Portal de Periódicos da CAPES

Short-time-scale variability in ventilation and export productivity during the formation of Mediterranean sapropel S1

2010; American Geophysical Union; Volume: 25; Issue: 4 Linguagem: Inglês

10.1029/2010pa001955

1944-9186

Tom Jilbert, Gert‐Jan Reichart, Paul R.D. Mason, Gert J. de Lange,

Geological formations and processes

PaleoceanographyVolume 25, Issue 4 Regular ArticlesFree Access Short-time-scale variability in ventilation and export productivity during the formation of Mediterranean sapropel S1 Tom Jilbert, Tom Jilbert [email protected] Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, NetherlandsSearch for more papers by this authorGert-Jan Reichart, Gert-Jan Reichart Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, Netherlands Alfred Wegener Institute for Polar and Marine Research, Biogeosciences, Bremerhaven, GermanySearch for more papers by this authorPaul Mason, Paul Mason Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, NetherlandsSearch for more papers by this authorGert J. de Lange, Gert J. de Lange Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, NetherlandsSearch for more papers by this author Tom Jilbert, Tom Jilbert [email protected] Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, NetherlandsSearch for more papers by this authorGert-Jan Reichart, Gert-Jan Reichart Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, Netherlands Alfred Wegener Institute for Polar and Marine Research, Biogeosciences, Bremerhaven, GermanySearch for more papers by this authorPaul Mason, Paul Mason Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, NetherlandsSearch for more papers by this authorGert J. de Lange, Gert J. de Lange Department of Earth Sciences–Geochemistry, Faculty of Geosciences, Utrecht University, Utrecht, NetherlandsSearch for more papers by this author First published: 31 December 2010 https://doi.org/10.1029/2010PA001955Citations: 19AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract [1] High-resolution laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) scanning of laminated sediments from the Urania basin is used to investigate short-time-scale variability in export productivity and redox conditions during the formation of eastern Mediterranean sapropel S1. Sedimentary enrichments of molybdenum (Mo), vanadium (V), and uranium (U) reflect deep-water redox conditions, most likely those near to the seawater-brine interface, while enrichment of Ba is related to biogenic barite and hence to export productivity. The enrichments of all four elements show strong variability on multidecadal to multicentennial time scales throughout S1. A partial decoupling of export productivity from redox conditions at the height of sapropel formation suggests that hydrographic changes, i.e., a variable ventilation rate of the eastern Mediterranean, played an important role in determining deep-water redox conditions. A pronounced switch is observed in the enrichments of redox-sensitive trace metals, from dominantly 300–600 year variability during early S1 to dominantly 100–300 year variability during late S1, indicating a change in the mean frequency of variability in the ventilation rate. The presence of a similar shift in the frequency of tropical and extratropical climate records at this time suggests that ventilation of the eastern Mediterranean was coupled to global climate variability. 1. Introduction [2] Few phenomena in the marine sediment record have been studied so thoroughly as the sapropels of the eastern Mediterranean. These highly conspicuous organic-rich layers, which punctuate the hemipelagic marls of Mediterranean sediment cores and land sections, have attracted paleoclimatologists from a wide range of disciplines since the pioneering work of Olausson [1961]. As a consequence, it is now well established that sapropels form rhythmically, in association with northern hemisphere insolation maxima driven by orbital precession [Hilgen 1991; Rossignol-Strick, 1985], and have done so since the Miocene [Cita and Grignani, 1982; Lourens et al., 1996; Thunell et al., 1984]. [3] During sapropel formation, an intensified African monsoon enhances Nile outflow [Rossignol-Strick et al., 1982], and activates dormant river systems elsewhere in North Africa [Rohling et al., 2002a]. Precipitation in southern Europe increases simultaneously [Bard et al., 2002; Kotthoff et al., 2008; Rohling and Hilgen, 1991] due to an intensification of moisture transport in the westerlies band. The resulting altered hydrologic balance of the eastern Mediterranean triggers the formation of organic-rich sediments, although a number of mechanisms may be involved. Evidence has been presented for enhanced export productivity during sapropels [Calvert, 1983; de Lange and ten Haven, 1983; Passier et al., 1999]. This implies an increased supply of limiting nutrients to the photic zone, either by enhanced river runoff, or, in the case of phosphorus, enhanced regeneration from the sediments [Slomp et al., 2002], and in the case of nitrogen, fixation from the atmosphere [Sachs and Repeta, 1999]. The fraction of carbon exported to the deep waters may be enhanced by the presence of a deep-chlorophyll maximum [Rohling and Gieskes, 1989] characterized by inefficient carbon recycling. On the other hand, changes in water buoyancy and circulation have been shown to isolate deep-water masses during sapropel formation [Rohling, 1994; Rohling and Bryden, 1994] and hence precondition the eastern Mediterranean for suboxia or anoxia, which may in turn intensify under elevated export productivity levels. Such deep-water sub/anoxia enhances the preservation of exported organic matter [see de Lange et al., 2008] providing an additional mechanism for carbon enrichment in the sediments. [4] Due to the sampling resolution of most studies, conclusions on the stability of the conditions leading to sapropel formation, or coupling between export productivity and the deep-water redox state, are largely based on centennially averaged data. Many sapropel sediments also show some evidence of syndepositional or postdepositional bioturbation, limiting their use as high-resolution archives for comparison with other regional and global records. Several studies have identified single multicentennial "interruptions" within sapropels [e.g., Kotthoff et al., 2008; Mercone et al., 2001; Meyers and Arnaboldi, 2005; Rohling et al., 2002a], which have been interpreted as large-scale ventilation events triggered by millennial-scale climate variability in the northern hemisphere [Rohling et al., 2002b]. Furthermore, observations of oxyphilic benthic formaminifera in sapropels from the Aegean Sea and offshore Libya [Casford et al., 2003] have raised the possibility that less intense but more frequent interruptions may occur throughout sapropel deposition. However, the precise duration of such events, and the roles in their occurrence of export productivity and physical ventilation changes, remain to be established. Improving the resolution of export productivity and redox proxy profiles within sapropels may hold the key to identify the processes controlling these events, and how they relate to climate variability on a global scale. [5] Here we investigate short-time-scale export productivity and redox variability during the most recent sapropel, S1, using geochemical proxies. We first establish the oceanographic and geochemical mechanisms responsible for short-time-scale variability in the enrichments of barium (Ba), molybdenum (Mo), vanadium (V) and uranium (U). Subsequently, we investigate the potential climatic drivers of these mechanisms. The early Holocene interval during which S1 was deposited is represented in high-resolution terrestrial climate records from a range of latitudes [e.g., Bar-Matthews et al., 2003; Fleitmann et al., 2003; Mayewski et al., 1997]. 2. Marine Geochemistry of Ba, Mo, V, and U in the Context of Sapropel S1 [6] Enrichment of Ba in sapropel sediments is traditionally thought to be caused by the preservation of barite (BaSO4), which precipitates within decaying organic matter during its descent through the water column [Bishop, 1988; Dehairs et al., 1980]. Due to the close correspondence of sedimentary Ba and organic carbon enrichments in unoxidized sapropels, the magnitude of this "biogenic"-Ba enrichment has been proposed to reflect export productivity (i.e., the organic-carbon flux to the seafloor) during sapropel formation [e.g., Pruysers et al., 1991; Thomson et al., 1995; van Santvoort et al., 1996]. However, a number of potential caveats are associated with the use of sedimentary barite enrichments for reconstructions of export productivity (see review of Paytan and Griffith [2007]). First, changes in the ambient dissolved-Ba concentration may lead to variable precipitation rates of barium in the water column [Dymond et al., 1992], or variable dissolution rates of biogenic barite in the water column and at the sediment-water interface [e.g., Eagle et al., 2003; McManus et al., 1998]. Second, barite may dissolve after deposition if pore waters become sulfate-depleted [e.g., van Os et al., 1991]. McManus et al. [1998] also provide evidence for regeneration of dissolved Ba from coastal margin sediments which show no significant depletion of pore water sulfate, suggesting that (labile) solid-phase species other than barite may be quantitatively important in the marine Ba cycle. Finally, in locations where the biogenic barite enrichment is low, changes in the detrital Ba/Al ratio may become significant if normalization to Al is used to estimate Ba enrichment [Reitz et al., 2004]. [7] Enrichments of Mo, V and U in sapropel sediments have been interpreted to reflect changing water column redox conditions during sapropel formation [Arnaboldi and Meyers, 2007; Gallego-Torres et al., 2010; Mercone et al., 2001; Nijenhuis et al., 1999; Warning and Brumsack, 2000]. The most commonly observed mechanism for sedimentary Mo, U and V enrichment in modern low oxygen environments is reductive transformation of the oxyanions MoO42−, UO2(CO3)34− and H2VO4− into stable solid phase species, either at, or by diffusion across, the sediment water interface [e.g., Crusius et al., 1996; Emerson and Huested, 1991]. Such transformations occur when the reduction potential (Eh) of pore waters decreases to critical values, as electron acceptors are used in organic matter remineralization. Pourbaix diagrams predict reductive precipitation of the three metals to occur close to the Eh level of iron oxide reduction, in the order U → V → Mo with decreasing Eh (Figure 1a, note simplified ion systems). Field and experimental evidence also suggests that the reduction of all three elements to insoluble forms may be accelerated in the presence of free sulfide [Helz et al., 1996; Klinkhammer and Palmer, 1991; Wanty and Goldhaber, 1992]. This appears to be particularly important for Mo; the oxyanion MoO42− undergoes rapid transformation to particle-reactive oxythiomolybdates during reaction with HS−, a phenomenon known as the "sulfide switch" [Helz et al., 1996]. MoO42− in oxic seawater also associates with iron and manganese oxides, hence Mo may be shuttled to the sediments by these carrier phases [e.g., Crusius et al., 1996], but will be released as they dissolve in suboxic conditions [Brumsack, 2006; van der Weijden et al., 2006]. Whatever the mechanism of trace metal removal to the sediments, persistent stagnation of water masses can also result in the depletion of MoO42−, [UO2(CO3)3]4− and H2VO4− [see Colodner et al., 1995; Emerson and Huested, 1991], potentially decreasing the rate of subsequent removal to the sediments by means of the "basin reservoir effect" [Algeo and Lyons, 2006]. Figure 1Open in figure viewerPowerPoint (a) Pourbaix Eh-pH diagrams for U, V, and Mo, as calculated from the FACT thermodynamic database with the software FACTSAGE [Bale et al., 2002], redrawn from Takeno [2005] with permission. The diagrams represent the simple system Element-H-O at a total solute concentration of 10−10 mol kg−1, standard temperature and pressure, thus simplifying the true dynamics of eastern Mediterranean seawater during sapropel formation. Dashed lines indicate Eh precipitation thresholds at seawater pH (8.2). The reductive precipitation thresholds of Mn and Fe oxides under the same conditions are shown to the right. (b) Map showing the coring site of PP44PC in the Urania basin, eastern Mediterranean, and locations of contemporaneous climate records used in this study. (c) Bathymetric detail of the Urania basin, with depth contours in kilometers. Contour interval is 0.1 km outside the basin and 0.02 km inside the basin. Shaded area indicates brine surface. 3. Materials and Methods Study Location [8] Piston Core PP44PC (35°14.20 N, 21°29.80 E) was recovered from 3400 m depth, within the Urania basin in the eastern Mediterranean (Figures 1b and 1c). The basin is a depression within the so-called "cobblestone topography" of the Mediterranean Ridge accretionary wedge [Kastens, 1981], and is filled with a brine of major ionic ratios similar to those of seawater, but with roughly six times higher salinity (Table 1). A number of such brine-filled basins have been discovered on the Mediterranean Ridge; their occurrence is related to the compressional deformation, and subsequent dissolution, of Messinian evaporite strata in the subsurface [Camerlenghi, 1990], or the consequent upward motion of relic brine [Vengosh et al., 1998]. Although the strength and major-ion composition of the brines vary between basins, all display a strongly stratified seawater-brine interface, coincident with a steep Eh gradient from seawater oxygen concentrations to complete anoxia over less than a meter [de Lange et al., 1990a]. Unfortunately, no data exist for Mo, U and V concentrations in the Urania brine. However, concentrations of Mo, U and V in the brines of the Bannock and Tyro basins show strong depletions across the seawater-brine interface. Concentrations of Mo and U decrease to values below the detection limit of the neutron activation analysis technique employed by van der Weijden et al. [1990], while the concentration of V drops to roughly 0.25 times the value in the overlying seawater (Table 1) [van der Weijden et al., 1990]. Such strong depletions are consistent with the reducing nature of the brines, and are also expected to be present in the Urania basin, which has the highest sulfide content of any of the basins yet discovered (Table 1). Table 1. Concentrations of Dissolved Constituents in Eastern Mediterranean Brine Basins and Typical Seawater Valuesa Brine Na Cl Mg K Ca SO4 HS CH4 Ba Mo U V Urania 3503 3729 316 122 32 107 16 5.56 - - - - Bannock 4235 5360 650 127 17 137 3.0 0.45 290 <20 <2 6.7 Seawater 528 616 60 11.4 11.6 31.8 2.6 × 10−6 1.5 × 10−6 50–150 160 16 32 a All values in mmol/L except for Ba (nmol/L) and Mo, U, and V (μmol/L). All data from van der Wielen et al. [2005], except Ba [from de Lange et al., 1990b] and Mo, U, and V [from van der Weijden et al., 1990]. Discrete Sampling and Dating [9] The S1 interval in PP44PC is finely laminated and visibly distinct from the rest of the core by its olive green color. Discrete samples were taken from the S1 interval at 3 mm resolution, freeze-dried, ground and analyzed for major and minor element concentrations by inductively coupled plasma–atomic emission spectroscopy (ICP-AES) (instrumental precision better than 5%), Organic carbon (Corg) content was analyzed on selected samples by thermal combustion elemental analysis (TCEA) after decalcification in 1M HCl (instrumental precision better than 3%). Age control within the S1 interval was established by AMS 14C dating of the organic-carbon residue of four samples, corrected to calendar years B.P. after calibration to the Marine04 curve [Hughen et al., 2004] (Figure 2c). The variable global reservoir age calculated by the ocean-atmosphere diffusion model of Marine04 was assumed, with no regional offset (i.e., ΔR = 0), as shown to be valid for the modern Mediterranean [Siani et al., 2000]. The age model was constructed using a linear best fit through the four dates, which falls within the 1-sigma uncertainty of each. The calculated sedimentation rate is 6.06 cm/kyr, corresponding to a discrete sampling resolution of 49.5 years (3 mm). The linearity of the sedimentation rate suggests that variability in ΔR between the four sampling intervals [see Siani et al., 2001] was negligible. Figure 2Open in figure viewerPowerPoint (a) Comparison of discrete-sample Ba/Al (open circles) with Corg (in weight %, filled circles) through the S1 interval in PP44PC. Horizontal dashed line represents global mean shale value of Ba/Al [Turekian and Wedepohl, 1961]. Note that data resolution is lower than the 3 mm core-sampling resolution because Corg analysis was not performed on all samples. b.s.f, below seafloor. (b) An x-y plot of the same data. (c) Radiocarbon dating of S1 in PP44PC. Dates are reported as calendar years B.P. after calibration to the Marine04 curve [Hughen et al., 2004]. The age model was constructed using a linear best fit through the four points, which falls within the 1-sigma uncertainties of all points (vertical error bars). Raw 14C ages indicated alongside data points. b.s.f, below seafloor. (d) Five-point running means of 20 μm resolution LA-ICP-MS scan data, plotted as element/Al after corrections described in section 3. Horizontal dashed lines represent global mean shale values [Turekian and Wedepohl, 1961]. Arrows inside vertical scales represent mean values of the complete series. Arrows outside vertical scales represent mean values in S1 in core LC21 [Mercone et al., 2001]. b.s.f, below seafloor. Resin Embedding and LA-ICP-MS [10] Across the S1 interval, sediment blocks were excavated in aluminum trays from the core surface, resin-embedded in a nitrogen-filled glove box [Jilbert et al., 2008] and polished. Major, minor and trace elements, including Ba, Mo, V, U and Al, were analyzed by LA-ICP-MS scanning. Each block was mounted on a stage within a sealed ablation chamber. A 193 nm Lambda Physik excimer laser beam (pulse repetition rate 10 Hz, diameter 80 μm, energy density 8 Jcm−2) was focused onto the sample, and ablated material was transported on He-Ar carrier gas to a quadrupole ICP-MS (Micromass Platform, measurement frequency 50 Hz). The stage was set in steady motion to create an ablation trace perpendicular to the laminations within the sapropel. The combination of measurement frequency and stage speed yields data points separated by ∼20 μm, with each point representing all the material ablated over each 20 μm increment of the stage motion. Due to the 80 μm diameter of the laser beam target, each point thus represents an average of a 100 μm interval. The sedimentation rate of 6.06 cm/kyr results in a LA-ICP-MS data resolution of 1.65 years (100 μm). Ion counts for each element were measured on specific isotopes to avoid mass overlap (Table 2). Repeat measurements of the standard material NIST610, before and after each sample trace, allowed instrumental precision for each element to be estimated (Table 2). Because ablation rate during LA-ICP-MS is nonconstant, precision estimates are made relative to a selected isotope (in this case 44Ca). To generate approximate concentration data, all LA-ICP-MS ion counts were corrected for background values in the carrier gas, for interelement fractionation effects by reference to the fractionation coefficients (ion counts/ppm) of NIST610, and for the natural abundance of the measured isotope with respect to all isotopes of the element in question. Element/Al ratios were then calculated from these corrected values. Table 2. Isotopes Measured by LA-ICP-MS, and Precision Relative to 44Ca as Estimated by Repeat Analysis of Standard Material NIST610 Element Na Mg Al K Ca Ca Ti V Mn Fe Co Ni Isotope 23 24 27 39 43 44 49 51 55 57 59 60 Precision (%) 3.4 3.4 6.9 1.8 0.8 - 3.6 2.0 1.4 1.3 1.7 3.2 Element Cu Zn Sr Zr Mo Mo Cd Cd Ba Ba Pb U Isotope 65 66 88 90 97 100 111 114 137 138 208 238 Precision (%) 4.7 4.0 3.17 9.1 1.2 1.1 8.1 9.1 2.1 2.3 9.1 11.9 Statistical Analysis of Time Series Data [11] Time series of LA-ICP-MS-derived element/Al enrichments from this study, and existing high-resolution climate records from the literature, were spectrally decomposed by means of Morlet wavelet analysis, using the online software of C. Torrence and G. P. Compo (http://ion.researchsystems.com/IONScript/wavelet/) [Torrence and Compo, 1998]. Prior to analysis, the 6700–10900 B.P. interval was selected from each series, low-pass filtered at a cutoff of 30 years, detrended and normalized to unit variance. For further analysis of the LA-ICP-MS element/Al time series, periodicity bands of elevated spectral power were identified from the corresponding wavelet spectra. The series were band-pass filtered at these periodicities, and running correlations were performed on pairs of band-pass-filtered series using a window and step size appropriate to the band pass. The running correlation procedure effectively calculates the coherence between two series (i.e., the covariance at a specified frequency, independent of the power of that frequency in the wavelet). 4. Results and Discussion Proxy Preservation [12] A positive linear correlation is observed between Ba/Al and Corg in the discrete sample series of sapropel S1 in PP44PC (Figure 2b), implying that the Ba enrichment in sapropel S1 sediments at this location is indeed related to biogenic barite, and that sedimentary barite provides a reliable estimate of the organic carbon flux to the seafloor. The magnitude of the Ba enrichment is greatly in excess of the detrital background for the eastern Mediterranean (close to the global average shale value of 0.0037, Figure 2a), indicating that normalization to Al offers a valid measure for biogenic barite enrichment. Furthermore, the sulfate content of the Urania brine is 107 mmol/L, roughly 3.5 times the seawater value (Table 1), preventing barite dissolution by undersaturation. However, these observations do not exclude the possibility that the relationship between export productivity and primary productivity (the f ratio [Laws et al., 2000]) may have varied during S1; hence, the Ba/Al data must be discussed strictly in terms of carbon fluxes to the seafloor (export productivity), and not primary productivity (see Eagle et al. [2003] for further discussion). [13] The close correspondence between Ba/Al and Corg in the uppermost part of S1 (Figure 2a) confirms that postdepositional oxic "burn-down" of S1 [van Santvoort et al., 1996] did not occur, consistent with an interpretation of persistent anoxic brine conditions in the Urania basin throughout the Holocene [Hübner et al., 2003]. This implies that no remobilization of redox-sensitive trace metals such as Mo, V and U has taken place due to the propagation of a burn-down front [see Thomson et al., 1995]. General Characteristics of LA-ICP-MS Profiles [14] The Al-normalized LA-ICP-MS profiles of Ba, Mo, V and U (Figure 2c) show that all four elements are enriched in the S1 interval in PP44PC, relative to global average shale values. The magnitudes of the mean enrichments of Mo, V and U (and the first-order shapes of the profiles) are similar to those observed in the unoxidized sapropel S1 interval from a core in the SE Aegean Sea [Mercone et al., 2001]. This observation is important as it implies that the source of the Mo, V and U enrichments in PP44PC is not specifically related to the Urania brine, although the brine may have been instrumental in subsequent preservation of the signals. The profiles of all four elements show some common features on the (multi)millennial scale, most notably the twin Gaussian structure dividing the sapropel into "early S1" and "late S1" intervals, with a saddle of low enrichments defining "mid-S1." This correspondence suggests that the carbon flux to the seafloor, and the severity of reducing conditions at the location of trace metal precipitation, broadly covaried on the (multi)millennial scale. However, on shorter time scales, the relationships between the enrichments of each element become more complex. Wavelet spectra of the LA-ICP-MS data confirm the existence of quasi-cyclic, multicentennial and multidecadal time scale variability in the enrichments of all four elements (Figure 3). Three useful arbitrary periodicity bands may be defined based on the results, namely, 300–600 years, 100–300 years and 30–100 years. Figure 3Open in figure viewerPowerPoint Morlet wavelet analysis of time series, performed using the online software of Torrence and Compo at http://ion.researchsystems.com/IONScript/wavelet/ [Torrence and Compo, 1998]. (left) Top to bottom: V/Al, Mo/Al, U/Al, Ba/Al (LA-ICP-MS count ratios in PP44PC). (right) Top to bottom: log non-sea-salt K+ in GISP2 [Mayewski et al., 1997], residual δ14C around a 2000 year running mean of the IntCal04 calibration line [Reimer et al., 2004], δ18Ocalcite of Qunf Cave stalagmite Q5 in southern Oman [Fleitmann et al., 2003]. All series were low-pass filtered at a cutoff of 30 years, detrended within the interval presented and normalized to unit variance prior to analysis. Wavelet power is scaled by the global wavelet spectrum (to the right of each plot). The cross-hatched region is the cone of influence, where zero padding has reduced the variance. The thick solid line contour is the 10% significance level, using a red noise (autoregressive lag 1) background spectrum. This 10% significance level is represented in the global wavelet power spectra by the dashed line. Note that the significance level is arbitrary and zones of high spectral power discussed in the text may fall outside the significance threshold. Horizontal dashed lines divide wavelet spectra into arbitrary 30–100 year, 100–300 year, and 300–600 year bands. Patterns of Multicentennial and Multidecadal Variability in Elemental Enrichments [15] High power in the multicentennial bands (300–600 years, 100–300 years) is observed in the wavelet spectra of all four elemental enrichments (Figure 3). A pronounced switch from dominantly 300–600 year to dominantly 100–300 year variability is observed in U/Al at ∼9 kyr B.P. This switch is also evident in V/Al, but its significance is damped by high power in the 30–100 year band throughout the V/Al profile. Damping of power in the multicentennial frequencies appears to be even greater in the case of Mo/Al, which is strongly dominated by 30–100 year variability throughout S1. Ba/Al shows intermittent high power in the 100–300 year band throughout S1, and an interval of high power in the 300–600 year band from ∼10.0–8.5 kyr B.P. We band-pass filtered the LA-ICP-MS data at 300–600 years from 10700 to 9000 years B.P. (roughly defining "early S1"), and at 100–300 years from 9000 to 6700 years B.P. ("late S1"), in accordance with the switch in power in the wavelet spectra (Figure 4b). Running correlation analysis of these series shows strong positive coherence between U/Al and V/Al during late S1, but negative coherence during early S1 (Figure 4b). Ba/Al and V/Al show positive coherence, which is relatively strong at the onset and termination of S1, and during the mid-S1 saddle, and relatively weak at times of fully developed sapropel conditions in early S1 and late S1 (Figure 4b). Figure 4Open in figure viewerPowerPoint (a) Detrended, normalized LA-ICP-MS Element/Al data used as input for subsequent band-pass filtering. (b) LA-ICP-MS Element/Al data, band-pass filtered at 300–600 years from 10700 to 9000 years B.P. and at 100–300 years from 9000 to 6700 years B.P., and running correlation analysis of selected pairs of series. Each data point represents the magnitude of Pearson's R for the linear correlation between two series, within a 1000 year window centered on the data point. Note that the 99% confidence threshold is sensitive to the resolution of resampling during detrending and normalization of data (2 years). (c) LA-ICP-MS Element/Al data, band-pass filtered at 30–100 years from 10700 to 6700 years B.P., and running correlation analysis of selected pairs of series. Each data point represents the magnitude of Pearson's R for the correlation between two series, within a 500 year window centered on the data point. Note that the 99% confidence threshold is sensitive to the resolution of resampling during detrending and normalization of data (2 years). [16] High power in the multidecadal band (30–100 years) is observed in all elemental enrichments throughout S1, with the exception of U/Al during early S1 (Figure 3). Running correlation analysis of the 30–100 year band-pass-filtered series shows that V/Al and Mo/Al display strong positive coherence throughout S1 (Figure 4c). Ba/Al also shows positive coherence with both V/Al and Mo/Al. However, the strength of coherence is much greater during the onset and termination of S1, and during the mid-S1 saddle, than during the fully developed sapropel conditions of early S1 and late S1 (Figure 4c). This loss of coherence between Ba/Al and V/Al in early S1 and late S1 is even more pronounced than in the m