Paleoceanography

Low-latitude control on seasonal and interannual changes in planktonic foraminiferal flux and shell geochemistry off south Java: A sediment trap study

Authors


Abstract

[1] Results from sediment trap experiments conducted in the seasonal upwelling area off south Java from November 2000 until July 2003 revealed significant monsoon-, El Niño–Southern Oscillation–, and Indian Ocean Dipole–induced seasonal and interannual variations in flux and shell geochemistry of planktonic foraminifera. Surface net primary production rates together with total and species-specific planktonic foraminiferal flux rates were highest during the SE monsoon-induced coastal upwelling period from July to October, with three species Globigerina bulloides, Neogloboquadrina pachyderma dex., and Globigerinita glutinata contributing to 40% of the total foraminiferal flux. Shell stable oxygen isotopes (δ18O) and Mg/Ca data of Globigerinoides ruber sensu stricto (s.s.), G. ruber sensu lato (s.l.), Neogloboquadrina dutertrei, Pulleniatina obliquiloculata, and Globorotalia menardii in the sediment trap time series recorded surface and subsurface conditions. We infer habitats of 0–30 m for G. ruber at the mixed layer depth, 60–80 m (60–90 m) for P. obliquiloculata (N. dutertrei) at the upper thermocline depth, and 90–110 m (100–150 m) for G. menardii in the 355–500 μm (>500 μm) size fraction corresponding to the (lower) thermocline depth in the study area. Shell Mg/Ca ratio of G. ruber (s.l. and s.s.) reveals an exponential relationship with temperature that agrees with published relationships particularly with the Anand et al. (2003) equations. Flux-weighted foraminiferal data in sediment trap are consistent with average values in surface sediment samples off SW Indonesia. This consistency confirms the excellent potential of these proxies for reconstructing past environmental conditions in this part of the ocean realm.

1. Introduction

[2] Preserved planktonic foraminifera in the geologic records are widely used to reconstruct past ocean conditions. A comprehensive use of this organism group requires a better understanding of the relationship between living foraminiferal distribution and oceanographic conditions. In addition to laboratory experiments [e.g., Erez and Luz, 1983; Spero and Lea, 1996; Bemis et al., 1998, 2000] and plankton tow studies [e.g., Fairbanks et al., 1980; Erez and Honjo, 1981; Williams et al., 1981; Bouvier-Soumagnac and Duplessy, 1985], sediment trap studies have contributed significantly to our understanding of the planktonic foraminiferal ecology within the past decades [e.g., Erez and Honjo, 1981; Thunell et al., 1983; Sautter and Thunell, 1991; Curry et al., 1992; Ortiz and Mix, 1997; Kuroyanagi et al., 2002]. Planktonic foraminiferal flux studies provide a unique opportunity to determine the possible impact of seasonal and depth distribution of foraminifera on sedimentary interpretations and thus, contribute to a better reconstruction of past oceanographic and climate changes in a particular region.

[3] Sediment trap studies also allow for comparing Mg/Ca ratio in planktonic foraminifera with the temperature data collected during the same experiment period. Mg content in shells of planktonic foraminifera have been recognized to be dependent on seawater temperature, either linearly [e.g., Hastings et al., 1998; Rosenthal and Lohmann, 2002] or exponentially [e.g., Nürnberg et al., 1996; Lea et al., 1999; Elderfield and Ganssen, 2000; Dekens et al., 2002; Anand et al., 2003; McConnell and Thunell, 2005]. These studies have provided different shell Mg/Ca to temperature relationships for a variety of species, size fractions, and temperature ranges.

[4] The Indonesian region is a key area along the return branch of the global conveyor belt and the only low-latitude pathway between two ocean basins, with major climatic importance on global scale. For instance, small changes in the sea surface temperature (SST) within the Indonesian Archipelago can result in significant changes in the thermocline structure of the Indian Ocean and the South Atlantic [Hirst and Godfrey, 1993], and in global heat exchange [Gordon, 1986; Gordon et al., 1992]. Continuous sediment trap series in and around the Indonesian Archipelago are rare, and almost entirely confined to the northern and eastern marginal seas, in particular to the South China Sea (see review by Wang et al. [2005]). In a recent sediment trap study from the South China Sea, Huang et al. [2008] observed partial dissolution of shell planktonic foraminifera altering their Mg/Ca ratios. The only sediment trap study from the southern part of the Indonesian Archipelago reports on the organic carbon, opal and carbonate fluxes off south Java between November 2000 and November 2002 and shows a strong control of the East Asian Monsoon (EAM) and El Niño–Southern Oscillation (ENSO) on these parameters on seasonal and annual time scales [Rixen et al., 2006].

[5] In this study, we examine in detail the seasonal and interannual variations in the flux of planktonic foraminifera through the water column off south Java using material collected by sediment traps over a period of almost 3 years. We also present paired stable oxygen (δ18O) isotopes and Mg/Ca measurements of Globigerinoides ruber sensu stricto (hereafter G. ruber s.s.), G. ruber sensu lato (hereafter G. ruber s.l.), Neogloboquadrina dutertrei, Pulleniatina obliquiloculata, and Globorotalia menardii. We further evaluate the habitat depth for each species on the basis of predicted and measured δ18O values, and Mg/Ca:SST equations for the first two species. We include new and previously published [Rixen et al., 2006] calcium carbonate (CaCO3) fluxes in the same sample set and compare trap fluxes with satellite-based surface productivity estimates and surface sediment data in the vicinity of the sediment traps [Mohtadi et al., 2007] in order to evaluate the link between primary production, seasonality, and foraminiferal export and preservation in this part of the Indian Ocean.

2. Background

[6] The Indonesian Archipelago is situated within the Indo-Pacific Warm Pool (IPWP) with annual mean SSTs exceeding 28°C. On interannual time scales, largest SST variability in the Indonesian Archipelago occurs in the coastal region off Java and Sumatra (>4°C) as a result of strong remote influence of the equatorial Indian Ocean via the Indian Ocean Dipole (IOD) (or zonal) mode [Saji et al., 1999; Webster et al., 1999]. The latter studies suggest the SST variability off Java and Sumatra to be a key component of the IOD with a great impact on regional climate, biological productivity, and economy. In addition, ENSO variability influences the hydrographic regime in this area by causing a weakened transport of the Indonesian Throughflow (ITF) and severe droughts in Sumatra and Java during El Niño events when the mean position of the Inter Tropical Convergence Zone (ITCZ) moves toward the equator [Ropelewski and Halpert, 1987; Gordon, 2005]. The ITF flows because of a pressure gradient between the Pacific and the Indian Ocean (Figure 1) and its volume is an important factor influencing water temperature in the depths of the thermocline in the region, but also SST in the upwelling areas south of Java and southwest of Sumatra [Ffield et al., 2000; Gordon, 2005; Tomczak and Godfrey, 1994]. Today, about one third of positive IOD events (cooling off Sumatra and Java) are associated with ENSO events, but they occur also during La Niña and ENSO neutral years [Ashok et al., 2003].

Figure 1.

Position of the sediment traps JAM1–3 (star) with the mean oceanographic surface (dashed arrows) and subsurface (solid arrows) currents. Gray stripes indicate the approximate spatial extension of the Java upwelling. Black dots symbolize the position of the surface sediment samples beneath the upwelling area [from Mohtadi et al., 2007]. ECC, Equatorial Counter Current; ITF, Indonesian Throughflow; LC, Leeuwin Current; SEC, South Equatorial Current; SJC, South Java Current.

[7] Another key element in the regional climate system is the EAM that is largely responsible for seasonal changes in the onshore precipitation and surface circulation of the Indonesian seas [e.g., Qu et al., 2005; Sprintall and Liu, 2005]. During the late northwest monsoon season (February–April), the precipitation rates over Indonesia are some of the highest in the world resulting in maximum riverine sediment discharge [Milliman et al., 1999]. Late in the southeast monsoon season (August–October), onshore precipitation is relatively low concomitant with maximum coastal upwelling off south Java and Sumatra (Figure 1). The cooling entrainment associated with upwelling is counterbalanced by the relatively warm ITF, especially the outflow from the Lombok Strait, which reaches its seasonal maximum during July–September, at the same time that the maximum upwelling occurs [e.g., Du et al., 2005]. Experiments with an ocean general circulation model suggest that another reason for the relatively small temperature reduction off Java and Sumatra is the formation of the barrier layer, representing an intermediate layer that separates the base of the mixed layer from the top of the thermocline [Lukas and Lindstrom, 1991]. The effect of salinity stratification caused by large rainfall and runoff sustains a shallow mixed layer, but a thick barrier layer, which impedes cold thermocline water from entering the mixed layer [Du et al., 2005; Qu et al., 2005]. These two mechanisms have been proposed to explain why the SST depression off Java and Sumatra is small compared with other eastern boundary upwelling regions.

[8] Although monsoon forcing and tidal mixing are important and explain a significant part of the mean seasonal SST cycle, the role of these processes become less important on interannual time scales. The interannual SST variability off Sumatra and Java appears to be largely driven by remote influence of the tropical Pacific (ENSO, ITF) and Indian (IOD) Oceans [Qu et al., 2005]. During El Niño or positive IOD years, reduced precipitation rates hinder the build-up of the barrier layer off south Java and W Sumatra making the erosion of this layer and the entrainment of cold, nutrient-rich subsurface water during the upwelling season easier. Additionally, a reduced ITF and abnormal winds in the SE monsoon season result in a more intense and extended upwelling period during the El Niño and positive IOD years [Susanto et al., 2001].

[9] Ocean currents in the study area move according to the wind regime. During the NW monsoon season, the South Java Current (SJC), derived from the Equatorial Counter Current (ECC), moves toward the southeast to meet the Leeuwin Current (LC), a narrow strip of warm, saline waters that originates in the eastern part of the Indonesian Archipelago [Tomczak and Godfrey, 1994]. The mixing of the SJC and the LC gives origin to the South Equatorial Current (SEC) that moves westward at ∼20°S (Figure 1). In the SE monsoon season, the SJC takes an opposite direction flowing northwestward and feeds the SEC without a significant contribution of the LC. Advection of fresher Java Seawater through the Sunda Strait and runoff from Sumatra and Java are responsible for the low-salinity “tongue” in the SJC with salinities as low as 32‰.

3. Material and Methods

3.1. Sediment Trap Moorings

[10] Two sediment traps have been deployed off south Java between November 2000 and November 2002 (JAM1–2, 8°17.5 S, 108°02.0 E, at 2200 m water depth) and November 2002 and July 2003 (JAM3, 8°16.1 S, 108°08.5 E, at 2460 m water depth). The traps were located about 600 m (JAM3) to 830 m (JAM1–2) above the seafloor signifying negligible influence of resuspension. Sampling intervals varied in general between 16 and 18 days (Table 1 and Figure 1). Before deployment, sample bottles were filled with seawater at 1800 m depth. To prevent degradation of trapped material mercury (II)-chloride (3.3 g l−1) was added to the cup water. Recovered samples were immediately refrigerated onboard at approximately 2–4°C.

Table 1. Sampling Dates and Durations in the Sediment Trap Time Series JAM1–3a
Sample NumberTrap CupExperiment Period Adjusted to the Settling Time and Life Cycle of Planktonic Foraminifera
SeriesCupOpenCloseDaysStartEndMonsoon
  • a

    The collection periods are adjusted to production and settling time of planktonic foraminifera (columns 6–8, see text for more details).

JAM-1D111 Nov 200029 Nov 20001824 Oct 200025 Nov 2000non
JAM-1D229 Nov 200017 Dec 20001811 Nov 200013 Dec 2000NW
JAM-1D317 Dec 20004 Jan 20011829 Nov 200031 Dec 2000NW
JAM-1D44 Apr 200122 Jan 20011817 Dec 200018 Jan 2001NW
JAM-1D522 Jan 20019 Feb 2001184 Jan 20015 Feb 2001NW
JAM-1D69 Feb 200127 Feb 20011822 Jan 200123 Feb 2001NW
JAM-1D727 Feb 200117 Mar 2001189 Feb 200113 Mar 2001non
JAM-1D817 Mar 20014 Apr 20011827 Feb 200131 Mar 2001non
JAM-1D94 Apr 200122 Apr 20011817 Mar 200118 Apr 2001non
JAM-1D1022 Apr 200110 May 2001184 Apr 20016 May 2001non
JAM-1D1110 May 200128 May 20011822 Apr 200124 May 2001non
JAM-1D1228 May 200115 Jun 20011810 May 200111 Jun 2001non
JAM-1D1315 Jun 20013 Jul 20011828 May 200129 Jun 2001non
JAM-1D143 Jul 200121 Jul 20011815 Jun 200117 Jul 2001non
JAM-1D1521 Jul 20018 Aug 2001183 Jul 20014 Aug 2001non
JAM-1D168 Aug 200126 Aug 20011821 Jul 200122 Aug 2001SE
JAM-1D1726 Aug 200113 Sep 2001188 Aug 20019 Sep 2001SE
JAM-1D1813 Sep 20011 Oct 20011826 Aug 200127 Sep 2001SE
JAM-1D191 Oct 200119 Oct 20011813 Sep 200115 Oct 2001SE
JAM-1D2019 Oct 20016 Nov 2001181 Oct 20012 Nov 2001SE
JAM-1D216 Nov 200124 Nov 20011819 Oct 200120 Nov 2001non
 
JAM-2D114 Dec 200130 Dec 20011626 Nov 200126 Dec 2001NW
JAM-2D230 Dec 200115 Jan 20021612 Dec 200111 Jan 2002NW
JAM-2D315 Jan 200231 Jan 20021628 Dec 200127 Jan 2002NW
JAM-2D431 Jan 200216 Feb 20021613 Jan 200212 Feb 2002NW
JAM-2D516 Feb 20024 Mar 20021629 Jan 200228 Feb 2002NW
JAM-2D64 Mar 200220 Mar 20021614 Feb 200216 Mar 2002non
JAM-2D720 Mar 20025 Apr 2002162 Mar 20021 Apr 2002non
JAM-2D85 Apr 200221 Apr 20021618 Mar 200217 Apr 2002non
JAM-2D921 Apr 20027 May 2002163 Apr 20023 May 2002non
JAM-2D107 May 200223 May 20021619 Apr 200219 May 2002non
JAM-2D1123 May 20028 Jun 2002165 May 20024 Jun 2002non
JAM-2D128 Jun 200224 Jun 20021621 May 200220 Jun 2002non
JAM-2D1324 Jun 200210 Jul 2002166 Jun 20026 Jul 2002non
JAM-2D1410 Jul 200226 Jul 20021622 Jun 200222 Jul 2002non
JAM-2D1526 Jul 200211 Aug 2002168 Jul 20027 Aug 2002SE
JAM-2D1611 Aug 200227 Aug 20021627 Jul 200223 Aug 2002SE
JAM-2D1727 Aug 200212 Sep 2002169 Aug 20028 Sep 2002SE
JAM-2D1812 Sep 200228 Sep 20021625 Aug 200224 Sep 2002SE
JAM-2D1928 Sep 200214 Oct 20021610 Sep 200210 Oct 2002SE
JAM-2D2014 Oct 200230 Oct 20021626 Sep 200226 Oct 2002SE
JAM-2D2130 Oct 200215 Nov 20021612 Oct 200211 Nov 2002SE
 
JAM-3D117 Nov 200214 Dec 20022830 Oct 200210 Dec 2002non
JAM-3D214 Dec 200231 Dec 20021726 Nov 200227 Dec 2002NW
JAM-3D331 Dec 200217 Jan 20031713 Dec 200213 Jan 2003NW
JAM-3D417 Jan 20033 Feb 20031730 Dec 200230 Jan 2003NW
JAM-3D53 Feb 200320 Feb 20031716 Jan 200316 Feb 2003NW
JAM-3D620 Feb 20039 Mar 2003172 Feb 20035 Mar 2003NW
JAM-3D79 Mar 200326 Mar 20031719 Feb 200322 Mar 2003non
JAM-3D826 Mar 200317 Apr 2003228 Mar 200313 Apr 2003non
JAM-3D917 Apr 200329 Apr 20031230 Mar 200325 Apr 2003non
JAM-3D1029 Apr 200316 May 20031711 Apr 200312 May 2003non
JAM-3D1116 May 20032 Jun 20031728 Apr 200329 May 2003non
JAM-3D122 Jun 200319 Jun 20031715 May 200315 Jun 2003non
JAM-3D1319 Jun 20036 Jul 2003171 Jun 20032 Jul 2003SE
JAM-3D146 Jul 200323 Jul 20031718 Jun 200319 Jul 2003SE

3.2. Planktonic Foraminiferal Analyses

[11] After splitting the individual samples into different aliquots, all planktonic foraminifera (and pteropods) were picked out from the respective wet samples using a small soft brush and then dried. The dried samples were then split into five size fractions, 63–150 μm, 150–250 μm, 250–355 μm, 355–500 μm, and >500 μm. The collection periods of each cup were corrected for the settling time and life cycle of planktonic foraminifera in order to derive species sensitivity to SST and surface productivity (Table 1). We applied a 2-week adjustment for foraminiferal production [Sautter and Thunell, 1991], and 4 days for the settling time of foraminifera to arrive at the sediment traps depth, assuming an average 500 m d−1 sinking velocity for shells larger than 150 μm [Takahashi and Bé, 1984].

3.2.1. Census Count and Fluxes

[12] Planktonic foraminifera were identified following the taxonomy proposed by Parker [1962], Kennett and Srinivasan [1983], and Hemleben et al. [1989]. N. dutertrei was distinguished from Neogloboquadrina pachyderma dex. primarily by the presence of an umbilical tooth, presence of more than four chambers per whorl, and a more pitted texture based on the description of Parker [1962]. In this taxonomic treatment, Neogloboquadrina p-d intergrade and Globigerinoides trilobus were included in N. dutertrei and Globigerinoides sacculifer, respectively. Determination of G. ruber s.s. and G. ruber s.l. follows the concept of Wang [2000], in which G. ruber s.l. corresponds to the more compact and higher trochospiral forms previously described as G. elongatus [d'Orbigny, 1826], G. pyramidalis [Van den Broeck, 1876], and G. cyclostomus [Galloway and Wissler, 1927].

[13] Flux estimates for planktonic foraminifera in individuals (ind.) m−2 d−1 were made considering the sample split, the duration of each collection period, and the size of the opening of the sediment trap (0.5 m2). Here we report on foraminiferal fluxes from the >250 μm size fraction (the “isotopic” size fraction), and >150 μm size fraction (the “census count” size fraction, Tables 13).

Table 2. Fluxes of the Most Abundant Planktonic Foraminifera Species in the >150 μm Size Fraction During the Entire Deployment Perioda
Sample NumberG. bulloidesG. falconensisG. rubescensG. calidaG. siphoniferaG. digitataG. adamsiH. digitataG. ruberG. sacculiferG. conglobatusO. universaS. dehiscensG. hexagonaN. dutertreiN. pachyderma sin.N. pachyderma dex.N. conglomerataP. obliquiloculataG. glutinataG. uvulaG. menardiiG. ungulataG. tumidaG. hirsutaG. theyeriG. scitulaFragmentsTotal
  • a

    Unit is ind. m–2 d–1. Note that total fluxes comprise all 37 identified species.

JAM-1
   D160737112005037218000257037025680500027932805
   D27491132000185400043001201443090005511379
   D37414254208520202140502250180002159225
   D4049122502011650404204209023200747240
   D5259942005520004405009016400209134
   D616021916000194120200230200214032000729374
   D72322075300035612270436421918160500202216653
   D81652511320402677000220220514140550204420518
   D92000000016000000020400700000231
   D100002200018200002020920400000443
   D112009124602010040700108248428250440007420481
   D12205121228000133120202128457732430410209516564
   D13122141120020809020220020212302415204912305
   D141441642802014220052253034927430117050251232612
   D1520149370201833000003222073030041500729458
   D1616020119020236430200620321441480172550011921795
   D1727764207521102813400001230841850101412422541218271096
   D1838920924452490308160421221314274711712405920522016711858
   D1996780939040165140204485757784011672009525816
   D20321437125220219414020484257439372200500423628
   D2125425122004098400002794147460680509923428
 
JAM-2
   D102001000024200004000400200002854
   D242026000162400240026008000610478
   D3000012000122000020641000006020860
   D4040620001101420001802262040040246223
   D584484000132304002200881012416000446291
   D600022000054122400602424421060180266273
   D71000454000364220220004800320602418217
   D86262438044180202001612602846300900602230641
   D9220818000581600004800026402800081214237
   D1012241416080822826207420050182160208214357
   D114226140207416220446042880600010012218
   D12162181024040156220604520182121621200214820410
   D1340104218260802523800048607612785021684184101638993
   D14564020441440403442121616282018841844841656076000420641972
   D156424364088000316684812128801003612484062401524836401956
   D16360561164414800035212400020212821620144168460128448442122
   D1712828846456016022836012881880264241001884212072008321756
   D18660242124810000031613680444428208364414853241040160012483748
   D191361060146006016224000812601842016184012040622241076
   D2022028108248002003125200001160296420288015608121628401816
   D21310821685642042604620010504162102823402240201042101723
 
JAM-3
   D151915702029601018081524022000127151
   D266881120028200004082621019000009136
   D38969602019110200921792360130000413174
   D42869285398087604223003815383602240202026644
   D5194112623000832304002641141134040000026299
   D6111581111000418080040110011080404015162
   D723111934230001732601901172038830450230004019567
   D8911910000204010015015094061000309172
   D932138320003280300335005027000008161
   D1038048400023000004011048044000030124
   D11138248000531106001709064611000826172
   D121138045458001816001500600600603001500002323742
   D1364838238330800683800082410120025668023000002331766
   D1444438306080801500000150688537501500000113892
Table 3. Species-Specific Fluxes of the Most Abundant Planktonic Foraminifera Species >150 μm (>250 μm) in the Sediment Trap Time Series JAM1–2a
Sample SeriesAverage Values >150 μm (>250 μm)
G. ruber s.s. (%)G. ruber s.l. (%)G. sacculifer (%)G. siphonifera (%)G. calida (%)G. menardii (%)N. dutertrei (%)P. obliquiloculata (%)N. pachyderma dex. (%)G. bulloides (%)G. glutinata (%)
  • a

    Average values for the NW monsoon, nonmonsoon, and SE monsoon periods are relative to the total flux of the respective species during the experiment period from November 2000 to November 2002. Relative abundances of planktonic foraminifera species for the entire experiment period are also calculated and compared to average values in surface sediments from the Java and Mentawai Basins [from Mohtadi et al., 2007].

  • b

    G. ruber total.

Trap JAM1–2
   NW monsoon12.4 (10.1)13.5 (10.7)6.4 (6.7)7.4 (7.2)9.5 (5.1)5.3 (4.4)4.1 (2.8)3.0 (3.2)1.9 (0.5)1.6 (1.1)4.2 (0.0)
   Nonmonsoon38.2 (44.0)44.5 (44.7)41.6 (51.9)40.8 (40.7)35.7 (39.4)33.9 (34.0)37.4 (38.6)50.7 (52.0)18.2 (17.2)12.0 (10.1)18.7 (26.4)
   SE monsoon49.4 (45.9)42.0 (44.6)52.0 (41.4)51.8 (52.1)54.8 (55.5)60.8 (61.6)58.5 (58.7)46.3 (44.8)79.9 (82.3)86.4 (88.8)77.1 (73.6)
   Entire period, relative to total fauna7.0 (6.7)15.2 (11.0)3.7 (3.6)5.1 (8.9)1.9 (1.4)9.9 (17.0)9.6 (16.0)5.8 (10.0)10.3 (11.0)8.7 (3.5)8.3 (0.4)
Surface samples
   Java Basin14.6b 7.63.63.414.28.25.27.517.03.8
   Southern Mentawai Basin18.9b 13.13.34.96.99.13.17.015.85.3
   Mean17.0b 10.73.54.210.18.74.07.216.34.6

3.2.2. Stable Oxygen Isotopes

[14] A Finnigan MAT 251 mass spectrometer was used to measure the δ18O composition of the planktonic foraminifera G. ruber s.s. and G. ruber s.l. from the 250–355 μm, N. dutertrei from the 355–500 μm, P. obliquiloculata from the 250–355 μm (or 355–500 μm), and G. menardii from the 355–500 μm (or >500 μm) size fraction (Table 4). Approximately 5–20 individual tests were picked for each measurement. The isotopic composition of the carbonate sample was measured on the CO2 gas evolved by treatment with phosphoric acid at a constant temperature of 75°C. For all stable isotope measurements a working standard (Burgbrohl CO2 gas) was used, which has been calibrated against PDB by using the NBS 18, 19 and 20 standards. Consequently, all isotopic data given here are relative to the PDB standard. Analytical standard deviation is about ±0.07‰ (Isotope Laboratory, Faculty of Geosciences, University of Bremen).

Table 4. Shell Geochemistry Data on Different Planktonic Foraminifera Species in the Sediment Trap Time Series JAM1–2a
Sample NumberG. ruber s.s. 250–355 μmG. ruber s.l. 250–355 μmN. dutertrei 355–500 μmP. obliquiloculata 250–355 (355–500) μmG. menardii >500 (355–500) μm
SeriesCupδ18O (‰PDB)Mg/Ca (mmol mol−1)δ18O (‰PDB)Mg/Ca (mmol mol−1)δ18O (‰PDB)Mg/Ca (mmol mol−1)δ18O (‰PDB)Mg/Ca (mmol mol−1)δ18O (‰PDB)Mg/Ca (mmol mol−1)
  • a

    P. obliquiloculata and G. menardii values in brackets are from the 350–500 μm size fractions.

JAM-1D1−3.044.94−2.824.85−2.332.23−2.40 −1.322.46
JAM-1D2−3.06 −3.05 −2.46 −2.34 −1.26 
JAM-1D3−2.87   −2.36   (−2.19) 
JAM-1D4−3.17 −2.78     −1.00 
JAM-1D5        −0.44 
JAM-1D6−2.755.82−2.945.51    (−2.20)(3.20)
JAM-1D7−2.935.67−2.975.40  −2.34 (−2.20)3.38 (3.27)
JAM-1D8−2.975.65−3.104.80  −2.37 −0.69(3.17)
JAM-1D9        −0.31 
JAM-1D10      −2.11   
JAM-1D11−3.48 −3.40 −2.363.05−2.52 −1.382.59
JAM-1D12−3.36 −3.395.47−2.263.26−2.432.83−1.452.89
JAM-1D13−3.26 −3.22 −2.12 −2.29 −1.14 
JAM-1D14−3.246.01−3.14 −2.493.16(−2.02)(2.92)−1.252.46
JAM-1D15−3.025.67−2.945.23−2.38 (−2.00)(2.67)(−1.74)(2.57)
JAM-1D16−2.89 −3.155.09−2.242.47−2.253.20−1.222.45
JAM-1D17−2.92 −3.044.65−1.813.45−2.022.33−1.012.18
JAM-1D18−3.034.69−2.92 −1.942.79(−1.94)(2.26)−1.12 
JAM-1D19−3.24 −2.74 −2.24 (−2.03) −0.332.09
JAM-1D20−3.054.64−3.06 −2.182.87(−2.34) −1.322.57
JAM-1D21−2.91 −2.96 −2.23 (−2.15) −1.602.68
JAM-2D1      −2.14   
JAM-2D2−3.43 −2.85   −2.33 (−1.88) 
JAM-2D3        (−1.86) 
JAM-2D4−3.09 −3.15 −2.60 (−1.75) (−1.76) 
JAM-2D5−3.25 −3.02       
JAM-2D6−3.52 −3.70   −2.14 −0.48(2.36)
JAM-2D7−3.52 −3.20   −1.572.41−0.302.39
JAM-2D8−3.52 −3.335.97−2.58 −2.652.63−0.86(2.79)
JAM-2D9−3.43 −3.43 −2.302.26  −0.702.23
JAM-2D10−3.36 −3.30 −2.312.83−3.662.75(−1.86)(2.10)
JAM-2D11−3.49 −3.41 −2.35 −2.18   
JAM-2D12−3.364.75−3.445.25−2.54 −2.28 −0.92 
JAM-2D13−2.97 −3.335.30−2.542.97−1.732.85−1.072.52
JAM-2D14−2.835.01−2.905.17−2.152.69(−1.78)(2.74)−1.212.19
JAM-2D15−3.13 −3.065.45−1.953.20(−1.77)(2.66)−1.332.27
JAM-2D16−2.794.20−2.564.94−1.872.99(−1.65)(2.85)−1.472.60
JAM-2D17−2.514.05−2.56 −1.622.71(−1.40)(2.35)−0.582.21
JAM-2D18−2.194.09−2.414.08−1.692.85(−1.48)(2.43)0.132.05
JAM-2D19−2.674.23−2.454.04−1.79 (−1.40) −0.561.96
JAM-2D20−2.494.21−2.65 −1.87 (−1.40) −0.782.01
JAM-2D21−2.98 −2.744.88−1.98 (−1.40) −0.622.13

3.2.3. Mg/Ca Analysis

[15] About 10 to 30 specimens were picked and gently crushed from the 250–355 μm (G. ruber s.s. and s.l., P. obliquiloculata), 355–500 μm (N. dutertrei, P. obliquiloculata, G. menardii), and >500 μm (G. menardii) size fraction, respectively (Table 4). Sample cleaning was applied according to the cleaning method introduced by Barker et al. [2003], which consisted of five water washes and two methanol washes followed by two oxidation steps for 10 min. each (with ∼1% buffered H2O2), and a weak acid leach (0.001M QD HNO3) before dissolving the foraminiferal calcite into 0.075M QD HNO3. The solutions were then centrifuged (10 min. at 6000 rpm), transferred into test tubes and diluted. Mg/Ca ratios were measured using a Perkin Elmer Optima 3300 R Inductively Coupled Plasma Optical Emission Spectrophotometer (ICP-OES) equipped with an auto sampler and an ultrasonic nebulizer U-5000 AT (Cetac Technologies Inc.) housed at the Faculty of Geosciences, University of Bremen. The Mg/Ca values are reported as mmol mol−1. Instrumental precision was determined after an external, in-house standard, which was run after every five samples. Relative standard deviation of the external standard was ±0.09% and the drift over a typical analytical session of 24 h was ±0.28%. Analytical precision of the samples was ±0.21% and their reproducibility (n = 39) ±0.09 mmol mol−1 (±2.66%).

3.3. Comparison With Surface Sediment Data

[16] The uppermost cm of seven multicore samples from the Java Basin (GeoB cores 10044–3, 47–1, 49–5, 50–1, 58–1, 59–1, 61–5, ranging between 1100 and 3350 m depth) and nine from the Southern Mentawai Basin (GeoB cores 10034–3, 36–3, 38–3, 39–3, 40–3, 41–3, 42–3, SO-189 cores 03MC and 09MC, ranging between 1000 and 2600 m depth) were averaged for each basin and compared to the trap data (Table 3). Core top data were obtained using the same methods and size fractions as described above for the sediment traps [Mohtadi et al., 2007]. All the sediment trap δ18O and Mg/Ca values were flux weighted by multiplying each value by the ratio of the flux rates of that cup to the total flux, and then summing the respective values to produce a single value for the NW monsoon, the nonmonsoon, the SE monsoon, and the entire experiment period. In this connection, only the flux rates of the respective component, species, or size fraction in the trap series were considered. The flux-weighted values for the entire experiment period are an approximation of a “core top” value, assuming that each core top is representative for the present-day conditions, or unless, seasonal flux patterns have remained similar to present over the time span represented by each core top.

3.4. Auxiliary Data

[17] In this study, we present CaCO3 fluxes in JAM3 sediment trap together with recently published JAM1–2 data [Rixen et al., 2006]. Sample processing and analytical methods are described elsewhere [Haake et al., 1993]. Satellite-based, 8-day average surface net primary production (NPP) data between 108°E and 108.30° E, and 8° to 8.30° S have been considered to reflect the surface productivity at the traps' position. Data were retrieved from the Ocean Productivity site of the Oregon University (http://web.science.oregonstate.edu/ocean.productivity). Local SST and precipitation for the deployment period were generated from the Advanced Very High Resolution Radiometer (AVHRR) data obtained from the National Oceanic and Atmospheric Administration (NOAA, http://www.ncdc.noaa.gov/oa/ncdc.html), and the Global Precipitation Climatology Center (http://www.dwd.de), respectively. The Indian Ocean Dipole Mode Index (DMI) and the Southern Oscillation Index (SOI) were obtained from Frontier Research Center for Global Change (FRCGC, http://www.jamstec.go.jp/frsgc/research/d1/iod/), and NOAA Climate Prediction Center (http://www.cpc.ncep.noaa.gov/data/indices/), respectively. Water column temperature and salinity data were obtained from CTD casts made in May 1999 and December 2001 (from Ocean Data View [Schlitzer, 2002]), and in September 2005 [Hebbeln et al., 2005]. The CTD profiles are used to assess the seasonal variability in hydrographic conditions.

4. Results and Discussion

[18] Climate data reveal that the deployment period of our sediment traps spanned over almost three monsoonal cycles and included, according to the DMI and SOI indices, a weak La Niña year (2001), a weak El Niño year (2002), and a weak positive IOD year (2003, Figures 2a2c). Highest onshore precipitation rates occurred during the NW monsoon period in 2001 (∼300 mm month−1) while the dry season during the SE monsoon period in 2002 was longer and more severe (Figure 2a, gray line). Likewise, SST depression is much stronger during the SE monsoon period in 2002 compared to 2001 (Figure 2a, black line). However, the strongest and most rapid SST depression occurred during the SE monsoon period in 2003.

Figure 2.

(a–d) Climatologic data (see text for references) and (e–h) measured parameters in the sediment traps during the deployment period between 2000 and 2003. Figure 2a shows monthly averaged SST (black line, at the trap site) and precipitation rates (gray line, over Indonesia). Figure 2b shows three-point moving average of the monthly mean Southern Oscillation Index (SOI). Figure 2c shows five-point moving average of the monthly mean Indian Ocean Dipole Mode Index (DPI). Figure 2d shows 2-week averaged, satellite-based net primary production (NPP) estimates at the trap site, in mg C m−2 d−1. Figure 2e shows total planktonic foraminifera flux rates in the >250 μm (gray line) and >150 μm (black line, including >250 μm) size fractions, in individuals (ind.) m−2 d−1. Figure 2f shows flux rates of pteropods from the >150 μm size fraction, in ind. m−2 d−1. Figure 2g shows total CaCO3 flux rates in the sediment traps samples, in mg m−2 d−1. Gray bars indicate SE monsoon (upwelling) season.

4.1. Flux Rates

[19] Total foraminiferal fluxes (TFFs) show a strong, monsoon-caused seasonality during the deployment period with enhanced fluxes during SE monsoon from July to October (Figure 2e). Higher fluxes are associated with increased NPP related to coastal upwelling and enhanced nutrient flux during these periods (Figure 2d) peaking during the maximum SST depression and upwelling intensity in September (Figure 2a). While the seasonal variability in the TFFs is driven by the monsoonal cycle, its amplitude is apparently controlled by ENSO/IOD, with much (slightly) higher flux rates during the SE (NW) monsoon in 2002 (2001). Remote sensing based NPP estimates imply that during the deployment period, marine productivity was highest (lowest) during the SE (NW) monsoon in 2003 (Figure 2d). This is also indicated by the TFFs that show highest late June–early July values within the deployment period in 2003 (Table 2 and Figure 2e).

[20] Total fluxes of pteropods show a similar pattern as for the TFFs throughout the deployment period (Figure 2f). The total carbonate fluxes show, however, an additional peak in March (Figure 2g) and cannot be attributed to planktonic foraminifera or pteropods. We tentatively attribute these peaks to coccolithophores fluxes.

[21] Among 37 identified species of planktonic foraminifera in the sediment trap samples, 27 species accounted for 97–100% of the TFFs (Tables 2 and 3), with only 10 species G. ruber (s.s. and s.l.), G. sacculifer, Globigerina bulloides, Globigerinella siphonifera, Globigerinella calida, Globigerinita glutinata, G. menardii, P. obliquiloculata, N. dutertrei, and N. pachyderma dex. constituting > 85% of the TFFs (Tables 2 and 3 and Figure 3). Species-specific flux rates exhibited patterns similar to the TFFs, with lowest fluxes during the NW monsoon season, and higher flux rates during the SE monsoon season, particularly in 2002 (Figure 3). The monsoonal driven seasonality is most prominent for G. bulloides, N. pachyderma dex., and G. glutinata fluxes, with peak values during the upwelling season, when > 75% of their species-specific total fluxes occur, and contribute to ∼40% of the TFF during the upwelling season (Tables 2 and 3 and Figures 3i and 3k). These species can be assigned as typical upwelling indicators in the study area, and have also been associated with enhanced nutrient supply during the upwelling season elsewhere [e.g., Schiebel et al., 2001; Peeters et al., 2002; Stoll et al., 2007].

Figure 3.

Species-specific flux rates of planktonic foraminifera during the deployment period. (a) Two-week averaged, satellite-based net primary production (NPP) estimates at the trap site, in mg C m−2 d−1. (b–k) Fluxes of the most abundant foraminifera species. Gray bars indicate SE monsoon (upwelling) season. Note the differences in scales between the species.

[22] Fluxes of the remaining planktonic foraminifera species show a less pronounced seasonal and interannual variability (Table 3 and Figure 3). Although most of these species show enhanced flux rates during the upwelling period (Figure 3), their relative fluxes are as high, or even higher, during the nonupwelling season (Table 3). Surprisingly, enhanced fluxes of some species occur prior to the upwelling season (Figures 3b3e), which might be related to (1) imprecise definition of the upwelling period by “SST anomaly” that might differ from “nutrient introduction” because of the presence of the barrier layer (see section 2), (2) inaccurate estimate of the planktonic foraminiferal shell production (see section 3.2), (3) limitation in the modeling of the NPP (for more details see http://web.science.oregonstate.edu/ocean.productivity), and (4) inadequate overlap of the NPP estimates (averaged over eight calendar days) with the sampling intervals (16 days at best). Finally, our experiment period covers rather normal years. The fact that the NPP, TFFs, and the species-specific foraminiferal flux rates are significantly higher in 2002 suggests ENSO to modulate their interannual variability (Figures 2 and 3). Likewise, the potential influence of IOD can be observed in the flux rates of many species that peak, simultaneously to the NPP peak, already in late June to early July 2003.

4.2. Shell Geochemistry

[23] It is noteworthy that some limitations to the foraminiferal shell geochemistry analyses exist in this study. First, the adjustment of the sampling periods for shell production (2 weeks) and settling time (4 days, Table 1) might not be valid for some species or periods. Second, some of the samples had an insufficient number of shells for Mg/Ca analysis, particularly during the NW monsoon season (see Tables 2 and 3). Therefore, this season is underrepresented in the Mg/Ca records (Table 4). On the same account, in some samples the δ18O and Mg/Ca analyses could be performed either on the 250–355 μm or the 355–500 μm size fraction (P. obliquiloculata), respectively on the 355–500 μm or the >500 μm size fraction (G. menardii, Table 4). Finally, the flux-weighted values in Table 5 are exploited by using the respective species fluxes from the >250μm size fraction that include, but do not inevitably represent the net fluxes of the much narrower size range used for the shell geochemistry analyses.

Table 5. Flux-Weighted Shell Geochemical Data in the Sediment Trap Time Series JAM1–2a
Sample SeriesAverage Values
G. ruber s.s δ18O, 250–355 μm (‰ PDB)G. ruber s.s. Mg/Ca, 250–355 μm (mmol mol−1)G. ruber s.l. δ18O, 250–355 μm (‰ PDB)G. ruber s.l. Mg/Ca, 250–355 μm (mmol mol−1)N. dutertreiδ18O, 355–500 μm (‰ PDB)N. dutertrei Mg/Ca, 355–500 μm (mmol mol−1)P. obliquiloculata δ18O, 250–355 (355–500) μm (‰ PDB)P. obliquiloculata Mg/Ca, 250–355 (355–500) μm (mmol mol−1)G. menardii δ18O, >500 (355–500) μm (‰ PDB)G. menardii Mg/Ca, >500 (355–500) μm (mmol mol−1)
  • a

    Average values for the NW monsoon, nonmonsoon, and SE monsoon are calculated considering the total flux of the respective species during the experiment period from November 2000 to November 2002. P. obliquiloculata and G. menardii values in brackets are from the 350–500 μm size fractions. For comparison, average values in surface sediments from the Java and Mentawai Basins are shown below.

  • b

    From Mohtadi et al. [2007].

Traps JAM1–2          
   NW monsoon−3.095.82−2.975.51−2.47(−2.14)−1.56 (−1.65)(3.29)
   Nonmonsoon−3.235.60−3.225.27−2.352.81(−2.25)2.69 (2.78)−1.25 (−1.93)2.55 (2.71)
   SE monsoon−2.824.30−2.784.73−1.932.92(−1.68)2.76 (2.51)−0.852.22
   Entire period−2.935.04−2.905.06−1.992.86(−1.84)2.71 (2.60)−1.012.41 (2.67)
Surface samples          
   Java Basin−2.91b4.69−2.694.48−1.57b2.36−1.16 (−1.41)1.98 (2.27)−0.78 (−0.66)1.99 (2.12)
   Mentawai Basin−2.97b5.36−3.045.46−1.98b2.72−1.82 (−2.01)−1.15 (−1.29)
   Mean−2.94b5.03−2.874.97−1.77b2.54−1.49 (−1.71)−0.96 (−0.98)

4.2.1. G. ruber s.s. and s.l.

[24] Shell δ18O values of G. ruber s.s. and G. ruber s.l. track the monsoonal SST cycle in the study area during the deployment period of JAM1–2 (Figures 4a and 4b). The δ18O records show lowest values during the nonupwelling period from March to May, an increase from May–July prior to the upwelling season, followed by another sharp increase from late July to the upwelling maximum in September, and a decreasing trend toward the NW monsoon season in December–February (Table 4 and Figure 4b). The δ18O values during the upwelling season in 2002 are on average ∼0.5‰ heavier than during the same period in 2001.

Figure 4.

(a) Monthly averaged SST (black line, at the trap site), and precipitation rates (gray line, over Indonesia). (b) Stable oxygen (δ18O) and (c) Mg/Ca data of G. ruber s.s. from 250 to 355 μm (black dots), G. ruber s.l. from 250 to 355 μm (gray dots), N. dutertrei from 355 to 500 μm (circles), P. obliquiloculata from 250 to 355 μm (white stars) and 355–500 μm (black stars), and G. menardii from 355 to 500 μm (white triangles) and >500 μm (black triangles) size fractions in the sediment trap time series JAM1–2. Gray bars indicate SE monsoon (upwelling) season.

[25] Shell δ18O values of G. ruber s.s. and G. ruber s.l. are virtually identical suggesting a similar habitat depth for both morphotypes. This is in contrast to the results from core top studies in the South China Sea showing ∼0.2‰ heavier δ18O values of G. ruber s.l. compared to G. ruber s.s. [Wang, 2000; Steinke et al., 2005], which have led the authors to infer different habitat depths for these morphotypes. Similar habitat depths of G. ruber s.s. and G. ruber s.l. in the study area are also indicated by their similar flux patterns during the experiment period (Figure 3b). However, it is noteworthy that δ18O values of these species differ at most in the late December to early January precipitation peaks during the NW monsoon seasons (Figures 4a and 4b) and could indicate that extreme surface conditions with respect to salinity or stratification might result in different habitat depths of these morphotypes.

[26] Likewise, shell Mg/Ca values of G. ruber s.s. and G. ruber s.l. are similar and show a strong covariance with the SST during the deployment period of JAM1–2 (Figures 4a and 4c). Mg/Ca values vary between 5 and 6 mmol mol−1 during the NW monsoon and nonmonsoon periods, and 4–5 mmol mol−1 during the SE monsoon season, respectively (Figure 4c). For both morphotypes, the Mg/Ca values are on average ∼0.5 mmol mol−1 lower during the SE monsoon in 2002 than during the same period in 2001. In light of the similarity between the δ18O and Mg/Ca values of G. ruber s.s. and G. ruber s.l., we combine these values in the following in order to generate a single value for G. ruber that can be compared to the values reported on this species in other sediment trap [Anand et al., 2003; McConnell and Thunell, 2005], core top [Dekens et al., 2002], or culture [Kisakürek et al., 2008] studies.

[27] The combined Mg/Ca data set exhibits a strong exponential relationship between the Mg/Ca ratio and measured SST at the traps' location (AVHRR data, see section 3.4.) according to the equation

equation image

where T is temperature in degree Celsius (Figure 5a). This paleotemperature equation is close to the equations derived from other sediment trap studies by Anand et al. [2003] in the Sargasso Sea (exponential values: 0.102, 0.09, 0.085, preexponential values: 0.34, 0.449, 0.48) and by McConnell and Thunell [2005] in the Gulf of California (exponential value: 0.081, preexponential value: 0.47, temperature range 25–33°C). In general, the G. ruber Mg/Ca:temperature relationship in our study (equation (1)) shows a good resemblance to other published studies with the values lying inside the other calibration curves (Figure 6). We infer that our data provide supporting evidence for these calibrations, in particular for those from the Anand et al. [2003] sediment trap study verifying the applicability of published G. ruber Mg/Ca:temperature relationships in the study area.

Figure 5.

Paleotemperature equations for G. ruber (250–355 μm) based on (a) Mg/Ca ratios and AVHRR daily SST data with (b) the resulting comparison between measured and predicted SST or based on (c) Mg/Ca ratios and δ18O-derived calcification temperatures, with (d) the resulting comparison between measured and δ18O-derived calcification temperatures.

Figure 6.

Comparison of the exponential relationship between G. ruber shell Mg/Ca ratio and temperature from different studies. Black dots, this study; black solid line, Dekens et al. [2002]; black dashed line, Kisakürek et al. [2008]; gray solid lines, three different relationships from Anand et al. [2003] for the 250–350 μm size fraction, for the same size fraction with an assumed exponential (A) value of 0.09, and for the 350–500 μm size fraction; gray dashed line, McConnell and Thunell [2005].

[28] An alternative method to calibrate the Mg/Ca:temperature relationship is the use of calcification temperatures calculated from the δ18O composition of planktonic foraminifera, which allows for the varying depth habitat of each species to be incorporated into the temperature calibration [Elderfield and Ganssen, 2000; Anand et al., 2003; McConnell and Thunell, 2005]. For this purpose, we use the δ18O:temperature equation of Bemis et al. [1998] for G. ruber:

equation image

where δ18Occ is the measured δ18O of G. ruber, and δ18Osw the oxygen isotope composition of the ambient seawater. Because the “Global Seawater Oxygen-18 Database” lacks any δ18Osw data from the eastern tropical Indian Ocean (see http://data.giss.nasa.gov/o18data), we calculated the δ18Osw using the relationship between δ18Osw and salinity (S) determined by Morimoto et al. [2002] for the western Pacific:

equation image

Salinities were not measured throughout the sample collection period. We thus calculated an average δ18Osw for the upper 30 m of the water column under different hydrographic settings at the trap site (Figure 7). Average δ18Osw values are approximately –0.24‰ relative to Standard Mean Ocean Water (SMOW). The δ18Osw values were converted to the PDB scale by subtracting 0.27‰ [Bemis et al., 1998]. Finally, Mg/Ca data from the sediment trap samples and the calcification temperatures estimated from δ18Occ (equation (2)) were used to develop a paleotemperature equation for G. ruber (Figure 5c). The Mg/Ca: δ18O-derived temperature relationship for G. ruber is:

equation image
Figure 7.

CTD-based (a) temperature and (b) salinity profiles and (c) the calculated seawater δ18O (δ18Osw), for the upper 200 m of the water column at the trap site. CTD casts were deployed during the NW monsoon (black solid line), nonmonsoon (dashed line), and SE monsoon (gray line) periods. SMOW, standard mean ocean water. Note the presence of the low-salinity cap and a deep thermocline during the NW monsoon.

[29] We note that the derived (pre-) exponential value in equation (4) is far too low (high) when compared with other G. ruber Mg:Ca:temperature calibrations [e.g., Lea et al., 2000; Dekens et al., 2002; Anand et al., 2003; McConnell and Thunell, 2005; Kisakürek et al., 2008]. Possible explanations for this offset could be (1) the inaccuracy in the 2-week approximation of foraminiferal shell calcification (section 3.2.), (2) imprecise estimate of the average δ18Osw that is based on only three CTD casts, (3) seasonal changes in the sea surface salinity and temperature not captured with our approach, (4) seasonal changes in the habitat depth of G. ruber, (5) small temperature variations during the deployment period resulting in a low signal-to-noise ratio, and (6) a combination of items 1–5. With data on hand, it appears that Mg/Ca in G. ruber records the temperature variation in the upper 30 m of the water column. However, data from the NE monsoon season are lacking, and the δ18O-derived calcification depths suggest a deeper habitat of this species during this period (Figure 8, circles).

Figure 8.

Temperatures derived from shell Mg/Ca values of G. ruber (250–355 μm) with daily SST from the AVHRR data at the trap site (gray line). Black dots represent Mg/Ca calibrated to the temperatures using the Anand et al. [2003] equation. Circles denote δ18O-based calcification temperatures after Bemis et al. [1998].

4.2.2. N. dutertrei, P. obliquiloculata, and G. menardii

[30] Shell δ18O of N. dutertrei and P. obliquiloculata display similar values in 2001 (on average ∼2.2‰, Figure 4b). During 2002, the δ18O values of N. dutertrei are on average ∼0.2‰ lighter than those of P. obliquiloculata (355–500 μm) in particular toward the end of the upwelling season in October and November. This indicates a similar habitat depth for both species in 2001 but a slightly different habitat of the larger P. obliquiloculata individuals in 2002. The δ18O of both species exhibit the same seasonal pattern with an average ∼0.9‰ enriched values compared to those of G. ruber s.s. and s.l., and a similar difference between the SE monsoon period in 2002 and 2001 as recorded by the latter species (∼0.5‰, Figure 4b). Shell δ18O data of G. menardii in the 355–500 μm size fraction are entirely confined to the nonupwelling periods and show slightly enriched values compared to N. dutertrei and P. obliquiloculata suggestive of a slightly deeper habitat of G. menardii (white triangles in Figure 4b). In contrast, shell δ18O data of G. menardii from the >500 μm size fraction are significantly enriched compared to other species, and exhibit the strongest fluctuations during the experiment period (black triangles, Figure 4b). The increase in δ18O with increasing size fraction presumably reflects increased calcification of more mature G. menardii individuals at greater depths [e.g., Erez and Honjo, 1981].

[31] Similar average Mg/Ca values can be observed during the SE monsoon periods of 2001 and 2002 for N. dutertrei (2.89 and 2.94 mmol mol−1), P. obliquiloculata (2.51 and 2.57 mmol mol−1), and G. menardii (>500 μm, 2.24 and 2.21 mmol mol−1, Table 4 and Figure 4c), and suggest that despite variable upwelling intensities, the thermal structure at the their habitat depth did not change significantly. This finding can be best explained by the hydrographic regime in the study area: the thick barrier layer in 2001 could not be eroded completely during the upwelling season, which led to a more damped SST depression in this year. In contrast, a less developed barrier layer caused by a relatively short precipitation season in 2002 (Figure 4a, gray line) was prone to be completely removed and enabled the ascent of cold subsurface water up to the surface and a more severe SST decrease during this period (Figure 4a, black line). This scenario is recorded by the rather small shell Mg/Ca decrease of the surface-dwelling G. ruber during the upwelling season in 2001 compared to 2002, whereas shell Mg/Ca values of N. dutertrei, P. obliquiloculata, and G. menardii recorded the proceeding upwelling at the thermocline depth, the habitat depth suggested for these species [e.g., Bé and Tolderlund, 1971; , 1977; Duplessy et al., 1981; Fairbanks et al., 1982; Thunell and Reynolds, 1984; Bé et al., 1985; Sautter and Thunell, 1991; Peeters et al., 2002] (see also below).

[32] We abstained to calibrate Mg/Ca:temperature relationships for these species, since no measured subsurface temperatures from the experiment period exist. In order to estimate the calcification depth of these species, we used the species-specific δ18O:temperature equations of Bouvier-Soumagnac and Duplessy [1985] derived from plankton tow samples in the Indian Ocean for N. dutertrei:

equation image

and for G. menardii (355–500 μm):

equation image

For G. menardii (>500 μm) we used the Spero et al. [2003] equation based on a culture study on G. menardii (600–850 μm):

equation image

For P. obliquiloculata, the equation of Russell and Spero [2000] based on a sediment trap study in the Pacific Ocean was used:

equation image

In order to calculate the δ18Osw, we used equation (3) and average salinities for 50–100 m water depth (upper thermocline depth) for the different hydrographic settings at the trap site (Figure 7). The estimated average δ18Osw values are ∼0.04‰ SMOW (−0.23‰ PDB) for the NW monsoon, approximately −0.22‰ SMOW (−0.49‰ PDB) for the nonmonsoon, and ∼0.16‰ SMOW (−0.11‰ PDB) for the SE monsoon season, respectively.

[33] According to the δ18O–derived calcification temperatures, P. obliquiloculata comprises the shallowest habitat among these species at the upper thermocline (Figure 9a, see also Figure 4). N. dutertrei and G. menardii (355–500 μm) thrive at slightly deeper depths at the upper thermocline, and the habitat of G. menardii (>500 μm) corresponds to the lower thermocline. Monthly averaged δ18O-derived calcification temperatures in the sediment trap samples compared with the mean temperatures at the trap site for the past ∼40years (WOA05 [Levitus and Boyer, 1994]) confirm the estimated habitat depths of the observed species (Figure 9b). Habitat depth of G. ruber corresponds to the surface, and appears to deepen during the NW monsoon season (December–February), when the mixed layer is deep, and shoals under the more turbid conditions during the peak of the upwelling season (August–September, Figure 9b). Conversely, the habitats of the thermocline species P. obliquiloculata, N. dutertrei, and G. menardii (355–500 μm) shoal during the NW monsoon and deepen during the SE monsoon possibly as a result of nutricline migration in these periods. Same pattern can be observed for G. menardii (>500 μm) except for the period between February and April, when its estimated habitat is deepest.

Figure 9.

(a) Estimated calcification temperatures of selected planktonic foraminifera versus the daily SST data during the experiment period (gray line) and (b) the monthly mean values at different depths (black lines) for the past 40 years at the sediment trap site.

4.3. Comparison With Surface Sediments

[34] Comparison between flux-weighted values from the sediment traps and average values in surface sediment samples off SW Java and Sumatra can be used for a better assessment and interpretation of past variability in our proxies (Tables 3 and 5). The good match between the flux-weighted CaCO3 contents in the sediment traps (11%) and the surrounding core tops from depths ranging between 1000 and 3350 m (∼10%, not shown) confirms that in the study area, the Carbonate Compensation Depth (CCD) is deeper than 2400 m (max. sediment traps depth), as also suggested by previous studies [Martinez et al., 1998; Ding et al., 2006; Mohtadi et al., 2007]. Planktonic foraminiferal abundances in core tops also represent flux-weighted sediment traps values except for Globorotalia inflata (almost absent in the sediment traps) and G. sacculifer (Table 3). Since no trend or systematical shift in the faunal assemblages is evident, the only sound explanation could be local variations in foraminiferal fauna, and the impact of sediment surface sample GeoB 10044–3 (3346 m water depth) on the average core top values, which is apparently affected by carbonate dissolution with 17% G. sacculifer and 14% G. inflata, [Mohtadi et al., 2007]. Altogether, abundances of planktonic foraminifera in the sediment trap time series appear to be well recorded by surface sediments in the study area.

[35] Likewise, flux-weighted δ18O and Mg/Ca values of G. ruber s.s. and s.l. for the entire deployment period are in good agreement with the average core top values (Table 5). This suggests that G. ruber shell geochemistry in core tops represents mean annual surface conditions in the study area, and can be used in sediment archives for reconstructing past changes in surface hydrography. On the other hand, flux-weighted δ18O and Mg/Ca values of N. dutertrei, P. obliquiloculata, and G. menardii for the entire experiment period tend to record slightly warmer conditions than in the adjacent core tops. One potential explanation for this offset is local variations in the hydrographic conditions. In addition, the deployment period did not cover the entire range of hydrographic variation in the study area, e.g., from the extreme years of ENSO and IOD, that might be well preserved in the core tops. Finally, the poor coverage of the NW monsoon and nonmonsoon periods by the geochemical (in particular Mg/Ca) data of these species in the sediment traps might have biased their estimated values for the entire experiment period.

5. Conclusions

[36] The long-term deployment of sediment traps has enabled us to determine processes controlling the fluxes of settling particles off south Java. The most prominent environmental factors controlling the particle fluxes are the monsoon system on seasonal, and ENSO/IOD on interannual time scales. In particular, the SE monsoon boosts primary production by inducing coastal upwelling and injection of nutrients from subsurface waters. The amplitude and the efficiency of upwelling is controlled by ENSO/IOD through an effective removal of the barrier layer and the low-salinity cap in this region during El Niño/positive IOD years.

[37] Both total and species-specific fluxes of planktonic foraminifera perfectly record these seasonal and interannual environmental changes, with highest fluxes occurring during the late SE monsoon season (August–October) and El Niño/positive IOD years, when primary production is highest. During the NW monsoon (December–February), fluxes of planktonic foraminifera are lowest. However, except for G. bulloides, N. pachyderma dex., and G. glutinata, flux-weighted values of planktonic foraminifera for the entire experiment period do not show a pronounced monsoon-related seasonality.

[38] The δ18O and Mg/Ca values of G. ruber s.s. and s.l. (N. dutertrei, P. obliquiloculata, and G. menardii) reflect mixed layer (thermocline) conditions during the deployment period. Predicted calcification temperatures derived from shell δ18O and Mg/Ca ratio suggest a habitat depth of 0–30 m for G. ruber, ∼60–80 m (60–90 m) for P. obliquiloculata (N. dutertrei) and 90–110 m (100–150 m) for G. menardii from the 350–500 μm (>500 μm) size fraction.

[39] Comparison of G. ruber shell Mg/Ca to the daily SST data reveals an exponential relationship and agrees with the published Mg/Ca:temperature calibrations [e.g., Dekens et al., 2002; Anand et al., 2003; McConnell and Thunell, 2005; Kisakürek et al., 2008]. In light of some limitations that exist to our study, e.g., a narrow temperature range and insufficient measure of salinity changes, we suggest the application of hitherto published equations in the study area.

[40] Comparison between flux-weighted faunal and geochemical data from the sediment traps and surface sediment samples off SW Java and Sumatra reveals a good match and suggest that modern environmental conditions are well recorded by surface sediments in the study area. In particular, G. ruber shell geochemistry in core tops obviously represents mean annual surface conditions. Nonetheless, it is noticeable that the NW monsoon season might be underrepresented in marine sediments from this region.

Acknowledgments

[41] We are grateful to M. Segl and B. Meyer-Schack for technical support. R. O'Malley provided the ocean productivity data. This study was supported by the German Bundesministerium für Bildung und Forschung through funding of the projects 03G0184A “PABESIA” and 03F0440A. The Center for Marine Environmental Sciences (MARUM) at the University of Bremen provided technical support. J.G. thanks RCOM for financial support through RCOM fellowship. The manuscript benefited from constructive comments of two anonymous reviewers. The data presented in this paper are also available in digital format (http://www.pangaea.de). This is a MARUM publication.

Ancillary