Millennial- to decadal-scale paleoenvironmental change during the Holocene in the Palmer Deep, Antarctica, as recorded by particle size analysis

Save this PDF as:

Size: px
Start display at page:

Download "Millennial- to decadal-scale paleoenvironmental change during the Holocene in the Palmer Deep, Antarctica, as recorded by particle size analysis"


1 PALEOCEANOGRAPHY, VOL. 17, NO. 3, 8004, doi: /2000pa000602, 2002 Millennial- to decadal-scale paleoenvironmental change during the Holocene in the Palmer Deep, Antarctica, as recorded by particle size analysis Nathaniel R. Warner 1 and Eugene W. Domack Department of Geology, Hamilton College, Clinton, New York, USA Received 30 October 2000; revised 18 January 2002; accepted 22 January 2002; published 13 September [1] A record of Holocene paleoenvironmental variability from the Antarctic Peninsula has been produced using sediment cores from Site 1098 gathered on Ocean Drilling Project Leg 178. The results provide an accurate and continuous record of biotic and sedimentologic responses to Holocene environmental change. Holes 1098B and 1098C were used to produce the first ultrahigh-resolution record of grain size variability for the past 13,000 years. We utilized samples every 2.5 cm (at 3 5 yr cm 1 ) which involved the analysis of 1466 samples from two holes at the site. Particle size was measured on uniform suspensions of dispersed particles using a laser diffraction method. The particles are mainly grains of silt (65 85%) and clay (15 35%) size. The analyses revealed that in most intervals, biotic particles in the form of diatom frustules (mostly spores) dominate the silt component, which is inversely reflected by magnetic susceptibility (MS). In intervals of the core that deviate from this relationship it was found that the MS signal correlated with medium to fine or coarse silt that was dominated by terrigenous rather than a biogenic (diatom) component. Spectral analyses of the variability in clay and medium-fine silt content over the last 9000 calendar years reveal cycles of 1800, 400, 200, 100, 70, 60, and 50 years. The analyses revealed that particle size, specifically medium-fine silt, is an excellent proxy for environmental change in the Palmer Deep region. Such changes may be forced by solar variability (the 200 and 400 year cycles), lunar tidal changes (the 1800 year cycles), and an as yet undetermined multidecadal forcing phenomena that operates in the SE Pacific sector of the Southern Ocean. INDEX TERMS: 3022 Marine Geology and Geophysics: Marine sediments processes and transport; 4207 Oceanography: General: Arctic and Antarctic oceanography; 9310 Information Related to Geographic Region: Antarctica; 4558 Oceanography: Physical: Sediment transport; 4267 Oceanography: General: Paleoceanography; KEYWORDS: Palmer Deep, grain size, Antarctic Peninsula, Holocene climate cycles Citation: Warner, N. R., and E. W. Domack, Millenial- to decadal-scale paleoenvironmental change during the Holocene in the Palmer Deep, Antarctica, as recorded by particle size analysis, Paleoceanography, 17(3), 8004, doi: /2000pa000602, Introduction [2] Despite increased and well-deserved attention toward climate/oceanographic variability during the Holocene [Broecker et al., 1999; Bond et al., 1997; demenocal et al., 2000], there remain few, if any, high-resolution records over this time span for the Southern Ocean [Hodell et al., 2001]. The Southern Ocean is a complex system that involves interplay of the Pacific, Indian, and Atlantic oceanic realms around the circum-antarctic [Barker et al., 1998] (Figure 1). Drilling of sedimentary sequences in the Palmer Deep (PD), Antarctica, during Leg 178 of the Ocean Drilling Program (ODP) was the first attempt at filling this gap in our paleoceanographic understanding of the circum-antarctic during the Holocene epoch. In a complimentary fashion, ice cores are already providing an understanding of climatic variability across regions of the Antarctic through the 1 Now at Department of Geology, Miami University of Ohio, Oxford, Ohio, USA. Copyright 2002 by the American Geophysical Union /02/2000PA Holocene (Taylor Dome and Siple Dome; Figure 1). Results from ocean drilling have not yet addressed the nature of century- to millennial-scale variability in the coastal Antarctic. Other sites of opportunity in the ODP program such as the Cariaco Basin, Santa Barbara Basin, and Sannich Inlet are excellent examples of the potential of the PD Sites. [3] The Antarctic Peninsula and the region surrounding PD have responded quickly to climate change over the last half century [Smith et al., 1999]. The western side of the Antarctic Peninsula has seen an increase in annual mean temperature of 2.5 C in the past 50 years [Jones et al., 1993]. The peninsula region is a prime area to study paleoenvironmental change (Figure 1). This is because of PD s proximity to the Antarctic Convergence, Antarctic Circumpolar Current (ACC), Antarctic Circumpolar Low (ACL) [Smith et al., 1996], and frontal boundaries between warm Circumpolar Deep Water (CDW) and Modified Weddell Seawater (MWW) [Ishman and Domack, 1994] (Figure 1). The PD also lies on a marine ecotone of contrasting surface water and sea ice conditions that regulate and result in contrasting productivity regimes [Domack and McClennen, 1996]. In addition, glacial climate systems are transitional across the region from polar to subpolar climates, in a south to north PAL 5-1

2 PAL 5-2 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE Figure 1. View of Antarctic and the Southern Ocean with location of the Palmer Deep (PD) site indicated by the diamond. Generalized flow of the Antarctic Circumpolar Current (ACC), East Wind Drift, and the Ross and Weddell Gyres are indicated by the arrows. Generalized position of the Antarctic Convergence (AC) is also shown by thick white line. Note the key position of the PD Site with respect to the ACC, AC, and overspill from the Weddell Gyre (the East Wind Drift in part). Also shown are the sites of ice cores at Taylor Dome (star) and Siple Dome (upside down triangle). See color version of this figure at back of this issue. transect gradient [Griffith and Anderson, 1989; Domack and Ishman, 1993; Kirby et al., 1998]. [4] Previous studies on marine sedimentation along the western edge of the Antarctic Peninsula have focused on the bays or fjords close to the continent [Kirby et al., 1998; Shevenell et al., 1996; Domack and Ishman, 1992, 1993]. More recent work has proven that these changes can also be observed in deep basins that lie along the inner continental shelf [Leventer et al., 1996; Kirby et al., 1998]. While deeper and more difficult to access, these inner shelf records record a more open oceanic signal removed from protection and local variability found within the fjords. [5] For the above reasons, Ocean Drilling Program Leg 178 drilled Sites 1098 Holes A C and 1099 (Holes A and B) in the PD (Figure 2). PD is an informal name given to an inner shelf depression of unusual depth and relief recently surveyed with swath mapping (SEABEAM 1 ) during cruise NBP of the RVIB N. B. Palmer (Figure 2). PD appears to have served as an intersection of distinct ice drainage systems and contains at least two distinct subbasins (Basins I and III) (Figures 2 and 3). These systems included those that fed off the Anvers Island ice cap from the northeast, valley glaciers that funneled into the system across the shelf from the west, and ice that entered the eastern end of the PD from the Bismark Straits and Southernmost Gerlache Strait (Figure 3). The convergence of flow probably helped to overdeepen the basin and funneled ice (perhaps in streaming fashion) out into a cross-shelf trough that extends some 200 km to the shelf break (Figure 3). [6] Site 1098 is located in the narrows of a 2 km wide perched basin (Basin I) that trends in a southwest to northeast direction (Figure 2). Acoustic imagery revealed a draped sequence of pelagic and hemipelagic sediments at this site [Rebesco et al., 1998; Scientific Party, 1999]. Although the contour trend does allow connection between Basins III and I, the floor of the two basins within PD are clearly separated by a narrow sill of 10 m relief (Figure 2). [7] The PD basin lies within a region of high productivity, and consequent high sedimentation rates. Biogenic productivity is higher in low-salinity warm layers near sea ice

3 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE PAL 5-3 Figure 2. Swath bathymetric map of the PD indicating locations of Sites 1098 (yellow diamond) and 1099 (green diamond) drilled during Leg 178 of the Ocean Drilling Program in Basins I and III are also indicated as are the water depths in meters. Contour interval is 50 m. See color version of this figure at back of this issue. margins and lower in cold water beneath sea ice [Leventer et al., 1996; Kirby et al., 1998]. Previous studies on sediments in the PD reveal the presence of a year productivity cycle preserved over the last 3200 years [Leventer et al., 1996]. Studies on Core 1098 B also reveal strong periodicity of MS in the 400, 200, and year frequency bands [Domack et al., 2001] (Figure 4). [8] This study aims to test the use of grain size as a paleoenvironmental proxy and to compare it to other proxies for the Holocene. The other proxies include diatoms [Taylor and Sjonneskog, 2002; Sjunneskog and Taylor, 2002; Leventer et al., 2002], oxygen isotopes [Shevenell and Kennett, 2002], organic matter/carbon isotopes [Dunbar et al., 2000], foraminifera [Ishman and Sperling, 2002], and magnetic mineralogy [Brachfeld et al., 2002]. [9] Our method chooses to utilize the total particle population of mineral and biotic grains and to treat the diatom frustules as part of this detrital fraction. Standard classification schemes of marine sediments do a poor job of characterizing the detrital size of the grains. Biosiliceous rich sediments are classified as either siliceous oozes (for those that contain >30% biogenic silica) or siliceous muds (for those that have <30% siliceous matter). This leaves a great deal of detail out of the picture. Fine-scale variations of the constituent components are then assessed by one of several time-consuming and, in some cases, error-rich methodologies, including (1) direct counting via microscopy of the biogenic component (this usually takes an experienced diatomist at least one working day per sample, not including preparation time) biogenic silica determinations via wet chemistry techniques that are associated with rather large errors of at least ±5%. [10] The results of D. Katz and E. W. Domack (manuscript in preparation, 2002) on these same cores suggested the utility of total particle size analysis in evaluating the relative contribution of biogenic (diatom) versus detrital (siliciclastic) constituents. Their results highlighted the potential use of the method in diatomaceous sediments that are rich in Chaetoceros spores, as the spores present an ideal particle to the laser beam without the pore character of most other diatom frustules (Figure 5a). Maxima in the medium to fine silt fraction (the average size of Chaetoceros spores are 5 8 mm) could then be used to rapidly asses the biogenic contribution to the sediment if detrital particles of similar size are accounted for and subtracted from the analyses. In fact, this proved to be an unnecessary step since the sediments lacking an overabundance of Chaetoceros spores were those that contained abundant detrital and diatom particles of widely different sizes not within the dominant 5 8 mm range (Figure 5b). The study of LoPiccolo [1996] is particularly

4 PAL 5-4 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE Figure 3. Regional view of the Graham Land Coast and Palmer Archipelago with locations of straits and bays mentioned in text. Bathymetric contours are labeled in meters. PD is outlined in a box. F, Faraday Station; P, Palmer Station. significant to our investigation. LoPiccolo [1996] investigated particle size variation within a 9 m long core from Andvord Bay, a coastal fjord some 100 km from the PD site (Figure 3). Here the sediments demonstrated the same periodicity in magnetic susceptibility (MS) and preserved organic carbon over the latest Holocene as that seen in the PD record [Domack et al., 1993; Leventer et al., 1996]. Particle size results from Andvord Bay demonstrated a strong correlative signal to the MS and TOC parameters, thus suggesting the utility of this approach to the much better dated ODP cores.

5 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE PAL 5-5 Figure 4. Stratigraphic succession of the composite record at Site 1098 (cores A C) with magnetic susceptibility (MS), mass accumulation rate (MAR), and gravel abundance versus age and composite depth in the core. Holocene paleoenvironmental intervals are also indicated. Note the change in scale for the MAR values prior to 9070 years B.P. From Domack et al. [2001]. See color version of this figure at back of this issue. [11] The changes observed in particle size represent changes in depositional environment, from hemipelagic silts and clays to pelagic seasonally open water conditions or from glacial marine to biogenic sedimentation, respectively. Particle size analysis in PD offers the opportunity to study transitions from biogenic sedimentation to terrigenous sedimentation and compare these to changes in MS and other paleoenvironmental proxies. 2. Methods 2.1. Core Collection and Sampling [12] Advanced piston cores (APC) 1098A, 1098B, and 1098C were recovered during Ocean Drilling Project Leg 178 from the Bellingshausen Sea on 13 March 1998 [see Scientific Party, 1999]. A composite depth scale was utilized from sites1098 A, 1098B, and 1098C [Acton et al., 2002] that eliminates core gaps (disturbances) that occur at the top of each APC core section. [13] The core was sampled every 2.5 cm at the Bremen core repository by the PD working group in August Grain size was measured using laser diffraction via a Malvern Mastersizer E 1 located at Hamilton College; 1413 samples from Core 1098B were analyzed using a 45 mm lens that resolves size distributions between 0.05 and 80 mm. Another 53 samples from 1098C ( m composite depth (mcd)) were analyzed in the same size range. Samples were prepared in order to disperse the clay particles using standard ultrasonic and dispersant suspension Data Analysis [14] The graphs produced by the Malvern include both frequency and cumulative particle diameters distributions in microns. Figure 5a shows a sample with high Chaetoceros spore content, while Figure 5b was from an interval with relatively low amounts of spores and a coarser content; observations confirmed by both scanning electron microscope (SEM) and standard diatom microscopy [Leventer et al., 2002]. The volume percent grain size of clay, finemedium silt, and coarse silt produced by the Malvern was normalized by eliminating the sand percentages. The fine fraction percentages were then processed under a five-point moving average smoothing function. The smoothing function reduces the sharp maxima produced by individual analyses and aids in observing broader-scale patterns in the data. Such smoothing, when applied to the age relation, results in averaging within year time binds. However, because of the ultrahigh resolution of our original

6 PAL 5-6 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE Figure 5. (a) A coarse skewed trimodal particle size distribution produced by the Malvern Mastersizer E. Modes are at 1, 7, and 11 mm. This graph represents a sample with a very high Chaetoceros resting spore content (7 mm). (b) A sample with a much lower content of Chaetoceros spores with a larger average grain size (25 mm) and more poorly sorted distribution, typical of samples with a higher terrigenous/biogenic ratio. (c) Well-sorted grain size distribution of a laminated interval rich in Chaetoceros spores from 35 to 42 mbsf (11,100 13,200 years B.P.). (d) Poorly sorted particle size distribution from a terrigenous rich lamina also from 35 to 42 mbsf (11,100 13,200 years B.P.). sampling (10 year intervals) we have avoided most of the problems commonly associated with high-frequency time series analysis [Crowley, 1999; Wunsch, 2000] Age Model and Spectral Analysis [15] The age model proposed by Domack et al. [2001] was applied to the upper 25 m of the core and employed a third-order polynomial (equation (1)) to regress age versus depth in the record (Figure 4). y ¼ 16:476 þ 556:18x 17:379 x 2 þ 0:38109 x 3 ; where y is age (calendar years B.P.), x is depth (m), and R 2 = The resultant ages were then plotted versus grain size variation with a resolution of 3 5 yr cm 1. The radiocarbon chronology employed 54 radiocarbon dates and is the highest resolved Holocene time series yet produced from Antarctica. Problems with reservoir age corrections were ð1þ overcome within PD by sampling a section with remarkably high sedimentation rates and with surface (modern) organic matter characteristics that match 14 C activity in biogenic calcite from living organisms. [16] Ages were applied during the spectral analysis to determine if there were any consistent patterns (periods) to the variation in grain size. The ages were plotted versus the unsmoothed grain size data for both medium-fine silt and clay to avoid any artifacts created by the five-point smoothing function. The analysis was done using a spectral software program AnalySeries 1.1 [Paillard et al., 1996]. A maximum entropy (high resolution) math function was chosen for accuracy on high-resolution data series [Haykin, 1983]. The maximum entropy method provides discernible maxima by making assumptions about continued (extrapolated) cyclicity outside the original data. For this reason the time series must also be analyzed using a spectral function,

7 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE PAL 5-7 Figure 6. Smoothed grain size content of medium-fine silt, silt, coarse silt, and clay plotted down core with ages. T 1,T 2, and T 3 are turbidites that were not sampled. such as the Blackman-Tukey method, although this produces a noisier spectrum. A Blackman-Tukey analysis was thus chosen as the secondary analysis and was compared to the maximum entropy to ensure confidence in the observed dominate frequencies. The data sets produced by the spectral analyses were then displayed with spectral power on the y axis (log scale) and frequency (cycles yr 1 )onthexaxis Temporal Analysis of Dominant Frequencies [17] The entire grain size set was divided into 2000 year intervals that were then analyzed using the maximum entropy method in order to produce an evolutive spectrum. This allowed for an accurate observation of the variability of the dominant frequencies (century to multidecadal only) throughout the Holocene. The data demonstrated very dominate multicentury periodicities in the last 2000 years and in the first 4000 years of the Holocene [Warner, 2000]. Only during a brief interval in the middle Holocene (6500 ± 100 years B.P.) was the spectral power weakened for the time periods. 3. Results 3.1. Core Description and Analysis [18] The top 23 m of sediment consists mainly of structureless to laminated diatom ooze/mud with scattered coarse sand and gravel [Scientific Party, 1999; Domack et al., 2001] (Figure 4). The lower 20 m also consists of diatom ooze/mud but with thin interbeds of pebbly mud and muddy diamicton and rhythmically laminated diatom ooze and sandy diatom mud [Leventer et al., 2002] (Figure 4). The cores from Sites 1098A, 1098B, and 1098C are described and documented by Scientific Party [1999] Particle Size Analysis [19] The results of the particle size (PS) analyses were used to calculate normalized proportions of clay, mediumfine silt, and coarse silt. Most of the particles are silt, with a range of 65 87%; most samples contain 70 80%. The average silt content decreases over the lower 10 m of the core (Figure 6). Normalized clay contents in 1098B and 1098C varied from 15 to 35% the majority of the samples contained 20 25%. Clay percentages average 22% throughout most of the core before increasing to averages of 33% in the lower 10 m of the core (Figure 6). [20] Throughout the length of the core, different relationships were observed between the particle size (PS) and MS, and so we divide the stratigraphy into five distinct intervals (sections 1 5), each with their own unique PS and MS characteristics Magnetic Susceptibility Versus Particle Size Section 1: 0 6 mbsf ( calendar years B.P.) [21] Magnetic susceptibility measurements over the interval of 0 6 m indicate fluctuations in MS of centimeter gram second (cgs) units. Average MS values gradually decrease through the first 6 m of the core. When MS and PS

8 PAL 5-8 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE Figure 7. Clay, medium-fine silt, and coarse silt plotted versus magnetic susceptibility from 0 to 6 mbsf. Note the correlation between clay and MS and the inverse relationship between MS and coarse silt. analysis are compared graphically, there is a strong correlation between the MS maxima and minima and corresponding fluctuations in clay content (Figure 7). Six minima over the interval of 0 6 m are visible in both smoothed and normalized clay content and MS values. Between the lows are usually three distinct maxima. The exception to this is the interval from 0.5 to 1.5 m ( calendar years B.P.) where there are pronounced MS maximum for a relatively small increase in clay content, whereas in the rest of the uppermost 6 m, clay content and MS are more closely related. [22] Smoothed and normalized silt percentages ranged from 73 to 81% and show cyclicity similar in frequency but not phase to the MS. This is because the PS classes are not independent variables; as silt content increases, clay content necessarily has to decrease. Maxima in coarse silt content correspond to MS maxima in the interval from 0.75 to 1.5m ( calendar years B.P.) but are inversely related (associated with MS minima) in all other intervals (Figure 7). [23] Medium-fine silt content varies from 59 to 65% and varies with MS over the interval of m ( years B.P.) but is inversely related over the other intervals (Figure 7). Coarse silt varies from 12% to a high of 21% at 1.25 m (675 years B.P.). Coarse silt corresponds to MS in the interval m ( years B.P.) but is inversely related over the remaining interval (Figure 7) Sections 2 and 3: mbsf and mbsf ( calendar years B.P.) [24] MS drops from 150 to 20 cgs units over the interval from 6 to6.75 mbsf ( years B.P.), and clay drops from 24 to 20% over this same interval (Figures 4 and 6). Through the rest of the interval, MS values fluctuate between 2 and 18 cgs units. Clay percentages drop slightly from those in the overlying interval, but maxima and minima still correspond to the lower MS maxima and minima. This pattern continues into section 3 with MS values fluctuating between 5 and 25 cgs units. Clay lows correspond to MS lows, and higher clay percentages correspond to high MS values. This section also produces a smoothed clay minimum of 16% at m (6735 years B.P.). Smooth silt varies from 72 to 82% and is inversely proportional to MS over the rest of the interval Section 4: mbsf ( and 10,400 10,680 calendar years B.P.) [25] Section 4 produces a smoothed clay maximum of 30% at m (9195 years B.P.), which then drops to

9 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE PAL 5-9 values between 20 and 24% (Figure 6). This section also produces an MS high of 1975 cgs units at 31 mbsf (Figure 4). This section was not sampled between and mbsf because of its turbidite character Section 5: mbsf (11,100 13,200 years B.P.) [26] Section 5 shows variability within smoothed clay percentages. Clay percentage minima gradually increase down core; clay minima increase from an average of 21 to 25% (Figure 6). MS values range between 50 and 500 cgs units and correspond to sharp maxima and minima in clay content through most intervals (Figure 4). Silt content varies from 74 to 69%, gradually decreasing in average value down core. Silt correlates with MS over the interval mbsf. This section of the core is rhythmically laminated with a sharp distinction of particle size distributions between the biogenic (diatom) lamina and the siliciclastic rich laminae (Figures 5c and 5d ). Biogenic lamina are well sorted and rich in fine- to medium-grained silt (Chaetoceros spores), while the siliciclastic lamina are poorly sorted and contain more clay and very fine grained sand most of which is of detrital origin [Leventer et al., 2002] First-Order Trends in PS Variation [27] The first-order trend in medium-fine silt correlates extremely well with the mass accumulation rate (MAR, from Domack et al. [2001]), with very broad maxima at roughly 6000 years B.P. followed by minima in the most recent portion of the record (Figure 8). Both the MAR and the medium-fine silt variation have a first-order trend that is of similar wavelength to solar insolation curves (dominated by the 11 kyr half cycle of precession) but are out of phase with the regional solar insolation record at this time that contains a maximum at 1000 calendar years B.P. [Berger, 1978]. This observation is consistent with the diatom record from Lallemand Fjord recently reviewed by Taylor et al. [2001] who noted the antiphase relationship between regional insolation and maximum in paleoproductivity. These authors suggested a forcing factor to regulate paleoproductivity related to advection of surface or deep water masses into the region, as opposed to changing regional insolation. This hypothesis is now supported with the isotopic work of Shevenell and Kennett [2002] on the PD cores, at least for the latest Holocene. A comparison of icerafted debris (IRD) content and medium-fine silt can be seen in Figures 4 and Spectral Analysis of PS Variability [28] Spectral analysis of the clay content variability (Figure 9) in the upper 8725 years B.P. (the 1098C data points) compares well in the higher (submillennial) frequency bands with an analysis done previously on the MS data alone [Domack et al., 2001]. However, our analysis revealed an additional cycle of 1800 years as well as the previously recognized maxima at 400, 200, 100, 80, 69, 61, and 51 years for clay (Figure 9a). [29] The second spectral analysis was done using the Blackman-Tukey method. Though this analysis is much noisier, maxima are still observable at the same periods as in the maximum entropy spectral analysis over the same time interval (Figure 9b). 4. Discussion 4.1. High-Frequency Variations in Grain Size [30] Fluctuations in both MS and grain size in the upper 6 m ( years B.P.) of core 1098B represent changes in biotic productivity and silciclastic sediment supply. The cyclic nature of the MS signal reflects a change in the dominant form of sediment being deposited [Leventer et al., 1996]. Periods of high MS reflect a hemipelagic environment rich in siliclastic, terriginous material. [Leventer et al., 1996]. Low MS units represent a depositional environment dominated by pelagic, biogenic (siliceous) sedimentation, reflecting an increase in regional productivity that reoccurs on a multicentury timescale [Leventer et al., 1996]. MS minima are correlated with laminated high productivity intervals and high silt content, reflecting high Chaetoceros spore contents in the core, while MS highs are correlated with structureless to bioturbated low-productivity intervals with high clay content [Domack et al., 2001]. The firstorder trend of mass accumulation rate (MAR) in core 1098B correlates with medium-fine silt through the mid-holocene ( years B.P.) (Figure 8). IRD does not increase over this same interval even as silt increases (Figure 4). This indicates that sediment accumulation is dominated by biogenic particulates (Chaetoceros spores) in the medium to fine silt range through the mid-holocene climatic optimum. Such an observation reinforces the use of grain size as an accurate reflection of paleoenvironmental change in this record. [31] The 200 year cycles of medium-fine silt and MS levels correspond to six high-productivity episodes in the late Holocene Antarctic record [Leventer et al., 1996] (Figure 7). These cycles reflect changes in sea surface conditions that are likely to be on a 400 year timescale with subordinate 200 year events as a harmonic. Cycles of this frequency are believed to be driven by variations in solar radiation output [Leventer et al., 1996; Domack and Mayweski, 1999]. These conditions lead to productivity blooms in low-salinity, warm surface water over the summer period as influenced by surface layer stability under reduced wind stress. This in turn leads to the high medium and fine silt content and low MS intervals because of the dominant sedimentation of diatom species; those that are adapted to high productivity conditions (such as Rhizosolenia spp., Corethron criophilium, and Chaetoceros). [32] Though clay percentages correlate with MS levels in the first 5.8 m ( years B.P.), rock magnetitc data indicate multidomain magnetite (the prime MS carrier) is >10 mm, so little or no magnetite is in the clay portion in the upper 5.8 m [Brachfield et al., 2002]. The MS-clay correlation is therefore a result of the inverse relationship between biotic flux and MS. The higher biotic productivity (higher medium and fine silt (spore) content) leads to a periodic dilution of the MS signal over this section at 200 year intervals (Figure 7). [33] The reduced MS signal over the interval of m below seafloor (mbsf) ( years B.P.) is associated with higher MAR and hence enhanced preservation of primary production. The MS signal is lower because of fewer coarse-grained magnetite grains that reflect either a change in terrigenous provenance [Brachfeld et al., 2002] or simple

10 PAL 5-10 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE Figure 8. Smoothed medium-fine silt percent plotted versus mass accumulation rate. Both show peaks at 6000 years B.P., and both show lows closer to years B.P. A first-order trend line has been added to both curves. Asterisks indicate maxima intervals of preserved total organic carbon flux [from Dunbar et al., 2000]. See color version of this figure at back of this issue. dilution by greater biogenic flux [Kirby et al., 1998; Leventer et al., 1996]. Fluctuations in clay content still correlate with these lower MS levels. Over this interval, MS fluctuations could reflect fluctuations in clay content as there is no coarse grained magnetite, so clay particles could contain a finer magnetic carrier [Brachfeld et al., 2002]. We suggest that clay input from distal sources controls the weak MS signals during periods of high productivity, such as the middle Holocene. [34] Settings >10 km from the coast do not vary significantly in their depositional facies from polar to subpolar areas in the Antarctic. This suggests that instead of climate, other variables such as productivity and bottom currents

11 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE PAL 5-11 a b Figure 9. (a) Spectral analysis, maximum entropy, high resolution of clay content ( years B.P.). Prominent peaks are shown at periods of 1800, 400, 200, 100, 80, 69, 59, and 52 years. (b) Spectral analysis, Blackman-Tukey (compromised) clay content ( years B.P.). Peaks are roughly if not perfectly aligned with the maximum entropy analysis. The combination of the two analyses reaffirms the presence of the cycles in clay content. may play a role in controlling accumulation at sites distal from glacial margins [Domack and Ishman, 1993]. The correlation between MAR and medium-fine silt (Figure 8) would suggest productivity cycles as a control. In support of this are the observations of Dunbar et al. [2000] on the preserved flux of total organic carbon. Episodes of maximum organic carbon preservation correspond to maxima in the medium to fine silt fraction (Figure 8). [35] Over the interval mbsf ( years B.P.), MS fluctuations correlate with silt content (Figure 7). In these intervals, MS levels are influenced by medium-fine and coarse silt that contains magnetite. This sudden drop and increase in silt comes not from biotic influence but more proximal glacial melt or ice rafting influences. [36] The grain size over the interval of mbsf (10,500 12,500 years B.P.) supports the hypothesis that this interval represents a deglaciation [Domack et al., 2001]. At the base of the core (41 m), during the colder periods, clay contents are high and silt is low, and the silt (Chaetoceros spores) content increases up core, representing more biogenic control over the depositional system, as less permanent sea ice is present. A problem with this hypothesis is that MS values do not decrease over this interval, as increased biogenic content should cause it to decrease. One explanation could be that at

12 PAL 5-12 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE Figure 10. Medium-fine silt and clay content with Bond events outlined. The events correspond to medium-fine silt lows back to 9600 years B.P. Correlation with clay content is not as apparent through the events. Further back in the record, the correlation of Bond cycles and medium-fine silt lows are not as strong. The discrepancy could be due to loss of signal, or the age model could need slight adjustments. first, terrigenous influence increased over this interval as well, which did not lead to any dilution of the signal. The increased temperatures would increase meltwater, which would carry the increased sediment into the PD. A progression from biogenic silt influence to terrigenous silt influence can be seen over the interval mbsf (11,125 11,280 years B.P.). Here silt drops dramatically and then increases in tandem with MS variability. The drop in silt content is interpreted as a decrease in regional temperature leading to a drop in productivity and lowered biogenic flux. This is followed by more terrigenous influence, first in clay, which tracks with MS, and then with a poorly sorted silt mode Lower-Frequency Variations in Particle Size [37] When the time interval in the spectral analysis is years B.P., the longer-term period (1800 years) becomes evident. However, the limited time span of only 9000 years is not fully adequate to assess a period of 1800 years as there are at most five cycles in the time series. The 1800 year cycle is evident only in the spectral analysis of clay content (Figures 9 and 10). Clay content shows a stronger periodicity than silt because of primary forcing due to its singular (terrigenous) source. Medium-fine silt is not a pure signal but rather reflects a composite signal from both terrigenous and the more constant biogenic (diatom) accumulation. It therefore does not show the strong 1800 year period in spectral analyses. [38] Bond et al. [1997] noticed a 1 2 kyr periodicity characterized by distinct cooling events (i.e., the Little Ice Age) and periodicity in the amount of IRD in deep-sea sediments of the North Atlantic Ocean basin at an average of 1476 ± 585 years. Evidence pointed toward a forcing

13 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE PAL 5-13 mechanism that operated independently of the glacialinterglacial cycle because it was evident in an entire record spanning 80 kyr [Bond et al., 1997] including the last 10 kyr. [39] A recent proposal by Keeling and Whorf [2000] suggests that the IRD forcing may come from an oceanic tidal mechanism for periodic oceanic cooling. The greatest astronomical tide raising forces occur when the Sun and Moon come in exact mutual alignment with the Earth at their closest respective distances [DeRop, 1971; Keeling and Whorf, 2000]. The 1800 year cycle represents the time for the reoccurrence of all four of these aspects. The Moon is closest to syzygy (new or full moon) and perigee (closest to the Earth). Together these cause perigean tides. The Moon must also be close to either of two points that lie on the ecliptic, while at the same time, the Earth is near perihelion, the point closest to the Sun [Keeling and Whorf, 2000]. The 1800 year cycle represents the time period for reoccurrence of the perhelion to be closely matched with perigean eclipses [Keeling and Whorf, 2000]. [40] The strong tidal forcing caused by the alignment of celestial bodies then cools the sea surface temperature by increasing vertical mixing in the oceans. The mixing causes exchange of deeper, cooler water with the surface. The evidence for this is in IRD and assemblage succession of cold water foraminifera in the Northern Hemisphere. ODP Core 1098B in the PD further supports this with what appears to be a response to climate and/or ocean temperature, evident in clay content on the order of the 1800 year trend (Figure 10). The relationship can clearly be seen as medium-fine silt minima correspond to the timing of the first three of the proposed Bond cycle events as determine by Keeling and Whorf [2000] (Figure 10). Although we question whether the 1800 year tidal period is of sufficient amplitude to effectively transfer to deep mixing, we propose that the tidal oscillator may play a role in regulating tidal pumping mechanisms and/or calving near ice shelves or tidewater glaciers (C. Keeling, personnel communication, 2001). This is supported by our own observations of sediment transport adjacent to AP glacial systems that contain midwater meltwater plumes of tidal and/or storm induced origin [Domack and Williams, 1990; R. Gilbert et al., Sediment trap records of glacimarine sedimentation at Muller Ice Shelf, Lallemand Fjord, Antarctic Peninsula, Arctic, submitted to Antarctic and Alpine Research, 2001]. Other periods of strong tides, 360, 180, and 90 years, do not show themselves in the spectral analyses, whereas there are persistent 77, 60, and 52 year trends (Figure 9). [41] A long-term Tasmanian tree ring record has significant warm season temperature variability with periods of 57, 77, and 200 years [Cook et al., 1996]. The 77 year trend is also apparent in a grain size record from the Gerlache Straight, though there is no 200 year cycle present at this site [Wright, 2000]. Also, Domack and Mayewski [1999] have tenuously correlated the 200 year cycle with both Greenland ice cores and tree ring data. This bipolar relationship indicates a global climate forcing mechanism, one that is reinforced with our observations on the grain size date over the entire Holocene in the PD. [42] The presence of a year cycle in Tasmanian tree rings is possibly attributed to solar sunspot number (Gleissberg cycle) or solar sunspot cycle length [Cook et al., 1996]. The shorter cycles in the record have previously been attributed to fluctuations in the atmospheric convergence line (ACL) by Cook et al. [1996]. The position of the ACL has been linked to sea ice coverage through cyclonic activity in the region. Greater storminess has been linked to regions of higher temperature and lower sea ice extent (Ackley and Keliher as discussed by Cook et al. [1996]) conflicting with other findings that support calmer weather and increased open water [Leventer et al., 1996]. An increase in intensity of the westerlies may cause an increase in surface water temperature and a decrease in sea ice extent [Cook et al., 1996]. This open water, however, would not cause an increase in production because the increased winds destabilize the surface layer via mixing. So, in fact, storm frequency increases ocean surface mixing and may not allow a summer warm layer of high productivity to persist. [43] CDW of the southern Ocean could have an important influence on sea ice extent. Sea ice extent is a major control for biogenic production. Therefore the intensity of upwelling of the CDW, driven by global climate forcing, can influence the timing and spatial extent of biogenic production in the PD region through the control of sea ice. CDW does not, however, have as great an effect on the waters of Gerlache Straight because of the blocking effect of the Palmer Archipelago and the dominance of Weddell Sea Water within the strait. It would appear that the 200 year cycle may be more closely linked to the upwelling of the CDW because of its presence in the PD core but not the Gerlache Straight stratigraphy. 5. Conclusions [44] On the basis of our observations outlined above we conclude the following. (1) Particle size is a good paleoenvironmental proxy, at least as good as MS in the PD sites and most likely in other Antarctic glacial marine sediments of similar character. (2) Medium to fine silt content parallels the MAR at Site 1098 because of the role of varying diatom abundance; hence MAR is likely a reflection of paleoproductivity across the PD that varied at millennial to multidecadal timescales. (3) Coherence of MS and PS is not continuous throughout the entire core, but strong agreement of the two proxies in terms of their temporal variance (assessed through spectral analyses) suggests solar variability as a forcing mechanism on paleoproductivity because of the strong 200 and 400 year frequencies. [45] Acknowledgments. This study was made possible through support from the National Science Foundation, Office of Polar Programs (grant OPP ), the Joint Oceanographic Institutions/U.S. Science Support Program (grant USSSP-235), and the Casstevens Fund for Undergraduate Research at Hamilton College. We would also like to thank Rob Dunbar, Amy Leventer, Scott Ishman, and Stefanie Brachfeld for all of their help on this project. Thanks also go out to the staff of the ODP Bremen Core repository, the PD working group, and the Antarctic CRC/Institute for Antarctic and Southern Ocean Studies (IASOS) in Hobart, Tasmania, for encouraging our pursuit of this study. We are also indebted to the shipboard participants who so enthusiastically embraced the PD objectives during Leg 178.

14 PAL 5-14 WARNER AND DOMACK: PALMER DEEP GRAIN SIZE AND HOLOCENE CLIMATE References Acton, G. D., C. J. Borton, and the Leg 178 Scientific Party, Palmer Deep composite depth scales for ODP Leg 178 Sites 1098 and 1099, Proc. Ocean Drill. Program Sci. Results, in press, Barker, P. F., P. J. Barrett, A. Camerlenghi, A. K. Cooper, F. J. Davey, E. W. Domack, C. Escutia, Y. Kristoffersen, and P. E. O Brien, Ice sheet history from Antarctic continental margin sediments: The ANTOSTRAT approach, Terra Antarct., 5, , Berger, A., Long-term variations of daily insolation and Quaternary climatic change, J. Atmos. Sci., 35, , Bond, G., W. Showers, M. Cheseby, R. Lotti, P. Almasi, P. demenocal, P. Proire, H. Cullen, I. Hajdas, and G. Bonani, A pervasive millennial-scale cycle in North Atlantic Holocene and glacial climates, Science, 278, , Brachfield, S., S. K. Banerjee, Y. Guyodo, and G. D. Acton, A year history of century to millennia-scale paleoenvironmental change magnetically recorded in the Palmer Deep, western Antarctic Peninsula, Earth Planet. Sci. Lett., 194, , Broecker, W. S., S. Sutherland, and T.-H. Peng, A possible 20th-century slowdown of Southern Ocean Deep Water formation, Science, 286, , Cook, E., B. M. Buckley, R. D. D Arrigo, Interdecadal climate oscillations in the Tasmanian sector of the Southern Hemisphere: Evidence from tree rings over the past three millennia, in Climatic Variations and Forcing Mechanisms of the Last 2000 Years, NATO ASI Ser., Ser. I, vol. 141, edited by P. D. Jones, R. S. Bradley, and J. Jouzel, pp , Springer-Verlag, New York, Crowley, T. J., Correlating high-frequency climate variations, Paleoceanography, 14, , demenocal, P., J. Ortiz, T. Guilderson, and M. Sarnthein, Coherent high and low-latitude climate variability during the Holocene warm period, Science, 288, , DeRop, W., A tidal period of 1800 years, Tellus, 23, , Domack,E.W.,andS.E.Ishman,Magnetic susceptibility of Antarctic glacial marine Sediments, Antarct. J. U. S., 27, 64 65, Domack, E. W., and S. E. Ishman, Oceanographic and physiographic controls on modern sedimentation within Antarctic fjords, Geol. Soc. Am. Bull., 105, , Domack, E., and P. Mayewski, Bi-polar evidence from late-holocene Antarctic marine and Greenland ice-core records, Holocene, 9, , Domack, E. W., and C. E. McClennen, Accumulation of glacial marine sediments in fjords of the Antarctic Peninsula and their use as late Holocene paleoenvironmental indicators, in Foundations for Ecosystem Research West of the Antarctic Peninsula, Antarct. Res. Ser., vol. 70, edited by R. Ross, E. Hofmann, and L. Quetin, pp , AGU, Washington, D. C., Domack, E. W., and C. R. Williams, Fine structure and suspended sediment transport in three Antarctic fjords, Contributions to Antarctic Research I, Antarct. Res. Ser., vol. 50, edited by D. H. Elliot, pp , AGU, Washington, D. C., Domack, E. W., T. A. Mashiotta, L. A. Burkley, and S. E. Ishman, 300 year cyclicity in organic matter preservation in Antarctic fjord sediments, in The Antarctic Paleoenvironment: A Perspective on Global Change, part 2, Antarct. Res. Ser., vol. 60, edited by J. P. Kennett and D. A. Warnke, pp , AGU, Washington, D. C., Domack, E. W., A. Leventer, R. Dunbar, F. Taylor, S. Brachfeld, C. Sjunneskog, and ODP Leg 178 Scientific Party, Chronology of the Palmer Deep site, Antarctic Peninsula: A Holocene paleoenvironmental reference for the circum- Antarctic, Holocene, 11, 1 9, Dunbar, R. B., A. C. Ravelo, E. W. Domack, A. R. Leventer, L. Anderson, D. A. Mucciarone, and S. Brachfeld, 13,000 years of decadal-tomillennial oceanographic variability along the Antarctic Peninsula: ODP site 1098, Eos Trans. AGU, 81(48), Fall Meet. Suppl., abstract OS51B-08, Griffith, T. W., and J. B. Anderson, Climatic control on sedimentation in bays and fjords of the northern Antarctic Peninsula, Mar. Geol., 85, , Haykin, S., Nonlinear Methods of Spectral Analysis, 2nd ed., Springer-Verlag, New York, Hodell, D. A., S. Kanfoush, A. Shemesh, X. Crosta, C. D. Charles, and T. P. Guilderson, Aburpt cooling of Antarctic surface waters and sea ice expansion in the south Atlantic sector of the Southern Ocean at 5000 cal yr B. P., Quat. Res., 56, , Ishman, S. E., and E. W. Domack, Oceanographic controls on benthic foraminifers from the Bellingshausen margin of the Antarctic Peninsula, Mar. Micropaleontol., 24, , Ishman, S. E., and M. R. Sperling, Benthic foraminiferal record of Holocene deep-water evolution in the Palmer Deep, western Antarctic Peninsula, Geology, 30, , Jones, P. D., R. Marsh, T. M. L. Wigley, and D. A. Peel, Decadal timescale links between Antarctica Peninsula and ice core oxygen-18, deuterium, and temperature, Holocene, 3, 14 26, Keeling, C. D., and T. P. Whorf, The 1,800-year oceanic tidal cycle: A possible cause of rapid climate change, Proc. Natl. Acad. Sci. USA, /pnas , Kirby, M. E., E. W. Domack, and C. E. McClennen, Magnetic stratigraphy and sedimentology of Holocene glacial marine deposits in the Palmer Deep, Bellingshausen Sea, Antarctica: Implications for climate change?, Mar. Geol., 152, , Leventer, A., E. W. Domack, S. E. Ishman, S. Brachfeld, C. E. McClennen, and P. Manley, Productivity cycles of years in the Antarctic Peninsula region: Understanding linkages among the Sun, atmosphere, ocean sea ice and biota, Geol. Soc. Am. Bull., 108, , Leventer, A., E. Domack, A. Barkoukis, B. McAndrews, and J. Murray, Laminations from the Palmer Deep: A diatom-based interpretation, Paleoceanography, 17(3), / 2001PA000624, LoPiccolo, M., Productivity and meltwater cycles in Andvord Bay, Antarctica: Evidence for high frequency paleoclimatic fluctuations, B. A. thesis, 51 pp., Hamilton Coll., Clinton, N. Y., Paillard, D., L. Labeyrie, and P. Yiou, Macintosh program performs time-series analysis, Eos Trans. AGU, 77(39), 379, Rebesco, M., A. Camerlenghi, L. DeSantis, E. W. Domack, and M. Kirby, Seismic stratigraphy of Palmer Deep: A fault-bounded late Quaternary sediment trap on the inner continental shelf, Antarctic Peninsula margin, Mar. Geol., 151, , Scientific Party, Leg 178 summary: Antarctic glacial history and sea-level change, Proc. Ocean Drill. Program Initial Rep., 178, 1 58, Shevenell, A. E., and J. P. Kennett, Antarctic Holocene climate change: A benthic foraminiferal stable isotope record from Palmer Deep, Paleoceanography, 17(3), /2000PA000596, Shevenell, A. E., E. W. Domack, and G. M. Kernan, Record of Holocene paleoclimate change along the Antarctic Peninsula: Evidence from glacial marine sediments, Lallemand Fjord, in Climate Succession and Glacial Record of the Southern Hemisphere, Pap. Proc. R. Soc. Tasmania, 130(2), 55 64, Sjunneskog, C., and F. Taylor, Postglacial marine diatom record of the Palmer Deep, Antarctic Peninsula (ODP Leg 178, Site 1098), 1, Total diatom abundance, Paleoceanography, 17(3), /2000PA000563, in press, Smith, R. C., S. E. Stammerjohn, and K. Baker, Surface air temperature variations in the western Antarctic Peninsula region, in Foundations for Ecological Research West of the Antarctic Peninsula, Antarctic Res. Ser., vol. 70, pp , AGU, Washington, D. C., Smith, R. C., et al., Marine ecosystem sensitivity to climate change: Historical observations and paleoecological records reveal ecological transitions in the Antarctic Peninsula region, BioScience, 49, , Taylor, F., and C. Sjunneskog, Postglacial marine diatom record of the Palmer Deep, Antarctic Peninsula (ODP Leg 178, Site 1098), 2, Diatom assemblages, Paleoceanography, 17(3), /2000PA000564, Taylor, F., J. Whitehead, and E. W. Domack, Holocene paleoclimate change in the Antarctic Peninsula: Evidence from the diatom, sedimentary, and geochemical record in Lallemand Fjord, Mar. Micropaleontol., 41, 25 43, Warner, N., Grain size analysis in the Palmer Deep, Antarctica: A high resolution paleoclimate record of the Holocene SE Pacific, Bellingshausen Sea, B. A. thesis, 112 pp., Hamilton Coll., Clinton, N. Y., Wright, W., The Schollaert sediment drift: An ultra high resolution paleoenvironmental archive in the Gerlache Strait, Antarctica, B. A. thesis, 72 pp., Hamilton Coll., Clinton, N. Y., Wunsch, C., On sharp spectral lines in the climate record and the millennial peak, Paleoceanography, 15, , E. W. Domack, Department of Geology, Hamilton College, Clinton, NY 13323, USA. N. R. Warner, Department of Geology, Miami University of Ohio, Oxford, OH 45056, USA.