The absolute abundances of planktic foraminifera of the 100–250 µm size class in the upper 500 m of the water column show a general increase from ca. 10 individuals per cubic metre (ind./m³) at the eastern stations to 29 ind./m³ at the sea-ice margin (station 54). Further westward we can see a drastic decrease (3 and 9 ind./m³, stations 75 and 71, respectively; Fig. 1).

The species assemblage is dominated by polar Neogloboquadrina pachyderma (sin.) and subpolar Turborotalita quinqueloba, with proportions of around 76 and 15% in the east and 90 and 5% in the west, respectively. Other planktic foraminifera species found in the samples (N. pachyderma [dex.] and Globigerina bulloides) contribute less than 9% to the faunal composition. The relative abundance of N. pachyderma (sin.) varies between 73 and 92% with an increasing trend towards the west, while T. quinqueloba shows a reverse tendency with the highest value of 23% in the eastern part of the strait and the lowest (5%) at the Greenland slope. Figure 3 shows the species composition at the stations in preselected depth intervals. N. pachyderma (sin.) contributes less than 70% to the total planktic foraminiferal fauna at two stations (at 8°E and 6°E), between 0 and 50 m water depth. The highest proportion of this species was found at station 75 (4°W).

The two dominant species, N. pachyderma (sin.) and T. quinqueloba, show similar trends in the depth distribution: both have maximum absolute abundances (ind./m³) between 0 and 100 m water depths (Fig. 4). N. pachyderma (sin.) has highest abundances near the surface between 0 and 50 m at stations 35 and 87 in the middle of the strait (4°E and 3°E, respectively) and under the ice cover at station 75. Highest amounts of T. quinqueloba in the upper 50 m were found at stations 127, 35 and 44 (8°E, 4°E and 0°E, respectively); at the remaining stations, the depth of 50–100 m yielded maximum abundances of this species.

Looking at the relative abundances (percentage of total population of the given species at the given station per depth interval; Fig. 4), both species show a trend to increasing percentages from the surface to the interval 50–100 m, followed by a decrease to 300–500 m. Only at the easternmost station (127), the population maximum of N. pachyderma (sin.) descends below 200 m, while the highest relative abundance of T. quinqueloba at all stations lies above 200 m.

The samples do not show any evidence of selective carbonate dissolution. Only few broken shells were found, and the two studied species do not show any significant difference in the degree of fragmentation.

In the absolute abundances of planktic foraminifera, we see a trend similar to that of the plankton samples: the station close to Svalbard (25) has a low value (<400 ind./g sediment), followed by a minor peak at 4°E (ca. 6500 ind./g). After a steady interval (ca. 3000 ind./g), a significant peak can be observed at the sea-ice margin (station 54, ca. 26 000 ind./g), followed by a drastic decrease to ca. 50 ind./g below the sea ice (Fig. 5).

Neogloboquadrina pachyderma (sin.) has highest proportions in the assemblage at the sea-ice margin (91%) and lowest (73%) in the east (7°E, station 25). Between these two stations, a steady increase can be detected. T. quinqueloba shows an exactly reverse trend with abundances between 8 and 24% (Fig. 6). Comparing the relative abundances of N. pachyderma (sin.) and T. quinqueloba on the sediment surface to those counted in the upper 100 m of the water column, the species composition in the depth interval of 50–100 m follows the trend of the core-top samples more precisely than the depth interval between 0 and 50 m mirrors the sediment surface (Fig. 6). All count data can be found in Supplementary Tables S1–S3.

The planktic foraminiferal species composition of the study area resembles the typical foraminiferal fauna previously reported from Arctic/sub-Arctic regions with the clear dominance of Neogloboquadrina pachyderma (sin.) and Turborotalita quinqueloba. A similar species distribution has been already shown for the Arctic Ocean (Bé 1960; Carstens & Wefer 1992) and for Fram Strait (Carstens et al. 1997; Volkmann 2000). Furthermore, despite the warming of Atlantic waters passing through the strait (Spielhagen et al. 2011), the polar species N. pachyderma (sin.) still clearly prevails in our samples, compared to the subpolar T. quinqueloba. Interestingly, at the sampled stations, T. quinqueloba showed relatively low values compared to the results of the above-mentioned studies (Fig. 3). Carstens et al. (1997) observed a range from 15 to 64% for this species, while the proportion of T. quinqueloba in certain samples collected by Volkmann (2000) in warm, Atlantic Water-dominated regimes reached even 93%. In our plankton samples, the relative abundance of T. quinqueloba varied between 5 and 23% and the sediment surface samples showed a very similar trend. These discrepancies point to a variable foraminiferal production found due to different sampling periods. Carstens et al. (1997) and Volkmann (2000) took their samples in August, while our samples were collected in late June/early July and therefore probably reflect an early summer plankton bloom. In the eastern North Atlantic Ocean, Schiebel & Hemleben (2000) found distinctly different planktic foraminiferal faunas in June and in August. In contrast to early summer, the late summer assemblage was dominated by species preferring warmer waters than other species present, just like the subpolar T. quinqueloba in the planktic foraminiferal fauna of Fram Strait. It has been shown for the western North Atlantic that, while the production of N. pachyderma (sin.) has two peaks during the season and one already in early summer, the production of T. quinqueloba reaches its maximum only in early autumn, after the second bloom of N. pachyderma (sin.) (Jonkers et al. 2010). Although sediment trap studies in the Northeast Water Polynya revealed a unimodal pattern of planktic foraminiferal production (Kohfeld et al. 1996), our data indicate that a temporal offset in the production maxima of the two species exists also at these northern latitudes. This could lead to the relatively low numbers of T. quinqueloba in our samples, collected in June/July compared to the studies conducted in August.

The variable species composition along the transect reflects the complex hydrography of Fram Strait. In the east, the WSC carries warm, saline Atlantic Water northwards. In this area, the subpolar T. quinqueloba had higher abundances and the polar N. pachyderma (sin.) had lower abundances than in the western part where the upper 200 m of the EGC consist of cold, low-saline Arctic outflow waters. Here, N. pachyderma (sin.) had a proportion of around 90%. This indicates that the horizontal distribution of the two species is primarily controlled by the distribution of different water masses.

The ratio between absolute abundances of planktic foraminifera in the open ocean, at the ice margin and in the ice-covered ocean was found to be ca. 2:4:1 (Fig. 1). Our data show a similar trend to that in the study of Carstens et al. (1997). The ice margin offers increased food supply for the foraminifera with higher primary production caused by diatoms that are the major food source for N. pachyderma (sin.) (Hemleben et al. 1989). Diatoms typically do not develop well under permanent ice cover due to the light limitation, but they prefer seasonally stratified water conditions that occur in the marginal ice zone (Smith et al. 1987; Williams 1993). Ice melting increases stratification and consequently the stability of the water column, which triggers phytoplankton blooms (Alexander 1980), and the enhanced phytoplankton biomass along the ice margin may have led to a major increase in absolute abundances of planktic foraminifera.

Both N. pachyderma (sin.) and T. quinqueloba had maximum absolute abundances between 0 and 100 m water depth. However, the relative abundances per depth interval (Fig. 4) showed that in the western Fram Strait, at the ice-covered stations more specimens of N. pachyderma (sin.) lived close to the surface (above 100 m), while in the eastern part of the strait the majority of the population was found deeper, between 50 and 200 m. Simstich et al. (2003) analysed plankton and several core-top samples from the Nordic seas, taken mainly south of our transect (80°N–62°N). They reported on a shallow apparent calcification depth in T. quinqueloba over most of the studied profile. If we assume that foraminifera calcify in the water depth where they are most abundant, these results are consistent with our findings. In their study, going from west to the east, the habitat of N. pachyderma (sin.) seems to sink deeper in the Atlantic Water off Norway, similar to what we observe from our samples taken from the warm and saline WSC. These results contrast with the findings of Carstens et al. (1997), who concluded that N. pachyderma (sin.) and T. quinqueloba follow the path of the Atlantic Water submerging below Polar Water and therefore live deeper in the western part of the strait. During our sampling period, in the western Fram Strait, the surface Polar water masses extended down to 200 m. As none of the studies on planktic foraminifera conducted in the same area (including the study of Carstens et al. 1997) revealed an average depth habitat below 200 m, the scenario described by Carstens et al. (1997) seems rather unlikely here. Furthermore, previous investigations revealed that food availability strongly affects the distribution of planktic foraminifera (Schiebel & Hemleben 2005). Thus, the depth habitat of these protozoa might be controlled by the position of the chlorophyll a maximum (Fairbanks & Wiebe 1980; Kohfeld et al. 1996) rather than by the water mass distribution. From the summer months in the Northeast Water Polynya, with oceanographic conditions similar to the western Fram Strait, Kohfeld et al. (1996) reported maximum chlorophyll a concentrations and consequently highest abundances of N. pachyderma (sin.) between 20 and 80 m water depth, which accords with the results presented here. Hirche et al. (1991) also found, for the same period of the year in Fram Strait, chlorophyll a maxima in the open ocean at greater depth than under the ice coverage. Assuming that the depth distribution of N. pachyderma (sin.) is indeed predominantly controlled by food availability in this area, the deeper depth habitat reported by Carstens et al. (1997) could be explained by different sampling periods. It is possible that during the above-mentioned experiments in late summer, the ice coverage was not as intact as in June/July and consequently, due to the higher translucency, the deep chlorophyll maximum might have descended/extended to deeper layers. Planktic foraminifera respond to the redistribution of chlorophyll within several days (Schiebel et al. 2001) and may therefore have followed the descent of the chlorophyll maximum. Another possible explanation for the different findings might be the sampling in different ontogenetic stages. N. pachyderma (sin.) and T. quinqueloba are known to reproduce on a synodic lunar cycle (Bijma et al. 1990; Schiebel & Hemleben 2005) and for reproduction they simultaneously descend from their average depth habitat, possibly to the greatest depth they inhabit during their ontogenetic cycle (Berberich 1996; Schiebel & Hemleben 2005). Volkmann (2000) also assumed that N. pachyderma (sin.) prefers shallower water depths under permanent ice coverage. In her study, this species showed maximum abundances in the narrow temperature layer between −1.5 and −1.8°C, while T. quinqueloba did not show such preference to a particular temperature range. In our study, the maximum abundances of N. pachyderma (sin.) and T. quinqueloba were found between 0 and 100 m water depth at every station and therefore in a very wide temperature range (from −1.7 to 7°C; Fig. 2). We therefore assume that the depth habitat of planktic foraminifera in Fram Strait during June/July is more influenced by the distribution of the food source than by temperature.

Kozdon et al. (2009) proposed that the average depth habitat of N. pachyderma (sin.) in the Nordic seas is controlled by water density and that this species is bound to a layer between 27.7 and 27.8 (potential density, σ_t). Along our studied transect during the sampling period, this isopycnal layer was found mainly between 0 and 100 m water depth. There are two exceptions. The first are the stations at 8°E, 7°E and 4°E, where this water mass reaches deeper. The second is the western part of the strait (west of ca. 3°W), on the Greenland shelf, where the whole layer submerged below 150 m. Regarding the relative abundances of N. pachyderma (sin.), our data show that in the eastern and central Fram Strait (2°W–8°E) during the sampling period a broad range from 34 to 81% of the population lived in the depth intervals including the specific density range. Extending the range to 27.85, we found a better correlation with values between 65 and 86% (Fig. 7). In contrast, in the west, in the cold and fresh water masses of the EGC, the depth habitat of N. pachyderma (sin.) was found in significantly shallower water depths than the layer of σ_t=27.7–27.8. This discrepancy of findings may be related to the nature of the samples and data used. While the water mass and foraminifer data of our study are from just a short time interval and may be representative of only one early summer plankton bloom, they are nonetheless consistent in being obtained at the same time and place (i.e., station and water depth). Kozdon et al. (2009) calculated the calcification depth of N. pachyderma (sin.) using a long-term oceanographic database and temperatures obtained from paired Mg/Ca and δ^44/40Ca measurements of shells picked from sediment surface samples. Such samples integrate foraminifer assemblages over the entire plankton growth seasons in a certain number of years, depending on sedimentation rates and bioturbation intensity. Furthermore, variable ages of the surface samples analysed (modern to more than 1000 yr ago; see Simstich et al. 2003) may play a more significant role than anticipated by Kozdon et al. (2009), who claimed that the samples “represent modern oceanographic conditions.” Recent work in the Norwegian Sea and Fram Strait (Sejrup et al. 2010; Spielhagen et al. 2011; Werner et al. 2011) has demonstrated that temperature variations of 2°C and more occurred in the Atlantic Water layer during the last one to two millennia. Particularly in areas with low sedimentation rates (i.e., in ice-covered regions), sediment surface samples may thus rather represent earlier and colder periods (e.g., the Little Ice Age) than just modern, warmer conditions of the comparatively short Industrial Period. For this reason, the isopycnals calculated by Kozdon et al. (2009) on the basis of data extracted from a hydrographic database may not always correspond to the average conditions during deposition of the related sediment surface samples. Our comparison of in situ sampled foraminifera and measurements avoid such potential bias and indicates a wider density range for the habitat of N. pachyderma (sin.).

Comparison of plankton tow results with those from the sediment surface has important implications for palaeoceanographic reconstructions because in such studies core-top samples are usually assumed to represent the recent foraminiferal fauna and therefore modern conditions. In our study area, the planktic foraminiferal species composition in the overlying water column is fairly well reflected in the sediment surface samples: the polar N. pachyderma (sin.) clearly dominates all the samples, with relative abundances of the subpolar T. quinqueloba varying between 8 and 24%. Again, the differences between plankton tow and sediment data can most likely be ascribed to the early summer sampling of the water column. Lateral advection of foraminifera is not considered a major factor because mean transport distances in Fram Strait are only 25–50 km for N. pachyderma (sin.) and 50–100 km for T. quinqueloba, as determined for a sediment trap at 1125 m water depth (von Gyldenfeldt et al. 2000). In the east and the west of the study area (where differences between plankton tow and sediment data are largest), the flow direction of warm and cold water masses is largely normal to the sampling transect and transport effects of subpolar and polar species on the species assemblages on the sea floor can be expected to be minor. Furthermore, the increase in abundances in the water column at the sea-ice margin compared to the open ocean and under the ice cover is also seen on the sediment surface (Fig. 5). Comparison of the relative abundances of N. pachyderma (sin.) and T. quinqueloba in the depth intervals of 0–50 m and 50–100 m to those from the sediment surface (Fig. 6) suggests that the sediment surface samples relate best to the fauna living between 50 and 100 m water depth. This hypothesis is supported by the fact that in the water column at almost all stations both species were most abundant in this depth interval. The congruence of the fauna found in the water column and on the sediment surface indicates that in this area processes that can influence the deposition of shells (e.g., lateral transport, selective dissolution) do not change significantly the distribution during settling. Since Fram Strait is relatively narrow and interannual west–east variability in the position of the average summer sea-ice margin can be high, effects of fluctuations related to centennial-scale climate changes may be more subdued in the sediment surface samples than in the Nordic seas. As a consequence, the large-scale oceanic regimes are more obvious from the faunal compositions. This suggests that planktic foraminifera in Fram Strait sediments provide a fairly reliable proxy for palaeo-water mass characterization.