consequent decline in birds such as the little auk (Alle alle), which feed on them (Karnovsky et al. 2003). Obviously, this is not yet the case in Kongsfjorden, as C. glacialis abundance increased in the sediment trap following the AW influx. C. glacialis is an Arctic shelf species (Conover 1988), hence its association with the shelf waters advected into the fjord. However, its distribution does appear to be correlated with temperature, as females and copepodite stage V tend to avoid water temperatures above 5°C (Kosobokova 1998; Niehoff and Hirche 2005). Similarly, Hirche (1997) questions whether C. hyperboreus is able to reproduce successfully in the warmer waters of the WSC. Therefore, in a future Arctic likely to be increasingly dominated by warmer AW, it is still possible that the Arctic Calanus species may decline in abundance. Finally, under ice free conditions such as those experienced during the winter of 2006, phytoplankton production will dominate over ice algae production, and grazing by zooplankton will reduce the export of primary production to the benthos, retaining it in the water column and driving the food web away from the benthic community (Carroll and Carroll 2003).