Geochemistry, Geophysics, Geosystems

Ecological and environmental factors affecting alkenone distributions in seawater and sediments



[1] The use of alkenones to estimate past sea surface temperatures has gained wide acceptance, but there are still some issues that need to be resolved before alkenone proxies can be used confidently in all environmental settings. A particular concern is defining what seawater temperature is recorded by the alkenone distributions preserved in sediments. Does it, for example, reflect a seasonal average or a snapshot of bloom conditions or some other combination of environmental effects? This paper reviews some of the environmental and ecological factors that might affect alkenone distributions and, in particular, the influence of depth and seasonality on alkenone production in the water column.

1. Introduction

[2] The alkenones comprise an unusual class of long-chain unsaturated methyl and ethyl ketones that are synthesized by a limited number of haptophyte microalgae [Volkman et al., 1980a; Marlowe et al., 1984; Conte et al., 1994a, 1995; Volkman et al., 1995]. These include the coccolithophorids Emiliania huxleyi and Gephyrocapsa oceanica, which are the two most common sources of alkenones in today's oceans and contemporary sediments.

[3] Work by Brassell et al. [1986] established that the proportion of the two main alkenones 37:2 and 37:3 varied in a systematic way with temperature and this led them to devise a simple expression such that U37k = ([37:2] − [37:4])/([37:2] + [37:3] + [37:4]), where [37:x] is the concentration of the C37 alkenone having x double bonds. This equation was later simplified to U37k′ = [37:2]/([37:2] + [37:3]) since the 37:4 alkenone is relatively rare in sediments [Prahl and Wakeham, 1987; Prahl et al., 1988].

[4] The U37k′ index in sediments deposited over the past 200,000 years has proven to be a very useful indicator of seawater temperature at the time of deposition. However, there have only been a few attempts to extend its use to much older sediments despite the fact that alkenones have been found in sediments as old as 100 million years B.P. [e.g., Farrimond et al., 1986]. Measurement of the U37k′ index has now become routine in most research groups in paleoceanography leading to a dramatic expansion in the number of published papers containing alkenone data.

[5] Unfortunately, despite the fact that it is now over 20 years since alkenones were first identified, there are still many issues that remain unresolved. The occasional mismatch between alkenone-derived sea surface temperatures (SSTs) and those derived from other proxies such as oxygen isotopes, trace element ratios, or foraminiferal assemblages indicates that the U37k′ index, like all proxies, must be applied with due recognition of its limitations. Questions have been raised about the sources of alkenones in some environments (particularly in lacustrine or older marine sediments), the appropriateness of various published calibrations [e.g., Sikes and Volkman, 1993; Müller et al., 1998], possible effects due to degradation [e.g., Sun and Wakeham, 1994; Hoefs et al., 1998; Teece et al., 1998; Gong and Hollander, 1999], the influence of environmental variables other than temperature [e.g., Epstein et al., 1998], the role of alkenones in the microalgal cell (there is growing evidence that they might function as storage products), and even whether alkenones can be measured with sufficient accuracy by all laboratories particularly when present in small amounts in sediments [Rosell-Melé et al., 1999].

[6] This review paper provides ideas and information relevant to the topic “Primary signal: Ecological and environmental factors” identified for Working Group 2 of the National Science Foundation (NSF) sponsored workshop on “Alkenone-based Paleoceanographic Indicators” held at Woods Hole Oceanographic Institution during October 1999.

2. Discussion

[7] In order to gain a better understanding of the environmental and ecological factors responsible for the alkenone distributions found in today's oceans and Recent sediments it is convenient to subdivide the research question into several main topics.

[8] These include the following:

[9] • Haptophyte speciation and ecology in modern and ancient oceans

[10] • Biogeography of haptophytes

[11] • Population dynamics, depths, and habitats

[12] • Temporal variation (seasonal evolution)

[13] • Other environmental influences on alkenone abundance

[14] • Related biomarkers and possible new proxies

[15] From a consideration of the topics one would hope to arrive at an assessment of the utility of alkenone paleothermometry through geological time, including the pre-Quaternary. The working group must consider the what, where, how, and why of alkenone-producing haptophyte abundances and alkenone production over time in aquatic ecosystems. Are ecological factors responsible for some of the “anomalous” SST estimates from modern and ancient sediments and the mismatch between values derived from different proxies? A fundamental question is: What is the environmental significance of the alkenone-derived temperature recorded in the sediments?

2.1. Haptophyte Speciation and Ecology in Modern and Ancient Oceans

[16] The Haptophyta includes 11 genera of unmineralized species having ∼80 representatives and a further 70 genera containing more than 200 species that produce calcium carbonate coccoliths [Thomsen et al., 1994]. There has been a resurgence of research interest in these organisms in recent years, resulting in a much better knowledge of their speciation and ecology. The literature is too extensive to review here, but some of the salient features that should be of interest to geochemists and paleoceanographers are highlighted. In oceanic waters, haptophytes can comprise a large proportion of the nannoplankton biomass [e.g., Chavez et al., 1990] and give rise to extensive blooms [e.g., Birkenes and Braarud, 1952; Holligan et al., 1983; Blackburn and Cresswell, 1993; Brown and Yoder, 1994]. The occurrence of coccolith oozes in the geological record testifies to the importance of these microalgae in the global carbon cycle. Haptophyte fossils first become abundant in the early Jurassic and reached their greatest abundance in the Late Cretaceous. Near the end of the Cretaceous the coccolithophores suffered a mass extinction with two thirds of the 50 genera disappearing. However, many new groups appeared in the Paleocene.

[17] The most important haptophyte species in modern oceans include those known to produce alkenones, i.e., Emiliania huxleyi and Gephyrocapsa oceanica. These genera belong to the family Noelaerhabdaceae, which also includes Reticulofenestra for which there are two known living species and many fossil forms. This genus has been suggested as a possible source of alkenones in ancient sediments [Marlowe et al., 1990]. Species from the Noelaerhabdaceae are often an important component of calcareous assemblages in sediments deposited since the Cretaceous [e.g., Bollmann et al., 1998]. Other species that synthesize alkenones such as Isochrysis spp. and Chrysotila lamellosa [Marlowe et al., 1984] occur mainly in coastal areas and are thus usually not significant contributors of organic matter in contemporary open-ocean environments [Ziveri and Broerse, 1997].

[18] There may be other significant microalgal sources of alkenones in oceanic waters, but we have no direct evidence for this. In some coastal waters and certain open-ocean water bodies it is quite likely that other haptophyte species may be important sources of organic matter, but not of alkenones. For example, the haptophyte Phaeocystis pouchetii is common in many marine water bodies including the North Sea and Southern Ocean where it can occur in bloom proportions, but this alga does not synthesize alkenones [Conte et al., 1994a; J. K. Volkman, unpublished data, 1990]. Coccolithus pelagicus is the main coccolithophore in the polar regions of the Norwegian Greenland Sea and the Northern Pacific [e.g., Baumann et al., 1997]. An earlier study of this alga failed to find alkenones [Volkman et al., 1980b], but a more recent study identified small amounts, although further work is needed to prove that this was not due to culture contamination or analytical errors (P. Ziveri and H. Kinkel, personal communication, 1999).

[19] There is still a need for a systematic examination of the lipid compositions of species in the Haptophyta given that perhaps only 10% of living species have been examined. The study of ancient environments is even more difficult since alkenone producers in the geological past may no longer be extant. The fossil record of Emiliania only extends back perhaps 220,000 years, and so there must be other sources of alkenones in more ancient sediments. E. huxleyi became dominant in haptophyte populations only in the last 50,000 to 70,000 years [Flores et al., 1997]. Marlowe et al. [1990] suggested that species of Reticulofenestra might contain alkenones based on the co-occurrence of coccoliths and alkenones, but this has not been confirmed.

[20] The proposition that species other than Emiliania might be significant sources of alkenones in some environments has implications for the use of U37k′ as a temperature proxy. Culture work with a single strain of G. oceanica suggested a relationship between U37k′ and temperature different from that of E. huxleyi [Volkman et al., 1995], but more recent research suggests that this may not be a significant problem for interpretation of U37k′ values in sediments [Conte et al., 1998; Müller et al., 1997, 1998]. However, it is clear that the alkenone distributions in Isochrysis have a very different temperature response [Brown et al., 1993]. The degree of uncertainty associated with using an Emiliania-based calibration in environments where other species may be an important source of alkenones cannot be known until additional species and strains have been examined systematically.

[21] Different species of alkenone-producing haptophytes contain different distributions of alkenones, alkenoates, and alkenes. Volkman et al. [1995] suggested that it might be possible to use features of these distributions to infer the likely source of the alkenones. Some support for this view is provided by the work of Sawada et al. [1996], who showed significant differences in alkenone distributions in sediments where the main contributing haptophyte was either E. huxleyi or G. oceanica. This idea has been contested: Conte et al. [1998] suggested that the alkenone distributions of these species cannot be reliably distinguished owing to significant intraspecies variation such that their alkenone distributions overlap. This raises a general question as to the significance of genetic and physiological differences between cultured and natural populations of haptophytes in terms of alkenone distributions and responses to environmental conditions. Prospects for distinguishing alkenone sources in ancient sediments may be better since quite distinct distributions of alkenones have been observed [Farrimond et al., 1986], including the occurrence of C41 and C42 alkadienones not found in either E. huxleyi or G. oceanica. Further studies comparing alkenone distributions with other indicators of species identity (e.g., coccolith morphology, other biomarkers) would be valuable.

2.2. Biogeography of Haptophytes

[22] Models on the ecological successions of coccolithophores [e.g., Westbroek et al., 1993], indicate that the long-held view that coccolithophores live only in the oligotrophic areas of the tropical oceans is wrong. Many species, including alkenone producers, are able to respond to increasing nutrient levels (e.g., upwelling) and favorable conditions (such as surface water stratification). In fact, alkenone-producing haptophytes are found in almost all of the world's oceans, with the possible exception of Antarctic polar regions covered by seasonal ice judging from the lack of alkenones found in sediments from such regions in the Southern Ocean [Sikes et al., 1997]. However, the occurrence of an unnamed marine-derived, alkenone-containing haptophyte in waters of Ace Lake in Antarctica [Volkman et al., 1988] suggests that even here some marine sediments covered by seasonal ice might contain alkenones. E. huxleyi is known to bloom in Norwegian fjords and weak summer blooms are found in northern polar waters that are ice covered in winter [Samtleben and Schröder, 1992]. Other haptophytes are known from polar environments, such as the mixotroph Chrysochromulina [Marchant and Thomsen, 1994], but those species of this genus that have been analyzed to date have not contained alkenones [Marlowe et al., 1984].

[23] To use alkenone paleothermometry effectively, we require an integrated picture of haptophyte ecology that answers such questions as the following:

[24] • Where are they the dominant phytoplankton?

[25] • What environmental factors influence haptophyte speciation and abundance?

[26] • In which types of water bodies are haptophytes found?

[27] • What effect do blooms of haptophytes have on the carbon speciation in seawater that might affect carbon isotope signatures?

[28] • How do we account for the fact that some species have different zonation in the water column?

[29] Ultimately, we would like to be able to determine from the abundance and composition of alkenones and related compounds preserved in sediments a picture of primary production at the time of deposition. A more difficult task is to convert the content of preserved alkenones in the sediments into an estimate of the absolute abundance of alkenone-producing haptophytes in the original phytoplankton population. This requires a knowledge of alkenone:TOC ratios in the microalgae (for which too few data are available) plus a reliable way to estimate the effects of postdepositional degradation (which is unfortunately lacking).

[30] The question of geographical zonation is interesting because if different species only occur in specific environmental zones then a calibration appropriate for that species can be used. Thus, while both E. huxleyi and Gephyrocapsa are cosmopolitan species, the latter tends to be dominant in warm equatorial coastal waters [Sawada et al., 1996; Bollmann et al., 1998]. Interestingly, the morphology of Gephyrocapsa coccoliths, in particular, the bridge angle and placolith length, seems to related to environmental gradients, and these features have been used to study past climate changes [Bollmann, 1997].

2.3. Population Dynamics and Population Depths

[31] In many of the earlier applications of U37k′, there was an assumption that the alkenone signal recorded the temperature of surface waters. Where haptophyte abundance is highest in surface waters, this is obviously a reasonable assumption, but several studies have confirmed that this will not always be the case. Prahl et al. [1993] showed that alkenones in the northeast Pacific were produced at the subsurface chlorophyll maximum during periods of water stratification rather than in the surface mixed layer. Ohkouchi et al. [1999] reported alkenone data for surface sediments taken along a latitudinal transect along 175°E from 48°N to 15°S covering a temperature range from 28.3°C in the tropical Pacific to 10.1°C toward 48°N. From the latitudinal trend they concluded that in midlatitudes (35°–19°N) the alkenones are produced in the thermocline waters, whereas in the high (48°–40°N) and low (10°N–2°S) latitudes they are produced in the surface mixed layer. Furthermore, Bentaleb et al. [1999] showed that alkenone production in the northwestern Mediterranean was highest at 30-m depth where primary production was highest. Interestingly, they demonstrated that the U37k′ signal was also the same at 5 and at 1100 m, indicating that the alkenones are mixed through the water column. Such observations demonstrate the need for a better understanding of how alkenone production varies with water depth in different oceans and whether such variations can be modeled simply. Improvements in the interpretation of temperature records from ancient sediments will probably require knowledge of paleolatitude and, perhaps, of other environmental factors.

[32] Some alkenone studies have led to widely different calibrations of U37k′ against temperature or have even demonstrated a lack of correlation. These include the eastern North Atlantic [Conte et al., 1992], Black Sea [Freeman and Wakeham, 1992], and northwestern Mediterranean [Ternois et al., 1997]. We need to determine whether different calibrations are really needed for different water bodies, as suggested by several authors [e.g., Sikes and Volkman, 1993; Cacho et al., 1999], or whether these discrepancies result from other factors such as variation in depth and seasonality of alkenone production. Two schools of thought are evident: One school accepts a general worldwide calibration such as that of Müller et al. [1998] and then interprets anomalous sediment U37k′ data as evidence for alkenone production at depth in the water column. The other asserts that different calibrations apply in different regions. Unfortunately, it is rare that sufficient data are available from water samples collected throughout the photic zone to determine the reasons for such anomalous U37k′ values in sediments [e.g., Ternois et al., 1997; Cacho et al., 1999].

2.4. Temporal Variation (Seasonal Evolution)

[33] Abundances of haptophytes, like other microalgae, vary seasonally. Moreover, the timing of maximum production varies in different parts of the world's oceans. The influences of such variations on U37k′ values have been evaluated in a limited way [e.g., Prahl et al., 1993; Sikes et al., 1997; Ternois et al., 1997; Bentaleb et al., 1999], but much more study is needed. In my view, data on temporal and spatial variations are essential to answer the fundamental question: “What temperature is recorded in the sediments?”

[34] Globally, U37k′ values tend to follow the annual mean of the surface water temperature [Müller et al., 1998], but this could be an artifact of the treatment of the data. It might be that alkenone-derived temperatures are close to the yearly average temperature in many regions simply because the spring temperature is often close to the yearly average (H. Kinkel, personal communication, 1999). At higher latitudes, high production is generally limited to the spring and summer. The effect of this pulsed production on sedimentary U37k′ values is illustrated by data presented by Sikes et al. [1997] from the Southern Ocean (Figure 1).

Figure 1.

Values of U37k′ determined in the sediment (solid symbols) and water column (shaded symbols) in the Southern Ocean, as determined by Sikes et al. [1997]. (a) Plot of the water column U37k′ values versus the temperature measured in those samples plus the sediment data versus summer SST for the overlying water [Levitus, 1994]. Dotted line is the linear correlation of Sikes and Volkman [1993] for the water column and solid line is the regression of the sediment data: the two are not significantly different. (b) Sediment data plotted versus winter SST [Levitus, 1994]. The solid line is the linear regression for the sediment data; the water column data and regression are the same as in Figure 1a. The water column and sediment regression are significantly different if the winter season is assumed to represent the sediment U37k′ values, thus providing support for the view that the U37k′ values in the sediment represent the period of maximum production of alkenones in late spring-summer rather than an annual mean temperature. Reprinted with permission from Elsevier Science.

[35] Sikes et al. [1997] found that the regression of sedimentary U37k′ against winter seawater temperatures derived from the Levitus [1994] data set was quite different from the seawater U37k′-T regression derived from actual water column samples determined previously [Sikes and Volkman, 1993]. However, there was a good correlation with summer water temperatures, consistent with the view that most of the alkenone production recorded in the sediments occurs in the warmer months of late spring and summer.

[36] In temperate regions, where two peaks in the annual production can occur in spring and in autumn, the alkenone-derived temperature should be the weighted mean of seawater temperatures at the two production peaks which, coincidentally, might approximate to the annual mean temperature. Production will also increase with upwelling or other processes that mix nutrient rich (and mostly colder) waters into the upper photic zone. Attempts to use sedimentary U37k′ values in such settings to follow yearly variability in seawater temperatures associated with El Niño events have not been very successful [e.g., McCaffrey et al., 1990]. Despite apparently good correlations of U37k′ with mean upper water temperatures in global data sets, these seasonality factors cannot be ignored in site-specific studies.

[37] Do we need to consider the mechanism by which alkenones reach the seafloor? If it is by the settling of zooplankton faecal pellets, then the effects of zooplankton grazing and their influence on the timing and magnitude of the flux of particles through the water column need to be considered. It is thought, however, that grazing by zooplankton has little affect on U37k′ values and therefore on alkenone-derived temperatures [Volkman et al., 1980b; Grice et al., 1998]. If, however, sedimentation is due to settling of a phytoplankton bloom, then the dynamics of production could be important. In blooms, half of the biomass is produced by the last doubling, and so the recorded temperature could be strongly influenced by the environmental conditions prevailing at that time. This could be important in regions of very high productivity (e.g., upwelling areas) and others where grazing by zooplankton is not the major factor limiting phytoplankton growth. Well-defined studies are needed to investigate how biochemical composition and isotope fractionation change in alkenone-producing microalgae as a phytoplankton bloom develops.

2.5. Other Environmental Influences on Alkenone Abundances and U37k′ Values

[38] Parameters other than temperature (e.g., nutrients) have been shown to influence the degree of unsaturation in laboratory cultures of microalgae [e.g., Popp et al., 1998a; Epstein et al., 1998; Riebesell et al., 2000]. It needs to be established whether such factors could account for some of the anomalous alkenone data in the field. Unfortunately, too few field data are available to adequately assess this. From culture experiments, growth rate and cell geometry are now considered important factors in addition to [CO2(aq)] in determining the isotopic composition of alkenones [e.g., Freeman and Wakeham, 1992; Popp et al., 1998b], but again field data confirming this are not yet available. Cell geometry should play a role, since haptophytes vary from extremely small cells (e.g., Gephyrocapsa ericsonii: 1–2 μm) to quite large cells (Gephyrocapsa oceanica: 8 μm). A further complication here is the way coccolithophorid cell sizes have varied over geological time [e.g., Kameo and Takayama, 1999].

[39] Questions requiring attention include the following:

[40] • Is there evidence from natural systems for the influence of other environmental factors on alkenone abundances and composition?

[41] • What environmental factors might we have to consider?

[42] • Do culture experiments provide a good approximation to natural aquatic systems? If not, which factors require special attention? What are the merits of different culturing systems (e.g., dilute batch culture, continuous culture, turbidostat) and how relevant are results from these experiments to different types of aquatic environments?

[43] • Can evidence be found for an effect of salinity or nutrient regime on alkenone abundances in seawater?

[44] • Where might such effects be most likely to affect the use of alkenones as paleothermometers (upwellings, polar regions, eutrophic areas)?

[45] Another intriguing possibility is that alkenone abundances might be affected by other environmental influences. For example, Rosell-Melé and Comes [1999] has proposed that variations in the relative abundance of the 37:4 alkenone might reflect changes in seawater salinity. Similar speculations that salinity might affect alkenone distributions can be found in earlier work [Conte et al., 1994b; Ficken and Farrimond, 1995], but this remains to be proven. In contrast, changes in CO2 concentration have been shown to have little systematic effect on alkenone distributions [Riebesell et al., 2000].

[46] Degradation in the water column is thought to have little effect on U37k′ values [e.g., Bentaleb et al., 1999; Sicre et al., 1999]. Alkenones are much more stable than other common phytoplankton lipids such as sterols or fatty acids, but it is clear that they do degrade in sediments [Hoefs et al., 1998]. Whether this leads to a change in U37k′ values is still in question. A priori one might expect the 37:3 alkenone to degrade faster than the 37:2 alkenone [Teece et al., 1998], but the available evidence for this is not compelling. More research is needed to identify the environmental conditions where degradation could lead to a change in U37k′ values.

2.6. Related Biomarkers and Possible New Proxies

[47] It is well recognized that the U37k′ index cannot be used effectively for seawater temperatures less than ∼7°C. Studies of particulate matter from the Southern Ocean [Sikes et al., 1997] suggest that at lower temperatures the abundance of long-chain alkenes increases, as does their degree of unsaturation, with decreasing temperature. Unfortunately, the changes observed were not consistent enough to support derivation of new temperature proxies. Further research on this topic is warranted. The abundance of the 36:2 fatty acid esters also appears to respond to temperature changes [Conte et al., 1992], but temperature proxies involving it and the alkenones are more likely to be affected by degradation effects, and thus they have not been widely adopted. Nonetheless, new research to develop other biomarker proxies for measuring temperature and environmental changes in the geological past should be encouraged.

3. Concluding Remarks

[48] In the 2 decades since alkenones were discovered in sediments and microalgae, much has been learned about their occurrence in aquatic ecosystems. Their value as proxies for seawater temperature at the time of deposition has been well established, but the picture is still not complete. More research on seasonality and depth of maximum production is needed to fully explain apparent discrepancies between U37k′ values in sediments and expected sea surface values. It also needs to be firmly established whether different calibrations are needed for different regions of the ocean, both now and in the past. Evidence for the possible effects of salinity changes, nutrient conditions, and degradation on U37k′ values is tantalizing, but still far from persuasive. Notwithstanding concerns about the reliability of absolute temperature values determined from sedimentary U37k′ values, inferred temperature changes over time seem realistic, and the technique has provided a new window through which to view climate changes over the past several hundred thousand years.


[49] I am grateful to many colleagues who have shared my fascination with the lipids in microalgae and their distributions in sediments. I thank John Sargent for supplying the original samples of North Sea sediments and Emiliania huxleyi from which I first isolated alkenones, Eric Corner and his team from Plymouth for microalgal cultures and Geoffrey Eglinton FRS for providing a stimulating environment at Bristol University in the late 1970s where my first work on alkenones was carried out. I also gratefully acknowledge helpful discussions with Elisabeth Sikes, John Farrington, James Maxwell, Fred Prahl, Ulf Riebesell, Geoffrey Eglinton, and many others over 2 decades. This review benefited from information and ideas provided by Elisabeth Sikes, Hanno Kinkel, Jella Bijma, Patrizia Ziveri, John Hayes, and an anonymous reviewer.