Planktic foraminiferal sedimentation and the marine calcite budget

Transcription

1 GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 16, NO. 4, 1065, doi: /2001gb001459, 2002 Planktic foraminiferal sedimentation and the marine calcite budget Ralf Schiebel ETH-Zentrum, Swiss Federal Institute of Technology, Geological Institute, Zurich, Switzerland Received 3 July 2001; revised 30 May 2002; accepted 5 June 2002; published 24 October [1] The vertical flux and sedimentation rate of planktic foraminiferal tests are quantified and a global planktic foraminiferal CaCO 3 budget is presented. Test and calcite flux rates are calculated according to the distribution of species obtained from multinet and sediment trap samples. Modern planktic foraminiferal population dynamics are discussed as a prerequisite for the quantification of the calcite budget, highlighting the importance of ecological, autecological (e.g., reproduction), and biogeochemical conditions that determine the presence or absence of species. To complete the open-marine, particulate CaCO 3 inventory, the contribution of coccolithophores, pteropods, and calcareous dinophytes is discussed. Based on the studied regions, the global planktic foraminiferal calcite flux rate at 100 m depth amounts to Gt yr 1, equivalent to 23 56% of the total open marine CaCO 3 flux. The preservation of tests varies on a regional and temporal scale, and is affected by local hydrography and dissolution. During most of the year (off-peak periods), many tests dissolve above 700-m water depth while settling through the water column, with on average only 1 3% of the initially exported CaCO 3 reaching the deepseafloor. Pulsed flux events, mass dumps of fast settling particles, yield a major contribution of tests to the formation of deep-sea sediments. On average, 25% of the initially produced planktic foraminiferal test CaCO 3 settles on the seafloor. The total planktic foraminiferal contribution of CaCO 3 to global surface sediments amounts to Gt yr 1, 32 80% of the total deep-marine calcite budget. INDEX TERMS: 1050 Geochemistry: Marine geochemistry (4835, 4850); 3030 Marine Geology and Geophysics: Micropaleontology; 4806 Oceanography: Biological and Chemical: Carbon cycling; 4863 Oceanography: Biological and Chemical: Sedimentation; KEYWORDS: calcareous plankton, population dynamics, ecology, particle flux, micropaleontology, Paleoceanography Citation: Schiebel, R., Planktic foraminiferal sedimentation and the marine calcite budget, Global Biogeochem. Cycles, 16(4), 1065, doi: /2001gb001459, Introduction [2] Planktic foraminifers (protists) and coccolithophores (gold-brown algae) are major contributors to the particulate inorganic calcite flux of the deep ocean, varying on regional and temporal scales [Deuser et al., 1981; Ziveri et al., 1995, 2000; Broerse et al., 2000]. Calcification and dissolution of CaCO 3 produces changes in the surface water carbonate system, and deep-water masses are affected by the descent and differential dissolution of tests [e.g., Berger and Piper, 1972; Berger et al., 1982; Dittert et al., 1999; Dittert and Henrich, 2000]. Thermodynamic dissolution is evident below the lysocline, and below the calcite compensation depth (CCD) only a minor proportion of calcite is preserved [Broecker and Peng, 1982]. In addition, a significant amount of calcite dissolution takes place some distance above the calcite lysocline [Anderson and Sarmiento, 1994; Lohmann, 1995; Schiebel et al., Copyright 2002 by the American Geophysical Union /02/2001GB a; Milliman et al., 1999]. This may be caused by the remineralization of organic matter and decreasing ph within microenvironments. The decrease in test flux rate far above the planktic foraminiferal lysocline has hitherto not been sufficiently explained. On the other hand, calcareous particles settle below the CCD because they sink faster than they can be dissolved, and CaCO 3 is well preserved in the surface sediments [cf. Archer, 1996]. Although the faunal record is biased, planktic foraminifers possess a high fossilization potential and have high biologic, ecologic, paleontologic, and stratigraphic significance [e.g., Bé, 1977; Vincent and Berger, 1981; Hemleben et al., 1989]. [3] Planktic foraminiferal test production and flux have an impact on and are affected by the turnover of oceanic and atmospheric CO 2 [Hay, 1985; Veizer, 1985; Wolf- Gladrow et al., 1999a, 1999b]. When planktic foraminifers produce calcite from bicarbonate (HCO 3 ) or carbonate (CO 2 3 ), CO 2 is released to the ambient water [Zeebe and Wolf-Gladrow, 2001]. As CO 2 is the second most important greenhouse gas after water vapor, the production 13-1

2 13-2 SCHIEBEL: MARINE CALCITE BUDGET Figure 1. Locations of multinet and sediment trap samples from this study (Tables 1 3), and of trap data from the literature (Table 5). A, Azores, AS, Arabian Sea, B, North Atlantic including BIOTRANS (47 N, 20 W), C, Caribbean, GNS, Greenland-Norwegian Sea. More detailed maps of the areas are given by Schiebel et al. [2002a, 2002b; Azores], A. Zeltner et al. (unpublished manuscript; Arabian Sea), Schiebel et al. [2001; BIOTRANS], and Schmuker and Schiebel [2002; Caribbean]. 1, Reynolds and Thunell [1985]; 2, Fischer et al. [1983]; 3, Ziveri et al. [1995]; 4, Thunell and Reynolds [1984]; 5, Thunell and Honjo [1981]; 6, Marchant [1995]; 7, Wefer et al. [1982]; 8, Fischer et al. [1988]; 9, Bishop et al. [1977]; 10, Fischer et al. [1996]; 11, Deuser and Ross [1989]; 12, Haake et al. [1993]. of planktic foraminifers and coccolithophores may affect the climate on seasonal to geological timescales [cf. Bramlette, 1958; Hay, 1985; Peterson and Prell, 1985; Henrich, 1986; Siegenthaler and Sarmiento, 1993; Denman et al., 1996; Archer et al., 2000; Schiebel and Hemleben, 2000]. This article discusses the population dynamics, production, and sedimentation of planktic foraminiferal tests and CaCO 3, and gives a first-order estimate Table 1. Samples From the Caribbean and North Atlantic (BIOTRANS Area Around 47 N, 20 W) Between 1988 and 1999 (Arq = BO Arquipelago (Horta, Azores)) Cruise Year Month Location No. of Samples Meteor June NE Atlantic N/14 W 20 E 9 Meteor August NE Atlantic N/20 W 28 E 35 Meteor June July NE Atlantic 72 N/10 W 7 E 217 Meteor July NE Atlantic 67 N/3 E 17 W 63 Meteor May NE Atlantic 57 N/20 W 9 Meteor May June NE Atlantic 57 N/22 W 22 Meteor October NE Atlantic N/16 34 W 18 Meteor January BIOTRANS 47 N/20 W 18 Meteor March April BIOTRANS Meteor March May BIOTRANS Meteor April May BIOTRANS Meteor May BIOTRANS Meteor May June BIOTRANS Meteor May June BIOTRANS Poseidon 200/ June BIOTRANS Meteor July August BIOTRANS Meteor July August BIOTRANS Meteor September BIOTRANS Meteor September BIOTRANS Meteor October BIOTRANS Poseidon 247/ January Azores Front N/20 32 W 91 Poseidon 231/ August Azores Front N/ W 44 ArqFCA97C 1997 August Azores Front N/29 32 W 20 Meteor August Azores Front N/29 32 W 51 Meteor March April Subtropical Atlantic 33 N/20 W 8 Meteor May Subtropical Atlantic 33 N/20 W 13 Meteor May June Subtropical Atlantic 33 N/22 W 5 Meteor March April Subtropical Atlantic 18 N/30 W 5 Meteor April May Caribbean N/61 79 N 129 Total 1492

3 SCHIEBEL: MARINE CALCITE BUDGET 13-3 Table 2. Samples From the Arabian Sea in 1995 and 1997 a Cruise Year Month Location No. of Samples Meteor February Red Sea 50 Meteor March Gulf of Aden, off Oman 75 Meteor May off Oman, 0 21 N, 65 E, Seychelles 13 Meteor July August off Oman, WAST, CAST, SAST 18 Meteor September October NAST, WAST, CAST, EAST, SAST 52 Sonne March 65 E, SAST, CAST, WAST 26 Sonne April May NAST, WAST, CAST, SAST 26 Sonne May June NAST, WAST, CAST, EAST, SAST 25 Total 285 a Stations are located at 20 N, E (NAST), N, E (WAST), N, E (CAST), N, E (EAST), and 10 N, 65 E (SAST). of the planktic foraminiferal contribution to the modern global marine carbonate budget. 2. Materials and Methods [4] A total of 1777 multinet samples (>100 mm mesh size) and 27 sediment trap samples (>20 >100 mm) from the North Atlantic Ocean, the Caribbean, the Red Sea, and the Arabian Sea are included in the present study (Figure 1 and Tables 1 3). The sampling methods used for obtaining planktic foraminifers and pteropods by multinet and sediment trap, as well as the processing methods, are described in detail by Schiebel et al. [1995]. Standardized water depth intervals hauled with the multinet were 0 100, 0 700, and m, with five depth intervals each (see Table 4), enabling a direct comparison of data. The longest and bestresolved time series were obtained with a multinet device from the Eastern North Atlantic (BIOTRANS station, 47 N, 20 W) between 1988 and 1996, January through October (Table 1). The data from the Red Sea [Auras-Schudnagies et al., 1989; Bijma and Hemleben, 1994] and from the northern North Atlantic, including the Greenland-Norwegian Sea and the Arctic Ocean [Jensen, 1998; Volkmann, 2000], are included in the data set of planktic foraminiferal CaCO 3 flux presented in this study. Daily to interannual data sets were combined, and regional data sets from different oceans were calibrated for their taxonomy and processing mode. [5] Sediment trap samples (Table 3), deployed by the Marine Chemistry working group of the IfM Kiel [Lundgreen, 1996] were analyzed and flux rates of planktic foraminiferal tests and the resulting CaCO 3 are compared Table 3. CaCO 3 Flux Rates Determined From Sediment Trap Material a Mooring Trap No. Water Depth, m Year Cup Bulk Weight Split Weight Time Interval CD, mm plf Tests, plf CaCO 3, 10 3 day 1 mg m 2 d 1 L2-92-A > > > L2-92-A > > > L2-92-B > > > > > > > > > L2-92-B > > > L2-92-B > > > > > > L > > L > a Samples were recovered by the marine chemistry working group at IfM Kiel from the North Atlantic at N, W (L2) and 54 N, 21 W (L3). Weight of samples (bulk and split) is brutto dry weight (mg), including sea-salt. Census data (CD) are collected from different size fractions (>20 mm) according to the presence of tests. CaCO 3 weight of planktic foraminifers ( plf ) is calculated according to test weight given by Schiebel and Hemleben [2000].

4 13-4 SCHIEBEL: MARINE CALCITE BUDGET Table 4. Planktic Foraminiferal Test CaCO 3 Flux Rates a Latitude Longitude Cruise Month Atlantic N 5 10 E MET21/4 June N 11 W MET21/4 June N E MET21/4 June N 0 5 E MET10/3 June N 0 5 W MET10/3 June N 5 10 E MET10/3 June N 5 10 W MET10/3 June N 5 10 W MET10/3 July N 5 10 W MET21/5 July N 0 W MET21/4 June N 0 W MET21/5 July N 0 W MET21/5 July N 3 E MET21/5 July N 5 E MET21/4 June N 4 W MET21/4 June N 9 14 W MET21/4 June N W MET10/2 May N W MET21/3 May N W MET10/2 June N W MET11/1 Oct N W MET11/1 Oct N 34 W MET11/1 Oct N 27 W MET11/1 Oct N 20 W MET27/2 Jan N 20 W MET21/1 2 March N 20 W MET21/1 2 April N 20 W MET6/7 April N 20 W MET6/7 May N 20 W MET10/2 May N 20 W MET12/3 May N 20 W MET21/1 2 May N 20 W MET21/3 May N 20 W MET12/3 June N 20 W POS200/6 June N 20 W MET10/4 Aug N 20 W MET21/6 Aug N 20 W MET26/1 Sept N 20 W MET36/5 Sept N 20 W MET36/5 Oct N 20 W MET36/6 Oct N 20.5 W POS247/2 Jan N 30 W POS247/2 Jan N 30 W MET 42/3 Aug N 30 W POS231/3 Aug N 30 W FCA97C Aug N 22 W POS247/2 Jan N 22 W MET21/3 May N 22 W MET36/2 May N 22 W MET36/2 June N 22 W POS231/3 Aug

5 SCHIEBEL: MARINE CALCITE BUDGET 13-5 Table 4. (continued) Atlantic

6 13-6 SCHIEBEL: MARINE CALCITE BUDGET Table 4. (continued) Latitude Longitude Cruise Month Atlantic 33.5 N 26 W POS247/2 Jan N 31 W POS247/2 Jan N 31 W MET 42/3 Aug N 31 W FCA97C Aug N 34.5 W POS231/3 Aug N 30 W MET10/1 April Arabian Sea and Red Sea 27 N 35 E MET31/2 Feb N 60 E So119 May N 65 E Met32/3 May N 58.5 E So119 May N 60 E Met31/3 March N 60 E So117 March N 60 E So118 April N 60 E So119 May N 60 E Met33/1 Sept N 69 E Met33/1 Oct N E Met31/3 March N 65 E So119 May N 65 E Met32/5 Aug N 65 E Met33/1 Oct N E Met31/3 March N 44 E Met31/3 March N 65 E So117 March N 65 E So118 April N 65 E So119 May N 65 E Met33/1 Oct Caribbean 19 N 63.7 W MET35/1 April N 63.6 W MET35/1 April N 65 W MET35/1 April N 67.5 W MET35/1 May N 64 W MET35/1 April N 63.6 W MET35/1 April N 65.5 W MET35/1 May N 65.5 W MET35/1 May N 79.2 W MET35/1 May N 64.2 W MET35/1 April N 62.5 W MET35/1 April N 62.2 W MET35/1 April N 61.5 W MET35/1 April N 61.2 W MET35/1 April a Expressed in mg m 2 d 1 and given as monthly averages for different water depths and locations in the North Atlantic, Arabian and Red Seas, and the Caribbean. Flux rates refer to a test size of >100 and >125 mm in the case of METEOR cruise 10-2 and SONNE cruise 119.

7 SCHIEBEL: MARINE CALCITE BUDGET 13-7 Table 4. (continued) Atlantic Arabian Sea and Red Sea Caribbean

8 13-8 SCHIEBEL: MARINE CALCITE BUDGET Table 5. Planktic Foraminiferal ( plf ) Test CaCO 3 Flux Rates (g m 2 yr 1 ) as Deduced From Published Sediment Trap Data a Location Water depth, m plf CaCO 3, Flux % of total CaCO 3, Flux Test Size, mm Loc. Reference 47 N, 20 W B Honjo and Manganini [1993] N, 22 W A Honjo and Manganini [1993] N, 21 W >0.45 Broerse et al. [2000] 34 N, 21 W > N, 145 W surface Reynolds and Thunell [1985] (Station PAPA) N, 145 W Thunell and Honjo [1987] 39.5 N, 128 W >150 2 Fischer et al. [1983] N, W >125 3 Ziveri et al. [1995] N, W Thunell and Reynolds [1984] N, W >100 5 Thunell and Honjo [1981] N, W >100 5 Thunell and Honjo [1981] S, W Marchant [1995] S, W sea surface Wefer et al. [1982] S, W 3880 < 0.01? 8 Fischer et al. [1988] N, W >53 9 Bishop et al. [1977] N, W Fischer et al. [1996] N, W N, W N, W > Deuser and Ross [1989] > N, E >20 12 Koppelmann et al. [2000] N, E Haake et al. [1993] a Numbers refer to the location (Loc.) as given in Figure 1. Honjo and Manganini [1993; 33 N, 22 W] and Haake et al. [1993] give planktic foraminiferal flux rates as part of bulk CaCO 3 flux rate. Broerse et al. [2000] refer to the coccolith and calcareous dinophyte flux rate, allowing for estimates of the planktic foraminiferal test flux rate. with flux rates calculated from multinet samples. Both methods have inherent advantages and disadvantages in temporal and spatial sampling resolution. The sediment trap methodology, including trapping efficiency, has been thoroughly discussed [see Lundgreen, 1996, and references therein; Scholten et al., 2001]. Census data are available from the Pangaea database at the AWI (Bremerhaven, Germany, Planktic foraminiferal calcite flux rates from key regions of the global deep ocean have been included from the literature (Figure 1 and Table 5). [6] Flux rates of planktic foraminiferal CaCO 3 are calculated from the multinet material according to the speciesspecific and size-specific settling velocity [Takahashi and Bé, 1984], test weight, and test concentration [cf. Schiebel and Hemleben, 2000]. As the vertical heave velocity of the multinet (0.5 m s 1 ) is much higher than the sinking velocity of tests ( m s 1 ), the sampled test assemblage is regarded as being at steady state. According to the depth distribution of live specimens and empty tests, the flux rate data provided here are based on the total planktic foraminiferal assemblage from the depth interval of m and below. Flux rate data refer to the lower depth limits of the MCN hauls. For calculation of CaCO 3 flux rates from the sediment trap material, including planktic foraminiferal specimens smaller than 100 mm, the weight of small specimens is calculated according to the allometric relation w = a + exp(c(x b)) where w (mg) denotes the weight of tests and x (mm) denotes test diameter, with a = 0.54, b = 73.88, and c = (n =9,r 2 = 0.997). 3. Planktic Foraminiferal Population Dynamics 3.1. Living Fauna [7] In the eastern North Atlantic (Table 1: BIOTRANS), low planktic foraminiferal abundance in late January and early February is assumed to represent the overall winter situation (Figure 2 and Table 4). The first increase in individual numbers occurs during March at depths of m, which is well displayed by the distribution of Globigerina bulloides (Figure 3), and possibly results from the enhanced availability of food. Maximum frequency in spring results from wind-driven water mixing, improved feeding conditions, and increased growth rates in the upper water column [Schiebel et al., 1995]. Due to mass sedimentation and scavenging, shallow-dwelling species also occur below the seasonal thermocline (Figure 2a). During

9 SCHIEBEL: MARINE CALCITE BUDGET Figure 2. Average annual distribution of (a) living specimens and (b) empty planktic foraminiferal tests per m3 in the upper 2500 m at BIOTRANS. Resolution of interpolated data (n = 432) is 14 days and 100 m. Black and gray levels correspond to >200, > , >30 100, >10 30, >5 10, and <5 specimens m 3. White = not sampled. Figure 3. Average annual distribution of G. bulloides at BIOTRANS. Note that the depth axis comprises only the upper 300 m of the water column. Deepest occurrence of G. bulloides corresponds to wind driven mixing [Schiebel et al., 1995]. During summer, G. bulloides is rare and increases in number again during fall [Schiebel et al., 2001]. Black = >120 specimens m 3, gray levels correspond to >60 120, >30 60, >15 30, and <15 specimens m 3. White = not sampled. Data are interpolated at 20 m and 14 days intervals. 13-9

10 13-10 SCHIEBEL: MARINE CALCITE BUDGET Figure 4. Average annual distribution of (a) living specimens and (b) empty planktic foraminiferal tests per m 3 in the upper 2500 m at WAST. Resolution of interpolated data (n = 41) is 14 days and 100 m. Gray levels correspond to >200, > , >30 100, >10 30, >5 10, and <5 specimens m 3. White = not sampled. spring, the so-called deep-dwelling species ascend to shallow waters and do not contribute to the stock of cytoplasmbearing specimens below 500 m. In contrast, during late spring and summer (May through August), deep-dwelling G. scitula and G. hirsuta live below 100 m and are a major part of the subsurface fauna. During summer (June July), highest standing stocks occur in the upper m, and Neogloboquadrina incompta and Turborotalita quinqueloba are major components of the shallow dwelling fauna. Large numbers of living and dead specimens in the fall result from improved feeding conditions due to redistribution of chlorophyll to surface waters, entrainment of nitrate, and increased primary production in the mixed layer [Schiebel et al., 2001]. During October, aphotic conditions occur, feeding conditions decline, and planktic foraminiferal production decreases. [8] Strong seasonality also prevails in the Arabian Sea (Table 2: WAST), where the highest planktic foraminiferal numbers are found during the late NE monsoon (March) and during the SW monsoon (June September) (Figure 4). During the monsoon seasons, the food availability for planktic foraminifers at WAST is improved compared with the intermonsoonal periods due to upwelling off Oman and resulting filaments that move toward the open ocean. [9] Most planktic foraminifers are restricted to the upper m [cf. Berger, 1969; Watkins et al., 1996; Hemleben et al., 1989] and to the upper 200 m under exceptional circumstances, e.g., periods of wind-driven deep mixing [Schiebel et al., 1995]. Cytoplasm-bearing specimens that are transported below their normal habitat by turbulent mixing can no longer subsist due to a lack of food or low radiation (in the case of symbiont bearing species), and are part of the vertical test flux. Below m water depth, the living:dead ratio (cytoplasm bearing specimens versus empty tests) decreases exponentially from 10 to 0.1 at m depth. Emerson et al. [1997] assume that significant amounts of empty, sinking tests are not present in the upper ocean. So-called deep-dwelling species such as G. truncatulinoides are found below 100 m, or at least below the mixed layer, during most of the year [cf. Hemleben et al., 1989]. However, deep-dwelling species are much less abundant than shallow-dwelling species [Bé, 1977] Concentration of Empty Tests in the Water Column [10] Increased numbers of empty and sinking tests generally result from increased growth rates in the surface

11 SCHIEBEL: MARINE CALCITE BUDGET Figure 5. (a) Planktic foraminiferal CaCO 3 flux rate (mg m 2 day 1 ) and (b) test flux rate (10 3 tests m 2 day 1 ) according to sediment traps moored at BIOTRANS between 1000 and 3530 m water depth (Table 3). Sampling intervals are marked by crosses. Differences between flux rates calculated from the trap material and multinet samples (Figure 8) may result from variations in the timing and the water depths of sampling (Tables 1 and 3). waters. The largest empty-test numbers within the upper 2500 m during spring at BIOTRANS (Figure 2b) result from mass sedimentation [cf. Anderson and Sarmiento, 1994]. An isolated, empty-test patch at m during July August (Figure 2b) is formed mainly by small ( mm) and slowly sinking tests of T. quinqueloba. A disproportionately large number of T. quinqueloba tests is also observed during September, when this species constitutes about 50% of the sediment trap assemblage (Figure 5) while T. quinqueloba makes up only 5 15% of the living fauna and 10% of the surface sediment assemblage. [11] It is conceivable that small tests (e.g., of T. quinqueloba) accumulate in mesobathyal waters at BIOTRANS due to deceleration as a result of increasing viscosity of the seawater. The viscosity of seawater increases with increasing salinity and decreasing temperature and pressure [Dietrich et al., 1975]. At BIOTRANS, the viscosity of seawater may increase as a result of decreasing temperature down to 1500 m [cf. van Aken, 2000]. Increased viscosity between 700 and 1200 m may also be linked to enhanced salinity, indicating the presence of Mediterranean Sea outflow water (MSW) that was repeatedly revealed by CTD measurements during the sampling (Table 1). Below the MSW, decreased viscosity causes increased velocity of sinking tests and the accumulation of small tests dissipates. [12] At WAST, the number of empty tests in the deepwater column is much higher than at BIOTRANS (Figures 2 and 4), although planktic foraminiferal standing stocks in surface waters are similar at both sites during the SW monsoon and during spring, respectively. This discrepancy may be due to the better preservation of settling tests within the oxygen minimum zone (OMZ) of the Arabian Sea [cf. Hermelin, 1992] than in the well-oxygenated water column of the eastern North Atlantic. 4. Fluxes of Planktic Foraminiferal Tests and CaCO Differential Test and CaCO 3 Flux Rates [13] As a consequence of ecological and autecological prerequisites, planktic foraminiferal test and calcite fluxes display complex, intermittent pulses [e.g., Sautter and Thunell, 1989; Bijma et al., 1994]. According to sediment trap investigations, the highest CaCO 3 flux rates at BIO- TRANS (1000 m depth) are observed during spring when test production in the surface waters is highest (Figure 5) [cf. Honjo and Manganini, 1993]. With increasing depth, a shift of maximum flux rates from spring toward summer (July August) takes place, reflecting the settling velocity of planktic foraminiferal tests. A comparison of CaCO 3 and test flux rates reveals that during times of maximum

12 13-12 SCHIEBEL: MARINE CALCITE BUDGET Figure 6. Planktic foraminiferal CaCO 3 flux rates (mg m 2 d 1 ) given as monthly averages between 100 and 2500 m. Data represent the lower level of each water depth interval. Panel (a) shows an enlarged view of the flux rates <20 mg m 2 d 1 given in (b). CaCO 3 flux rates (Figure 5a) only moderate test flux rates (Figure 5b) occur at 1000 m depth. In contrast, during June July, the highest test flux rates occur when the CaCO 3 flux rate at around m is moderate. Maximum test flux rates at 2000 m during June July (Figure 5b) are mainly caused by the small sized species T. quinqueloba. The difference between test and calcite flux rates indicates that comparatively few large tests dominate the spring CaCO 3 flux pulse and that a large number of small tests, which settle through the water column at comparatively low velocity, constitute the CaCO 3 flux pulse during the late summer [cf. Deuser, 1987]. The summer test pulse obtained by the sediment traps and multinet sampling (Figure 2b) mainly consists of tests mm in size. Both large and small tests predominantly result from enhanced spring production Planktic Foraminiferal CaCO 3 Flux Modes [14] In the eastern North Atlantic at 47 N 57 N, maximum test production at depths >200 m during the spring bloom (Figures 6 and 7 and Table 4), causes highest flux rates within the upper 500 m, and distinct flux pulses in the deeper water column (Figure 7; 1500 m depth). During May, flux rates decrease but flux pulses occasionally still occur in the deeper water column. In the Azores region, low flux rates during August point toward oligotrophic conditions, and high flux rates and a distinct CaCO 3 flux pulse at 1000 m depth during January are due to large numbers of G. truncatulinoides [Schiebel et al., 2002a, 2002b]. Comparatively, balanced flux rates in the Caribbean display a less distinct seasonality than in the high latitudes. [15] Differential sinking velocities of planktic foraminiferal tests result in different settling tracks that can be traced through the water column (Figure 8). Following times of biological mass production in March April and September, a vast number of large, quickly settling tests occurs in the deep-water column during May and October. At the same time, many small tests ( mm) occur in deep waters that are scavenged by larger particles. Nevertheless, the majority of the small tests settle through the water column much more slowly than assumed from the test size and weight alone [cf. Takahashi and Bé, 1984]. Consequently, a cloud of slow-sinking small tests is present during summer (June August) at around 1000 m depth (Figures 2 and 8). Deep-dwelling G. scitula contribute a minor part to the enhanced calcite budget at depths, and other deepliving species are virtually absent during summer at BIO- TRANS. Slightly enhanced CaCO 3 flux rates at m during summer are caused by small sized and slow settling T. quinqueloba and N. incompta and have no significant impact on the test cloud at 1000 m depth. In autumn, increased production causes CaCO 3 flux pulses and tests are scavenged from the cloud of tests [Schiebel et al., 2001], similar to the particle dynamics described by Thomsen and McCave [2000] from the bottom nepheloid layer. [16] In the Arabian Sea, relatively high flux rates occur in the water column below 500 m depth, particularly in the upwelling area off Oman and in the southern Arabian Sea

13 SCHIEBEL: MARINE CALCITE BUDGET Figure 7. Average daily planktic foraminiferal CaCO3 flux rates from different months at BIOTRANS (multinet data). Note that the strongest decrease in CaCO3 flux rates occurs in water depths above 700 m. Increasing test flux rates in deep waters may result from deceleration due to disaggregation and accumulation of small tests [cf. Berger and Wefer, 1990; Ransom et al., 1998]. Particles of different size and settling velocity are captured with differential efficiency [e.g., Deuser, 1986]. (Figure 6). At WAST, a seasonally pulsed CaCO3 flux is observed (Figure 9) corresponding to the late stages of the NE and SW monsoons during March and July September, respectively [cf. Rixen et al., 2000]. In the deep-water column, maximum planktic foraminiferal flux rates are delayed and were observed in April at 1500 m, and during August September at and 2500 m depth (Figure 9). The seasonally increased CaCO3 flux rates at Figure 8. Planktic foraminiferal flux rate (CaCO3 mg m 2 d 1) at BIOTRANS between 100 and 2500 m, according to monthly average flux data from different years. May was sampled over 5 years. No samples were obtained during February, November, and December (Table 1). The paths of relatively fast (a) and slow (b) settling tests are indicated by arrows. Black and gray levels correspond to flux rates >60, >30 60, >10 30, >3 10, >1 3, and <1 mg m 2 d 1. White = not sampled.

14 13-14 SCHIEBEL: MARINE CALCITE BUDGET Figure 9. Planktic foraminiferal CaCO3 flux rates (mg m 2 d 1) at WAST in the Arabian Sea display pulsed events. The upper and lower limit of the oxygen minimum zone is at about 120 and 1200 m depth [Hermelin, 1992]. Black and gray levels correspond to flux rates >60, >30 60, >10 30, >3 10, and >1 3 mg m 2 d 1. White = not sampled. Note that flux rates of <1 mg m 2 d 1 were not observed in the Arabian Sea, in contrast to the North Atlantic. WAST are much higher than at BIOTRANS, and likely result from better preservation of tests within the OMZ than in the well oxygenated water column, respectively Mass Sedimentation of Planktic Foraminiferal Tests [17] Disproportionally high flux rates of >1 g CaCO3 m 2 1 d at m depth in the Arabian Sea occurred during March 1995 around the new moon (Figures 10 and 11, and Table 4). This CaCO3 flux pulse was mainly caused by large tests of G. sacculifer (> mm). Although G. sacculifer is frequent in the Arabian Sea [Auras-Schudnagies et al., 1989; Naidu and Malmgren, 1996; Conan and Brummer, 2000], mass sedimentation of large tests of G. sacculifer was observed only once during the study presented here. Production and flux of G. sacculifer are related to the synodic lunar cycle [Almogi-Labin, 1984; Bijma et al., 1990, 1994; Erez et al., 1991]. Therefore although the observed mass flux event does not display the average sedimentation ( steady particle rain ), mass sedimentation is characteristic of the deep-marine environment [cf. Anderson and Sarmiento, 1994], though under-represented in sediment trap and multinet samples. The temporal resolution of most deep traps ( 1000 m water depth) is too low to record single mass flux events. Shallow sediments traps ( 1000 m depth) have a low trapping efficiency [Scholten et al., 2001] and are not included here. Mass flux events may not be detected by sediment traps even if the test flux rates double or triple within a short time-interval, because the relatively small amount of large planktic foraminiferal tests yield no statistical significance. In contrast, the quantity of large tests obtained by multinet sampling is much larger than in trap samples and variation in standing stocks and flux rates is statistically significant. However, mass flux events are highly unpredictable and met only by chance with net hauls. [18] Compared with the live fauna, surface sediments of the Arabian Sea and other tropical to subtropical ocean basins contain a disproportionately high portion of large tests [Peeters et al., 1999]. These tests may result from events like the mass sedimentation previously described. Pulsed flux events seem to yield a major contribution to the deep-sea sediment accumulation and remove bicarbonate on long-term timescales from the upper ocean and transfer it to deep-sea sediments [cf. Wefer, 1989; Berger and Wefer, 1990]. Mass sedimentation of large tests, however, requires the presence of species that have the autecological prerequisites to form large tests [Brummer et al., 1987; Hemleben et al., 1987; Caron et al., 1990] and that are adapted to specific ecologic conditions [Bijma et al., 1990; Huber et al., 2000; Kemp et al., 2000]. To understand such processes, detailed knowledge of population dynamics is crucial Planktic Foraminiferal CaCO3 Flux Rates [19] Calcite flux rates determined from the total set of multinet and sediment trap data (Tables 1 3), including literature data (Table 5), range between <0.001 and >2000 mg m 2 d 1 (Figure 10). Flux rates span more than four orders of magnitude in each multinet depth interval, and the upper to lower quartile of flux rates covers about one order of magnitude (Figure 11). Flux rates within the upper 100 m of the water column are similar for each of the five investigated depth intervals and clearly result from complete vertical mixing. The most significant decrease in flux rates takes place between 100 and 700 m depths. Between 700 and 2500 m, and possibly below 2500 m, only small changes in flux rates occur (Figures 6, 7, and 11). Outliers with flux rates of >1000 mg m 2 d 1 between 1000 and 2500 m (Figure 11) result from mass sedimentation of G. siphonifera and G. sacculifer Dissolution of Planktic Foraminiferal Tests [20] The most significant decrease in planktic foraminiferal test flux rates between 100 and 700 m takes place at depths where thermodynamic calcite dissolution does not occur [cf. Broecker and Peng, 1982]. This decrease in flux

15 SCHIEBEL: MARINE CALCITE BUDGET where the empty-test patch occurred in the eastern North Atlantic during July August (Figure 2), bacterially mediated dissolution of tests may be lower than above or may have already ceased. [21] Dissolution of planktic foraminiferal tests within aggregates of marine snow is unlikely, although marine snow consists of organic matter and microbes, and contains planktic foraminiferal tests [Ransom et al., 1998]. Sedimentation of marine snow aggregates is probably too fast to allow for significant calcite dissolution (H. Jansen, University of Hamburg, written communication, October 2001). To conclude, dissolution of planktic foraminiferal tests in waters that are supersaturated with respect to calcite is hitherto not sufficiently explained Planktic Foraminiferal Flux Versus Pteropod, Calcareous Dinophyte, and Coccolith Flux [22] The planktic foraminiferal, pteropod, and coccolith contribution to the total calcareous particle flux is assessed to estimate the composition of the total planktic carbonate budget (cf. Table 5). Fischer et al. [1996] state that each Figure 10. CaCO 3 flux rates determined from multinet samples (crosses; n = 1864; Tables 1, 2, and 4). Rates between the sea surface and 4500 m water depth span more than eight orders of magnitude. Data are given for the lower level of each water depth interval. Low flux rates of <0.1 mg m 2 d 1 determined from multinet samples in the North Atlantic occur during late summer and winter. Flux rates determined from sediment traps below 1000 m have ranges similar to the upper limit of multinet data. Trap data (circles) from this study are on average slightly higher than data deduced from the literature (diamonds). rates possibly results from increased bacterially mediated decomposition of cytoplasm and a decreasing ph in the microenvironments within foraminiferal tests [Schiebel et al., 1997b; cf. Turley and Stutt, 2000]. Bacterial activity is limited by oxygen, and bacterially mediated decomposition of cytoplasm is possibly less significant in low-oxygen environments than in well-oxygenated waters. According to Milliman et al. [1999], calcite dissolution far above the lysocline is probably biologically mediated, and takes place within the guts of grazers [cf. Jansen and Wolf-Gladrow, 2002] or is related to organic matter degradation. Both grazers and remineralization depend on oxygen and, therefore, calcite preservation is assumed to be better in lowoxygen environments, as demonstrated by differential flux modes in the North Atlantic and Arabian Sea (Figure 6). In the upwelling area off Oman and in the central and southern Arabian Sea, high flux rates of planktic foraminiferal tests within the prominent OMZ point toward a high degree of calcite preservation. The low oxygen content within the OMZ of Arabian Sea waters may permit limited decomposition of organic material and less dissolution of calcite than in the eastern North Atlantic. At m depth, Figure 11. Boxplots of planktic foraminiferal CaCO 3 flux rates, including multinet data obtained from the North Atlantic, Arabian Sea, and the Caribbean (Tables 1 and 2). Only depth intervals with n > 50 are included. Note that the upper to lower quartile of flux rates in water depth between 0 and 200 m may exceed one order of magnitude. Boxes cover the upper and lower quartile, with horizontal lines for the upper and lower adjacent values. Outliers are marked by crosses.

16 13-16 SCHIEBEL: MARINE CALCITE BUDGET Figure 12. Calculated primary productivity and exported primary production (g C m 2 yr 1 ) obtained from the studies of Berger et al. [1988] and Berger [1989]. Five out of six intervals of (a) different productivity are well correlated with (b) the planktic foraminiferal calcite flux rates (g CaCO 3 m 2 yr 1 ). For comparison see Figure 1 and Table 4. group, planktic foraminifers, coccoliths, and pteropods, constitutes about one-third of the total carbonate flux off Cape Blanc, West Africa [see also Kalberer et al., 1993]. High flux rates of pteropods seem to be restricted to distinct pulses with high temporal variability, indicating patchy distribution, and are not yet well understood [cf. Fischer et al., 1983; Wefer and Honjo, 1985; Fischer et al., 1996; Honjo, 1996]. On a global scale, pteropods account for 10% of the total planktic carbonate production [Fabry, 1990], and, due to high sinking velocity (R. Schiebel, unpublished data), most of the shells may arrive at the seafloor. In deep-sea sediments, pteropods are only sporadically present due to their dissolution susceptible, aragonitic shells [e.g., Almogi-Labin et al., 1986; Auras-Schudnagies et al., 1989]. [23] Calcareous dinophytes constitute probably only a minor part of the flux within the calcareous fine fraction in modern oceans [Broerse et al., 2000]. In the South Atlantic, calcareous dinophytes form up to 3.5% of deepmarine sediments (A. Vink, oral communication, Bremen University, 2001), the only number on modern sediment available to date. [24] The proportion of planktic foraminifers and coccoliths within the total carbonate flux varies on large regional and temporal scales (Panama Basin [Honjo, 1982]; Equatorial Pacific [Dymond and Collier, 1988]; Sargasso Sea [Deuser et al., 1995]; Southern California [Ziveri et al., 1995]). Broerse [2000] estimates that coccoliths account for 4 38% of the total CaCO 3 flux rate in the different ocean basins, with a global mean of only 12% [cf. Beaufort and Heussner, 1999]. Planktic foraminifers are assumed to constitute between 2 and 100% of the calcareous particle flux within different regions of the world s ocean (Table 5) The Global Planktic Foraminiferal CaCO 3 Flux [25] The sample locations within the Atlantic, Caribbean, and Arabian Sea (Figure 1) cover a wide range of productivity regimes (Figure 12) [Berger et al., 1988; Berger, 1989; Antoine et al., 1996] and allow for a first-order estimate of the global planktic foraminiferal calcite budget. According to Berger [1989] and Milliman [1993], provinces of varying primary productivity and calcium carbonate sedimentation mainly differ with respect to latitude and individual hydrographic conditions (e.g., upwelling). An estimate of global planktic foraminiferal calcite flux rate is possible considering population dynamics, regional calcite flux rates, and hydrography. Regional variations in productivity and carbon export are identified on a global scale by Berger et al. [1988] and coincide well with the regional variability of planktic foraminiferal production and test flux rate (Figure 12). [26] Detailed maps of estimated primary productivity created from satellite chlorophyll observations [Antoine et al., 1996] appear to be similar to the maps of vertically exported primary production given by Berger et al. [1988] and Berger [1989], although they are not linearly proportional and differ remarkably in detail. For example, in the upwelling area off Oman (Arabian Sea), the rate of primary productivity and exported primary production differ significantly, and the shallow (100 m) planktic foraminiferal