TROPICAL NORTH ATLANTIC SEA SURFACE TEMPERATURE RECONSTRUCTION FOR THE LAST 800 YEARS USING MG/CA RATIOS IN PLANKTIC FORAMINIFERA.

TROPICAL NORTH ATLANTIC SEA SURFACE TEMPERATURE RECONSTRUCTION FOR THE LAST 800 YEARS USING MG/CA RATIOS IN PLANKTIC FORAMINIFERA A Thesis Presented to The Graduate Faculty of The University of Akron In Partial Fulfillment of the Requirements for the Degree Master of Science Matthew A. Abahazi May, 2009 i

TROPICAL NORTH ATLANTIC SEA SURFACE TEMPERATURE RECONSTRUCTION FOR THE LAST 800 YEARS USING MG/CA RATIOS IN PLANKTIC FORAMINIFERA Matthew A. Abahazi Thesis Approved: Accepted: Advisor Dr. Lindgren L. Chyi Dean of the College Dr. Chand Midha Faculty Reader Dr. Lisa E. Park Boush Dean of the Graduate School Dr. George R. Newkome Department Chair Dr. John P. Szabo Date ii

ABSTRACT Global warming is an important scientific and cultural problem for our time. As such, determining the historical sea surface temperatures is critical to our understanding of past, present and future climate change. In order to assess the patterns of prior climate change and establish reliable records upon which future predictions of climate change can be based, it is necessary to measure and calibrate climate change proxies. The samples used for this study come from a 56.5 cm box core retrieved in 1990 from 450 m depth in the Cariaco basin. Consecutive 1mm-thick samples were taken and processed for the planktic foraminifera species Globigerina bulloides for Mg/Ca analyses to estimate sea-surface temperature (SSTs) during the winter/spring upwelling season. The Mg/Ca values were obtained using an ICP-AES and later used in comparison with historical instrumental records to derive an equation to reconstruct sea-surface temperatures during the last 800 years. The record derived from this core indicates a high amount of variability in seasurface temperatures over this 800 year interval. The Medieval Warm Period had considerable temperature variability that was previously not well-documented. Average sea-surface temperatures over this time period are 26.2ºC with slight cooling toward the end of the period. The Little Ice Age is marked by the coldest temperatures of the Late Holocene record, averaging approximately 25.6ºC. The coldest period within the LIA is iii

coincident with the Maunder Minimum in the core record. Modern temperatures as recorded by instrumental records and from the proxies recovered from the core is increasing at rates greater than they have in the past 800 years represented in this sediment-core record. This study is one of the first of its kind to present a highresolution, sea surface temperature record for the tropics. The proxy equation can be used in the region to further delineate climate variability in the Late Holocene. iv

TABLE OF CONTENTS Page LIST OF FIGURES... vii CHAPTER I. INTRODUCTION...1 II. MAGNESIUM/CALCIUM RATIOS...4 Sea surface temperatures from Mg/Ca ratios...4 Other proxies for determining paleotemperature...4 History of the Mg/Ca ratio technique...8 Magnesium incorporation into calcite...10 Calibration and equations...11 Caveats of technique...21 Use of the Mg/Ca technique...24 Benefits of the Mg/Ca technique...26 III. STUDY AREA...27 Physical Characteristics...27 Climate of the Cariaco Basin...33 Previous studies in the Cariaco Basin...34 Importance of the Cariaco Basin...44 IV. MATERIALS AND METHODS...45 v

Materials...45 Methods...45 Importance of calibration and comparison...49 Age Model...50 V. TWENTIETH CENTURY CALIBRATION...54 Importance of Calibration...54 Sea-surface temperature equation derivation...55 Reconstructing Late Holocene SSTs Using the Mg/Ca Calibration Equation...55 Large-scale correlations...58 VI. PRE-TWENTIETH CENTURY COMPARISON...68 Estimated sea-surface temperature change results...68 Comparison to other studies...70 Spectral analysis and forcing mechanisms...85 Summary of comparisons...89 VII. CONCLUSION...91 Conclusions...92 REFERENCES...94 vi

LIST OF FIGURES Figure Page 1. Temperature dependence of distribution coefficient (λ Mg ) on temperature (Oomori et al., 1987)....12 2. Compositional profiles showing the variation of Mg/Ca molar ratios through each analyzed chamber wall of an individual test of G. ruber (Eggins et al., 2003)....13 3. Mg/Ca uptake in the subpolar, foraminifera Globigerina bulloides and the subtropical, spinose, symbiont-bearing planktonic foraminifera Orbulina universa compared to growth temperature (Lea et al., 1999)....15 4. Mg temperature calibration results from culturing experiments with live Globigerina bulloides and core-top samples (Mashiotta et al., 1999)....16 5. Mg/Ca ratio for G. bulloides cultures where Mg/Ca = 0.56*e 0.10*T where T= temperature ( C). The standard error in the single-species core-top calibrations translates to about ±0.6 C (Elderfield and Ganssen, 2000)....18 6. Mg/Ca:temperature exponential relationship for 212-355μm size fraction of G. bulloides (McConnell and Thunell, 2005)....20 7. Location of the Cariaco Basin with the position of the ITCZ during the boreal winter and summer months. Regional surface currents are denoted by arrows. (after Dürkoop et al., 1997)....28 8. Location of the Cariaco Basin with local-rivers entering the basin. Second figure (lower left) shows depth isobaths (meters) by color, star represents core location (after Black et al., 1998)....29 9. Temperature and salinity profiles of the western and eastern sub-basins, respectively. Temperature and salinity above the sills have similar characteristics with the open Caribbean, but below the sills there is an abrupt change in both temperature and salinity (after Peterson et al., 1990)....31 vii

10. Temperature (ºC), salinity ( ), and dissolved oxygen (μm) between November 1995 and August 1998 from the eastern basin (Astor et al., 2003).... 32 11. Bulk Ti content as a three point running mean of 2-mm resolution measurements at ODP hole 1002C and 1002D in the Cariaco Basin during the past 2,000 years (Haug et al., 2003)....42 12. Location of core PL07-73 BC in the Cariaco Basin and core PL07-71 BC which was used for age model (Black et al., 1998)....46 13. A Soutar type box-core with photograph of a core (Peterson, 1991)....47 14. Tie points between cores using percent G. bulloides as applied to the well-established age model of PL07-71 BC that was calibrated using varve, 210 Pb, and accelerator mass-spectrometry 14 C based data (Black et al., 2007)....51 15. Tie points between cores using percent G. ruber (pink) as applied to the well-established age model of PL07-71 BC that was calibrated using varve, 210 Pb, and accelerator mass-spectrometry 14 C based data (Black et al., 2007)....52 16. Tie points between cores using percent G. crassiformis as applied to the well-established age model of PL07-71 BC that was calibrated using varve, 210 Pb, and accelerator mass-spectrometry 14 C based data (Black et al., 2007)....53 17. Comparison over period of overlap of G. bulloides Mg/Ca (black line) to Hadley MAM instrumental SSTs (gray line) (Black et al., 2007)....56 18. Mg/Ca values from PL07-73BC were plotted with the SST s from Jones et al., 1999. Equation of regression line through data is Mg/Ca= 0.048*e 0.173xT (r 2 =.313) (Black et al., 2007)....57 19. Temperature reconstruction from sediments derived from PL07-73BC with 5 point running average....59 20. G. bulloides temperature record from Cariaco Basin sediments (gray) with 95% confidence limit of temperatures (black)....60 21. Twentieth century SST s over the period of instrumental overlap (1870-1990), smooth line represents 5 year running average....61 22. PL07-73BC SST data as compared to the temperature, salinity, δ 18 O from the Sargasso Sea. The Great Salinity Anomaly is noted by the vertical bars during the 1960 s (Keigwin, 1996)....64 viii

23. Multitaper method spectral analysis of Mg/Ca derived SST temperature data (Black et al., 2007)....65 24. Comparison between the temperature record from the Cariaco basin and the Niño 3.4 anomalies for August and February....66 25. Cariaco basin temperature record compared to Sr/Ca ratios and δ 18 O from sclerosponges from Jamiaca (Haase-Schramm, 2003)....67 26. Cariaco temperature record over period without instrumental overlap....69 27. The entire Cariaco basin temperature record illustrating major climate events....72 28. Comparison of the temperature record of this work (black) to the G. bulloides abundance record (gray) of Black et al., 1999....75 29. Comparison of the temperature record of this work to the G. bulloides and G. ruber (pink) δ 18 O record of the nearby core PL07-71BC of Black et al., 1999....76 30. Comparison of temperature record of this work to that of δ 18 O record of the ostracod Cytheridella boldi, and Heterocypris communis from Curtis et al., 1999....78 31. Comparison of the temperature record of this work to %Ti smoothed with a 3 point running mean of Haug et al., 2001....79 32. Comparison of record of this work to that of SST s (Aug-Oct) of the northeastern Caribbean reconstructed from δ 18 O using an Artificial Neural Network. Middle record offset by nearly 100 years from this thesis, bottom record after 100 year correction added (Nyberg et al., 2002)....81 33. Comparison of SST of Cariaco Basin to a faunal record of subtropical August SST s off of West Africa (demonecal et al., 2000)....83 34. Comparison of SST of the Cariaco basin to %CaCO 3, %S, δ 18 O from Physocypria sp. and δ 18 O Pyrgophorus sp. from Lake Chichancanab, Mexico (Hodell et al., 1995)....84 ix

CHAPTER I INTRODUCTION With the growing emphasis of global climate change, there is a greater need for high-resolution temperature records from all over the world. A wide variety of locations can give a much more accurate view of climate, because the record from any one location may be disconnected from the entire climate system over a period of time. Some of the ways in which scientists study climate consist of a wide variety of data and methodologies, including geophysical surveys, sediment and ice cores, tree rings and isotopes from corals and other organisms (Officer et al., 1957; Lidz et al., 1969; Overpeck et al., 1989, Briffa and Melvin, 2009). Our current reconstruction of past climates has relied heavily upon records from the Arctic and Temperate regions. However, these records may not entirely reflect what is occurring near the tropics and can potentially be disconnected from regional climate patterns existing in the Equatorial belt. A greater emphasis on the tropics is needed in order to have a more complete climate history. The tropics were originally thought to have little to no climate variability, but recent studies have shown that this region can exhibit considerable variability on both a yearly and decadal scale. In addition, the response of the tropics to climate change in comparison to the poles has also varied. Co-occurrence of high-latitude cooling and low- 1

latitude aridity is the typical pattern of long-term climate trends during the Pleistocene and occurred during various intervals throughout the Holocene. However, there were times when the tropics were wet when the poles were cool, suggesting a decoupling of the poles from the tropics (Mayewski et al., 2004). Thus, there is a need for a better resolved Holocene tropical climate record. In addition to the lack of tropical climate records, many earlier studies were not of a significant resolution to be able to accurately describe the climate of any one area, including the tropics. Thus, it was fairly common to have a sampling frequency in the hundreds of years. Whereas this method would allow for a basic knowledge of the climate, it may not account for rapid climate events. As Mayewski et al., (2004) notes, climate change that occurs during the Holocene typically was rather rapid as compared to earlier times. Sampling frequency has a large effect over the climate that is able to be described in a particular area. The whole Mayan civilization collapsed as a result of rapid climate change that occurred over an approximately 200-year period (Hodell et al., 2001, and Haug et al., 2003). One of the best places to obtain high-resolution tropical climate change records is in the Cariaco basin. The Cariaco basin is located along the southern edge of the tropical North Atlantic. It was originally created as a graben that was in-filled with waters from the Atlantic (Lidz et al., 1969; Morelock, 1972; Schubert, 1982). The basin is isolated from the Atlantic by high sills that surround it. Water above the sills is essentially the same as the Atlantic; however water below this level becomes anoxic (Richards, 1965; Lidz, 1969). The anoxic bottom water of the Cariaco minimizes bioturbation within the sediment. The sills act as a barrier to mixing of the water column, but during the winter 2

months, there is an increase in the wind strength that creates upwelling (Richards and Vaccaro, 1956). As a result of the upwelling, the primary productivity increases dramatically. The sedimentation rate within the basin is considerably high (1mm/year) due in part to the upwelling intensity (Black et al., 1999; Peterson et al., 2000). Previous studies involving the Cariaco basin include geochemical, sediment (sediment trap as well as coring), geophysical, and paleoceanographic studies (Officer et al., 1957; Lidz et al., 1969; Overpeck et al., 1989, Hughen et al., 1996). Other studies have examined chemical and biological anaerobic processes (Peterson et al., 1991; Thunnel et al., 2000). Each of these types of studies helps to create a balanced view of the climate of the Cariaco basin that can then be applied to the Tropical North Atlantic. When the Mg/Ca proxy data of this study is compared to the Hadley SST data of Rayner et al., (2003), an equation for calibration of the Mg/Ca paleothermometer can be created. This equation can then be utilized throughout the length of the core to acquire a more accurate temperature record of the Cariaco and the tropics. This record can also be compared to other nearby records in order to create salinity and precipitation history of the tropics. High resolution records will allow us to understand what the true past climate history of a region and can then be compared to other areas of the world to achieve a more accurate climate record. 3

CHAPTER II MAGNESIUM/CALCIUM RATIOS Sea surface temperatures from Mg/Ca ratios Magnesium to calcium ratios are an effective proxy for reconstructing past sea surface temperatures (SST). However, that was not always the case. Initial studies attempting to use this technique (Emiliani, 1955; Cronblad and Malmgren, 1981; Delaney et al., 1985) found no link between calcification temperature and the Mg/Ca ratios of the foraminiferal shells. However, more recently, through laboratory and plankton-tow studies the Mg/Ca ratios of planktic foraminiferal calcite have been found to be primarily a function of seawater temperature; and salinity does not contribute significantly to these ratios. (Mashiotta et al., 1999; Lea et al., 1999; Toyofuku et al., 2000; Rosenthal et al., 2002; Anand et al., 2003). Therefore it is possible to reconstruct the sea surface temperatures of past climates using this technique. Other proxies for determining paleotemperature Of the variables within the climate system, SST are the most significant because they control the atmospheric circulation and is a contributing factor to deep ocean circulation (Henderson, 2002). Thus, a proxy that can be used to reconstruct temperatures for a large portion of the oceans in the past is extremely useful. There are 4

several other proxies involving oceanic organisms used for the reconstruction of SSTs, including transfer functions, δ 18 O, alkenones, and Sr/Ca, however, they are considered to be less reliable than Mg/Ca ratios (Nürnberg et al., 2000). Transfer functions are used to correlate species assemblages to a variety of paleoclimate variables, including temperature. Once the population is quantified, species in a given sample are grouped into assemblages that correlate to specific temperature ranges. The census data are then inputted into the derived equation, and temperature is then calculated (Imbrie and Kipp, 1971). Transfer functions were used extensively in the Climate Long Range Investigation and Mapping (CLIMAP) project (1976), that reconstructed a warming of the tropics by as much as 1-2 C during the last glacial maximum (LGM) and cooling at high latitudes (CLIMAP, 1976). However, more recently developed paleotemperature proxies do not produce the same SST results for the tropics (Elderfield and Ganssen, 2000; Nürnberg, et al., 2000). Paired oxygen isotopes, Mg/Ca data and alkenone paleothermometry indicate a cooling of approximately 2-3 C in the tropics during the LGM (Nürnberg, et al., 2000; Schmidt et al., 2004), suggesting that there are considerable problems associated with the transfer function approach. One of the problems with the transfer function technique is that microplankton population variability can be driven by climate variables other than temperature. Biological controls such as nutrient variability can affect the population dynamics and may not always yield accurate results. Taphonomic variations may also influence the reliability of the transfer function s ability to reconstruct climate. For example, in the study by Chen et al., (1999) reconstructing rapid millennial-scale changes from a core in 5

the South China Sea, considerable variability in carbonate preservation of some species in the study was observed. They determined the relative changes in the preservation by calculating the percentage of whole foraminiferal shells versus the fragments. Principal component analysis (PCA) was used to determine a composite preservational index of all of the species. It was found that changes in preservation correlate to glacial-interglacial fluctuations, as well as high frequency oscillations (Chen et al., 1999). This study reveals both the potential and the problem with preservation when using foraminifera as climatological proxies. Whereas, preservation should be addressed for all studies using foraminifera, it is not always a valid proxy for determining past climate conditions. Oxygen isotope temperature studies also have been used extensively in climate studies (Lin et al., 1997; Mashiotta et al., 1999; Elderfield and Ganssen, 2000; Schmidt et al., 2004). One of the main limitations of δ 18 O as a proxy for temperature is that the δ 18 O value of a sample is a function of both the δ 18 O content of the water, salinity and/or global ice volume controlled, in which the calcite was precipitated as well as the temperature. These limitations may be minimized in various ways such as knowing the δ 18 O content of the water or by making comparisons to other proxies that can be used to determine the water chemistry (Elderfield and Ganssen, 2000; Nürnberg et al., 2000). In 1997, Lin et al., reconstructed surface water conditions for the last 28 kyr using oxygen isotopes from four different species of foraminifera from the Cariaco Basin, Venezuela. They found that all of the species showed a simultaneous cooling during the last glacial interval of 3-4 C. Lin et al., (1997) corrected for the δ 18 O content of the water by using Fairbanks (1989) mean δ 18 O isotopic composition of the whole ocean as the values expected from deglaciation meltwater, isotopically light water. Independent Mg/Ca 6

ratios can also be applied to the δ 18 O records from a given area to determine salinity at the time of calcification. Schmidt et al., (2004) combined Mg/Ca and oxygen isotopes in to reconstruct the salinity of the tropical Atlantic and examined its effect on North Atlantic thermohaline circulation during the last glacial cycle. It was found that the salinity of the Caribbean region fluctuated between more saline conditions during the relatively cold oxygen isotope stages 2, 4, and 6 and fresher conditions during relatively warm oxygen isotope stages 3 and 5 (Schmidt et al., 2004). Rühlmann et al., (1999) used alkenones in coccoliths to reconstruct SSTs. Alkenones are highly resistant organic compounds, such as ketones, produced by coccolithophorid algae (Bradley, 1999). The underlying principal upon which alkenone paleothermometry is established is that as water temperature decreases, the production of alkenones and ketones within the cell of coccolithophorid algae increases. The increase in ketone concentration allows the plant to adjust its buoyancy in response to temperature-induced changes in the surrounding water s density (Rühlmann et al., 1999). It has been shown that SST estimates from this method are in general agreement with other proxies (Rühlmann et al., 1999; Sachs et al., 2000; Herbert et al., 2001). Some difficulties with alkenone paleothermometry are that it should be paired with other proxies that can estimate sedimentation rate, because the signal may be altered due to mixing (Sachs and Lehman, 1999; Sachs et al., 2000). Alkenone paleothermometry may also be of limited use in areas that have large variations in salinity and nutrients (Sachs et al., 2000). Other trace metal based paleothermometry proxies have also been developed such as Sr/Ca ratios. The Sr/Ca ratios can be derived from several sources, including corals 7

and foraminifera. The Sr/Ca ratios in coral skeletons have been particularly useful for studying both seasonal and El Niño variability (Gagan et al., 2000; Ayliffe et al., 2004). However, Sr/Ca ratios have limitations as a paleothermometer as well. Potential problems include growth-rate effects, long-term changes in seawater Sr/Ca and geographic limitations to only tropical oceans (Henderson, 2002). Stoll and Schrag (1998) studied Sr and Ca of sea water using coupled numerical models to examine the effect of sea level changes on these elements. They found that cycles of glacial/interglacial sea level changes influence ocean Sr concentrations directly through weathering and large carbonate accumulation rates. It was found that the variation of the Sr concentration could produce up to 1.5 C error in paleotemperatures from coral Sr/Ca ratios since the LGM (Stoll and Schrag, 1998). Lea et al., (1999) noted that the Sr/Ca ratios in foraminifera are not nearly as robust as that of the Mg/Ca ratios because not as much Sr is added into the calcite lattice as Mg. Delaney et al., (1985) found that the Sr/Ca ratio of the foraminiferal shell is a function of the ratio of the water in which it grew in and therefore changing sea water Sr/Ca correlates to changing shell Sr/Ca. History of the Mg/Ca ratio technique The potential use of marine carbonate Mg content as a paleotemperature proxy was first noted by Clarke and Wheeler (1922). They recognized that since the Mg/Ca ratio of the oceans stays relatively constant due to long residence times, and that the content of the Mg proportion of carbonates varies with temperature, a growth temperature could be determined from marine organisms. Clarke and Wheeler (1922) were able to determine with some accuracy, although not with the precision that is possible today, the 8

Mg content of foraminifera and other Mg bearing biota. It was noted that as the seawater temperature increased, the Mg content of the foraminifera also increased. In the mid-1980 s, several studies investigated the links between Mg content of carbonates and growth temperature (Cronblad and Malmgren, 1981; Delaney et al., 1985). Both Delaney et al., (1985) and Cronblad and Malmgren (1981) note a slight temperature dependence of the Mg/Ca ratios of foraminifera, but that the correlation was very low and inconsistent. They concluded that there is some other environmental factor that controls the incorporation of Mg in the calcite of the shell. Cronblad and Malmgren (1981) noted that the dissolution effect, or the dissolution of carbonate as it falls through the water column, is what was controlling the variation of Mg within their samples, and that the differential susceptibility from the interglacial to the glacial would explain the trends that were seen in their data. In the late 1990 s and early 2000 s, a suite of studies was published that used live culture and core-top studies to demonstrate the sensitivity of carbonate Mg/Ca ratios to temperature (Nürnberg, 1995; Rosenthal et al., 1997; Mashiotta et al., 1999; Lea et al., 1999; Elderfield and Ganssen, 2000; Toyofuku et al., 2000; Rosenthal et al., 2002; Anand et al., 2003; Russell et al., 2004). From these studies, it was recognized that there were large exponential increases in the ratio of shell Mg/Ca of approximately 10 ± 1% per C. (Nürnberg, 1995; Nürnberg et al., 1996a, b; Rosenthal et al., 1997) Subsequent calibrations of the Mg/Ca ratios in foraminifera continued via culturing and core-top studies, further refining the equations derived from individual foram species (Mashiotta et al., 1999; Lea et al., 1999; Elderfield and Ganssen, 2000; Toyofuku et al., 2000; Rosenthal et al., 2002; Anand et al., 2003). More recently, sediment trap studies have 9

been utilized to calibrate the Mg/Ca temperature relationship by comparing instrumental SST s from the study site with Mg/Ca values of the foraminifera in the traps (McConnell and Thunell, 2005). A detailed discussion of the application of the calibrations will be provided below. Magnesium incorporation into calcite Magnesium incorporation into the crystal lattice of inorganic carbonates has been examined by several studies and found to be thermodynamic in nature (Lea et al., 1999; Erez, 2003; Russell et al., 2004). Koziol and Newton (1995) noted that the ΔH for the substitution of Mg into calcite is 21 kj/mol, which is equivalent to a ~3% increase of Mg/Ca per C between 0 C and 30 C. Several inorganic precipitation studies also note a 3% increase of Mg/Ca per C (Katz, 1973; Burton and Walter, 1987; Oomori et al., 1987). An inorganic precipitation study by Oomori et al., (1987) noted that when the Mg/Ca molar ratio of the parent solution reached a threshold of 0.2-1.0 (depending upon water temperature), aragonite would be precipitated. However, no matter how much Mg was added to the solution prior to reaching this threshold for a given temperature, the same amount of Mg would be precipitated with the calcite at that temperature. As the temperature increased, there was an increase in the Mg content of the precipitated calcite (Oomori et al., 1987). It was noted that the incorporation of Mg into calcite is strictly temperature dependent as the Mg concentration of the solution does not affect the eventual Mg concentration of the crystallized calcite. Oomori et al., (1987) found that the distribution coefficient of Mg 2+ ions increases with increasing temperature as would 10

be expected (Fig. 1). The equation for the distribution coefficient is: λ Mg = 0.00066 * T C + 0.0042, equivalent to a ~3% increase in the concentration of Mg 2+ / C increase. Live foraminifera aquarium based culture studies have also been performed to calibrate the Mg/Ca temperature relationship and to investigate the mechanism of Mg uptake into the foraminifer shell (Lea et al., 1999; Erez, 2003). Lea et al., (1999) determined there was an approximately 8 to 10% exponential increase of Mg/Ca / C over a narrow 9 C temperature range, thus showing that the incorporation of Mg into the calcite crystal lattice is biologically mediated beyond simple thermodynamics. Erez (2003) noted that Mg incorporation into the calcite test of the organism may act as an inhibitor to spontaneous precipitation of calcite. Thus the foraminifera may use Mg incorporation to have greater control on the location and creation of new shell material. Eggins et al., (2003) used laser ablation, inductively-coupled, plasma mass spectrometry to systematically measure the composition of successive foraminifera shell layers. They were able to resolve geochemically-distinct layers by ablating the shell wall at a rate of 0.15 µm per pulse with each sample. The trace element composition of the shell was collected simultaneously and showed that the obtained Mg/Ca ratios were consistent with other methods. They also found that Mg enrichment occurs at the outside of each successive test wall (Eggins et al., 2003) (Fig. 2). Calibration and equations Laboratory and plankton tow studies of Mg/Ca ratios in planktic foraminiferal calcite have found that the Mg/Ca ratio is primarily a function of seawater temperature; salinity does not contribute significantly to these ratios (Mashiotta et al., 1999; 11

0.10 λ = 0.000898 x ( C) + 0.0348 Katz (1973) 0.08 Distribution Coefficient 0.06 0.04 0.02 λ = 0.00066 x ( C) + 0.042 (Oomori et al., 1987) 0.00 0 10 20 30 40 50 60 70 80 Temperature ( o C) Figure 1. Temperature dependence of distribution coefficient (λ Mg ) on temperature (Oomori et al., 1987). 12

Figure 2. Compositional profiles showing the variation of Mg/Ca molar ratios through each analyzed chamber wall of an individual test of G. ruber (Eggins et al., 2003). 13

Elderfield. and Ganssen, 2000; Toyofuku et al., 2000; Rosenthal et al., 2002; Anand et al., 2003). The Mg/Ca ratio to temperature conversion equations are exponential, allowing for more accurate temperature estimates at higher temperatures. As the temperatures increases, so does the Mg content of the foraminiferal shell (Mashiotta et al., 1999; Lea et al., 1999; Elderfield. and Ganssen, 2000; Toyofuku et al., 2000; Rosenthal et al., 2002; Anand et al., 2003). The Mg/Ca ratios for benthic species were used to determine bottom-water temperatures during the Quaternary, but the calibration equations are less robust at lower temperatures, and the foraminifer tests are more susceptible to dissolution at depth (Martin et al., 2002). Despite this, their study showed that there was an approximate 1 C change in bottom water temperatures between the glacial and interglacial times in the deep Atlantic. Several studies have established equations for converting Mg/Ca ratios to sea water temperature. Lea et al., (1999) utilized live planktic foraminifera laboratory cultures with controlled ph and salinity and found that these variables exert only minor influences on the ratio of Mg/Ca. The Mg/Ca of cultured Globigerina bulloides showed strong temperature dependence with an Mg/Ca increase of 160% over a 9 C temperature range (Lea et al., 1999). The fit to the data (Fig. 3) for G. bulloides: Mg/Ca =0.528*e 0.102*T, where T= C temperature, with an r 2 = 0.93. The error, as stated by Lea et al., 1999, for this equation is ± 1.1 C, consistent with other proxy methods. Mashiotta et al., (1999) used a combination of live culturing and core top studies to establish a Mg/Ca temperature equation for G. bulloides. They found that the equation that fit to the data (Fig. 4) for G. bulloides is also exponential: Mg/Ca= 0.474*e 0.107*T, with an r 2 = 0.98. The standard error for this equation is ± 0.8 C. This 14

Figure 3. Mg/Ca uptake in the subpolar, foraminifera Globigerina bulloides and the subtropical, spinose, symbiont-bearing planktonic foraminifera Orbulina universa compared to growth temperature (Lea et al., 1999). 15

Figure 4. Mg temperature calibration results from culturing experiments with live Globigerina bulloides and core-top samples (Mashiotta et al., 1999). 16

differs from Lea et al., (1999) because it compared the cultured foraminifera to core tops from the southern ocean where temperatures are much colder than can be cultured in a laboratory. It should be noted that G. bulloides originally collected from surface waters between 18-22 C are not easily cultured below 15 C, and in order to extend the calibration curves and better define it, would require core-top samples, that creates a more accurate temperature equation. Elderfield and Ganssen (2000) created calibrations for eight different species using a combination of live culturing and core-tops. They found that a single equation can be used for all of eight of the species although it should be noted that the study deliberately excluded species that have high residual Mg content. Two different equations were formulated in this study. The first equation is an average of all species used, where Mg/Ca = 0.52*e 0.10*T. The second equation is for G. bulloides which is Mg/Ca = 0.56*e 0.10*T (Fig. 5). The standard error for both equations is ± 1.1 C, which is consistent with other proxy methods. Anand et al., (2003) created Mg/Ca temperature equations for nine planktic foraminifera species from a six-yr sediment trap located in the Sargasso Sea. The results clearly show that species-specific equations should be used when making temperature estimates. It was also found that Globigerinoides ruber (white) and G. ruber (pink) have similar Mg/Ca ratios at lower temperatures, but at higher temperatures G. ruber (white) has higher values (Anand et al., 2003). It should be noted that several common species, specifically G. bulloides, Globiberinella aequilateralis and Orbulina universa, have a significantly higher Mg/Ca value at a given temperature than most of the other species they studied. They found that the equation for G. aequilateralis is 17

Figure 5. Mg/Ca ratio for G. bulloides cultures where Mg/Ca = 0.56*e 0.10*T where T= temperature ( C). The standard error in the single-species core-top calibrations translates to about ±0.6 C (Elderfield and Ganssen, 2000). 18

approximately the same as the calibration for G. bulloides from prior studies: Mg/Ca = 0.532*e 0.090*T. This study showed that intra-annual variation in planktic foraminiferal Mg/Ca as well as δ 18 O could be observed for all species studied from the sediment traps during this study (Anand et al., 2003). A recent study (McConnell and Thunell, 2005) located in the Guaymas basin, Gulf of California utilized more than four years (August 1992 through October 1997) of instrumental SST measurements and bi-weekly sediment trap samples that were collected concurrently. The species G. bulloides and G. ruber were further investigated to calibrate the Mg/Ca temperature relationship, and to compare with oxygen isotope and alkenone temperature measurements from the same set of samples. McConnell and Thunell (2005) derived several equations for both species, taking into account different variables such as size fraction, and the temperature range of the sample core/sample location. However, only the equation for G. bulloides will be presented here because it is the only species studied for this thesis. The equation for the size range of 212-355 µm and temperature range between 16-31 C fit to the data (Fig. 6) is also exponential: Mg/Ca= 1.20*e 0.057*T, with an r 2 = 0.90. The average error for this equation is ± 0.38 C, which is more accurate than other proxy methods. It should be noted that the equations developed in this study greatly differ than those previously shown, (Lea et al., 1999; Elderfield and Ganssen, 2000) because the previous equations were developed over much shorter temperature ranges and shorter time periods than the McConnell and Thunell (2005) study. The time series sediment trap samples taken, are at known temperatures and that the Guaymas Basin undergoes large seasonal temperature ranges (~16-33 C) that are also 19

Mg/Ca (mmol/mol) Sea-surface Temperature ( C) Figure 6. Mg/Ca: temperature exponential relationship for 212-355μm size fraction of G. bulloides (McConnell and Thunell, 2005). 20

considerably higher than any previous aquarium based culture, sediment trap or core-top study (McConnell and Thunell, 2005). Caveats of technique As with most climate proxies, there are other factors that contribute to the degree of accuracy. When determining foraminiferal Mg/Ca ratios certain issues should be taken into account, especially the fact that species-specific calibration equations should be used to determine the SSTs because each species calcifies their shells differently. Even within the same species, foraminifera may occupy different depths during different parts of the year, and thus record different temperatures throughout its lifetime (Anand et al., 2003; Eggins et al., 2003; Lea, 2003). Gametogenic calcite may also be added, which is often incorporated when the foraminifer is living at the thermocline and associated colder temperatures, resulting in a lower than normal Mg/Ca ratio in that portion of the shell (Eggins et al., 2003). Toyofuku et al., (2000) compared both cultured and natural benthic foraminifera and found that both groups correlated well between each other, but that the group that was collected in the natural setting showed greater variability due to changes in seasonal temperatures. Lea (2003) noted that a preferential loss of the less robust individuals at depth would typically shift the mean value of the Mg/Ca ratio to lower values, and thus falsely indicate colder temperatures of calcification. Generally speaking, the more delicate, thinner-walled, spinose foraminifera calcify at shallower depths, thus creating the opportunity to more accurately assess the surface or near surface conditions (Lea et al., 2003; Anand et al., 2003). 21

Rosenthal and Boyle (1993) demonstrated that Mg/Ca ratios of species-specific foraminifera tests decreased with depth, suggesting that partial dissolution can influence the Mg/Ca content of the outer calcite layer. They also noted that there was a decrease in the shell Mg/Ca for both spinose and nonspinose species, but there was a greater effect for the nonspinose species. Dissolution can occur when the organism dies and passes through the mixed layer, where CO 2 is readily available to dissolve the shell (Rosenthal and Boyle, 1993; Brown and Elderfield, 1996). Through laboratory culture studies, Russell et al., (2004) found that shell Mg/Ca decreases in G. bulloides by 16 ± 5% per 0.1 unit increase in ph below ambient ph of 8.2. However, above ambient ph there is no significant change in the Mg/Ca of the shell of the foraminifera. Lea et al., (1999) found that shell Mg/Ca ratios are affected by the ph of the water, which is equivalent to - 0.6 C / 0.1 ph increase. Anand et al., (2003) noted that post-depositional dissolution may occur as a function of shell chemistry heterogeneity that would result in the preferential loss of the Mg rich portions of the shell. Several studies utilized a correction factor for the dissolution effect that is based on the location of the cores at the time of retrieval and the water depth (Lea et al., 1999; Mashiotta et al., 1999). As such, locations that have high sedimentation rates and are shallow generally do not require a correction factor (Anand et al., 2003) Brown and Elderfield (1996) found that with increasing depth there was a decrease in the Mg/Ca of the shell, and that dissolution preferentially removed the Mg rich outer gametogenic calcite. This study also found that the majority of the dissolution occurred above the lysocline for some species, and that other species tended to be more 22

resistant to dissolution. Elderfield and Ganssen (2000) used a core-top calibration paired with the δ 18 O records of the same shells to correct for the dissolution effect. Rosenthal and Lohman (2002) determined the size-normalized shell weight and Mg/Ca composition compared against the pressure corrected carbonate ion concentration to correct for dissolution. Therefore, it was determined that with an increase in depth, there is a similar increase in the dissolution of the shell (Rosenthal and Boyle, 1993; Brown and Elderfield, 1996). Salinity variations do have a small effect on shell Mg/Ca, equivalent to an increase of 0.6-0.8 C per unit salinity, depending upon the species (Nürnberg et al., 1996; Lea et al., 1999). Estimations of salinity can be made by comparing paired Mg/Ca ratios and stable oxygen isotope measurements on the same foraminifera. Groenveld et al., (2003) compared Mg/Ca and oxygen isotopes from foraminifera from the western Caribbean and found that maximum salinities occurred during times of high Mg/Ca values. It was also noted that salinity changes of up to 4 practical salinity units (psu) increased Mg/Ca ratios by 20-30% (Groenveld et al., 2003). Another possible source of error in Mg/Ca temperature estimates is related to the preparation technique. The preparation for the determination of the Mg/Ca content of the foraminifera is fairly simple, however the great number of steps and transfer of the foraminifera create many chances for error. Martin and Lea (2002) studied the effects on Mg/Ca ratios using different cleaning procedures on fossil benthic foraminifera. It was found that with more rigorous cleaning procedures, the Mg content of the foraminifera was lowered. The cleaning procedure for this work was developed by Boyle (1981) and later modified at the University of South Carolina, which produces excellent results in 23

interlaboratory comparisons (Rosenthal et al., 2004). A complete description of the preparation technique is described in appendix A. Use of the Mg/Ca technique The application of foraminifer shell Mg/Ca ratios for SST reconstructions have been used in a variety of locations around the world (Mashiotta et al., 1999; Rosenthal and Lohmann, 2002, McConnell and Thunell, 2005). Below is a general overview of some of the more notable studies using Mg/Ca paleothermometry. Mashiotta et al., (1999) reconstructed SSTs in the Sub Antarctic Indian and Pacific Oceans for the last 300 kyr using the Mg/Ca of planktic foraminiferal calcite. Correlations were made to planktic δ 18 O records and showed that changes in SST led changes in δ 18 O. Seawater δ 18 O records were calculated from the foraminiferal δ 18 O and Mg/Ca, and suggest that over the past 300-kyr seawater δ 18 O was 1 more positive during glacial episodes than interglacial episodes, thus indicating changing ice volume on the continents (Mashiotta et al., 1999). Interestingly, Mashiotta et al., (1999) found that the δ 18 O record was influenced by temperature change ~40-60%. Lea et al., (2003) reconstructed SSTs for the last 25,000 yr using Mg/Ca ratios and δ 18 O records derived from Cariaco Basin sediments. Comparisons were made to ice core gas records from the GISP2 air temperature proxy record and atmospheric methane records. It was found that during the Younger Dryas temperatures dropped in the Cariaco Basin by approximately 3-4 C and the temperature change was nearly simultaneous with the ice-core gas records within ± 30-90 yrs. 24

Lea et al., (2000) also compared Mg/Ca ratios and δ 18 O of foraminifera from two cores in the equatorial Pacific, one western and one eastern, and to an ice-core deuterium isotope record from Vostok, Antarctica. It was found that changes in temperature of both the Antarctic air and sea surface of the equatorial Pacific tend to lead changes in δ 18 O by approximately 3,000 yrs. Thus the SST change of the tropical Pacific Ocean would have played a major role in driving climate change through both the cooling and moisture addition to the atmosphere (Lea et al., 2000). This study shows that the tropics may play a major role in driving ice age climate. Schmidt et al., (2004) performed paired Mg/Ca and δ 18 O analyses on foraminifera collected from two Caribbean cores to reconstruct tropical surface ocean hydrography over the last 125,000 yr. The study found that the Caribbean alternated between relatively lower salinity during warm isotope stages (3 and 5) and higher salinity values during cold isotope stages (2, 4 and 6). Higher salinities in the Caribbean during glacial stages suggest a reduction in salt export to the North Atlantic and hence a likely shutdown or significant reduction in the North Atlantic s thermohaline circulation (Schmidt et al., 2004). Tripati et al., (2003) reconstructed SSTs from the extinct planktic foraminifer from the genus Morozovella sp. a species thought to be similar to Globigerina sacculifer (Rosenthal et al., 2000). Calculated SSTs were influenced by corrections for dissolution, paleo-sea water Mg/Ca values, as well as the Mg/Ca temperature equation. However, they determined that warming occurred during the early Eocene (54.8-49.0 million yrs ago) with maximum values between 51 and 48 Ma. However, their findings do not compare well with the δ 18 O records from this site (Tripati et al., 2003). A major 25

complicating factor in this study is that an extinct species whose Mg/Ca temperature relationship could not be directly calibrated was used, thus creating a situation that is basically a variation on the transfer function modern analog problem. Stott et al., (2004) used paired δ 18 O and Mg/Ca from foraminiferal calcite to reconstruct salinity and thus eventually the oxygen isotope composition of western tropical Pacific waters over the past 10,000 yr. They found that the SST s of the western tropical Pacific have decreased ~ 0.5 C and that sea surface salinities have also declined ~1.5 psu over the Holocene (Stott et al., 2004). The reason for the apparent decrease in temperature and salinity is attributed to either a general decrease for the Pacific Ocean as a whole, or that the modern salinity gradient in the tropics recently developed (Stott et al., 2004). Benefits of the Mg/Ca technique It is clear that foraminiferal Mg/Ca ratios can be used to accurately reconstruct both surface and deep ocean temperatures in the past. Additionally, the combination of paired Mg/Ca and δ 18 O analyses provides a powerful proxy for reconstructing not just temperature histories, but salinity time series as well. The salinity data can be used for estimating variables such as regional patterns of evaporation and precipitation, and provides insight on the state of the global thermohaline circulation. 26

CHAPTER III STUDY AREA Physical Characteristics The Cariaco basin is located along the northern coast of Venezuela (Fig. 7 and 8) in the southern Caribbean Sea. The basin is located between two fault zones and is a small east/west trending pull-apart basin that is structurally similar to the fault bounded basins of the California borderlands (Lidz et al., 1969; Morelock, 1972; Schubert, 1982). The Cariaco basin is the second largest anoxic marine basin in the world after the Black Sea, and is 160 km long and 40 km wide. The basin consists of two sub-basins, each approximately 1400 m deep, separated by a central saddle that rises to approximately 950 m depth (Richards, 1975). There are also two smaller depressions, the Araya and the Margarita basins (550 and 410 m water depth, respectively) located north of the eastern sub-basin (Maloney, 1966). The surface water flow of the Caribbean Sea is dominated by the east to west flow of the Caribbean Current that carries water from the equatorial Atlantic into the Gulf of Mexico. The prevailing northeasterly trade winds and the North Atlantic Equatorial Current drive the Caribbean Current. Water from the open southern Caribbean Sea then flows over the surrounding sills into the Cariaco basin (Richards, 1975; Maloney; 1966). 27

NAEC N CC Cariaco Basin ITCZ (summer) NECC ITCZ (winter) NBC NBUC SEC BC - Brazil Current, CC - Caribbean Current, NBC - North Brazil Current, NBUC - North Brazil Undercurrent, NAEC - North Atlantic Equatorial Current, NECC - North Equatorial Countercurrent, SEC - South Equatorial Current BC surface currents trade winds Intertropical Convergence Zone (ITCZ) Figure 7. Location of the Cariaco Basin with the position of the ITCZ during the boreal winter and summer months. Regional surface currents are denoted by arrows. (after Dürkoop et al., 1997). 28

W 70 o CariacoBasin 65 o 60 o Rio Tuy Rio Guera Rio Manzanares N 11 N TORTUGA BANK TORTUGA BANK Rio Tamanaco Rio Unare W S E Cariaco Basin CABO CABO CODERA CODERA UNARE UNARE PLATFORM N 10 N 66 65 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 Figure 8. Location of the Cariaco Basin with local-rivers entering the basin. Second figure (lower left) shows depth isobaths (meters) by color, star represents core location (after Black et al., 1998). 29

Along the surrounding shallow sills there are two depressions; one located half-way between the islands of Tortuga and Margarita along the northeastern corner of the saddle (120 m), and the other located at the western end of the basin at Farallon Centinela (147 m) (Fig. 8). The two depressions limit horizontal water exchange from the Caribbean Sea to the deeper part of the Cariaco basin (Lidz, 1969). Above the sills, the water has similar geochemical characteristics to the open Caribbean Sea (Richards, 1965). Temperature and salinity profiles from the western (10 40.84 N, 65 19.94 W) and eastern sub-basin (10 33.26 N, 64 47.05 W) as measured during June, 1990 demonstrate the stratification of the basin water (Fig. 9) (Peterson et al., 1990). Temperature and salinity profiles from the eastern sub-basin (10.5 N, 64.66 W) that were collected during monthly cruises from November 1995 to August 1998 also demonstrate these trends (Fig. 10) (Astor et al., 2003). Temperature decreases rapidly from the surface to a nearly constant 17 C below sill depth (500 m). During the nonupwelling season, the temperature profiles indicate a stable water column and a shallow mixed layer (Astor et al., 2003). Maximum salinities of approximately 36.8 can occur at depths of ~ 60 m in both sub-basins and decreases to an almost constant 36.2 below 500 m (Peterson et al., 1990; Astor et al., 2003). Sill depth and seasonal fluctuations in coastal upwelling cause rapid shifts of temperature and salinity at similar depths of approximately 150 m (Peterson et al., 1990). The strong pycnocline that results from pronounced changes in temperature and salinity in the upper water column restricts vertical mixing within the Cariaco basin (Scranton et al., 1987; Astor et al., 2003). The deep Cariaco water below 300 m, is anoxic and is a result of the combination of limited 30

Figure 9. Temperature and salinity profiles of the western and eastern sub-basins, respectively. Temperature and salinity above the sills have similar characteristics with the open Caribbean, but below the sills there is an abrupt change in both temperature and salinity (after Peterson et al., 1990). 31

ity Temperature Salinity Dissolved Oxygen Figure 10. Temperature (ºC), salinity ( ), and dissolved oxygen (μm) between November 1995 and August 1998 from the eastern basin (Astor et al., 2003). 32

horizontal and vertical circulation and intense seasonal upwelling (Richards and Vaccaro, 1956; Richards, 1975). Prior studies of the basin suggest that the depth of the oxic/anoxic layer fluctuates with time (Scranton et al., 1987; Astor et al., 2003). Lin et al., (1992) theorized that the depth of the oxic/anoxic boundary shifts vertically as a result of seasonal changes in ventilation and/or productivity. Astor et al., (2003) showed that the boundary fluctuated nearly 100 m during 1997. Recent data indicate that the boundary depth is between 225 m and 250 m (Black, pers com., 2003). It was suggested by Holmen and Rooth (1986) that seasonally produced warm hypersaline waters over the shallow Unare platform and/or warm water spilling over the sill from uppermost Caribbean Sea may create small scale ventilation events. Climate of the Cariaco Basin Within the Cariaco basin, the annual migration of the Intertropical Convergence Zone (ITCZ) creates seasonal cycles of upwelling and precipitation (Fig. 7) (Redfield, 1955; Richards, 1960, 1975; Wust, 1964; Hastenrath, 1978; Aparicio, 1986). The ITCZ is a belt of low pressure located along the tropics that is a result of solar heating of Earth s surface. As warm air rises, it creates lower pressure at the surface. Adiabatic cooling of these air masses results in cloud formation and increased precipitation. During the boreal winter months of January through March, the ITCZ is at its southernmost position of the year (0º to 5º S) and causes strong easterly trade winds to blow along the coast of Venezuela (Fig. 7) (Hughen et al., 1996). The increase of wind speed and persistence over the Cariaco basin creates Ekman pumping and a net offshore transport of waters that is then replenished by the colder nutrient-rich upwelled waters, causing the 33

primary productivity of the basin to increase dramatically. The winter months are also the local dry season, as the ITCZ and its associated rainfall is deflected further south during this time. Beginning around July when the ITCZ is at its most northerly position (about 5º N), the local rainy season begins, which increases the discharge of rivers that drain into the southern Caribbean Sea (Fig. 7) (Aparicio, 1986). Whereas the ITCZ influences the climate of the Cariaco basin throughout the year, the climate of the entire South American Continent is controlled by the ITCZ. This makes the Cariaco basin important in the understanding of the ITCZ because it sits in a sensitive location along the continent as well as the greater Carribbean Sea. Previous studies in the Cariaco Basin Previous studies involving the Cariaco basin include a wide variety of disciplines. Investigations encompass geophysical, sediment (sediment trap as well as coring), geochemical, and paleoceanographic studies (Officer et al., 1957; Lidz et al., 1969; Overpeck et al., 1989). Other studies have examined biological and chemical anaerobic processes (Peterson et al., 1991; Thunnel et al., 2000). Geophysical works Officer et al., (1957) completed one of the first seismic profiles of the structure of the Caribbean crust that included the Cariaco basin. Lidz et al. (1968) completed depth recordings of the Cariaco and concluded that the El Pilar fault zone extended along the axis of the basin. In 1971, Ball et al., completed a more detailed geophysical study and found that the basin was created by the north-south extension of the Venezuelan margin. Schubert (1982) further described the basin as part of the Caribbean-South American 34

plate boundary, forming from the east-west tension in an overlap area between two en echelon right-lateral strike-slip fault zones, the El Pilar and Moron. Peterson et al. (1990) completed a high-resolution bathymetric study as part of the 1990 cruise (Fig. 8). The survey was completed using 5 km grid spacing, providing much greater resolution than any previously published bathymetric maps of the basin. Sediment studies Heezen et al. (1958, 1959) conducted the first studies describing the character of the sediments from the basin. They concluded that the basin became anoxic 10,900 YBP based on radiocarbon dates from organic matter. Needham (1971) used the cores from the 1957 Heezen expedition and further described and characterized the sediments. In the early 1960 s, Woods Hole Oceanographic Institution collected a suite of 24 cores from the basin. Athearn (1965) described these cores and completed clay and grain-size analyses. Within the Cariaco basin, two major sediment zones have been described based on the presence or absence of laminae and color variations (Heezen et al., 1958, 1959; Athearn, 1965; Lidz et al., 1969; Needham, 1971). The upper 9 m of sediment within the basin typically consists of laminated grayish-green silty clays (Athearn, 1968; Lidz et al., 1969). Anoxic conditions are suggested by the rarity or absence of benthic fauna as well as the presence of sediment laminations (Peterson et al., 1991). The uppermost 2 m of sediments are further distinguished by their faint laminations. The laminae consist of alternating light-dark couplets on millimeter to sub-millimeter scales (Peterson et al., 1991). Hughen et al., (1996) recognized that the light laminae that are rich in planktic 35

foraminifera, diatoms, and coccoliths were deposited during the local dry upwelling seasons, and the dark laminae, consisting mainly of terrigenous silts and clays were deposited during the local rainy non-upwelling seasons. Laminae counts and 210 Pb data from the upper sediment region confirm that these are true annual varves (Peterson et al., 1991; Hughen et al., 1996). The lower sedimentary unit consists of gray to yellowish-brown silty clays that are bioturbated and contain benthic fauna, suggesting deposition under oxic conditions (Athearn, 1968; Lidz et al., 1969; Peterson et al., 1991). The boundary between the two sedimentary units is usually distinct and has been dated at 12,600 14 C years B.P. by Accelerator Mass Spectrometry (AMS) 14 C (Overpeck et al., 1989; Peterson et al., 1991; Lin, 1992; Lin et al., 1997). Sediment-trap studies are one of the more recent ways to examine the sediments of the basin. As part of a study beginning in 1995, the Carbon Retention in a Colored Ocean (CARIACO) project placed sediment traps in the eastern sub-basin. Four traps were placed at different depths, both above and below the oxic/anoxic interface (Thunell et al., 2000). Scranton et al. (1997) found that a large portion of the primary production is recycled by bacteria within three main horizons: 1) the euphotic zone, 2) the region just above the oxic/anoxic interface and 3) a layer 100 m below the interface. As would be expected from the influence of seasonal upwelling, the locations of these layers tend to migrate vertically (Scranton et al., 1997). Thunell et al. (2000) measured organic-carbon fluxes during different levels of primary productivity over a six-month period. They found that of the total organic carbon produced at the surface, only1 2% reaches the seafloor, and within the sediment, only 50% of this is preserved. 36

Geochemical One of the first studies of the sediment geochemistry in the basin was by Lidz et al. (1969). As part of the study, nitrogen, organic carbon, CaCO 3 and the coarse sediment fraction (>63 μm) were compared from various cores. They found that the coarse-fraction content and surface sediment CaCO 3 increased away from the shore, but nitrogen content decreased. Downcore CaCO 3 records were shown to range from 10-30% (Lidz et al., 1969). There have been many organic geochemistry studies performed on Cariaco basin sediments, including kerogen, organic δ 13 C, and total carbon studies. Eckelmann et al., (1962) made measurements of the organic δ 13 C of sediments and found that the total organic matter of the δ 13 C ratios of the sediments below the oxic/anoxic interface ranged from -22.1 to -23.1, whereas values above the interface were heavier, ranging from - 20.1 to -20.6. Thunell et al. (1997) notes that under the current conditions of anoxia, the carbon that is being deposited is primarily of marine origin. It was suggested by Needham (1971) that organic components of the oxic layer contain more carbon of terrestrial origin than the sediments deposited under anoxic conditions. Using alkenone biomarkers, Werne et al. (2000) observed a shift in the population of primary producers from a diatom-dominated assemblage to a coccolithophoriddominated assemblage over timescales of one thousand years or more. Geochemical analyses of the planktic communities found within the sediment were used as evidence that sedimentation is mainly of marine origin within the basin (Eckelmann et al., 1962; Peterson et al., 1995). Increased diatom abundance during the Younger Dryas indicates a 37

period of more intense upwelling, whereas increased coccolithophorids during the Holocene indicate reduced upwelling (Werne et al., 2000). Werne et al., (2000) inferred that the productivity and ecosystem dynamics from the Cariaco basin can be correlated to areas throughout the tropical oceans, creating important links to the global climate system and the carbon cycle. Goñi et al. (2003) used sediment traps at varying depths as well as core data to determine that organic carbon and opal burial over the past 5,500 years was similar to that found in the sediment traps at the deepest depths. Burial fluxes of CaCO 3 were four times higher than those of the deepest CARIACO project traps but were similar to that of the shallowest trap (Goñi et al., 2003). Burial rates of long-chain C 37 alkenones (Σ Alk 37 ) were significantly decreased at burial as compared to all sediment traps, indicating a large decline in the flux of these compounds upon deposition (Goñi et al., 2003). Paleoclimate and hydrographic The Cariaco basin is an exceptional recorder of paleoclimate history because its sediments can be used on a variety of time scales and with a variety of proxies, to reconstruct past climate history of northern Venezuela and the northern tropical Atlantic Ocean. What follows is an overview of significant paleoclimate research that has been completed from the Cariaco basin. Overpeck et al. (1989) used population variation of planktic foraminifera to reconstruct the late Quaternary climate history within the Cariaco. Abrupt changes in the foraminifera populations were observed at 13,000, 12,600, 11,000, and 10,000 years B.P. that correlated to tradewind intensity and lake level records of aridity from the southeast 38

US coastal plain region (Overpeck et al., 1989). As part of the study, a global circulation model was used and it was determined that an increase in tradewind intensity could affect the foraminifera population. It was hypothesized that the increase in the tradewind intensity could be a result of a melt-water induced glacial cooling in the Gulf of Mexico (Overpeck et al., 1989). Peterson et al. (1991) completed a foraminiferal census study on the same cores as Overpeck et al. (1989) using a similar temporal resolution of 17-33 years. They found a similar up-core transition that went from a G. ruber-dominated assemblage, to a G. bulloides-dominant assemblage that was dated at 12,600 years B.P. This transition is coincident with the up-core change in lithology and correlates to the first large sea-level rise associated with the melting of the Northern Hemisphere ice sheets. Peterson et al. (1991) also noted the presence of laminated sediments, high abundances of G. bulloides, and biogenic opal that they inferred indicated the initiation of strong upwelling as the basin progressed into more open marine conditions. Lin (1992) and Lin et al. (1997) completed foraminifera census and oxygen isotope data from two cores taken during the PLUME-7 cruise in 1990. A highresolution paleoclimate reconstruction was generated for the Cariaco basin for the last 28,000 years. The δ 18 O record obtained from G. ruber during the last glacial period showed decreased inter-specific differences, and indicated surface water cooling and/or a reduced vertical thermal gradient in the water column as a result of the shallower than present glacial sills (Lin et al., 1997 ). Hughen et al. (1996) utilized grey scale reflectance and laminae thickness as a proxy to determine paleoproductivity in the Cariaco basin. They found that darker 39

laminae are consistent with increased rainfall and lighter laminae are consistent with arid/upwelling conditions. Interestingly, the grayscale variability pattern observed in Cariaco basin cores is similar to δ 18 O records from Greenland ice cores. Hughen et al., (1996) surmised that there is a common forcing mechanism during both tropical and high-latitude climate variability, although the exact mechanism is unknown. Black et al. (1999) compared the G. bulloides abundance data to Comprehensive Ocean-Atmosphere Data Set (COADS) zonal wind speed data in order to calibrate proxy to instrumental data and reconstruct the upwelling history and tropical Atlantic trade wind variability through the past 825 years. The data indicate that with increasing tradewinds, there is a decrease in the SST within the basin, and an associated increase in the population of G. bulloides due to the increase in upwelling intensity (Black et al., 1999). Comparing the records between Cariaco and the North Atlantic, it is apparent that when the tradewinds are more intense over the Cariaco basin, SSTs are cooler in the North Atlantic (Black et al., 1999). A possible explanation could be found in an earlier study by Hastenrath and Greischar (1993) who noted a southward shift of the ITCZ could occur when the cross-equatorial SST gradient in the North Atlantic is cooler relative to the South Atlantic. The inter-hemispheric SST gradient would thus cause an increase in surface pressure over the North Atlantic and shift the ITCZ southward (Hastenrath and Greischar, 1993). Yarincik et al. (2000) used Al/Ti and K/Al ratios of bulk sediment from Ocean Drilling Program (ODP) core 1002 to describe wind-blown and hemipelagic sources of deposition over the last 578 kyr. Comparisons were made to planktic foraminiferal δ 18 O and found that the ratios vary in phase with major climate changes. During interglacial 40

periods, Al/Ti and K/Al values are significantly higher than during glacial periods (Yarincik et al., 2000). High K/Al values during the glacial periods are consistent with lower sea level resulting in a higher relative proportion of the Cariaco basin s water being derived from local rivers that drain into the basin. During glacial times, the Al/Ti ratio decreases, suggesting that greater portions of eolian dust/rutile from the northern Sahara are being blown into the basin (Yarincik et al., 2000). In 2000, Peterson et al. performed major elemental (Fe, Ti, Ca and CaCO 3 accumulation) analyses and color reflectance from ODP core 1002 for the last 90,000 years B.P. They found that productivity in the basin and increased precipitation/river discharge are linked to warm interstadial climate events. When the data are compared to the Greenland Ice Sheet Project (GISP) II δ 18 O signal, the records are almost identical, suggesting a strong teleconnection between the tropics and the high latitudes (Peterson et al., 2000). The position of the Atlantic ITCZ has varied through time, affecting regional precipitation changes over South America (Peterson et al., 1991; Hughen et al., 1996). Haug et al. (2001, 2003) reconstructed the precipitation history of the Southern Caribbean using bulk sedimentary iron (Fe) and titanium (Ti) from Cariaco basin cores. Fe and Ti are continental in origin and can thus be used as an indicator of fluvial input. Relatively high percentages of Ti or Fe indicate increased precipitation such as during the Medieval Warm period. Lower relative values of both Ti and Fe, and hence decreased precipitation prior to the Medieval Warm period at ~ 800 A.D. is coincident with the collapse of the Mayan civilization (Fig. 11) (Haug et al., 2003). They further argue that during this time the position of the ITCZ was located further south than it is today, 41

0.28 0.24 0.20 %Titanium 0.16 0.12 0.08 Increased precipitation 0.04 0.00 1200 1400 1600 1800 2000 Years A.D. Figure 11. Bulk Ti content as a three point running mean of 2-mm resolution measurements at ODP hole 1002C and 1002D in the Cariaco Basin during the past 2,000 years (Haug et al., 2003). 42

enhancing precipitation over the Amazon and central South America, but potentially causing drought in Central America. Hodell et al. (1995) studied Lake Chichancanab within the Yucatan peninsula and found that precipitation and evaporation conditions were coincident with Haug et al. (2001, 2003). Some of the possible causes of the lack of precipitation are a weakening of the El Niño/ Southern Oscillation (ENSO), weaker South American Monsoon, decreased ITCZ associated rainfall, or increased solar output (Hodell et al., 2001; Haug et al., 2003; Lachniet et al., 2004). Planktic foraminiferal oxygen isotope data were utilized by Tedesco and Thunell (2003) to reconstruct tropical Atlantic climate for the mid to late Holocene. Over the past 6,000 years there have been six major increases in δ 18 O as a result of both salinity increases and SST decreases. The increases in δ 18 O coincide with increasing aridity within the Caribbean region (Tedesco and Thunell, 2003). Concurrent with these drier conditions is the onset of wetter conditions in the South American highlands. The contrasting climate conditions for these regions is associated with an average southward migration of the ITCZ that resulted in increased precipitation towards the south, and a decrease in precipitation towards the north over the basin. Black et al. (2004) reconstructed SST ITCZ precipitation related salinity variations over the Caribbean and tropical north Atlantic over the last 2000 years using δ 18 O from two planktic foraminifera species. Stable-oxygen isotope data from seasonally-representative planktic foraminifera were calibrated to instrumentallymeasured Atlantic SSTs for the interval spanning A.D. 1871-1990 (Black et al., 2004). Between 1949 and 1990, δ 18 O was highly correlated to tropical Atlantic SSTs, but correlations over the earlier interval of instrumental overlap were considerably weaker. 43

The low δ 18 O-SST correlations observed when the older instrumental SST data were included suggests that the quality of the instrumental data decreased or variations in seasurface salinity (SSS) overprint the temperature effect on the δ 18 O (Black et al., 2004). Importance of the Cariaco Basin The Cariaco Basin is an important recorder of the past climate of the tropics due to its location on the southern margin of the Caribbean and high sedimentation rate. As part of this project, I used Mg/Ca ratios derived from Cariaco Basin sediments to create a SST history for the tropical Atlantic. The reconstructed record has near-annual resolution and spans the last 800 years. Comparisons may be made to other records of similar resolution such as the δ 18 O record of Black et al. (2004) to reconstruct the salinity history for the tropical Atlantic. The records can then be compared to higher latitude records in order to describe possible correlations. 44

CHAPTER IV MATERIALS AND METHODS Materials The samples used for this study come from a 56.5 cm box core retrieved on a cruise aboard the R/V Thomas Washington in 1990 (Peterson et al., 1990). A series of 104 sediment cores (box, giant gravity, and piston) were collected during that time. Box core PL07-73BC (Fig. 12) was taken from a depth of 450 m at 10º 45.98 N and 64º 46.20 W using a Soutar type box core (Fig. 13). Methods Consecutive 1 mm thick samples were taken by scraping a clean glass slide along the exposed end of the working half of the core. A one millimeter marked scale was utilized in order to guide the sampling procedure. Each sample was then freeze dried to allow for easier processing of the samples (after Black et al., 1999). The small sample size was not a limitation in this analysis because the Cariaco basin has such high sedimentation rates that allow for a resolution of 1 to 1.5 years for any given sample. Two-thirds of each sample was placed into distilled water to disaggregate the sample and sieved through a 63 μm mesh also using distilled water. A one liter aliquot was collected from the < 63 μm fraction, dried, and saved for nannofossil and clay 45

TORTUGA BANK 11 N TORTUGA PL07-73BC MARGARITA FARALLON CENTINELA CUBAGUA PL07-71BC ARAYA CABO CODERA CUMANA LAGUNA DE TACARIGUA UNARE PLATFORM PIRITU BARCELONA 10 N 66 RIO UNARE 65 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 Figure 12. Location of core PL07-73 BC in the Cariaco Basin and core PL07-71 BC, which was used for age model (Black et al., 1998). 46

500 520 540 Length (mm) Figure 13. A Soutar type box-core with photograph of a core (Peterson, 1991). 47