Geochemistry, Geophysics, Geosystems

Review of alkenone calibrations (culture, water column, and sediments)



[1] I review the status of alkenone unsaturation parameters as proxies for growth temperature and the potential for pCO2 estimates using the carbon isotopic composition of alkenones. Culture studies allow investigators to manipulate parameters that could influence the relative proportions of C37 and C38 ketones in producing organisms and to assess the possible influence of genetic variability on the calibration of unsaturation to growth temperature. Culture studies suggest large variability in the relation of unsaturation indices to growth temperature and hint that growth rate may play a role along with temperature in controlling unsaturation. Water column studies match the unsaturation index measured in particulate matter to ambient temperature, providing a snapshot of unsaturation-temperature relationship. In comparison to culture work, assemblages of regional water column data sets give more consistent relations of the unsaturation index to temperature, although the argument can be made for regional rather than basin-wide or global calibrations. A much simpler picture emerges from analyzing a global array of core top sediments. Regressions show that the sedimentary unsaturation indices follow a linear trend that is best correlated to mean annual temperature in the upper 10 m of the overlying water column. The regression is indistinguishable from the original Prahl et al. [1988] culture calibration of the U37k′ index to growth temperature. Exceptions occur at high latitudes, where the effects of highly seasonal production and the possible influence of salinity may affect the unsaturation index in sediments. Core top analyses also fail to reveal the variability of other alkenone parameters (for example, the ratio of C37 to C38 ketones) which have been found in culture studies. Much of the variance that appears in culture and water column calibrations is thus apparently removed in sediments, perhaps by temporal averaging. The promising paleo-pCO2 approach based on the isotopic composition of alkenones awaits the arrival of a significant number of core top and stratigraphic studies that will determine the extent to which physiological factors compromise the alkenone paleobarometer.

1. Introduction

[2] Our goal is to explain how sedimentary data encode paleoenvironmental properties. This simple statement helps us weigh the significance of various calibration studies that relate alkenone parameters to both environmental and genetic variability of the producing organisms. Looking upward from sediments, we view data from sediment traps, from particulates in the water column, and from cultures as models for the record ultimately buried in sediments. Growth temperature, salinity, pCO2, growth rate, and the species and strains of alkenone producers present at the time of deposition may influence this record. Various types of studies generate hypotheses on the relative importance of these and other variables in setting alkenone parameters. Only the sedimentary record, however, can allow us to determine the plausibility and reliability of insights developed from process studies. This review will conclude that many hypotheses generated by model systems above the sediment column fail to explain the apparent simplicity in what is now a large inventory relating alkenone unsaturation indices in recent marine sediments to near-surface ocean temperatures.

[3] In the main body of this paper, I will discuss consistencies and conflicts in our understanding of the relation of alkenone unsaturation to upper ocean temperature, proposed by Brassell et al. [1986] to be a useful paleoceanographic proxy. The application of alkenones to paleotemperature estimation leads the paleo-pCO2 approach considerably; I therefore devote a short concluding section to the latter proxy. I define six significant questions related to the paleotemperature proxy which can be approached in different ways by culture, water column, sediment trap, and core top studies.

[4] 1. Is there a single equation appropriate to the global ocean relating U37k′ to growth temperature? The use of “growth temperature” rather than sea surface temperature (SST) is intentional. The promise of the U37k′ proxy is the mechanistic link of the index to the temperature of the water in which the producing organisms grew. Identified factors which could undermine the use of a global equation are genetic differences between strains of Emiliania huxleyi and Gephyrocapsa oceanica and physiological state (logarithmic growth/late stage/stationary stage).

[5] 2. How predictably are growth temperature and SST related? We allow the possibility that growth temperature may vary from SST when and where significant alkenone synthesis (restricted, as far as is known, to a subset of haptophyte algae) occurs below the ocean mixed layer. Exactly how this temperature relates to SST is a question of haptophyte ecology, not of the immediate calibration of the index. We allow the possibility that growth temperature may vary from SST when and where significant alkenone synthesis (restricted, as far as is known, to a subset of haptophyte algae) occurs below the ocean mixed layer. Further, if fluxes of alkenones vary during the annual cycle, as they almost inevitably must, we must determine by how much this seasonal weighting biases the alkenone index away from recording a specified oceanic property such as mean annual SST.

[6] 3. Does the calibration of the U37k′ index become nonlinear near the extremes of 0.0 and 1.0 of the index? If this is case, then we will have to work hard to pin down the details of the calibration curve in these areas or admit significant uncertainties in paleoceanographic reconstructions in very warm and cold waters.

[7] 4. Does the tetraunsaturated ketone vary in response to temperature and/or salinity?

[8] 5. Are there markers within the alkenone/alkenoate system that allow us to distinguish variations in strains/species, growth phase, and physiological state? The typical suite of alkenone biomarkers includes C36 alkenoates (diunsaturated and triunsaturated) and C38 ethyl and methyl esters (diunsaturated and triunsaturated). Should these be used these to modify our interpretation of the better-known C37 sedimentary alkenone variations?

[9] 6. To what degree does the hypothesized preferential loss of the C37:3 ketone in early diagenesis compromise the alkenone paleotemperature proxy (negative view given by Prahl et al. [1989] and Teece et al. [1994]; positive view given by Hoefs et al. [1998])?

[10] One can approach these questions at considerably different scales of time, genetic, and environmental variability. Culture studies allow the experimentalist to eliminate many confounding variables in the natural environment (genetically mixed populations, varying nutrient availability, different depth habitats, etc.) to isolate the influence of factors such as growth temperature, physiological state (exponential versus late log versus stationary growth), and nutrient availability on alkenone and alkenoate systematics. Despite the elegance of the experimental approach, results from such studies must be viewed as models for what may exist under natural conditions, not necessarily as calibrations. By collecting particulate material in the water column, either by filtering material in the euphotic zone or collecting falling particles in sediment traps and relating their alkenone composition to time series of near-ocean temperatures, we move closer to relating the growth environment of the alkenone-producing algae to the signals sent to the sediment. This comes at the cost of unknown genetic variability in the natural populations and some ambiguity in the actual time and depth of alkenone synthesis and hence the appropriate growth temperature-alkenone/alkenonate relations. Near-surface sediments now provide a global database for examining the paleoenvironmental information contained in the preserved record of alkenones and alkenoates. The time averaging inherent in sedimentation means that one is integrating temporal, physiological, and genetic influences on scales not approached in the laboratory or in field studies. Correlations to environmental variables rely on statistical relations rather than demonstrable processes.

[11] As this review will make clear, significant barriers remain to uniting these perspectives on alkenone and alkenoate synthesis, transport, and deposition into a coherent framework. I trace below how different analytical perspectives answer the questions posed above, in some cases successfully and in other cases not.

2. Approaches to Calibrating the Influences of Temperature, Genetic Variability, and Growth Rate on Alkenone and Alkenonate Parameters

2.1. Culture Studies

[12] Culture studies allow the experimentalist to vary factors that may influence alkenone and alkenoate distributions under controlled conditions. Experience shows that rather different results can be obtained from using batch or continuous culture methods on the same strain of alkenone-producing algae [Popp et al., 1998], from the phase of growth from which alkenones are harvested [Conte et al., 1998; Epstein et al., 1998, unpublished data, 1999], and from different laboratories culturing the same strain (see results of culturing E. huxleyi strain VAN556 by Conte et al. [1998] compared to Prahl et al. [1988] or to data presented by Sawada et al. [1996] showing differences between two laboratories). In batch culture a nutrient medium is provided to a strain inoculate. After a period of rapid (exponential) growth, the cell density approaches a limit and may even decline. The investigator can harvest cells at various times during the sequence to determine alkenone and alkenoate concentrations. Continuous cultures maintain the organisms in the exponential growth phase by supplying nutrients. Growth of alkenone-producing haptophyte algae in chemostats [Popp et al., 1998] represents a particularly sophisticated manipulation, as these cultures grow in a medium of constant (low) nutrient availability. It is not immediately clear whether batch or continuous growth models better represent natural conditions or whether the sinking flux of alkenones and alkenoates in the ocean comes from populations in exponential, late logarithmic, or stationary growth state.

[13] All culture studies confirm the first-order dependence of U37k′ and other alkenone [Conte et al., 1998] unsaturation parameters on growth temperature but produce results that conflict in many ways. The study of Prahl et al. [1988] demonstrated that haptophytes adjust their unsaturation to temperature changes on a timescale of days; culture work by Conte et al. [1998] suggested rapid adjustment of alkenoate/alkenone ratios to changes in growth temperature as well. However, culture calibration studies suggest very large variations in the relation of unsaturation to growth temperature that may depend on genetic and physiological factors [Conte et al., 1995, 1998]. Twenty-four strains of alkenone-producing species cultured by Conte et al. [1995] at 15°C gave U37k′ values that ranged from 0.3 to 0.55. Only one of these approached the value of ∼0.56 appropriate for the Prahl et al. [1988] temperature calibration, and indeed, Conte et al. [1995] obtained a U37k′ value of ∼0.4 at 15°C for VAN55, the strain used by Prahl and Wakeham [1987] and Prahl et al. [1988]. Volkman et al. [1995] suggested that cultures of G. oceanica produce a significantly different relation of unsaturation to growth temperature; however, these results do not agree with G. oceanica cultures grown by Sawada et al. [1996] or Conte et al. [1998]. Figure 1 gives a suggestion of the large range of temperature equations generated by culture studies to date.

Figure 1.

Representative regressions of U37k′ versus growth temperature obtained by different laboratories for strains of Emiliania huxleyi and Gephyrocapsa oceanica. Note especially different results obtained by Prahl et al. [1988] (denoted by 1) and Conte et al. [1998] (2) on the same strain of E. huxleyi and by Sawada et al. [1996] (3) and Volkman et al. [1995] (4) for G. oceanica. Conte et al., [1995, 1998] report additional relations from strains of E. huxleyi, not shown here.

[14] It also seems clear that the growth phase significantly influences the unsaturation index of haptophytes grown in culture. Both Conte et al. [1998] and Epstein et al. [1998] found changes in the unsaturation index between log, late log, and stationary phases of growth. Epstein et al. [1998] proposed that nutrient availability, which would control growth rates of cultured and natural populations, could significantly affect the calibration of unsaturation to growth temperature. Both investigations found increasing alkenone concentrations (pg cell−1) in late logarithmic and stationary phase growth. Conte et al. [1998] also documented very large ranges in the ratios of alkenoates to C37 and C38 ketones (0.–2.8) and in the ΣC37/ΣC38 ketone ratio depending on growth phase. Comparison of these parameters to field data led Conte et al. to conclude that natural populations most closely resemble late log of stationary populations grown in the laboratory. These results were obtained from batch culture. Popp et al. [1998] used chemostats to control steady state growth rates, which they argue may be a better model for natural systems. The latter study found no significant dependence of U37k′ to growth rate at constant temperature.

[15] A number of studies have investigated whether genetic and/or physiological differences create “fingerprints” in other aspects of alkenone/alkenoate systematics that might allow investigators to distinguish past variations in species/strain production in sediments. Volkman et al. [1995] and Sawada et al. [1996] suggested that the proportions of C37 to C38 ketones, or alkenoate/alkenone ratios, might relate to the proportions of E. huxleyi to G. oceanica at the time of production. Further culture work by Conte et al. [1998] does not support either suggestion (see also section 2.4).

[16] Nearly all culture calibration studies predict higher growth temperatures than postulated by the Prahl et al. [1988] regression for the same unsaturation index, although several (G. oceanica cultures of Sawada et al. [1996] and Conte et al. [1998]) fall very close to the Prahl et al. relation. If the apparent bias of many strains of E. huxleyi and G. oceanica shown in Figure 1 is correct, field and sediment studies applying the Prahl et al. [1988] relation might frequently overestimate temperatures. As we evaluate water column, sediment trap, and core top data, we should assess whether the large range of possible U37k′-temperature relations suggested by culture studies, whether of genetic or physiological origin, demonstrably affect the accuracy of a unified calibration relation. One would expect to find that different haptophyte biogeographic regions produce distinct calibrations of unsaturation to temperature and to see the influence of nutrient availability in offsets between upwelling and nonupwelling regions.

[17] The following questions arise. (1) Can we improve our knowledge of the genetic variability of natural assemblages? Can we associate better regional field calibrations with strains grown in culture to determine whether proposed genetic controls suggested by cultures explain deviations (if such exist) from a global calibration of to growth temperature? (2) What type of growth most represents the natural environment? Does the flux of alkenones to sediments derive from populations in exponential, late log, or stationary state? (3) Do field and sediment studies suggest that batch or continuous cultures are more appropriate models for the natural environment? How does the variability observed in culture, not only of the U37k′ index but other unsaturation ratios, and in the relative proportions of C37 and C38 ketones and in C36 alkenoates compare to natural variability? Can we identify what must be done to make alkenone systematics in cultures more closely resemble field and sediment data?

2.2. Particulate Material

[18] By studying alkenone parameters in particulate matter collected in the photic zone we lose the ability to manipulate potential genetic or physiological influences, but we gain the ability to compare alkenone systematics to temperatures in the natural setting. One can also develop an understanding of how the depth of alkenone production and the seasonal abundance of alkenones in the upper water column can vary in the ocean. One generally assumes that the measured water temperature is the same as the temperature in which the alkenones and alkenoates were synthesized. This may not always be correct for particles sinking through a temperature-stratified water column. In fact, while the temperature of alkenone synthesis might be higher than the temperature in which the particles are collected, it could almost certainly not be lower. A potential temporal offset also exists between the time of alkenone synthesis and the measurement of ambient water temperatures.

[19] In contrast to culture studies, relationships between U37k′ and temperature in suspended particulate organic carbon show much more agreement with the Prahl et al. [1988] calibration equation (Figure 2). Several large-scale compilations of water column unsaturation ratios have been published [Brassell, 1993; Sikes et al., 1997] as well as regional studies in the North Atlantic and Mediterranean [Conte et al., 1992; Conte and Eglinton, 1993; Sikes and Volkman, 1993; Ternois et al., 1997]. Data have been variously interpreted as requiring regional calibrations of growth temperature [Conte and Eglinton, 1993; Ternois et al., 1997] or as requiring modifications of the original Prahl et al. [1988] culture-based U37k′ equation to a more appropriate relation based on a water column particulates equation [Brassell, 1993; Sikes and Volkman, 1993]. The arguments favoring regional calibrations are based on regression estimates of a small number of samples and are not yet compelling. Most values of U37k′ observed in particulate organic matter do not deviate by more than a few degrees from the expected Prahl et al. [1988] relation but generally lie in the field of warmer-than-predicted growth temperatures at a given U37k′. In the case of the Black Sea, however, Freeman and Wakeham [1992] determined a water column U37k′ that would underestimate SST by >5°C. It now appears that the Black Sea represents a special case of mixing of brackish water haptophyte alkenone-producing algae with open ocean varieties.

Figure 2.

Reported regression equations relating U37k′ measured on filtered particulates or sediment trap material to water temperature. 1, Sikes and Volkman [1993]; 2, Conte and Eglinton [1993]; 3, Goni et al. [2001]; 4, Prahl and Wakeham [1987]; 5, Ternois et al. [1997]; 6, Prahl et al. [1988] culture equation for reference.

[20] The apparent systematic difference in slope of the U37k′-temperature relationship between water column particulates [Ternois et al., 1998] and core top sediments [Muller et al., 1998] merits more attention. An example of this unsatisfying link comes from Cacho et al.'s [1999] comparison between surficial sediment data and suspended particles in the western Mediterranean Sea. The authors find a close agreement between spring suspended particulate U37k′ values and water column temperatures according to the Ternois et al. [1997] regional calibration. However, surficial sediment U37k′ values closely agree with the annual average near-surface temperatures that one would obtain with the Prahl et al. [1988] culture calibration or the Muller et al. [1998] global core top calibration.

[21] The following questions arise. (1) Can systematic relations be developed between alkenone parameters and physiological state in field studies? Answering this will require time series studies through blooms and the seasonal cycle of production in different regions. (2) What conditions control the actual export of alkenone material produced in the euphotic zone to depth, and how do these influence the alkenone signature imparted to sediments? (3) How does the variability of alkenone parameters other than U37k′ compare to culture and sediment studies? (4) What environmental variable controls the abundance of the C37:4 ketone, which appears to have little relation to temperature [Sikes and Volkman, 1993]? Virtually no measurements on water column particulates exist in regions of the high-latitude Northern Hemisphere where the C37:4 ketone appears in abundance in surficial sediments [e.g., Rosell-Mele, 1998].

2.3. Sediment Traps

[22] Sediment trap material provides a valuable view of the U37k′ and quantity of alkenones transiting to the seafloor. Few time series experiments have been reported to date, although data from the Gulf of California [Goni et al., 2001] and off the coast of Angola will soon appear. Ternois et al. [1997] synthesized the results from a shallow sediment trap with water column material to assess potential biases in the alkenone unsaturation index as functions of variations in the depth and season of production in the Mediterranean Sea. Their results demonstrate that alkenone flux can vary significantly within the year. During late spring and early summer, sediment trap U37k′ values gave colder-than-surface temperatures, consistent with water column observations of subsurface production. A high flux of “warm” U37k′-bearing alkenones in the late fall coincided with an episode of high productivity within the mixed layer. A several year follow-up study confirms this seasonal bias [Sicre et al., 1999].

[23] There are some indications that the U37k′ value found in sediment traps and water column particulates may reflect in some instances “relict” material not representative of current near-surface temperatures [Thomsen et al., 1998; Sicre et al., 1999]. At least some of the scatter in water column data may reflect a memory effect of previously synthesized alkenones lingering in a form out of equilibrium with temperature. Thomsen et al. [1998] have compared U37k′ in sediment traps at 1840 and 1960 m in the Norwegian Sea to demonstrate the presence of reworked alkenones with anomalous U37k′ values in the lower trap. The lower trap presumably intercepted material suspended near the seafloor and transported down slope. Another memory effect may come from variable heterotrophic pressure, which may cause a time offset between alkenone production and final flux out of the photic zone [Sicre et al., 1999].

[24] The following questions arise. (1) Can an array of sediment traps be used to assess biases in the annually integrated U37k′, and can these biases be modeled in a predictive way? (2) Can coccolith species or morphotype variability in time series sediment traps be correlated to “anomalies” in unsaturation ratio or other alkenone/alkenoate parameters? (3) How well do sediment trap alkenone parameters agree with core top values in the same region?

2.4. Core Tops

[25] Core top material provides the benefit of temporal and spatial averaging of all the factors that may influence alkenone systematics. This comes at costs: the time averaging varies with sedimentation rate and bioturbation, and much information related to the original production of alkenones in the surface ocean is lost.

[26] One cannot, however, fail to be impressed by the systematic distribution of core top U37k′ and other alkenone parameters in the ocean (Figure 3). Large data sets [Rosell-Mele et al., 1995; Sonzogni et al., 1997; Herbert et al., 1998] show strong convergence with the original Prahl et al. [1988] temperature calibration, using mean annual surface (0–10m) as the temperature reference. The most recent compilation by Muller et al. [1998] synthesized results of over 300 core top analyses from the different ocean basins, determined by various laboratories. The resulting calibration does not differ statistically from the original Prahl et al. [1988] culture and water column line. Sediment U37k′-temperature relations scatter less than water column calibrations of the unsaturation index to ambient temperature and far less than the diversity seen in cultures (Figure 1). A statistical comparison now needs to be performed to determine whether core top and water column regressions to temperature differ beyond the uncertainties of regression parameters.

Figure 3.

Updated global core top calibration of Muller et al. [1998], courtesy of P. Muller. The regression of U37k′ to mean annual surface (0–10 m) temperature is statistically identical to the Prahl et al. [1988] culture calibration of unsaturation to growth temperature.

[27] The success of core top temperature calibrations indicates that physiological state, genetic variability, and depth and seasonality of production play secondary roles to the control on the sedimentary U37k′ index exerted by mean annual near-surface temperature. In most cases, these factors produce errors at the level of 1.5°C or less in the global core top calibration. In this observer's opinion, core top data cannot be reconciled with the large variations in the U37k′ index attributed to genetic or physiological factors by some culture studies. This does not indicate that the culture data are wrong in a technical sense, but that their results cannot always be extrapolated to the natural environment [Popp et al., 1998].

[28] The core top data do permit one to observe the influences of seasonal and subsurface production suggested by water column and sediment trap studies [Prahl et al., 1993; Ternois et al., 1997], although these generally amount to deviations from mean annual SST of no more than 1°–2°C. Surficial sediments should reflect the weighting function of alkenone production throughout the annual cycle, the integrated production temperature (IPT) concept of Conte et al. [1992]. Ternois et al. [1996] and Sicre et al. [1999] report that core top U37k′ values in the western Mediterranean are consistent with an annually integrated average weighted toward autumnal season of maximum production in the study area. U37k′ values from Southern Ocean sediments suggest that most alkenone production and flux originates in the summer months [Sikes et al., 1997]. U37k′ values from sediments underlying regions of seasonally varying upwelling conditions at midlatitudes show surprisingly little influence of seasonality, however [Herbert et al., 1998]. Evidence for subsurface alkenone production comes from gyre locations. For example, Prahl et al. [1993], Doose et al. [1997], and Herbert et al. [1998] all report core top U37k′ values lower than mean annual SST in gyre locations in the eastern North Pacific, consistent with subsurface fluorescence maxima [Prahl et al., 1993], and maximal production during the late winter and early spring in the region. Ohkouchi et al. [1999] obtained a similar gyre bias in a survey of core tops in the central North Pacific Ocean.

[29] Little support for the large range in alkenone parameters observed in culture experiments comes from the sediment realm. As one example, Figure 4 plots the frequency distribution of the ratio of total C37 ketones (ΣC37) to total C38 ketones (ΣC38) in a core top data set generated by our laboratory. The data set is weighted to samples along the California margin but also includes samples from the North Atlantic, equatorial Pacific, central Pacific gyre, Peru margin, and western Pacific. The tight cluster of core top values around a mean of just over 1.0 contrasts with the extreme heterogeneity of culture results (Figure 4). The values we report here are similar to the mean of ∼1.2 determined by Rossell-Mele et al. [1994] in North Atlantic core top and by Sonzogni et al. [1997] in Indian Ocean sediments and the average of ∼1.15 reported by Sawada et al. [1996] from the Sea of Japan. Core top data do not therefore encourage the idea that regional variations in strain composition or the fraction of production due to E. huxleyi and G. oceanica [cf. Volkman et al., 1995; Sawada et al., 1996] can be distinguished by alkenone systematics.

Figure 4.

Comparison of the chain-length index (ΣC37/ΣC38) of alkenones in core top material analyzed at Brown University. Sediment data cluster closely around a value of 1.0, unlike the variability in the same parameter observed in culture experiments (arrows indicate mean values of culture data).

[30] The recently published study by Hoefs et al. [1998] challenges the general assumption that alkenone unsaturation indices remain unchanged during early diagenesis in sediments. The authors measured a preferential loss in the C37:3 ketone in the oxidation front of a turbidite that reexposed organic matter to oxic degradation. This study flies in the face of a similar study by Prahl et al. [1989] that found no significant change in the U37k′ index with diagenesis. Laboratory degradation experiments [Teece et al., 1994] and studies of the U37k′ index during herbivory [Grice et al., 1998] also fail to demonstrate appreciable alteration of the biomarker ratios as alkenones are processed through the food chain. The Hoefs et al. [1998] study may need to be reconsidered in light of a recent analytical critique by Grimalt et al. [2000] that suggests that very low alkenone concentrations may cause chromatographic artifacts that bias the accurate determination of the U37k′ index in the Hoefs et al. [1998] material.

[31] Core top data sets still leave significant room for improvement in several important regards. The important question of nonlinearity at the high and low ends of the U37k′-temperature relation remains unsettled [Sikes et al., 1997; Sonzogni et al., 1997], although the global sediment compilation of Muller et al. [1998] provides no support for a nonlinear relationship. One would expect that calibrations on the extreme ends of the temperature range become difficult. Analytical difficulties grow as the diunsaturated ketone becomes a minor peak at the low end of the index; similar difficulties pertain to the detection of the C37:3 ketone in the face of chromatographic interferences in sediments under very warm ocean surface waters. In addition, it also likely that seasonal production biases become very large in high-latitude waters, emphasizing the need to combine modern-day ecological information with core top data. We also have little consensus on the temperature significance of the tetraunsaturated C37 ketone [see Rosell-Mele et al., 1995; Rosell-Mele, 1998; Sikes and Volkman, 1993; Sikes et al., 1997; Ternois et al., 1998].

[32] The following questions arise. (1) Can one demonstrate biases in the core top unsaturation parameter consistent with large-scale variations in the season of maximum production derived from satellite observations? (2) Can one detect differences in alkenone unsaturation that can be related to the relative abundances of E. huxleyi versus G. oceanica coccoliths in the sediment? Initial indications of core top [Herbert et al., 1998] and paleoceanographic studies [Muller et al., 1997] suggest not. (3) How are alkenone parameters (U38mek, U38ethk, AA/LCK, etc.) other than U37k′ distributed in sediments [see Conte and Eglinton, 1993; Rosell-Mele et al., 1995]? (4) How do alkenone concentrations and fluxes in sediments correspond with estimates of haptophyte productivity in surface waters? (5) Can we identify key supporting information in sediments which will help us correct potential biases in the alkenone index as a proxy for SST?

3. Calibration of the Alkenone Paleo-pCO2 Proxy

[33] The carbon isotope approach to determine paleo-pCO2 requires knowledge of multiple factors in addition to the isotopic composition of the alkenones. These include temperature, the δ13C of dissolved inorganic carbon (DIC), and the specific growth rate of the alkenone-producing organisms. Most of our understanding of the proxy comes from laboratory culture experiments. Conspicuously lacking are quantities of sediment trap and core top data to assess the reproducibility of our methods of paleo-pCO2 reconstruction in the modern ocean and to analyze potential biases which might affect sediment records. One would like to document that the method is accurate to better than 80 ppm, the approximate range of glacial-interglacial CO2 fluctuations known from ice cores. One problem is that surface pCO2 gradients are fairly small in the modern ocean, and regions of anomalously high and low pCO2 may also have unusual nutrient properties (e.g., eastern equatorial Pacific) that may complicate calibrating the proxy. Despite the excitement generated by Pagani et al.'s [1999] reconstruction of pCO2 from 8 to 25 Ma the published record of recent sediments is limited to studies by Jasper and Hayes [1990] and Jasper et al. [1994]. While both records are intriguing, neither come from the oligotrophic, stable regions that would be ideal for pCO2 reconstruction. In the case of the equatorial Pacific record published by Jasper et al. [1994], the deep-dwelling foraminifera N. dutertrei provided the δ13C of DIC. Most paleoceanographers would not regard this species as ideal for reconstructing the surface carbon isotopic composition. Recent core top data indicate that the Cd/Ca ratio of planktonic foraminifera, proposed as a method of paleonutrient reconstruction, may instead record ocean temperatures [Rickaby and Elderfield, 1999]. In addition, paleoceanographers are currently evaluating the hypothesis that the seawater carbonate ion concentration can significantly affect planktonic foraminiferal δ13C [Spero et al., 1997].

[34] The following questions arise. (1) Can we constrain better the depth habitats of alkenone producers and foraminifera? These are important variables since there may be significant subsurface gradients in both pCO2 and the δ13C of DIC. (2) What are the “real-world” reproducibilities of the components of the paleo-pCO2 reconstruction? What do these uncertainties imply for errors in reconstructing past CO2 values? (3) Can core top data convincingly identify regions of undersaturation and supersaturation in the modern ocean? (4) Can sediment records consistently reproduce the 400 kyr record of past CO2 now available from the Vostok ice core?

4. Summary

[35] The most significant limitations to using alkenone paleotemperature proxies come from ecological factors not uncertainties in calibrations due to genetic or physiological factors. Compilations of water column (particulate) and core top sediment data favor global, linear relations of the U37k′ index to growth temperature, although one obtains slightly different calibrations depending on which type of data one chooses. The uncertainties come from the interpretation of when in the year and where in the water column the temperatures determined originate. I take the optimistic view that these are problems that we can solve and that they will not undermine the use of the organic proxy for most paleotemperature reconstructions. The alkenone pCO2 proxy has not arrived at a similar state of coherency and simplicity. One can hope that larger field and core top data sets will show, as was true for the unsaturation index, that potential complexities will not overwhelm the usefulness of this alkenone-based proxy.