KRISTINA L. FAUL 1,A.CHRISTINA RAVELO 2, AND M. L. DELANEY 2

Transcription

1 Journal of Foraminiferal Research, v. 30, no. 2, p , April 2000 RECONSTRUCTIONS OF UPWELLING, PRODUCTIVITY, AND PHOTIC ZONE DEPTH IN THE EASTERN EQUATORIAL PACIFIC OCEAN USING PLANKTONIC FORAMINIFERAL STABLE ISOTOPES AND ABUNDANCES KRISTINA L. FAUL 1,A.CHRISTINA RAVELO 2, AND M. L. DELANEY 2 ABSTRACT In the hydrographically complex eastern equatorial Pacific Ocean (EEP), the distinction between changes in productivity and changes in upwelling is important to the study of the causes and implications of changes in paleoproductivity during the Last Glacial Maximum (LGM). We studied seven EEP coretops representing a gradient of increasing primary productivity from west to east. Comparison of the coretop data indicates calcification depth and temperature for each planktonic foraminiferal species may change depending on the vertical position of hydrographic features such as the degree of stratification of the water column, as well as associated biological parameters such as the depths of the photic zone and the chlorophyll maximum. Because these biological parameters are related to primary productivity, calcification depth and temperature patterns for each species are somewhat different for high and low productivity regions in the EEP. We use the relationship between modern surface hydrography and coretop planktonic foraminiferal abundances and isotopic composition to interpret upwelling and productivity changes in the EEP over the last 20,000 years. While data indicate higher primary productivity and lower SSTs, they do not indicate that there was greater upwelling at the location of our site during the LGM relative to present. INTRODUCTION The balance between upwelling of carbon dioxide-rich waters and the draw down of carbon dioxide by biological productivity in surface waters of the eastern equatorial Pacific (EEP) affects the ocean-atmosphere exchange of carbon dioxide. At present, this region is a source for carbon dioxide because the oceanic partial pressure of carbon dioxide is greater than the atmospheric partial pressure of carbon dioxide (Keeling, 1968). Greater primary production in equatorial regions has been suggested as a cause of lower atmospheric carbon dioxide levels during the last glacial maximum (LGM) relative to present (Broecker, 1982; Berger and Keir, 1984; Sarnthein and others, 1988; Mix, 1989a, 1989b). Higher primary productivity (Pedersen, 1983; Sarnthein and others, 1988; Lyle and others, 1988; Mix, 1989a, 1989b; Pedersen and others, 1991) has been attributed to greater LGM upwelling rates (Sarnthein and others, 1988; Lyle and others, 1992). Despite higher primary pro- 1 Earth Sciences Department, University of California, Santa Cruz, CA Ocean Sciences Department and Institute of Marine Sciences, University of California, Santa Cruz, CA K. L. Faul is correspondence author. ductivity, others have found evidence for lower glacial nutrient utilization in the EEP (Farrell and others, 1995) and greater glacial pco 2 levels in the surface ocean (Jasper and others, 1994) relative to present, thereby implying that the rate of nutrient delivery to the surface water exceeded that of nutrient utilization. Both greater nutrient availability and higher pco 2 are thought to be due to greater upwelling during the LGM (Jasper and others, 1994; Farrell and others, 1995). The study of the causes and implications of changes in paleoproductivity would greatly benefit from the development of proxies and techniques that could distinguish between changes in productivity versus changes in upwelling. Most indicators focus on changes in surface biological production. Upwelling could be monitored by reconstructing the upper water column temperature structure. The foundations for understanding how planktonic foraminifera from deep sea sediments could be used for reconstructing thermocline depth and seasonality come from past plankton tow and sediment trap studies. For example, in the northern Panama Basin (5 20 N, 85 W), studies of relative abundances and stable isotopes of planktonic foraminifera from plankton tows (Fairbanks and others, 1982) and sediment traps (Curry and others, 1983; Thunell and others, 1983) showed that species are stratified within the photic zone because of ecological needs and temperature preferences. In the central equatorial Pacific, living planktonic foraminiferal species assemblages respond to primary productivity, temperature, and mixed layer depth (Watkins and Mix, 1998). Plankton tow and sediment trap work from outside the tropics in the Pacific Ocean provide an excellent understanding of species ecological preferences and seasonal fluxes in the San Pedro Basin (Sautter and Thunell, 1991; Thunell and Sautter, 1992) and off the coast of Oregon (Ortiz and Mix, 1992; Ortiz and others, 1995, 1996, 1997). Applying this understanding to the interpretation of deepsea coretop sediment samples has been done for some regions of the tropical Atlantic (Ravelo and others, 1990; Ravelo and Fairbanks, 1992) and tropical Pacific (Andreasen and Ravelo, 1997), but never for the hydrographically complex and important region of the EEP. It is clear from past studies of plankton tows and sediment traps that planktonic foraminiferal depth stratification is linked to chlorophyll concentrations, and that transfer functions relating foraminiferal assemblages in the sediments to global maps of modern productivity can be used to estimate past productivity (Mix, 1989a). However, no past studies have directly compared water column chlorophyll data to the foraminiferal record in coretops. This study compares surface water column properties, productivity, and upwelling with coretop foraminiferal isotopic values and abundances. Oxygen isotopic values of depth-stratified planktonic foraminiferal shells can be used 110

2 EAST PACIFIC UPWELLING AND PRODUCTIVITY RECONSTRUCTIONS 111 FIGURE 1. Locations of sites used in this study. Surface currents in the eastern equatorial Pacific Ocean are indicated by black arrows: the Equatorial Countercurrent (ECC), the South Equatorial Current (SEC), and the Peru Current (PC). The Equatorial Undercurrent (EU), indicated by the open arrow, is a subsurface current. Sites are grouped into regions based on hydrography and productivity differences: solid circles are low productivity Region 1 sites (RC and RC13-138), open circles are intermediate productivity Region 2 sites (RC23-12 and RC23-20), and solid triangles are high productivity Region 3 sites (RC11-238, RC9-69, and VM17-44). Low productivity Region 1 sites are primarily influenced by the SEC, whereas the high productivity Region 3 sites are primarily influenced by the PC. The relative influence of the PC on Region 2 varies seasonally. to reconstruct the thermal structure of the surface waters, and therefore upwelling. Relative abundances of species known to proliferate during the season of high biological production are used to understand changes in productivity. To demonstrate how to apply these modern relationships to past times, and how to begin to distinguish changes in productivity from changes in upwelling, we have generated foraminiferal isotope and abundance records from a site in the EEP for the last 20,000 years. BACKGROUND AND STRATEGY FORAMINIFERAL STUDIES In the tropical Pacific, sub-surface hydrographic features such as the annual average mixed layer and thermocline depth are highly correlated to the distribution of planktonic foraminiferal abundances in coretop sediments (Andreasen and Ravelo, 1997). Although these correlations were made with annual average thermocline depth, the seasonality of upwelling and thermocline depth also influence planktonic foraminiferal assemblages. For example, in the Panama Basin, species fall into two seasonal flux groups: a maximum flux of Neogloboquadrina dutertrei, Neogloboquadrina pachyderma (dextral), and other deep dwellers during winter upwelling months, and a maximum flux of surface spinose species such as Globigerinoides ruber and Globigerinoides sacculifer in the summer months with strong thermal stratification (Thunell and others, 1983; Thunell and Reynolds, 1984). In the San Pedro Basin, maximum production of N. pachyderma (dextral) occurs during the preupwelling spring bloom season, maximum production of Globigerina bulloides occurs during the upwelling season, and maximum production of N. dutertrei occurs in the postupwelling season when the thermocline is well-developed (Sautter and Thunell, 1991; Thunell and Sautter, 1992; Sautter and Sancetta, 1992). Other factors such as light (i.e., photic zone depth) and food availability (i.e., primary productivity, chlorophyll maximum depth) may influence planktonic foraminiferal distributions (Fairbanks and Wiebe, 1980; Ortiz and others, 1995). In the Oregon coastal upwelling zone, asymbiotic foraminifera like N. pachyderma (dextral) and G. bulloides are most abundant near the coast where food is plentiful and turbidity is high, while G. ruber and N. dutertrei, classified by Ortiz and others (1995) as photosynthetic symbiotic foraminifera, are most abundant offshore where food is limited and turbidity is low (Ortiz and others, 1995). Off the Oregon coast, total foraminiferal shell flux parallels organic carbon flux (Ortiz and Mix, 1992). Thus, relative and absolute foraminiferal abundances may be useful in reconstructing physical and biological factors like the depth of the photic zone and the depth of the chlorophyll maximum. HYDROGRAPHY In the EEP the interactions of four currents control hydrography: the South Equatorial Current (SEC), the Equatorial Counter Current (ECC), the Peru Current (PC), and the Equatorial Undercurrent (EU) (Fig. 1). The SEC, driven by the southeast trade winds, flows from east to west at the surface along the equator (Fig. 1). The ECC flows at the surface from west to east where the trade winds are weakest: between the westward flowing SEC and the westward flowing North Equatorial Current, driven by the northeast trades (Fig. 1). The northerly component of the winds drives the PC, which flows northward along the coast of Peru and is advected to the northwest into the SEC (Wyrtki, 1966) (Fig. 1). Coastal upwelling of cool, nutrient-rich waters occurs along the Peruvian coast south of the equator. The EU flows eastward under the surface at about 100 m depth along the equator west of the Galapagos to replace surface water driven westward by the trades (Wyrtki, 1966). Upwelling along the equator occurs across the Pacific; in the East Pacific, equatorial upwelling is at least partially of cold, nutrientenriched waters of the EU (Stevenson and Taft, 1971; Pak and Zaneveld, 1973; Leetma, 1982).

3 112 FAUL, RAVELO, AND DELANEY FIGURE 2. Organic carbon mass accumulation rates (MARs) in the eastern tropical Pacific. Our sites represent the gradient in organic carbon MARs across the EEP: Region 3 sites (RC9-69, RC11-238, VM17-44) are located in the area of highest rates (15 24 mg org C cm 2 kyr 1 ), Region 2 sites (RC23-12 and RC23-20) are located in an area of slightly lower rates (12 15 mg org C cm 2 kyr 1 ), and Region 1 sites are in an area of lower rates (3 12 mg org C cm 2 kyr 1 ). Organic C MARs roughly correspond to the relative influence of the upwelling and lateral advection of nutrients in cold tongue of the Peru Current. Contour units are mg cm 2 kyr 1. Stars indicate core locations. Figure modified from Lyle (1992). SITE SELECTION Mean annual thermocline depth throughout the EEP is nearly constant (about 55 m), yet primary productivity increases from west to east, as evidenced by water column (Chavez and Barber, 1987) and sediment (Lyle, 1992) organic carbon measurements (Fig. 2). We use this primary productivity gradient to define three regions of varying primary productivity. Far west of the Galapagos (Region 1), mean primary productivity is 0.5 gc m 2 day 1, between the Galapagos and the west coast of South America (Region 2) it is twice as high (about 1 gc m 2 day 1 ), and in the coastal upwelling region (Region 3) within 150 km of Peru it is even higher (2.25 gc m 2 day 1 ) (Chavez and Barber, 1987). Consequently, the chlorophyll maximum depth, and thus the turbidity of the surface water and photic zone depth, is variable across regions depending on productivity. Our chlorophyll maximum depth and photic zone depth data (Table 1) are not well constrained because they represent a few cruises, and are not annual averages or seasonal compilations. Because annual averages or seasonal compilations of these variables are currently unavailable, we use the data presented here as our best estimates of these parameters. No past studies have directly compared water column chlorophyll data to the foraminiferal record in coretops, so this comparison is a first order study. The primary productivity gradient is also recorded in the sediment record: in Region 1 the organic carbon mass accumulation rates are less than or equal to 12 mg cm 2 kyr 1, whereas in Region 2 they are between 12 and 18 mg cm 2 kyr 1 and in Region 3 they are greater than or equal to 18 mg cm 2 kyr 1 (Fig. 2). We focus on the differences between the end members, Region 1 and Region 3. Low primary productivity in Region 1 (RC and RC13-138) results in a deep chlorophyll maximum and photic zone relative to the other regions (Table 1). Because this region is primarily influenced by Sub-Tropical Surface Water delivered by the SEC, it has a relatively large seasonal sea surface temperature (SST) range (15-28 C) and high salinity (Wyrtki, 1966). Site RC (3195 m) is above the lysocline (3200 m) on the western flank of the East Pacific Rise (Adelseck and Anderson, 1978). Site RC (2655 m) is well above the calcite compensation depth (CCD) in the Guatemala Basin (3700 m) (Lyle, 1992). Region 2 (RC23-12 and RC23-20), influenced by divergent upwelling of water from the EUC and advection of water from the ECC, has higher primary productivity than Region 1, and on average a shallower chlorophyll maximum and photic zone than Region 1 (Table 1). The Tropical Surface Water of Region 2 is characterized by low salinity and high SSTs (approximately greater than or equal to 25 C) with less than 5 C seasonal range (Wyrtki, 1966). Temperature and salinity profiles (Levitus, 1982), monthly averages interpolated at a 1 by 1 spacing, do not resolve differences in hydrography between RC23-12 and RC23-20 that are apparent from the water column data. At the longitude of these sites (about 84 W), there is a steep productivity gradient between the equator and about 2 N. At RC23-12, half a degree farther north than RC23-20, chlorophyll values are

4 TABLE 1. Site characteristics. Site Latitude a ( W) a Longitude Water depth (m) Mean annual SST ( C) b Annual SST range ( C) b Mean annual surface salinity b Mean annual mixed layer depth (m) c Thermocline depth range (m) d Mean annual thermocline depth e (m) Photic zone depth (m) f Deep chlorophyll maximum depth g (m) Region 1: Low Organic Carbon Accumulation Rate Western Sites, Sub-Tropical Surface Water RC S RC N Region 2: Moderate Organic Carbon Accumulation Rate Northern Sites, Tropical Surface Water RC N RC N Deep chlorphyll maximum (mg chl/m 2 ) h Region 3: High Organic Carbon Accumulation Rate Southeastern Sites, Equatorial Surface Water RC S RC S VM S a Latitude, longitude, and water depth for each site from Lamont Doherty Earth Observatory records. b Hydrographic data for each site from Levitus (1982). Annual SST range calculated as the difference between highest and lowest monthly average SSTs for each site. c Mean annual mixed layer depth defined as the depth at which the mean annual temperature is 1 C lower than mean annual SST (Levitus, 1982). Precise to within 10 m. d Thermocline depth range from monthly thermocline depths. Monthly thermocline depth determined by steepest slope of temperature with respect to depth (Levitus, 1982). e Precise to within 10 m. f Photic zone depth defined as depth where E o (PAR) 1% (Barber, R. M., written communication, 1997). Nearest data available for RC are integrated over 2.9 S to 10.1 S and 94.9 W to W; for RC13-138, at 1 N, 95 W in December; for RC23-12, at 1.5 N, 85 W in March; for RC23-20, at 1.0 N, 85 W in April; for RC , at 1 S, 85 W in April; for RC9-69, at 2 S, 85 W in April; for VM17-44, at 3 S, 85 W in April. Calculated as depth of 1% light with observed chlorophyll values. g Depth of deep chlorophyll maximum (Barber, R. M., written communication, 1997). h Chlorophyll level at depth of deep chlorophyll maximum (Barber, R. M., written communication, 1997). RC and VM17-44 have surface (0 m depth) chlorophyll maximums of 0.38 mg chl/m 2 and 0.68 mg chl/m 2 respectively. EAST PACIFIC UPWELLING AND PRODUCTIVITY RECONSTRUCTIONS 113

5 114 FAUL, RAVELO, AND DELANEY higher and the photic zone is 30 meters shallower than at RC The Levitus (1982) data are more representative of RC23-20 than RC Core RC23-12 is below the reported CCD in the Panama Basin (3200 m) (Lyle, 1992), so foraminiferal isotopic and relative abundance records from RC23-12 must be interpreted with caution. The Equatorial Surface Water of Region 3 (RC11-238, RC9-69, and VM17-44), influenced by seasonal advection of cool waters from the PC and by divergent equatorial upwelling (Fig. 1), has the highest productivity, and on average, the shallowest photic zone depths relative to the other regions. Within error the chlorophyll maximum depth of Region 3 is indistinguishable from Region 2 (Table 1). The hydrographic data (Table 1) for each site in Region 3 indicate that the depth of the photic zone and deep chlorophyll maximum is variable between sites because data are from different single days. Seasonal temperature data indicate that there is strong seasonal upwelling at these sites (Levitus, 1982). Mean annual thermocline depth estimates precise within 10 meters (Table 1) are shallower in Region 3 (50 m) than in Regions 1 and 2 (60 m). Region 3 has the highest productivity as indicated by studies of the modern water column (Chavez and Barber, 1987), as well as the highest organic carbon accumulation rates in surface sediments (Fig. 2) (Lyle and others, 1988). Nevertheless, due to the seasonality and interactions involving coastal and equatorial divergent upwelling, as well as outcropping of the equatorial undercurrent, there are heterogeneities even within each region resulting in areas within Region 2 that may have higher upwelling and/or biological productivity than areas within Region 3. For example, the hydrography at RC9-69 in Region 3 appears to be most similar to the hydrography at RC23-12 in Region 2 (Table 1). Records from VM17-44 below the CCD (3200 m) in the Peru Basin (Lyle, 1992) must be interpreted with care. Our downcore site, RC (to 20 ka) in the Peru Basin, will be used to reconstruct hydrographic changes in Region 3 over the last 20,000 years. STRATEGY Our approach is to determine the relationship of foraminiferal oxygen isotope and relative abundance patterns to modern productivity and hydrographic conditions including thermocline depth, photic zone depth, and chlorophyll maximum depth, and to track past hydrographic changes in the EEP. In the first part of our study, we determine calcification depths by calculating the calcite 18 O values in equilibrium with the water column, assuming that the eleven species or morphotypes of planktonic foraminifera we measured calcified in oxygen isotopic equilibrium in the water column. We count the relative abundance of species and compare them to the vertical distribution of species determined from the isotopes to find relationships that can be used to reconstruct water column characteristics in the past. Because the temperature and salinity data we use are averaged over a 1 by 1 grid, and the chlorophyll concentration and photic zone depth data that we have represent only one day out of the full seasonal cycle, it is difficult to make quantitative calibrations between coretop samples and water column hydrographic parameters. Thus, we look for patterns in the Site Sample depth (cm) TABLE 2. AMS- 14 C ages a (ka) Radiocarbon dates. AMS- 14 C error (kyr) Calibrated age b (ka) RC cm cm cm cm RC9-69, 0 cm RC23-12, 0 cm RC23-20, 0 cm VM17-44, 0 cm a AMS- 14 C dates obtained on tests of the planktonic foraminiferal species N. dutertrei from the Center for Accelerator Mass Spectrometry, Lawerence Livermore National Laboratory. b Corrected using Stuiver and others (1998). data that allow us to distinguish between different hydrographic regimes (particularly between end member Regions 1 and 3). The understanding we gain from interpreting the coretop data in the context of a comparison between the three hydrographic regions will then be applied to a record of isotopic values and abundances of various species from a site to evaluate its history of upwelling and productivity over the last 20,000 years. METHODS Ages for the coretop sediments and stratigraphic control for the downcore study are from radiocarbon dating of foraminiferal tests (Table 2). We obtained radiocarbon dates on 50 to 100 N. dutertrei shells from several coretops and selected depths in core RC at the Center for Accelerator Mass Spectrometry (CAMS), Lawerence Livermore National Laboratory. We corrected all dates for a global oceanic reservoir age of 400 years (Bard, 1988). We converted reservoir corrected radiocarbon dates to calibrated age using the Radiocarbon Calibration Program Revision 4.1 of Stuiver and others (1998). Coretop ages range from 2,000 to 4,000 years (Table 2). These ages are expected as a result of mixing due to bioturbation in the upper several centimeters of the sediment column. By assuming linear sedimentation rates between radiocarbon dates in the downcore section, we interpolated ages between dated samples. We used 5 cm 3 coretop samples from all sites (Table 1), and5cm 3 samples every 5 cm from 0 to 100 cm from the downcore site (RC11-238). We disaggregated samples using a sodium metaphosphate solution and wet sieved them to separate the 63 m fraction. We picked eleven species or morphotypes of planktonic foraminifera (G. ruber ( m), G. sacculifer (without a sac-like final chamber) ( m), G. sacculifer (with a sac-like final chamber) ( m), O. universa ( m), G. bulloides ( m), P. obliquiloculata ( m), G. menardii ( m), G. tumida ( m), N. dutertrei ( m), N. pachyderma (dextral) ( m) (except for downcore record), and G. inflata ( m)) and benthic Cibicidoides spp. ( m) (for downcore record) when present in the sample. We avoided smaller size fractions of each species in order to minimize size dependent fractionation effects greatest for juveniles (e.g., Ravelo and Fairbanks, 1995). We measured the 18 O and 13 C val-

6 EAST PACIFIC UPWELLING AND PRODUCTIVITY RECONSTRUCTIONS 115 ues of various species from each sample on a VG Prism or a VG Optima stable isotope mass spectrometer equipped with a common acid bath in the University of California Santa Cruz Stable Isotope Laboratory. We measured the 18 O and 13 C values of single G. sacculifer (without saclike final chamber) tests from the RC coretop sample. Results (Table 3 for coretops and Table 4 for the downcore record) are reported in relative to VPDB and calibrated via NBS-19. Our precision is 0.08 for 18 O values and 0.05 for 13 C values based on replicates of an internal calibration standard. We estimated the 18 O value of calcite precipitated in equilibrium with water at each site from 0 to 300 meters for each month of the year using temperature and salinity data from Levitus (1982) and the paleotemperature (t w ) equation of Bemis and others (1998) developed for O. universa under low light conditions: t w ( s w ), where s and w are the oxygen isotopic composition of the calcite sample and the seawater, respectively. We calculated w using the relationship determined for the EEP by Fairbanks and others (1982): w (0.26 salinity) Many species have maximum fluxes during a specific season of the year. However, it is impossible for us to assume which season would be most appropriate for our calculation of calcification depth based on a comparison of measured and predicted 18 O values of calcite. Therefore, we determined a possible calcification depth range for each species at each site as the overlap between the measured 18 O value and the seasonal range of the predicted 18 O value of equilibrium calcite. For the 18 O values of the foraminifera that are lower than equilibrium values at the sea surface, we assumed a surface (0 m) calcification depth. Because there are only slight differences in the w at the minimum and maximum calcification depths, the calculated minimum and maximum calcification temperature were typically within a few tenths of a degree; thus, we use the average calcification temperature in all interpretations. We use the Bemis and others (1998) O. universa (low light) equation because it calibrates well to plankton tow data. For comparison, for species from RC11-238, the O. universa (low light) equation of Bemis and others (1998) predicts temperatures less than or equal to 1 C lower than the Erez and Luz (1983) equation. There is little difference between the two equations predicted calcification depths for the surface dwellers (e.g., G. ruber, G. sacculifer (without a final sac-like chamber)). The Bemis and others (1998) equation predicts calcification depths 50 m or more deeper for deep dwellers (e.g., P. obliquiloculata and N. pachyderma (dextral)) relative to the Erez and Luz (1983) equation. Using the Bemis and others (1998) G. bulloides (12 chamber version) equation we calculated unreasonably deep calcification depths (300 to 400 m) and cold calcification temperatures (approximately 11 C) for this species for several coretops (RC13-113, RC13-138, RC9-69, and RC11-238). This implies that this equation is not applicable to our data. We estimated percent relative abundances for eleven planktonic foraminiferal species or morphotypes (G. ruber, G. sacculifer (without a sac-like final chamber), G. sacculifer (with a sac-like final chamber), Orbulina universa, G. bulloides, Pulleniatina obliquiloculata, Globorotalia menardii, Globorotalia tumida, N. dutertrei, N. pachyderma (dextral), and Globorotalia inflata) from each sample using a one-quarter split of the 150 m size fraction. We based percent relative abundances on the number of specimens of a given species per 300 specimens counted yielding a minimum error in percent abundance of approximately 0.3%. We calculated the number of each species per gram using the weight of the original sample and species count data. We were unable to calculate foraminiferal flux accumulation rates for the coretops since coretop sedimentation rate data were unavailable. We calculated foraminiferal flux accumulation rates for the downcore samples using the number per gram data, sedimentation rates, and determinations of dry bulk density (DBD) using the equation of Murray (1987) based on measured percent calcium carbonate (CaCO 3 ) in the samples. The standard error estimate for the relationship of DBD to percent CaCO 3 is gm/cm 3. We measured weight percent CaCO 3 on all samples from the downcore record (Table 4) using a UIC, Inc. Coulometrics Model 5012 CO 2 Coulometer in the University of California at Santa Cruz Institute of Marine Sciences Marine Analytical Laboratory. Relative standard deviations on the means for multiple determinations of a pure CaCO 3 standard and samples run in duplicate within a given analytical run were less than 1%. The detection limit for weight percent CaCO 3 depended on sample size; for typical sample sizes of 5 10 mg, the detection limit is wt %. RESULTS AND DISCUSSION GROUNDTRUTHING RELATIONSHIPS BETWEEN HYDROGRAPHY AND FORAMINIFERAL INDICATORS In the modern ocean, planktonic foraminiferal species preferred ecological habitats, and therefore their vertical distribution in the water column, depend on vertical variations in hydrographic parameters such as temperature, food availability, chlorophyll concentration, and light levels (e.g., Fairbanks and others, 1980; Fairbanks and Wiebe, 1980; Fairbanks and others, 1982; Curry and others, 1983; Thunell and others, 1983; Thunell and Reynolds, 1984; Sautter and Thunell, 1991; Thunell and Sautter, 1992; Ortiz and others, 1995, 1996, 1997; Watkins and Mix, 1998). Consistent with these previous studies, we find the calcification depths of species that have algal symbionts, such as G. sacculifer (without a sac-like final chamber) and G. ruber, are restricted to upper levels of the water column where light levels, and not nutrient or chlorophyll concentrations, are highest (Fig. 3 5). For other species, calcification depths may be influenced by the position of the deep chlorophyll maximum and the thermocline. When the thermocline, the bottom of the photic zone, and the chlorophyll maximum are coincident and deep, G. bulloides, N. dutertrei, G. tumida, and N. pachyderma (dextral) calcify in close proximity to each other both in temperature and depth just below the photic zone (Region 1; Table 1, Table 3, Fig. 3). In the most intensely productive region where the photic zone and the chlorophyll maximum are shallowest (Region 3), N. pachyderma (dextral) and N. dutertrei calcify deeper in the thermocline at similar temperatures to Regions 1 and 2, but G. bulloides (except at RC9-69) and G. tumida calcify at shallower depths and warmer temperatures (Region 3; Table 1,

7 116 FAUL, RAVELO, AND DELANEY TABLE 3. Planktonic foraminiferal abundances, stable isotopes, and calculated calcification depths and temperatures. Species and site Relative % species 13 C( ) 18 O( ) Calcification Depth Range (m) Midpoint of Calcification Temperature Range ( C) G. ruber RC RC RC RC RC RC VM Average G. sacculifer (without a sac-like final chamber) RC RC RC NA RC RC RC VM Average G. sacculifer (with a sac-like final chamber) RC RC RC RC RC RC VM NA NA NA NA Average O. universa RC RC RC NA NA NA NA RC NA NA NA NA RC RC VM NA NA NA NA Average G. bulloides RC RC RC NA NA NA NA RC RC RC VM Average G. menardii RC RC RC RC RC RC VM Average G. tumida RC RC RC RC RC RC VM Average

8 EAST PACIFIC UPWELLING AND PRODUCTIVITY RECONSTRUCTIONS 117 TABLE 3. Continued. Species and site Relative % species 13 C( ) 18 O( ) Calcification Depth Range (m) Midpoint of Calcification Temperature Range ( C) N. dutertrei RC RC RC RC RC RC VM Average P. obliquloculata RC RC RC RC RC RC VM NA NA NA NA Average G. inflata RC NA NA NA NA RC NA NA NA NA RC RC NA NA NA NA RC NA RC NA VM NA NA NA NA Average NA NA N. pachyderma (dextral) RC RC RC RC RC RC VM Average TABLE 4. Measurements from Core RC Depth (cm) Age (ka) CaCO 3 (%) 13 C G. sacculifer 18 O N. dutertrei 13 C 18 O 13 C G. tumida 18 O 13 C G. bulloides 18 O 13 C G. ruber 18 O Cibicidodies spp NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA C 18 O

9 118 FAUL, RAVELO, AND DELANEY FIGURE 3. Modern monthly 18 O profiles of equilibrium calcite (gray curves) predicted using temperature and salinity data (Levitus, 1982) at (a) RC and (b) RC (Region 1) plotted vs. depth in the upper 150 meters of the water column. The 18 O values for 11 species of planktonic foraminifera in the coretop are plotted as vertical bars. The overlap between the measured 18 O values of the foraminifera and the predicted 18 O values of the water column is the implied calcification depth range. G. ruber is plotted as a dot in coretop RC because its 18 O value only intersects the lowest predicted 18 O values of the water column. Below the calcification depth figure, the relative abundance of 11 species when present are plotted. The bottom of the photic zone and the chlorophyll maximum are coincident with the relatively deep thermocline under well stratified Region 1 conditions. This leads to a distribution of species where G. bulloides, N. dutertrei, G. tumida, and N. pachyderma (dextral) calcify in close proximity to each other both in temperature and depth just below the bottom of the photic zone. Table 3, Fig. 5), in some cases as shallow as the chlorophyll maximum. Overall, comparison of the coretop data from the three regions indicates that the calcification depth of each species is not necessarily constant, nor is calcification temperature. Rather, both may change depending on the vertical position of hydrographic features and the availability of food. Calcification depth differences based on hydrographic features may be subtle. High productivity regions (Region 3) are different from much lower productivity regions (Region 1). However, the subtle differences between Regions 1 and 2 and Regions 2 and 3 may not be well detected in coretops and/or the sediment record. By interpreting species oxygen isotopic values in the context of the hydrographic region, we can define what environmental parameters may control calcification depths, and how to use foraminiferal isotopic signatures and abundances to interpret significant environmental changes. Our interpretations are significantly informed by comparing our coretop results to those of other studies. Specifically, studies in the northern California Current region demonstrate that distinct planktonic foraminifera populations are associated with gyral waters (low nutrients, relatively warm water), midway waters (offshore upwelling: warm oligotrophic thin mixed layer above upwelled sub-surface nutrients), and nearshore waters (coastal upwelling: nutrient-rich cold surface water) (Ortiz and Mix, 1992; Ortiz and others, 1995). In the San Pedro Basin (a southern California borderland basin), planktonic foraminifera species have distinct seasonal fluxes. Some species only proliferate in the upwelling season (e.g., G. bulloides), and other species fluxes are highest in the post-upwelling season (e.g., N. dutertrei) (Sautter and Thunell, 1991; Thunell and Sautter, 1992). The differences seen in the plankton tow standing stock of different species that span the nearshore to gyre waters of the north are analogous to the seasonal differences in flux in the San Pedro basin, which are related to seasonal changes in the water column temperature and nutrient structure. For example, the highest standing stock of G. bulloides is located in the nearshore upwelling plankton tow in the northern California Current (Ortiz and others, 1995) where the photic zone depth is at a minimum and chlorophyll concentrations are highest. Analogously, maximum flux of G. bulloides is in the up-

10 EAST PACIFIC UPWELLING AND PRODUCTIVITY RECONSTRUCTIONS 119 FIGURE 4. Same as Figure 3 for the coretops from Region 2 (a) Site RC23-12 and (b) Site RC Notice that when the deep chlorophyll maximum is shallow as in Region 2, G. bulloides calcifies at shallower depths relative to Region 1 (and its relative abundance in the assemblage is higher) while N. dutertrei and G. tumida calcify deeper in the thermocline. Note that Levitus (1982) hydrography for both sites is the same due to resolution limitations. Hydrography is probably more accurate for RC welling season sediment trap in the south (Sautter and Thunell, 1991; Thunell and Sautter, 1992), and its isotopic values indicate calcification in the warm (low 18 O value) nutrient enriched (low 13 C value) surface mixed layer that develops during the upwelling season. In both the northern California Current and the San Pedro basin studies, G. bulloides calcifies over a wide range of temperatures depending on the season or depth of the chlorophyll concentration maximum; calcification occurs in deeper colder water when upwelling is less intense and the water column is well stratified, and in warmer surface water when upwelling is more intense. Our coretop results from the EEP are consistent with these previous studies: G. bulloides calcifies at a wide range of temperatures (16.8 C 3.4) and depths (0 210 m) (Fig. 3 5; Table 3) depending on the intensity of upwelling and the intensity of production. At the sites where we would expect upwelling and biological productivity to be relatively strong near the Peruvian upwelling system (VM17 44) and closest to the equator (RC23-20), G. bulloides is most abundant and its 18 O and 13 C values relative to N. dutertrei are relatively low (Fig. 6). Low 18 O values indicate calcification in relatively warm water, and low 13 C values indicate calcification in relatively nutrient enriched water, typical of regions where nutrient rich water upwells to the surface at least seasonally. At other sites (RC9-69, RC11-238, RC13-113, RC13-138) with slightly lower relative abundances of G. bulloides, higher 18 O and 13 C values indicate deeper calcification indicative of slightly more stratified conditions where the chlorophyll maximum is below the surface at colder temperatures on the annual average (Fig. 6). Note that the range in hydrographic properties and G. bulloides relative abundances at our sites is much smaller than in either of the California margin studies, making the relationships between hydrographic conditions and foraminiferal abundances more difficult to resolve. Thus, we rely on previous groundtruthing studies in combination with core top results to interpret the downcore data presented in the next section. N. dutertrei, a thermocline dweller, dominates the assemblages in Region 3 where there are cool sub-surface temperatures influenced by equatorial upwelling or advection of cool waters from the Peru Current (Fig. 5). For example, N. dutertrei calcifies consistently along the 15 C isotherm (Table 3) often in association with the lower thermocline (Fig. 3 5). G. tumida calcifies near the bottom of the photic zone regardless of temperature (Ravelo and Fairbanks, 1992). Hence, the difference in 18 O values between N. dutertrei

11 120 FAUL, RAVELO, AND DELANEY FIGURE 5. Same as Figure 4 for the coretops from Region 3 (a) Site RC11-238, (b) Site RC9-69, and (c) VM Seasonal temperature data represented by the 18 O values of equilibrium calcite indicate that there is strong seasonal upwelling at these sites. Low 18 O values of G. bulloides and higher G. bulloides relative abundances at VM17-44 are indicative of active upwelling. In this most intensely productive region where the photic zone and the chlorophyll maximum are shallowest, N. pachyderma (dextral) and N. dutertrei calcify deeper in the thermocline at similar temperatures to Regions 1 and 2 but G. bulloides (except for at RC9-69) and G. tumida calcify at shallower depths and warmer temperatures. and G. tumida reflects the relative difference in depth between approximately the 15 C isotherm (often associated with the lower thermocline at our sites) and the bottom of the photic zone (Fig. 3 5). In Region 3, where biological production is highest, G. tumida calcifies in warmer water at shallower depths than N. dutertrei because the photic zone is shallow relative to the thermocline depth (Table 1, Fig. 5). In contrast, in Region 1 where biological production is lowest, the depth stratification of these two species may even be reversed, reflecting the fact that the photic zone is on average deeper than the thermocline (Table 1, Fig. 3). In contrast, G. ruber, a warm-water surface dweller, is slightly more abundant in Region 2 where there are high SSTs than in the other regions (Fig. 4). The patterns of abundance of the warm surface water G. sacculifer (with and without a sac-like final chamber) vary with those of G. ruber across regions. G. ruber and G. sacculifer (with and without a sac-like final chamber) 18 O values indicate calcification in the surface layer or upper thermocline with one exception (G. sacculifer (without a sac-like final chamber) at RC23-12) (Fig. 3 5). We do not have well resolved hydrographic information for RC23-12, located slightly north of RC There are steep temperature gradients (about 1 C per 1 latitude) from the equator northward to about 1.5 N in this region. Higher SSTs at RC23-12 relative to RC23-20 due to surface temperatures increasing sharply northward across the equator would potentially produce lower 18 O values of surface dwelling foraminifera, and thus higher calcification temperatures at RC23-12 relative to RC The 18 O values of G. menardii indicate that it calcifies in the mid- to lower thermocline (Fig. 3 5). Because O. universa, G. inflata, and P. obliquiloculata are trace species in all the coretops in this study, their oxygen isotopic values cannot be expected to be representative of their preferred depth ecology. Differences between sites for O. universa, G. inflata, P. obliquiloculata, and G. menardii relative abundances do not seem to be systematically related to hydrographic variations. A RECORD OF PAST THERMOCLINE STRUCTURE AND PRODUCTIVITY: RC We interpret changes in biological production based on foraminiferal flux and assemblage changes. Since foraminiferal flux rates depend on sedimentation rate estimates (6.4

12 EAST PACIFIC UPWELLING AND PRODUCTIVITY RECONSTRUCTIONS 121 FIGURE 6. Comparison of 18 O and 13 C values of G. bulloides relative to N. dutertrei at the each of the modern coretop sites (closed circles). Circle size is proportional to G. bulloides % relative abundance at each site. A 18 O difference value of 0 means that G. bulloides and N. dutertrei calcified at the same depth. Downcore data are also given: Site RC at 10 ka (cross), 19 ka (open circle). The G. bulloides isotopic values at 10 ka at RC is similar to the G. bulloides isotopic values at the modern sites (RC23-20 and VM17-44) where G. bulloides relative abundance is high. cm/kyr for RC11-238), we discuss trends (e.g., modern vs. LGM) in foraminiferal flux rather than specific events. Several lines of evidence indicate that total flux changes in core RC are primarily driven by changes in productivity rather than dissolution intensity. Core RC (2573 m) is well above the present Peru Basin CCD (3200 m) (Lyle, 1992). In the equatorial Pacific, resistant species percent (a dissolution indicator for foraminifera) is not correlated with core depth above 3800 m, indicating that dissolution is not primarily controlling the assemblages (Andreasen and Ravelo, 1997). In addition, we measured 18 O values of single G. sacculifer (without a sac-like final chamber) specimens to check for dissolution effects. G. sacculifer (without a saclike final chamber) 18 O values (average value ; range from 0.60 to 1.91 ) from the RC coretop produce a range similar to the surface seasonal range of predicted 18 O equilibrium values ( 1.0 to 2.0 ) at this site (Fig. 5). These similar ranges indicate that the 18 O values of these surface dwelling foraminifera are not strongly affected by dissolution. Individuals show a slightly larger range ( 1.3 ) than the predicted monthly 18 O equilibrium values ( 1.0 ) because they may record more short-term variability. Thus, total flux changes in core RC are primarily driven by changes in productivity rather than dissolution intensity. In addition, several other studies using various indicators have found evidence for higher productivity during the LGM relative to present (Pedersen, 1983; Sarnthein and others, 1988; Lyle and others, 1988; Mix, 1989a, 1989b; Pedersen and others, 1991). Therefore, it is likely that higher total foraminiferal flux in the LGM in core RC implies greater productivity during the LGM relative to present. Earlier studies (Thunell and others, 1983; Thunell and Reynolds, 1984) indicate that N. dutertrei proliferates during the winter upwelling season in the Panama Basin and in the post-upwelling season when the thermocline is welldeveloped in the San Pedro Basin (Sautter and Thunell, 1991; Thunell and Sautter, 1992; Sautter and Sancetta, 1992). Annually, N. dutertrei calcifies at approximately 15 C in a well developed mid- to lower thermocline, regardless of upwelling conditions. In the coretops, N. dutertrei s 18 O values are consistent with shallow calcification depths in months of greater upwelling (Fig. 5). Based on previous studies and our coretop data, we use N. dutertrei as a monitor of the sub-surface thermocline. High N. dutertrei relative abundances, especially from 10 ka to the present, indicate a well developed sub-surface thermocline was probably present from the LGM to modern but was more prevalent in the Holocene (Fig. 7c). Higher absolute fluxes of N. dutertrei as well as other species during the LGM relative to modern fluxes imply greater biological production relative to present (Fig. 7d). Our foraminiferal 18 O records support little change in the local thermocline. Most planktonic foraminiferal species show gradual deglaciation trends with glacial-interglacial 18 O changes within the range of the whole ocean 18 O shift due to ice volume changes of 1.0 (Schrag and others, 1996) to 1.2 (Fairbanks, 1989) (Fig. 7a). This indicates little difference between LGM and modern calcification temperatures or depths (Fig. 7a). The benthic foraminiferal 18 O record (Fig. 7a) has a glacial-interglacial difference of 2.0, reflecting a combination of colder bottom water temperatures and the global 18 O ice volume signal. N. dutertrei, calcifying in the thermocline, and G. tumida, calcifying at the bottom of the photic zone, record close to the same 18 O values relative to each other throughout the record (Fig. 7a). At least seasonally, the modern close relationship between thermocline and the bottom of the photic zone was maintained. Thus, at least an annual average higher biological production does not appear to have been accompanied by greater upwelling during the LGM relative to present. Our study implies lower SSTs during the LGM relative to present in the absence of changes in the thermocline. A smaller 18 O gradient between G. sacculifer (without a saclike final chamber) and N. dutertrei during the LGM relative to modern may be interpreted as indicating either a reduced surface to thermocline thermal gradient (shoaling of the thermocline), or reduced SSTs for the LGM relative to present. Patrick and Thunell (1997) concluded the thermocline was shoaled and upwelling was more intense in the nearby Panama Basin during the LGM. However, this interpretation requires N. dutertrei to have calcified at warmer temperatures during the LGM than it does today. We find little evidence to support this from our data, since N. dutertrei calcifies so consistently along the 15 C isotherm in this region today, and because it calcifies in a small temperature range between seasons in other regions (e.g., the Panama Basin (Curry and others, 1983) and the San Pedro Basin (Sautter and Thunell, 1991)). An alternate interpretation of the reduced 18 O gradient is that G. sacculifer (without a sac-like final chamber) calcified at the surface in SSTs cooler (by about 2 C) than present, and N. dutertrei calcified at temperatures and depths similar to present, implying that SSTs were cooler during the LGM without a change in the thermocline temperature. G. bulloides isotopic data combined with abundance data