U.S. Geological Survey


Last Interglacial Sea-Surface Temperature Estimates from the California Margin: Improvements to the Modern Analog Technique

By Harry J. Dowsett and Richard Z. Poore

U.S. Geological Survey, 955 National Center, Reston, VA 20192

Available only on the World Wide Web
Version 1.0

[NOTE: All tables and appendixes are in PDF format and require Adobe Acrobat Reader. If Acrobat Reader is not installed on this computer, it can be downloaded free of charge from the Adobe Web site. High-quality PDFs of each figure are also available for viewing or printing.]


Total faunal analyses of planktic foraminifer assemblages are used to derive sea surface temperature estimates for the last interglacial from Ocean Drilling Program Sites 1018 and 1020 off northern and central California. Foraminifer assemblage data were transformed to sea-surface temperature (SST) estimates by using the modern analog technique (MAT). In order to improve our ability to estimate SST in this area, the coretop calibration data base used in the MAT was augmented by 13 new age-validated coretop assemblages from the U.S. Pacific Margin.

Introduction and Background

We are currently analyzing Ocean Drilling Program (ODP) Leg 167 material from the California Margin to better understand the characteristics of the climate record within Marine Isotope Stage 5 (MIS 5) and other warm intervals of the late Pleistocene. The ODP 167 cores selected for study recovered several relatively undisturbed, high-accumulation-rate marine sequences that allow decadal- to century-scale sampling. In addition, the sediments off of the California Margin contain marine microfossils and a significant terrigenous component. Well-preserved pollen assemblages provide an opportunity to directly correlate marine and continental records preserved in the marine cores. Our long-term goal is to assemble and analyze a number of climate proxies from cores along the California Margin in order to examine the range and rate of natural climate variability and the relationship between marine and terrestrial climate in this area during previous interglacials such as MIS 5 and MIS 11 (for example, Poore and others, 1999, in preparation).

The purpose of this paper is to document the planktic-foraminifer-based SST record from MIS 5 at two sites along the California Margin (ODP 1018 and 1020). Total faunal analysis of planktic foraminifer assemblages is a widely used proxy for SST. However, a transfer function must be created to transform foraminifer assemblage census data into estimates of SST. There are two widely used types of multivariate transfer functions: the traditional factor analytic transfer function (Imbrie and Kipp, 1971) and the modern analog technique (MAT) (Hutson, 1979). Both methods rely on the availability of a modern calibration data set. Workers have assembled and used successfully a global planktic foraminifer data set for over 25 years (for example, Imbrie and Kipp, 1971; Kipp, 1976; CLIMAP, 1981; 1984; Prell, 1985; PRISM, 1994 [see also Dowsett, 1991; Dowsett and others, 1996, and so on]; Ortiz and Mix, 1997). Factor analytic transfer functions work best when the data being used for paleoenvironmental estimates come from within the geographic region covered by the calibration data. Thus, workers have developed "ocean-specific" equations for different regions (see CLIMAP, 1981). The MAT searches the entire calibration data set for the nearest analog (in a multivariate sense) to the fossil sample being analyzed. Searching a global data set allows patterns to emerge that might otherwise have gone unnoticed (for example, upwelling samples from many regions match the downcore sample, suggestive not only of a specific temperature or salinity range but also of a specific type of oceanic environment).

Figure 1 illustrates the distribution of the 1,145 coretop samples in the modern "CLIMAP" database (see also Prell, 1985). The lack of samples from the northeastern Pacific region precludes the development of a North Pacific factor analytic transfer function. Instead, foraminiferal assemblages from the California Margin can best be transferred to SST estimates by using the MAT. However, MAT SST estimates will still be hampered by lack of modern northeastern Pacific calibration samples. Therefore, before applying MAT, we decided to first examine and incorporate additional coretop samples from the Pacific Margin of North America into the global calibration data set to provide more accurate estimate of MIS 5 paleoceanographic conditions.

Figure 1. Location of coretop samples containing planktic foraminifera

Figure 1. Locations of 1,145 coretop samples with planktonic foraminifer census data. Shaded box shows the area of this study. Note the lack of coretop samples from the northeastern Pacific in the Prell (1985) data set.

View a high-quality PDF version [195K] of this figure in a new window.

The scope of this paper then is twofold. First we generate new calibration data for the Pacific Margin of North America, and then we apply the improved global calibration data set via the MAT to generate SST estimates for two MIS 5 sequences off the California coast.

Coretop Calibration Data

MAT depends upon a large modern database that captures much of the variability present in modern faunas. The global coretop data set consists 1,145 coretop samples distributed between 64.55° S. and 76.82° N. that represent surface conditions between -1.78° and 29.87° C. These data have been summarized by Prell (1985). Faunal analyses are from the work of Kipp (1976), CLIMAP (1981), Hutson and Prell (1980), Cullen and Prell (1984), Thompson (1976, 1981), Parker and Berger (1971), Coulbourn and others (1980), and unpublished data. We have calibrated all coretop localities to winter and summer temperature by using the SST climatology of Reynolds and Smith (1995).

In contrast to the modern coretop calibration data set summarized by Prell (1985), the data set of Ortiz and Mix (1997) shows 39 coretop samples from the northeastern Pacific region. Our initial evaluation of these 39 samples determined that three of the calibration points represent sediment trap data, not coretop data. Most of the remaining data points are coretop data from Coulbourn and others (1980), which were excluded from the coretop synthesis of Prell (1985). Inspection of the faunal data of Coulbourn and others (1980) suggests that a large number of the samples have "glacial" assemblages (see discussion below) and do not represent modern or Holocene samples. Thus, we conclude that these data require additional evaluation before they can be incorporated into the Prell (1985) coretop data.

To improve our ability to estimate reliable late Pleistocene SST along the California Margin, we set out to locate, count, and date coretop material from the Pacific Margin of North America. We culled through core repositories around the world and located approximately 75 potential coretop samples. After available core logs were evaluated (screening out cores with missing tops or obvious evidence of pre-Holocene age), this number was reduced to 36 samples, which were acquired from various repositories and processed at the U.S. Geological Survey in Reston, Va.

Samples were dried at temperatures of less than or equal to 50° C, disaggregated in deionized water, and wet sieved at 63. Samples were then dry sieved at 150 and thereby separated into <63-, 63- to 150-, and >150- sized fractions. In order to conform with the existing coretop database, we used only the >150- sized fraction for this study. If necessary, samples were split by using a CARPCO device1 to obtain a 300-specimen subsample. All specimens were identified to species level and fixed on a 60-square micropaleontological slide for counting.

For a variety of reasons (high dissolution, too few planktics, pre-Pleistocene fauna, evidence of reworking, and so on), a number of samples were determined to be unsuitable for inclusion in the coretop database. Of the original 36 samples processed and examined, 16 samples were considered for further analysis. Core names, sample intervals, locations, observed winter and summer SST, and faunal census for the 16 samples are given in tables 1 and 2.

Fig. 2. Location of 13 new modern coretop samples

Figure 2. Location of 13 new modern coretop samples (§) generated by this study and the location of ODP 1018 and 1020 (l). SamplesY6908-5, Y6910-3GC, F2-92-P29TW, BC-14 and BC-47 have confirmed Holocene ages.

View a high-quality PDF version [17K] of this figure in a new window.

The final test for inclusion in the coretop database was determination of a near-modern age for each sample. Several of these samples were located adjacent to each other and had very similar faunal assemblages. We decided that accelerator mass spectrometry (AMS) 14C dating of one of the samples within a geographic group would be sufficient to indicate the age of adjacent samples in that group. We devised a strategy whereby eight samples were dated by using AMS; the results are shown in table 1. Note that the dates listed in table 1 are uncorrected for reservoir effect. The limited amount of calcareous material available in the coretop samples required us to use various mixtures of planktic and benthic foraminifers for the AMS dates, precluding application of a standard planktic and benthic reservoir correction to our dates. In addition, determining a reservoir correction requires assumptions about the apparent age of the water in which the foraminifers calcify. Application of the standard reservoir correction of 2,380 for benthic foraminifer dates used by Ortiz and others (1997) for Pacific Margin cores would yield negative ages for two of our Baja coretop assemblages. Thus, for the purposes of this study, we report raw 14C dates because they are sufficient to establish our samples as late Holocene. Three samples were significantly older and represent last glacial maximum (LGM) conditions. The LGM foraminiferal assemblages were dominated (>70 percent) by Globigerina bulloides and sinistrally coiled Neogloboquadrina pachyderma with no N. dutertrei. Conversely, coretop samples that are AMS dated as Holocene are not dominated by these taxa. Ortiz and others (1997) showed similar findings in their comparison of LGM samples and sediment trap samples from 42° N. on the California Margin.

After deleting the LGM samples, we were left with 13 coretop samples of Holocene age (fig. 2). These were added to the modern coretop data set and represent the only age verified modern planktic foraminifer coretop data for this region of the world.

ODP Leg 167

Ocean Drilling Program (ODP) Leg 167 occupied a series of drill sites along the California Margin of North America. The 13 sites of Leg 167 are arranged in a series of depth and latitudinal transects to aid in the study of the origin and development of the California Current and the Neogene paleoclimatic evolution of the eastern Pacific.

Two sites, 1018 and 1020, have been selected for a multiproxy study of MIS 5e.

Site 1018 is located at 36.99° N., 123.28° W., approximately 75 km west of Santa Cruz, Calif., south of Guide Seamount, in 2,477 m of water. The site is on a sediment drift that rises 400 m above the adjacent sea floor. The sediments are characterized by olive-gray clay with silt. Modern August and February SST at Site 1018 are 15.30° and 12.39° C, respectively.

Planktonic foraminifers from the MIS 5 interval of Site 1018 are generally poorly preserved, and many samples have too few individuals to analyze quantitatively. The faunal data that we used for the MAT are given in table 3.

Fig. 3. Surface isotherms during summer and winter off California Margin

Figure 3. Surface isotherms during summer and winter off the California Margin.

View a high-quality PDF version [276K] of this figure in a new window.

Site 1020 is located at 41.00° N., 126.43° W. on the eastern flank of the Gorda Ridge approximately 170 km west of Eureka, Calif., in 3,038 m of water. The site is on a hill that rises 50 m above the surrounding sea floor. Site 1020 is ideally located to monitor the strong summer upwelling associated with the California Current system. Modern August and February SST at Site 1020 are 15.83° and 11.13° C, respectively.

Planktonic foraminifers from the MIS 5 interval of Site 1020 are better preserved than those of Site 1018. The faunal data are shown in table 4.

Figure 3 characterizes the summer and winter SST over Sites 1018 and 1020. During summer, upwelling is strong, and the isotherms close to the coast are oriented longitudinally so that warmer waters are found to the west. Summer temperatures at the two sites are almost identical (gradient 0.53° C). During winter, upwelling is greatly reduced, and the isotherms off the California Margin are oriented latitudinally. At this time, the surface temperature gradient between the two sites is 1.26° C.

Modern Analog Technique

The MAT quantifies faunal changes within deep-sea cores in terms of modern oceanographic conditions (Hutson, 1979). The method uses a measure of faunal dissimilarity to compare downcore samples to each sample in a modern oceanographic database or coretop calibration set. Working with late Pleistocene planktic foraminifers, Hutson (1979) originally used cosine-theta distance to match modern Indian Ocean samples to core samples of late Pleistocene age and then used a weighted average of sea-surface temperature and salinity associated with the closest analogs of each core sample to derive downcore environmental estimates. Overpeck and others (1985) investigated the responsiveness of eight dissimilarity coefficients to palynological changes caused by differences in modern vegetation and applied the technique to late Quaternary pollen diagrams from eastern North America. Their analyses suggested that, although all coefficients give roughly similar results, signal-to-noise measures performed better than unweighted or equal weight measures of dissimilarity (Overpeck and others, 1985).

We follow the methodology of Dowsett and Robinson (1997) and use the squared chord distance (SCD) measure:


where dijis the SCD between two multivariate samples i and j and pikis the proportion of species k in sample i. SCD values can range from 0.0 to 2.0; a value of 0.0 indicates identical proportions of species within the samples being compared. See Dowsett and Robinson's (1997) discussion of the technique.

Reorganization of Faunal Data

The coretop database was simplified by combining pink and white varieties of Globigerinoides ruber into one category; combining Globigerinoides sacculifer with and without a sacklike final chamber into one category; combining sinistral and dextral coiling Globorotalia truncatulinoides into one category; combining all subspecies of Globorotalia tumida and Globorotalia menardii into a single complex; and deleting from the coretop database those taxa represented by just a few individuals in only a small number of the coretop samples. This procedure resulted in the 27 taxonomic categories listed in table 5. In addition, once 11 duplicate samples were deleted from the original 1,145 modern samples, the new coretop database (with 13 new Pacific Margin) contained 1,147 faunal analyses.

Faunal data from Leg 167 samples were reorganized to conform to the same 27 categories. Both coretop and Leg 167 fauanl counts were recalculated to percentages prior to undergoing the MAT procedure.

SST Estimation

The SCD coefficient was used to make comparisons between each downcore sample and every sample in the coretop calibration data set. Following the work of Dowsett and Robinson (1997), we chose a 0.15 SCD cutoff. For each downcore sample, we located all coretop samples having an SCD of 0.15. SST was then estimated by taking a weighted average of the SST associated with the set of coretop samples exhibiting SCDs of 0.15. Standard deviations were calculated for the same set of samples so that each downcore SST estimate has its own error bar. Both cold season and warm season SSTs were estimated by means of this procedure.


Fig. 4. Downcore SST estimates

Figure 4. Downcore SST estimates for ODP 1018 and 1020. Horizontal bars at SST estimates represent 2s.

View a high-quality PDF version [18K] of this figure in a new window.

Following the procedure outlined above, we analyzed downcore samples from ODP 1018 and 1020 and each of the 1,147 modern coretop samples for a total of 56,203 SCD measurements (appendixes A, B). Of the 49 downcore samples analyzed, 11 failed to meet the criteria of having at least one coretop sample within a distance of 0.15 unit. These samples are designated in tables 6 and 7. The remaining SST estimates are plotted against age in figure 4.

Although, in many cases, nearest analogs to downcore samples were found outside the North Pacific, the additional 13 new coretop samples from the Pacific Margin were involved in 91 percent of the SST estimates generated by the MAT.

The SST estimates show that surface-water temperatures were at maximum values within MIS 5e between about 127 and 124 ka, consistent with the timing of insolation maxima and ice-volume minima, according to the Milankovitch theory (Imbrie, 1985; Imbrie and others, 1984). This finding suggests that changes in maximum summer insolation at 60° N. is a primary mechanism for glacial-interglacial cycles in Earth's climate. At ODP 1018, maximum summer and winter temperatures exceed modern temperatures; summer increases are up to 2° C, whereas winter temperature increases do not exceed 1° C. In contrast, maximum winter estimates at 1020 are essentially identical to modern values, whereas maximum summer SSTs are 1 to 3° C above modern values. The fact that the pattern of SST changes seen at the warmest intervals of MIS 5 match diatom and foraminifer assemblage data suggests that upwelling along the coast of central and northern California was less intense or more sporadic during the early warm parts of MIS 5e (Poore and others, in preparation). Thus, during the summer season, isotherms were not deflected as far south as they are today (see fig. 3), and overall atmospheric circulation was probably more zonal.

SST estimates from ODP 1020D indicate that summer and winter temperatures underwent a rapid decline of about 5° C near the end of MIS 5e and remained cold throughout MIS 5d. The temperature decline near the end of MIS 5e is a rapid transition, most of the 5°-C temperature decline occurring within 1,000 years.

Only a few samples from MIS 5c of ODP 1020 had assemblages that yielded reliable SST estimates. The limited number of data points suggests that summer SSTs may have rebounded to near-modern values for brief intervals. The two samples available from the MIS 5c interval of ODP 1018 yield temperature estimates far below the maximum values found in MIS 5e.

In summary, the results of this study show that continued work to improve the modern coretop calibration dataset is warranted. An expanded calibration data set coupled with quantitative analyses of planktic foraminifer assemblages can lead to better estimates of past oceanographic and climatic conditions from deep-sea sediments along the California Margin.


John Barron and Stacey Verardo (U.S. Geological Survey) and Joseph Ortiz (Lamont-Doherty Earth Observatory) provided thoughtful reviews of this manuscript. We thank Warren Smith (Scripps Institution of Ocenaography), Tim Herbert (Brown University), Bobbi Conard (Oregon State university), and Rusti Lotti Bond (Lamont-Doherty Earth Observatory) for their help in obtaining coretop material from the California Margin. We especially thank Paula Quinterno (U.S. Geological Survey, retired) for her help in obtaining material from central California Margin cores F2 92-P3 and F2 92-P29. Marci Robinson, Sarah Dunn, and Aaron McMahon (U.S. Geological Survey) provided technical assistance during this study.

Taxonomic Notes

The following taxa were identified from Site 1018 and Site 1020 samples:

Globigerina bulloides d'Orbigny
Globigerina falconensis Blow
Globierinella aequilateralis (Brady)
Globigerinita glutinata (Egger)
Globigerinoides conglobatus (Brady)
Globigerinoides ruber (d'Orbigny)
Globigerinoides sacculifer (Brady)
Globorotalia hirsuta (d'Orbigny)
Globorotalia inflata (d'Orbigny)
Globorotalia menardii (Parker, Jones, and Brady)
Globorotalia pumilio Parker
Globorotalia scitula (Brady)
Globorotalia truncatulinoides (d'Orbigny)
Globorotalia tumida (Brady)
Globorotaloides hexagona (Natland)
Neogloboquadrina dutertrei (d'Orbigny)
Neogloboquadrina pachyderma (Ehrenberg)

We separated N. pachyderma into sinistral-coiling and dextral-coiling categories. In addition, specimens having more than four chambers in the last whorl that were transitional to large typical forms of N. dutertrei were tabulated under the informal category of "dupac" and considered to be transitional forms between N. dutertrei and N. pachyderma.
Orbulina universa d'Orbigny
Pulleniatina obliquiloculata (Parker and Jones)
Turborotalita quinqueloba (Natland)


CLIMAP, 1981, Seasonal reconstruction of the Earth's surface at the last glacial maximum: Geological Society of America, Map and Chart Series MC-36, p. 1-18.

_________1984, The last interglacial ocean: Quaternary Research, vol. 21, p 123-224.

Coulbourn, W.T., Parker, F.L., and Berger, W.H., 1980, Faunal and solution patterns of planktonic foraminifera in surface sediments of the North Pacific: Marine Micropaleontology, vol. 5, no. 4, p. 329-399.

Cullen, J.L., and Prell, W.L., 1984, Planktonic foraminifera of the northern Indian Ocean: Distribution and preservation in surface sediments: Marine Micropaleontology, vol. 9, p 1-52.

Dowsett, H.J., 1991, The development of a long-range foraminifer transfer function and application to Late Pleistocene North Atlantic climatic extremes: Paleoceanography, vol. 6, no. 2, p. 259-273.

Dowsett, H.J., and Robinson, M.M., 1997, Application of the Modern Analog Technique (MAT) of sea surface temperature estimation to Middle Pliocene North Pacific planktonic foraminifer faunas: Palaeontologia Electronica, vol. 1, no. 1 [http://www-odp.tamu.edu/paleo/1998_1/dowsett/issue1.htm].

Dowsett, H.J., Barron, J., and Poore, R., 1996, Middle Pliocene sea surface temperatures: A global reconstruction: Marine Micropaleontology, vol. 27, nos. 1-4, p. 13-25.

Hutson, W.H., 1979, The Agulhas Current during the Late Pleistocene: Analysis of modern faunal analogs: Science, vol. 207, no. 1, p. 64-66.

Hutson, W.H., and Prell, W.L., 1980, A paleoecological transfer function, FII-2, for Indian Ocean planktonic foraminifers: Journal of Paleontology, vol. 54, no. 2, p 381-399.

Imbrie, J., 1985, A theoretical framework for the Pleistocene ice ages: Journal of the Geological Society of London, vol. 142, no. 3, p. 417-432.

Imbrie, J., Hays, J.D., Martinson, D.G., McIntyre, A., Mix, A.C., Morley, J.J., Pisias, N.G., Prell, W.L., and Shackleton, N.J., 1984, The orbital theory of Pleistocene climate: Support from a revised chronology of the delta 18O redord, in Berger, A.L., Imbrie, J., Hays, J., Kukla, G., ans Saltzman, B. (eds.), Milankovitch and climate, pt. 1, Understanding the response to astronomical forcing: Dordrecht, D. Reidel, NATO ASI Series, vol. 26, pts. 1, 2, p. 269-305.

Imbrie, J. and Kipp, N.G., 1971. A new micropaleontological method for paleoclimatology: Application to a Late Pleistocene Caribbean core. The Late Cenozoic Glacial Ages. New Haven, Yale University Press. 71-181.

Kipp, N.G., 1976, New transfer function for estimating past sea-surface conditions from sea-bed distribution of planktonic foraminiferal assemblages in the North Atlantic: Investigations of late Quaternary Paleoceanography and Paleoclimatology: Geological Society of America Memoir 145, p. 3-41.

Ortiz, J., and Mix, A., 1997, Comparison of the Imbrie-Kipp transfer function and modern analog temperature estimates using sediment trap and coretop foraminiferal faunas: Paleoceanography<, vol. 12(2): 175-190.

Ortiz, J., Mix, A., Hostetler, S., and Kashgarian, M., 1997, The California Current of the last glacial maximum: Reconstruction at 42° N based on multiple proxies: Paleoceanography, vol. 12, no. 2, p. 191-205.

Overpeck, J.T., Webb, T., III and Prentice, I.C., 1985. Quantitative interpretation of fossil pollen spectra: Dissimilarity coefficients and the method of modern analogs: Quaternary Research, vol. 23, no. 1, p. 87-108.

Parker, F.L., and Berger, W.H., 1971, Faunal and solution patterns of planktonic foraminifera in surface sediments of the South Pacific: Deep Sea Research, vol. 18, no. 1, p. 73-107.

Poore, R., Dowsett, H., Barron, J., Heusser, L., Ravelo, C., and Mix, A., in press, Multiproxy record of the last interglacial off central and northern California, USA: ODP Sites 1018 and 1020: Paleo 3.

Poore, R., Dowsett, H., Barron, J., Heusser, L., Ravelo, C., Mix, A., and McMahon, A., 1999, Microfossil and stable isotope data from the last interglacial records of ODP 1018 and 1020: U.S. Geological Survey Open File Report OFR-99-397, 19 pp.

Prell, W.L., 1985, The stability of low-latitude sea-surface temperatures: An evaluation of the CLIMAP reconstruction with emphasis on the positive SST anomalies: Washington, D. C., Department of Energy, Office of Energy Research, Office of Basic Energy Sciences, Carbon Dioxide Research Division, DOE/ER/60167-1 (under contract DE-ACO2-83ER60167), 60 p.

PRISM, 1994, PRISM 8° x 10° Northern Hemisphere paleoclimate reconstruction: Digital data: U.S. Geological Survey Open File Report 94-281, 23 p.

Reynolds, R.W. and Smith, T.M., 1995, A high-resolution global sea surface temperature climatology: Journal of Climate, vol. 8, no. 6, p. 1571-1583.

Thompson, P.R., 1976, Planktonic foraminiferal dissolution and the progress towards a Pleistocene equatorial Pacific transfer function: Journal of Foraminiferal Research, vol. 6, no. 3, p. 208-227.>

_________1981, Planktonic foraminifera in the western North Pacific during the past 150,000 years: Comparison of modern and fossil assemblages: Palaeogeography, Palaeoclimatology, Palaeoecology, vol. 35, nos. 2-4, p. 241-279.

Approved for publication October 6, 1999.

Return to Bulletins

This site is https://pubs.usgs.gov/bulletin/b2171/
Maintained by the Eastern Publications Group Web Team
Last revised 10-07-99