Journal of Geophysical Research: Oceans

Preferential sinking export of biogenic silica during the spring and summer in the North Water Polynya (northern Baffin Bay): Temperature or biological control?

Authors


  • This is a commentary on DOI:

Abstract

[1] Short-term free-drifting particle interceptor traps were deployed at 28 stations from April to July 1998 in the North Water Polynya (northern Baffin Bay). The amount, composition, and vertical transformation of the organic material sinking out of the euphotic zone were assessed. Clear seasonal sedimentation patterns were apparent throughout the Polynya. Maximum sedimentation occurred during the month of June, at which time high sedimentation of intact diatom cells and empty frustules was observed. In July, abundant resting spores and zooplankton feces were sinking out of the euphotic zone. Vertical transformation of the sinking material, between 50 m and 100 m, revealed a consistent loss of ∼30–35% for particulate organic carbon and nitrogen and a loss of ∼10% for biogenic silica. A temperature model of silica dissolution was used to assess the role of temperature in controlling the low loss of biogenic silica observed in the upper water column. Model results show divergence of modeled rates from in situ loss rates at times when the biosiliceous fraction of the sinking material was high. This indicates that biological factors played a key role in reducing biogenic silica dissolution in the North Water Polynya. Sedimentation of intact cells, abundant resting spores, and feces all contributed to enhance preservation of silica in the sinking material. These results suggest that the North Water Polynya is a sink for biogenic silica and emphasize the significance of biological processes in controlling the silica pump in the marine environment.

1. Introduction

[2] Diatoms are considered to play a major role in the downward export of organic carbon from the euphotic zone in the ocean [Goldman, 1988]. In both hemispheres, spring diatom blooms are usually periods of highest new production and sinking export of organic carbon [e.g., Billet et al., 1983; Wilson et al., 1986]. The significant role played by diatoms in the export of organic matter can be explained by a combination of factors. First, rapid sinking of intact diatom cells [e.g., Davies and Payne, 1984] and spores [e.g., Lutter et al., 1989] has been observed upon the termination of blooms, possibly reflecting the life cycle of certain species [Smetacek, 1985]. Second, because of their large size, diatoms serve as food for mesozooplankton and macrozooplankton grazers and are incorporated into rapidly sinking feces [e.g., Schrader, 1971; González et al., 1994]. Third, diatoms can sink as part of organic aggregates that have sedimentation rates orders of magnitude higher than individual cells [Alldredge and Gotschalk, 1989; Kiørboe et al., 1994].

[3] In addition to their significant role in the downward export of carbon from the euphotic zone, sedimenting diatoms are responsible for most of the removal of biogenic silica from surface waters, which gives rise to the “silica pump” concept [Dugdale et al., 1995; Dugdale and Wilkerson, 1998]. However, the coupling between primary production at surface and sedimentation at depth does not depend only on the amount and type of primary producers but also on the transformation of the sinking material during its descent, which itself depends on a combination of physical and biological processes [e.g., Peinert et al., 1989; Kiørboe, 1996]. Global budgets for biogenic silica show that less than 50% of the production by diatoms is exported below 100 m because of dissolution in surface waters [Nelson et al., 1995]. Silica dissolution is strongly temperature-dependent [Kamatani and Riley, 1980; Hurd and Birdwhistell, 1983] but also depends on other factors, as discussed by Ragueneau et al. [2000]. For example, packaging of diatom frustules into rapidly sinking feces [Tande and Slagstad, 1985], bacterial colonization of diatom frustules [Biddle and Azam, 1999], differential dissolution of diatom species [Gersonde and Wefer, 1987; Kohly, 1998], and availability of aluminium (Al) [Van Bennekom et al., 1991; Van Beusekom et al., 1997] all influence biogenic silica dissolution and export. In the Southern Ocean, low temperatures have been suggested to play a significant role in the slow dissolution of biogenic silica observed in the upper 100 m, while the importance of biological processes is yet to be assessed [Nelson and Gordon, 1982; Nelson et al., 1995]. Recent findings that iron (Fe) limitation increases the Si:N uptake ratio [Hutchins and Bruland, 1998; Takeda, 1998] could be of high significance for silica export in the Southern Ocean and other Fe-limited areas [e.g., Martin et al., 1989; De Baar et al., 1995; Coale et al., 1996]. In the Arctic, Measures [1999] reports high Fe concentrations in the surface mixed layer, which makes it unlikely that Fe could be limiting. Measures suggests that transport by ice-rafted sediments and delivery through ice melt is a significant process for the supply of trace elements in the Arctic.

[4] In the Arctic and Antarctic, very low winter irradiances and the presence of ice for most of the year provide for a very short period of primary production. Diatom blooms usually develop upon stratification of the surface layer after ice melt [Smith and Sakshaug, 1990], and there is a strong seasonality in phytoplankton production and export [e.g., Cripps and Clarke, 1998]. In the high Arctic, the phytoplankton bloom typically takes place in July or August [Grainger, 1979; Hsiao, 1988, 1992] and is followed by maximum sedimentation in August [Atkinson and Wacasey, 1987; Hsiao, 1987]. Polynyas, which are defined as mesoscale areas (10,000–90,000 km2) of open water surrounded by ice [Stirling, 1997], are considered to be areas of high biological production [Smith and Gordon, 1997; Stirling, 1997]. The North Water Polynya (NOW) is the largest recurring polynya in the Arctic. It is considered to be one of the most productive areas in the Arctic, as well as the most important polynya in terms of biological significance [Stirling, 1980, 1997]. Open water in early spring in the NOW is thought to enhance localized production at a period when production remains low in surrounding ice-covered areas [Lewis et al., 1996].

[5] Rapid transfer of primary-produced particles to the seafloor was observed in the Northeast Water Polynya (NEW) [Bauerfeind et al., 1997]; however, sedimentation has never previously been investigated in the NOW. We hypothesize high sinking fluxes of biogenic particles in the NOW on the basis of the positive relationship between sedimentation and productivity in boreal regions [Wassmann et al., 1991], the evidence of rapid sedimentation in the NEW [Bauerfeind et al., 1997], and the persistence of a silicate anomaly in deep Baffin Bay waters and the NOW. Dissolved silicon concentrations in bottom waters of the NOW (∼30–40 μM) are more than 2 times higher than at similar depths in central Arctic basins and the Labrador Sea [Jones and Coote, 1980; Tremblay et al., 2002a]. Moreover, higher silicic acid concentrations were measured in sediment pore water, suggesting a dissolved silicon efflux from the sediments [Jones et al., 1984]. This scenario would be consistent with locally enhanced deposition of biosiliceous particles and subsequent silica dissolution and diffusion from sediment pore water.

[6] Sediment traps have been widely used to estimate the sinking flux of organic material in the ocean [e.g., Honjo, 1990; Olesen, 1995; Asper and Smith, 1999; Blain et al., 1999] and to assess the composition of that flux [e.g., Bathmann et al., 1991; Kohly, 1998]. Comparison of particle flux estimates derived from surface-tethered drifting traps with estimates using 234Th suggest that the former may underestimate or overestimate particles fluxes [Buesseler, 1991]. Some studies showed good agreement between fluxes measured with shallow sediment traps and 234Th calculated fluxes [Coale and Bruland, 1987; Murray et al., 1996]. Results by Murray et al. [1996] suggest that shallow traps moored within the euphotic zone may overtrap, while fluxes measured from traps installed below the euphotic zone agreed well with 234Th flux estimates. A recent comparison of surface-tethered traps and neutrally buoyant ones showed consistency in particulate organic carbon and nitrogen fluxes but discrepancy in the 234Th flux [Buesseler et al., 2000]. While the issue of sediment trapping efficiency is subject to debate [e.g., Boyd and Newton, 1997; Rivkin et al., 1996, 1997] and requires further study [Gust et al., 1994; Buesseler et al., 2000], sediment traps provide unique insights as to the composition of the material leaving the euphotic zone.

[7] This study was designed to test the hypothesis of a high sinking export of diatoms from the euphotic zone of the NOW. Specific objectives were to (1) characterize the spring-summer sedimentation regime during 1998, (2) assess the vertical transformation of the biogenic material sinking out of the euphotic zone during its descent, and (3) investigate the significance of biogenic silica export in the Polynya with respect to temperature control on silica dissolution.

2. Material and Methods

2.1. Study Area

[8] The North Water Polynya is located between Greenland and Ellesmere Island, northern Baffin Bay. The Polynya is under the influence of Arctic water, entering Nares Strait (78.5°N) and flowing south along the coast of Ellesmere Island, and Baffin Bay water, which moves northward along the Greenland coast. The circulation dynamics in the Polynya are described by Bourke et al. [1989] and Melling et al. [2001], and the seasonal extent of the different water masses is well described by Tremblay et al. [2002a]. The bathymetry of the North Water Polynya is complex, with an average depth of 400 to 500 m and maximum depths in the center basin (Figure 1).

Figure 1.

Bathymetry of the North Water Polynya. The isobaths are in meters.

[9] The North Water owes its existence to a recurring ice bridge that forms in Nares Strait and acts as its northern boundary, preventing ice flowing south to enter the Polynya. Ice formed within the Polynya drifts southward owing to strong winds and currents. The Polynya starts to open toward the south in early spring and reaches its maximum extent in July, before its southern boundary disappears. Ice cover distribution in the NOW during the spring and summer of 1998 (during this study) is described by Mundy and Barber [2001] and Wilson et al. [2001], and nutrient distributions are presented by Tremblay et al. [2002a]. In addition, Sea-viewing Wide Field-of-view Sensor (SeaWIFS) data analyses for 1998–2000 show that intense spring blooms taking place in May or early June are a recurring feature in the Polynya [Bélanger, 2001].

2.2. Sampling

[10] Sampling was carried out from April 9 to July 20, 1998, on board CCGS Pierre Radisson. During that period, short-term drifting particle interceptor traps were deployed at 28 stations. In April and May, the traps were regularly moored from drifting ice floes (referred to as ice moorings), while in June and July, moorings were deployed from the ship (referred to as free-drifting moorings). The traps were deployed for a period of ∼12–24 hours, except for station C33 (36 hours), and the average distance traveled by the drifting moorings was ∼10 km. Details for each deployment are given in Table 1.

Table 1. Characteristics of Free-Drifting Sediment Trap Moorings in the North Water Polynya in 1998
StationLegDate InDuration, daysBegin Latitude, °NBegin Longitude, °WEnd Latitude,a °NEnd Longitude,a °WDistance Traveled,a kmSpeed,a cm s−1Mooring Type
  • a

    NA is data not available.

Ice 21April 100.5078° 36.0′74° 20.6′NANANANAIce
Ice 7 (A33)1April 180.6477° 00.0′76° 13.0′NANANANAIce
Ice 15 (A49)1April 230.9876° 17.0′74° 40.8′76° 15.5′74° 35.4′3.13.7Ice
Ice 20 (A27)1April 260.6277° 21.0′73° 48.0′NANANANAIce
Ice 221April 280.8976°19.4′77° 24′.7′76° 17.0′77° 30.0′4.86.2Ice
B1022May 70.4076° 36.2′76° 29.6′76° 35.6′76° 30.9′1.13.2Ice
B222May 180.4877° 21.0′76° 35.0′NANANANAIce
B342May 191.0476° 58.5′75° 30.9′76° 46.1′75° 55.2′23.826.5Drifting
B442May 280.4876° 22.9′77° 24.8′76° 21.777° 22.32.35.6Ice
C493June 51.0276° 16.9′74° 58.3′76° 17.3′74° 53.0′1.71.9Drifting
C23June 71.0178° 13.6′74° 39.3′78° 10.9′74° 55.4′6.37.3Drifting
C143June 100.8177° 49.0′75° 26.8′77° 47.9′74° 43.4′11.115.8Drifting
C183June 110.4877° 50.1′73°16.2′77° 51.6′73° 23.1′3.37.9Drifting
C273June 120.5977° 21.0′74° 02.6′77° 20.9′74° 29.4′6.913.6Drifting
C223June 130.6577° 21.0′76° 35.2′77° 21.0′76° 35.2′1.83.2Drifting
C333June 141.5077° 01.0′76° 06.1′77° 00.1′76° 15.2′2.92.3Drifting
C403June 180.5877° 00.3′72° 15.8′76° 59.4′72° 23.7′2.75.4Drifting
C543June 200.7976° 18.0′71° 48.7′76° 23.0′72° 04.4′10.315.0Drifting
C503June 220.5576° 16.5′74° 09.0′76° 10.0′73° 31.1′16.133.8Drifting
C443June 230.5776° 24.7′77° 23.1′76° 17.1′77° 23.4'′14.128.8Drifting
D14June 301.0575° 13.4′74° 58.8′75° 06.7′74° 58.8′12.413.7Drifting
S44July 31.0676° 18.1′74° 23.3′76° 20.2′74° 50.5′8.69.4Drifting
S54July 70.7776° 25.0′77° 20.5′76° 11.5′77° 18.1′25.037.8Drifting
N24July 111.1278° 21.5′74° 19.0′78° 15.1′74° 05.4′12.312.7Drifting
E24July 141.1878° 01.4′73° 50.8′77° 54.1′73° 42.8′13.713.4Drifting
E14July 160.9877° 00.5′72° 39.6′77° 07.1′73° 06.6′14.216.8Drifting
D24July 180.9676° 59.6′75° 0.9′76° 58.0′74° 49.4′4.35.1Drifting
S24July 201.0076° 16.9′71° 59.1′76° 10.0′71° 41.6′13.715.8Drifting

[11] Free-drifting traps deployed from the ship were positioned at depths of ∼50, 100, and 150 m, and traps attached to ice floes were at 1, 15, and 50 m. Shallow depths for moorings attached to ice floes were to reduce excessive drag and current shear when ice drift and surface currents have different directions than deeper currents. In the present paper, results from all 50-m moorings are used for the analysis of seasonal sedimentation patterns, and results from the free-drifting moorings (from 50 to 150 m) serve for the analysis of the vertical transformation of the sinking material. Euphotic depths (ED) (1%) ranged from an average of 46 m (standard deviation (SD) = 14 m) in April and May to 27 m (SD = 21 m) in June and 28 m (SD = 15 m) in July. For all free-drifting moorings except stations C2 (ED = 75 m) and D1 (ED = 50 m) the shallowest trap (50 m) was placed well below the euphotic layer. Mixed layer depths at the trap stations were ∼30 m in April and May and were 10–15 m in June and July (see Tremblay et al. [2002] for complete analysis of mixed layer depths in the NOW in 1998). Sinking fluxes thus represent the actual flux out of the surface layer and the euphotic zone.

[12] The traps used for the free-drifting moorings were Plexiglas cylinders with a baffle, a 6.5-cm internal diameter, and of an aspect ratio (height:diameter) of 9. A series of cylinders (four or six) were installed at each sampling depth in order to collect enough material to perform subsequent analyses. Upon deployment the traps were filled with Whatman GF/F filtered seawater with added brine to increase the salinity by ∼5 and create a dense layer. No poison or preservative were used. Upon recovery the traps were covered with a tight lid and placed vertically in a cold chamber (0°C) in the dark to allow for the material to settle during 8 hours. After the settling period the trap supernatant was gently removed and the lower fraction of the trap volume (∼500 mL) was kept. The trap samples from the same depth were pooled together and subsequent analyses were performed on the pooled sample.

[13] For moorings deployed from the ice, the traps were attached to a tripod solidly anchored in the ice. The traps were Plexiglas cylinders of 10-cm diameter, with an aspect ratio of 7. As for the free-drifting moorings, the traps were filled with Whatman GF/F filtered seawater with added brine, without poison or preservative, upon deployment. After recovery the total volume of the trap was gently mixed and transferred to a dark container for subsequent analyses. All traps (drifting plus underice) were thoroughly rinsed with distilled water between each deployment.

2.3. Analyses

[14] For all samples, analyses were performed after prescreening the trap material through a 500-μm mesh. Very few swimmers (<5 in most samples) were observed and retained during the prescreening procedure. Trap subsamples were used for analysis of the particulate material. Duplicate subsamples were filtered onto Whatman GF/F glass fiber filters for standard fluorometric determination of chlorophyll a (chl a) and pheopigments. Pigments were determined on board the ship using a R010 Turner Designs fluorometer after 24-hour extraction in acetone 90% at 4°C in the dark [Parsons et al., 1984]. Particulate organic carbon and nitrogen (POC and PON) were determined on duplicate subsamples filtered onto Whatman GF/F precombusted filters, which were dried (at 60°C for 24 hours) during the expedition, using a Perkin-Elmer Model 2400 CHN analyzer. Biogenic silica (BioSi) was determined on subsamples filtered onto Nuclepore polycarbonate 0.8-μm filters using an all-plastic filtration unit. The filters were frozen at −80°C until analysis, using the hydrolysis method of Paasche [1980]. With this method, hydrolysis is carried out using a 0.5% Na2CO3 solution and heating at 85°C for 2 hours. After cooling, the solution is neutralized with 0.5 M HCl to the turning point of methyl orange (pH 3–4). An aliquot of the solution is then used for the colorimetric determination of silicate [Parsons et al., 1984]. Blank correction was made by carrying out the same procedure using a blank filter. Microscopic examination of our samples did not reveal the presence of lithogenic material, and Paasche's [1980] method should give less than 0.5% recovery of mineral silica. The average percent variation among duplicate or triplicate biogenic silica samples, for all particle interceptor trap samples (number of samples n = 44), was 5.4%.

[15] Cell and feces enumeration were performed on samples from the 50-m depth, which had been preserved with buffered formaldehyde. Cell enumeration was done under the inverted microscope [Lund et al., 1958]. For each sample a minimum of 400 cells were counted except for a few samples from April for which a minimum of 200 cells were counted. Feces were also enumerated under the inverted microscope; the total number of feces in each sample was counted. The length and width of complete or broken feces were measured; feces volume was then calculated using the appropriate geometric equation. Total feces volume was estimated for each sample. All regressions presented are reduced major axis Model II regressions, which accounts for the fact that both variables are subject to error due to analytical measurement [Sokal and Rohlf, 1981].

3. Results

[16] Figures 2 and 3 show sinking fluxes of chl a and POC, respectively, at 50 m for all stations sampled in the Polynya from April to July. Because of the limited number of trap deployments during April and May, these two months are pooled together on the same map (Figures 2a and 3a). Chl a fluxes remained <5 mg m−2 d−1 in April and May, except for station B44, which was sampled at the end of May (May 28, see Table 1) and showed chl a fluxes of ∼10 mg m−2 d−1 (Figure 2a). Sinking fluxes of chl a were quite high at all stations in June (mean plus or minus standard deviation; 6.1 ± 5.8 mg m−2 d−1), with highest recorded values of ∼16 mg m−2 d−1 and 19 mg m−2 d−1 at stations C49 and C18, respectively (Figure 2b). In July, sinking fluxes of chl a were <3 mg m−2 d−1 (1.1 ± 0.7 mg m−2 d−1) at all stations (Figure 2c).

Figure 2.

Sinking fluxes of chl a at 50 m in the North Water Polynya in (a) April and May, (b) June, and (c) July 1998.

Figure 3.

Sinking fluxes of particulate organic carbon (POC) at 50 m in the North Water Polynya in (a) April and May, (b) June, and (c) July 1998.

[17] Sinking fluxes of POC at 50 m showed a seasonal pattern matching that of chl a, with highest values in June (Figure 3b). However, the magnitude of the seasonal change for POC was less than for chl a, as showed by the average POC flux in June (46.2 ± 26.3 mmol m−2 d−1) (Figure 3b) which was slightly more than twice the average for July (20.0 ± 10.4 mmol m−2 d−1) (Figure 3c), while average chl a fluxes were more than 5 times higher in June than July (Figures 2b and 2c). Average chl a and POC sinking fluxes during the three periods were statistically different (Kruskall-Wallis test of means; probability p < 0.01 [Sokal and Rohlf, 1981]), indicating that seasonal variations exceeded spatial variations.

[18] Clear seasonal changes in composition ratios of the material in the 50-m traps were observed for April and May, June, and July (Figure 4). The percent chl a in total sinking pigments decreased from an average of ∼65% in April and May to 50% in June and 34% in July (Kruskall-Wallis, p < 0.01) (Figure 4a). BioSi:POC and BioSi:PON molar ratios showed a strong seasonal increase (Figures 4b and 4c), with average values of 0.1 and 1.0, respectively, in April and May and increasing to 0.4 and 4.6, respectively, in July (Kruskall-Wallis, p < 0.01). The POC:PON molar ratios in the traps did not show any significant seasonal trend, with an average of ∼9–10 for the 3 periods (not shown; Kruskall-Wallis, p > 0.1).

Figure 4.

Seasonal trends in composition ratios of the sinking material at 50 m: (a) Percent chl a in total pigments, (b) biogenic silica (BioSi):POC ratio, and (c) BioSi:particulate organic nitrogen (PON) ratio. Mean plus standard deviation are shown (number of stations visited N = 9 for April and May, N = 11 for June, and N = 8 for July).

[19] Seasonal changes were also observed in the composition of the sinking material (Figure 5). The number of intact diatoms (pennates plus centric) in the trapped material drastically increased from April and May to June, and decreased in July (Figure 5a). Diatom resting spores were not observed in April and May, while they were found at 4 stations (C40, C54, C50, and C44) out of 11 in June (Figure 5b). In July the spores were present in high numbers at all stations but one (D1) (Figure 5b). When looking at relative numbers, the proportion of total cell numbers (excluding empty frustules) made up by the resting spores increased dramatically during the season, with 62% of total cell numbers in July compared with 5% in June and 0% in April and May (Figure 5c). The abundance of empty diatom frustules also showed a strong seasonal pattern, and very high numbers were observed in the material trapped during the month of June (Figure 5d). Empty diatom frustules were observed at only two stations out of eight in April–May, while they were present at all stations in June and July, although in lower numbers during the latter month (Figure 5d). Zooplankton feces volume increased fivefold from April and May to June, and increased by the same factor from June to July (Figure 5e). All observed changes were statistically significant (Kruskall-Wallis, p < 0.01).

Figure 5.

Seasonal trends in the composition of the sinking material at 50 m: (a) intact diatom cells (pennates plus centric), (b) diatom resting spores, (c) percent resting spores in total algal cells, (d) empty diatom frustules, and (e) total feces volume. Mean plus standard deviation are shown (number of stations visited N = 9 for April and May, N = 11 for June, and N = 8 for July). For each period, n denotes the number of stations where the variable occurs.

[20] Figure 6 shows regressions of sinking fluxes of particulate material at 100 m as a function of the same fluxes at 50 m for all free-drifting trap data (number of stations where the variable occurs n is 19). The internal consistency of the observed decrease in flux, irrespective of station location or sampling period, is indicative of robust trends which are unlikely to be caused by hydrodynamic biases [e.g., Gust et al., 1994]. All regressions are highly significant, and the regression slopes are used as estimates of the relative loss of each element from 50 to 100 m. Accordingly, ∼30% of the sinking POC (slope = 0.71 in Figure 6a), 36% of the sinking PON (slope = 0.64 in Figure 6b), and 33% of the total sinking pigments (slope = 0.67 in Figure 6c) were lost between 50 and 100 m. Regression slopes for POC, PON, and total pigments were not significantly different (tests for equality of slopes, p > 0.1). The regression slope for chl a (0.54, Figure 6d) was significantly lower than those of total pigments, POC and PON, indicating a higher loss of chl a (46%) compared with these elements. For BioSi the loss estimated was 11% (slope = 0.89 in Figure 6e), a value much lower than the loss observed for all other elements.

Figure 6.

Sinking fluxes of material at 100 m versus sinking fluxes at 50 m: (a) POC, (b) PON, (c) total pigments, (d) chl a, and (e) biogenic silica. For reference, solid lines show regression slopes (model II), and dashed lines are 1:1 slopes. Regression statistics are shown.

[21] Figure 7 shows regressions of BioSi versus PON and POC versus PON for all free-drifting stations. Regression lines are shown for each sampling depth (50, 100, and 150 m), and for the combined data at all depths. Slopes of the regression BioSi versus PON are 3.2, 4.6, and 4.4, at 50, 100, and 150 m, respectively (Figure 7a, equations not shown). The regression slope for combined data, irrespective of depth, is 3.7 (Figure 7a). The slope for BioSi versus PON significantly (p < 0.05) increased from 50 m to 100 m and 150 m but did not change (p > 0.1) from 100 to 150 m. Regressions for sinking POC versus PON did not reveal any significant difference among depths (p > 0.1) with an average slope for all depths of 11.1 (Figure 7b).

Figure 7.

Sinking fluxes of (a) BioSi versus PON and (b) POC versus PON for free-drifting stations at all depths. Solid lines are regression slopes for combined (all depths) data. Statistics shown are for combined (all depths) data.

4. Discussion

4.1. Seasonal Sedimentation Patterns

[22] The extensive short-term particle interceptor trap deployments performed during the 4-month expedition in the North Water Polynya provided a clear definition of seasonal sedimentation patterns, despite the rather loose coverage during the months of April and May (nine stations sampled; see Table 1 and Figures 2a and 3a). The seasonal trend analysis includes results from traps moored at 50 m, either free drifting (almost exclusively in June and July, see Table 1) or attached to ice floes (April and May, see Table 1). One might argue that this difference in sampling and the fact that the trap design was slightly different for the two types of moorings introduced bias in the seasonal trend analysis. We did not perform intercomparisons of trap sampling efficiency in the field; however, aspect ratios of the trap cylinders point to similar sampling efficiencies [e.g., Blomqvist and Håkanson, 1981; Butman, 1986; Baker et al., 1988]. In addition, when comparing the material collected by the drifting trap in May (B34 in Table 1) with ice-moored traps during the same period (B22 and B44 in Table 1), we observed very similar species composition for all moorings (C. Michel, M. Gosselin, and C. Nozais, unpublished data, 1998). These observations, together with the very consistent seasonal trends in elemental ratios (Figure 4), suggest that the seasonal patterns are valid even though most moorings were attached to ice floes during the months of April and May.

[23] The distribution of Arctic and Baffin Bay water masses in the NOW influences the timing and magnitude of primary production [Tremblay et al., 2002b], as well as the species composition of phytoplankton within the Polynya [Booth et al., 2002]. Sedimentation is influenced by a number of factors, including the timing and magnitude of primary production [e.g., Honjo, 1990] and the size structure and composition of primary producers [Goldman, 1988]. Spatial heterogeneity in sinking fluxes and in the composition of the sedimenting material was thus expected to occur in the Polynya and was indeed observed (Figures 2 and 3). This aspect is not the focus of the present paper and will be addressed separately (C. Michel, M. Gosselin, and C. Nozais, unpublished data, 1998). Although the Polynya does not behave uniformly, the strongly significant seasonal trends observed in the present paper reveal general seasonal changes that did override spatial variations.

[24] Perhaps one of the most significant seasonal trend observed was the strong pulselike episode of sedimentation in the Polynya, with a period of maximum sedimentation during the month of June followed by a rapid decrease in July (Figures 2 and 3). This is somewhat different than what was observed by Bauerfeind et al. [1997] in the NEW polynya where highest sinking fluxes of POC were observed in August. We also observed much higher sinking export of organic material in the NOW compared with the NEW polynya [Bauerfeind et al., 1997]. This agrees with the intense phytoplankton bloom [Mei et al., 2002] and high new production [Tremblay et al., 2002b] observed in the NOW, while limited new production [Smith et al., 1997] and high water column remineralization [Daly, 1997] is reported for the NEW polynya. While the chl a sinking flux had started to increase at the end of May (station B44 in Figure 2a) and was also somewhat higher on the Greenland side of the Polynya in April (station A27 in Figure 2a), in June all stations showed elevated chl a sinking fluxes, except northwestern stations C2 and C14 (Figure 2b). These two stations, which were under the influence of the southwestern flow of Arctic water entering the Polynya, showed very low production rates compared with other stations [Tremblay et al., 2002b]. Maximum chl a sinking fluxes in June (>15 mg m−2 d−1) compare with high values reported during high sedimentation episodes following diatom blooms in coastal environments [e.g., Bodungen et al., 1986; Olesen, 1995]. Chl a sinking fluxes in July (average of ∼1 mg m−2 d−1, Figure 2c) compare with other values reported for the Arctic under ice cover [Michel et al., 1996] or in open waters [Hsiao, 1987]. The strong seasonal pulse of sedimentation in the Polynya is typical of the seasonality observed in high-latitude environments, where high spring production by large diatoms is usually followed by rapid sedimentation of either intact cells or feces.

[25] The phytoplankton bloom was fully taking place in the southeastern part of the Polynya in May and entered a declining phase in June [Tremblay et al., 2002b]. High chl a fluxes and the high numbers of diatoms collected in the sinking material in June (Figures 2 and 5) clearly indicate that phytoplankton was sedimenting out of the euphotic zone during that period. Our results also show that high sedimentation of intact algal cells extended over all of the month of June but was finished in July, at which time both chl a and cell-sinking fluxes had decreased (Figures 2 and 5a). Since bottom ice algal biomasses were quite low in the Polynya in June (5.8 ± 9.5 mg chl a m−2 [Gosselin et al., 1999]) and accounted, on average, for <3% of the suspended stock, it is safe to assume that suspended phytoplankton contributed the bulk of the algal biomass sinking out of the euphotic zone during that period. In addition, on the basis of SeaWIFS data which showed that intense phytoplankton blooms in May or early June are a recurrent feature in the NOW [Bélanger, 2001], it is reasonable to infer that the seasonal sedimentation pattern observed is typical for the area.

[26] While the bulk fluxes (Figures 2 and 3) point to a period of maximum sinking export from the euphotic zone and a significant sedimentation of intact diatoms in June (Figure 5a), looking into elemental ratios for the trapped material provides additional insights on the seasonal sedimentation cycle in the Polynya. The seasonal decrease in the chl a:pigments ratio in the traps (Figure 4a) points to an increased contribution of sinking feces, which was indeed observed (Figure 5e). The lag evidenced between the period of maximum sedimentation of feces (July, Figure 5e) and that of intact algae (June, Figures 3b and 5a) agrees well with other observations following periods of high diatom blooms, when zooplankton grazing does not control phytoplankton biomass accumulation [Wassmann et al., 1991, and references therein]. Other interesting seasonal changes in elemental ratios are the parallel increases in BioSi:POC and BioSi:PON ratios in the trapped material. Average values for these two ratios in April and May were 0.13 and 1.0, respectively (Figures 4b and 4c), which closely match values observed for phytoplankton (Si:C = 0.13, Si:N = 1.12 [Brzezinski, 1985]). The threefold to fourfold increase observed in June and July (Figures 4b and 4c) indicates a strong seasonal increase in silica-rich material sinking out of the euphotic layer.

[27] BioSi is almost exclusively found in diatom frustules and can be observed in sinking material in the form of intact cells, empty frustules, or in feces produced by zooplankton feeding on diatoms. All of these biosiliceous particles contributed, in variable proportions, to the material sinking out of the euphotic zone in June and July. High numbers of intact diatom cells and empty frustules were observed in June, and feces were abundant in July (Figure 4). In addition, after mid-June, Chaetoceros resting spores were observed in the sinking material at all stations, except at the southernmost station D1, which is located outside the Polynya. In July, resting spores made up the bulk of total cell numbers sinking out of the euphotic zone (Figure 4c). Diatom resting spores are known to be highly silicified [Hargraves and French, 1983] and resist gut passage, contrary to diatom frustules which are broken during gut transit [Porter, 1976]. It follows that the production and subsequent sinking of Chateoceros resting spores in the Polynya was probably a determinant factor in the seasonal magnification of biosiliceous export. We cannot rule out the possibility that our biogenic silica extraction method underestimated silica concentrations when there were high numbers of resting spores in the sinking material. If this had been the case, the seasonal magnification of biosiliceous export would be increased. Since Fe limitation can increase Si:N uptake rates [Hutchins and Bruland, 1998; Takeda, 1998], this factor could be considered to have induced the seasonal change observed. Fe concentrations were not measured during our sampling. However, the high Fe concentrations observed in the Arctic Ocean, and in the vicinity of the North Water (>3.5 nM [Measures, 1999]), preclude any inference of Fe limitation. In addition, the fact that melting of rafted ice might play a significant role in Fe supply in surface waters of the Arctic [Measures, 1999] suggests high Fe concentrations in the NOW and, possibly, seasonally increasing surface concentrations with increased ice melt.

[28] Resting spore formation in natural populations is usually associated with nutrient depletion [see Garrison, 1984, and references therein], and in laboratory experiments, nitrogen limitation has been shown to be a major trigger for spore formation [Hargraves and French, 1983 and references therein]. In the Polynya, Tremblay et al. [2002b] showed that nitrogen exhaustion was the main factor responsible for the termination of the bloom in June. Our results suggest that nitrogen limitation also induced formation of resting spores, followed by their sedimentation after mid-June.

[29] The temporal analysis provides a clear image of the seasonal sedimentation cycle in the Polynya during the spring and summer. The suite of events agrees with pulselike sedimentation episodes following diatom blooms. However, the period of high sedimentation in the Polynya preceded that in surrounding ice-covered areas by ∼2 months. This early sedimentation of algal material in the NOW likely provides an early food input to the benthic fauna. Besides, abundant Chaetoceros resting spores after mid-June support the life history hypothesis of Smetacek [1985]. In the Polynya, resting spores may provide a seed population for a phytoplankton bloom later during the season, at a time when nutrients would be replenished at the surface. Increasing evidence of the presence of sinking Chaetoceros resting spores in polar environments [e.g., Leventer and Dunbar, 1996; Kohly, 1998] suggests that this species undergoes massive encystment in these environments. This life cycle strategy can have a profound influence on the type and amount of material being exported from the euphotic zone, as well on the dynamics of the phytoplankton bloom.

4.2. Vertical Transformation of Sinking Material

[30] The regression analyses revealed a similar loss (30–35%) of POC and PON between 50 m and 100 m (Figures 6a and 6b), which was also corroborated by the tight coupling in POC and PON sinking fluxes at all depths (Figure 7b). Moreover, BioSi versus PON regressions (Figure 7a) revealed that the loss of PON mainly occurred between 50 and 100 m and that additional loss from 100 to 150 m was negligible or, at most, did not exceed that of BioSi. Because of the strong covariance in sinking POC and PON and the small BioSi loss observed between 50 and 100 m, we can infer that the main loss of POC and PON occurred in the uppermost 100 m. These results agree with those of Andreassen and Wassmann [1998], who observed, using short-term drifting traps in the Barents Sea, a decrease of POC and PON in the upper 140 m and no further decrease at deeper depths.

[31] POC and PON removal at the surface depends on microbial recycling and zooplankton grazing. Estimating the relative share of these processes is beyond the scope of this paper. Nevertheless, higher loss of chl a (46%) compared with that of total pigments (33%), between 50 and 100 m, indicates a transfer of chl a to pheopigments, which likely reflects grazing activity and feces production. Besides, similar loss rates for POC, PON, and total pigments between 50 and 100 m suggest that remineralization was essentially removing pigmented material (algae and feces) from the sedimentation pool. A corollary is that pigmented material accounted for the bulk of the organic sinking flux. The strong regression BioSi versus PON (Figure 7a), low POC:pigments ratios in drifting traps (range of 11–144, g/g), and abundant algae and feces (Figures 5a, 5b and 5e) in the sinking material all support this interpretation.

[32] High removal of POC and PON from the sinking material during its descent, compared with the negligible loss of BioSi, points to differential sedimentation of silica (Si) in the Polynya. Indeed, the sinking material was Si-enriched as it sank to greater depths, as shown by the higher BioSi:PON ratio at 100 m and 150 m compared with 50 m (Figure 7a). Contrary to carbon which undergoes biological remineralization during its descent, the loss of BioSi is a function of its dissolution rate, which itself depends on both physical and biological factors. Silica dissolution is strongly temperature dependent [Kamatani and Riley, 1980; Hurd and Birdwhistell, 1983] but also depends on a variety of factors (see review by Ragueneau et al. [2000]). For example, the dissolution of diatom frustules varies among species and depends on surface area and frustule structure. Biological coatings on healthy diatoms reduce their dissolution [e.g., Lewin, 1962], while bacterial colonization of dead diatoms has an opposite effect [Biddle and Azam, 1999]. The role of Al in decreasing silica dissolution rates is also reported by several authors [e.g., Van Bennekom et al., 1989, 1991; Van Beusekom et al., 1997]. Physical or biological factors that increase the sinking rate of diatoms contribute to lower biogenic silica dissolution during descent by reducing their residence time in undersaturated water relative to silica dissolution.

4.3. Significance of Temperature for BioSi Dissolution

[33] In order to assess if the low water temperature in the Polynya was the main controlling factor for the low loss of BioSi in sinking particles, we compared in situ BioSi loss rate estimates from 50 and 100 m with rates calculated using a temperature dissolution model [Gnanadesikan and Toggweiler, 1999]. Equations for the temperature model are presented below, and in situ loss rates were estimated by applying the standard BioSi loss rate observed between 50 and 100 m (11%; see Figure 6e) to fluxes measured at 50 m. The same BioSi loss rate (11%) was applied to fluxes measured from ice moorings (April and May, except station B34), even if there were no traps at 100 m for these moorings. The validity of this extrapolation is supported by (1) comparable BioSi loss rates between 15 and 50 m for ice moorings (14% loss of BioSi, r = 0.93, n = 8) and (2) consistent results for ice moorings and free-drifting moorings before mid-June (see Figure 7). The equation for in situ loss rates estimates is as follows:

equation image

where dF/dz is the average Si dissolution rate (in mmol m−3 d−1) between 50 and 100 m, F50m is the sinking flux of BioSi measured at 50 m (in mmol m−2 d−1), and Δz50–100 m is the depth interval between 50 and 100 m (50 m).

4.3.1. Model of Si dissolution

[34] We estimated temperature-dependent specific dissolution rates (R) based on the Arrhenius function developed by Gnanadesikan [1999], as in (2). These rates were then used to estimate Si dissolution for biosiliceous particles sinking out of the euphotic zone (50 m), as in (3) [Gnanadesikan and Toggweiler, 1999]. The Arrhenius function for the specific Si dissolution rate is

equation image

where R is the Si specific dissolution rate (in d−1) and T is the average in situ temperature, in Kelvin, in the 50- to 100-m depth layer. Temperature was measured with a conductivity-temperature-depth profiler (Falmouth ICTD) mounted on a Rosette sampler [see Melling et al., 2001]. Temperature in the 50- to 100-m depth layer did not show any significant trend during the season and was, on average, −1.5°C (range from −1.8° to −1.2°C).

[35] We estimated the dissolution rate of Si between 50 and 100 m with

equation image

where dF/dz is the average Si dissolution rate (in mmol m−3 d−1) between 50 and 100 m, R is the specific Si dissolution rate (in d−1) as estimated from (1), Wfall is the sedimentation rate of biosiliceous particles sinking out of the euphotic zone, in m d−1, as estimated from (4), and F50 m is the sinking flux of BioSi measured at 50 m (in mmol m−2 d−1).

[36] The sinking rate of biosiliceous particles (in m d−1) was computed for each station as follows:

equation image

where zeup is the depth of the euphotic layer (in meters) (1% irradiance level), F50 m is the sinking flux of BioSi measured at 50 m (in mmol m−2 d−1), and BioSi (z) is the suspended BioSi biomass at depth z (in mmol m−3), estimated by multiplying suspended POC by the slope of the regression BioSi versus POC for suspended biomass (slope of regression = 0.14 ± 0.004, r = 0.98, n = 53; (J.-É. Tremblay, unpublished data, 1998)).

4.3.2. Model results

[37] Figure 8 compares Si dissolution rates obtained from the temperature model with in situ loss rates for the biosiliceous material leaving the euphotic zone, on a station by station basis, for all stations where sinking fluxes (at 50 m) and suspended biomass in the euphotic zone were concurrently measured. Overall, the temperature model yields values and trends that agree well with in situ observations. Si dissolution rates estimated from the temperature model closely match in situ loss rates in April and May and at stations visited in early June. Thereafter, Si dissolution rates estimated from the temperature model are almost invariably higher than in situ loss rates (Figure 8). The close correspondence between modeled and in situ rates at stations visited before mid-June (and at station D1) brings up two conclusions. First, it validates model results. Second, it corroborates the hypothesis that low temperature can be a determinant factor for low Si dissolution in polar regions [e.g., Nelson and Gordon, 1982]. Conversely, the divergence in estimates observed in June and July provides evidence that factors other than temperature significantly reduced in situ Si dissolution rates in the NOW.

Figure 8.

Comparison between Si dissolution rates, between 50 and 100 m, estimated from the temperature model of Gnanadesikan and Toggweiler [1999] and in situ rates. Stations are positioned along the x axis according to sampling date.

4.3.3. Ecological significance of model results

[38] Different factors could have contributed in two different ways to reduce in situ Si dissolution of the sinking biosiliceous particles: (1) by increasing their sinking rates and (2) by decreasing Si dissolution rates of the particles themselves. On the one hand, estimated sinking rates ranged from 0.4 to 5 m d−1 (2.3 ± 1.3 m d−1) and did not show any significant seasonal trend, so that the first of these two mechanism can be dismissed. On the other hand, the observed seasonal changes in the composition of the sinking material point to a reduction of dissolution rates, as discussed below.

[39] The temperature model does not take into account the effect of ambient silicate concentrations on dissolution rates, as in the equation of Hurd and Birdwhistell [1983]. Since surface silicate concentrations (upper 150 m) decreased seasonally [Tremblay et al., 2002a], the effect would be to increase dissolution rates. Consequently, this factor cannot explain the divergence between in situ and modeled rates. Conversely, the incorporation of trace elements within the opaline matrix can significantly decrease silica solubility [Ragueneau et al., 2000]. Al is considered to play a key role [Ragueneau et al., 2000, and references therein], and seasonally increasing Al/Si ratios in diatom frustules could potentially decrease their dissolution rates [Van Bennekom et al., 1991]. However, concentrations of Al in the mixed layer of the Arctic Ocean are rather low (7.8 ± 5.0 nM, n = 45 [from Measures, 1999]) so that high dissolution rates of biogenic silica could be expected, as for the Antarctic [Van Bennekom et al., 1991]. It thus appears that none of these factors is likely to explain the lower in situ dissolution rates compared with modeled rates during June and July.

[40] Spores are heavily silicified and resist dissolution [Hargraves and French, 1983] and thus most likely contributed to significantly reduce Si dissolution rates of the sinking material after mid-June. Yet, during the first half of June, spores were not observed in the sinking material, and in situ Si dissolution rates were nevertheless lower than modeled rates. Siliceous contributors to the sinking flux during that period were dominated by intact diatoms and empty frustules. Since the biological coating on healthy diatoms protects them from dissolution [e.g., Kamatani, 1982] and bacterial colonization on lysed cells increases Si dissolution [Biddle and Azam, 1999], low dissolution rates during the first half of June suggest that the sedimenting diatoms were healthy cells. Indeed, microscopic examination revealed sinking assemblages dominated by long chains of intact diatoms, with no apparent colonization by bacteria. Accordingly, estimated sinking rates corresponded to those of healthy phytoplankton cells [Smayda, 1971]. Abundant sinking feces in July could also have enhanced silica preservation because of the presence of the peritrophic membrane; however, the high recycling rate of organic material would likely have counteracted this effect. It follows that sedimentation of fresh phytoplankton cells from the euphotic zone in June, together with abundant resting spores in July, most likely contributed to slow silica dissolution in the sinking material.

5. Conclusions

[41] This study provides the first data on sedimentation, as well as vertical degradation of the sinking material, in the NOW. The results show clear seasonal trends in the composition and amount of sinking organic matter during the spring and summer. In addition, the model approach combined with the observed seasonal trends point to the significance of biological processes associated with the diatom bloom in reducing Si dissolution rates.

[42] Specifically, this study clearly shows that preferential sinking export of BioSi was taking place during the spring and summer in the NOW. Three pelagic processes controlling this preferential export were apparent: (1) high sinking export of biosiliceous material from the euphotic zone, (2) high recycling of organic carbon and nitrogen in the upper 100 m, and (3) low dissolution of biogenic silica during its descent. Furthermore, the temperature model showed that temperature was not the only factor controlling Si dissolution in the sinking material and that biological factors played a key role. All results converge to show high biosiliceous sedimentation during the spring and summer in the NOW, suggesting that the polynya can act as a sink for biogenic silica. Estimates of Si dissolution rates at depth, using in situ temperatures close to bottom, generate values that are up to 50% higher than at surface. This supports the hypothesis that high sedimentation of biogenic silica, followed by dissolution at depth, could contribute to explain high dissolved silicon concentrations in deep waters of the NOW and Baffin Bay.

Acknowledgments

[43] This project was supported by a Research Network grant and by an individual research grant from the Natural Sciences and Engineering Research Council (NSERC) of Canada to M. G. During the expedition the first author's salary was paid from the central budget of the International North Water Polynya Study (NOW). The authors gratefully acknowledge the Canadian Coast Guard officers and crew of the CCGS Pierre Radisson for their invaluable support during the expedition. We also acknowledge the skillful support of the Polar Continental Shelf Project. We are especially indebted to M. Gagnon for his invaluable help during the expedition, S. Pesant, M.-È. Garneau, M. Holst, P. Larouche, P. Lee, C. J. Mundy, and A. Weise for technical assistance in the field or in the laboratory. We also thank J.-Y. Anctil, B. LeBlanc, and A. Gagné for logistical assistance, D. Bérubé for CHN analyses, M. Simard for cell and feces identification and counts, and D. Hamel for providing the bathymetric map. I. Walsh kindly provided the drifting trap cylinders. Special thanks go to J. Fauchot and J.-É. Tremblay for providing suspended data on biogenic silica and to Y. Gratton and B. VanHardenberg for providing water temperature data. We thank J.-É. Tremblay and two anonymous reviewers for helpful comments on the manuscript. This is a contribution to the research programs of the International North Water Polynya Study and the Institut des sciences de la mer de Rimouski.

Ancillary