A Cretaceous-Palaeogene stable isotope record

Palaeoclimate records of the Mesozoic and Cenozoic eras display a long-term trend from extremely warm conditions during the mid-Cretaceous optimum to cooler climates of late Palaeogene and Neogene time. The Neogene climate record is reasonably well known on a time scale of c. 50-100 ka resolution or better, based on extensive surveys of benthic and planktonic stable isotope stratigraphies from all the major ocean basins save the Arctic. However, Palaeogene and Cretaceous climates are currently poorly resolved owing to the scarcity of high deposition rate sections and the frequent poor preservation of microfossils upon which various faunal and geochemical climate proxies are based. Consequently, most of our highly resolved records of Cretaceous and Palaeogene climate come from stable isotope analyses of bulk carbonates. The uncertain role of diagenesis in these isotopic data has made it difficult to draw firm conclusions about many aspects of ocean climate and circulation.

To improve the situation we have combined stable isotope records from Leg 171B sites with previously published datasets to illustrate long-term trends in ¹⁸O and ¹³C for the surface and deep ocean (Fig. 9). These data include records of planktonic foraminifera that record the most negative ¹⁸O as well as benthic species to show the vertical ¹⁸O and ¹³C gradients in the oceans. Currently, there are insufficient data to determine which of the features of these isotope records reflect global trends rather than regional trends for the interval before Campanian time. However, Campanian and younger trends have been duplicated in a number of localities suggesting that the basic patterns are of global significance. Trends in mid-Cretaceous records have been illustrated in several low- to moderate-resolution datasets from both deep-sea and outcrop sections (e.g. Clarke & Jenkyns 1998; Stoll & Schrag 2000). Our foraminifer isotopic records broadly mirror patterns in the bulk sediment stable isotope data. It is likely that parts of the Albian and Cenomanian records will ultimately be found to differ in amplitude and absolute values from deep Pacific or Indian Ocean records. The North Atlantic was partly a silled basin during Albian and Cenomanian time and was part of the Tethys seaway, suggesting that it may have been filled with intermediate waters or even surface waters from elsewhere in the world's oceans.

Cretaceous climatic optimum

Benthic ¹⁸O in early Albian time was broadly similar to that during the Cretaceous and early Palaeogene time. However, Atlantic intermediate waters (c. 1500 m water depth at ODP Site 1049) approached the same ¹⁸O values as planktonic foraminifera during early Albian time. The convergence of planktonic and benthic ¹⁸O records suggests that the North Atlantic basin was either filled with overturning surface waters or that the planktonic foraminifera chosen for analysis grew in subthermocline waters. We think it most likely that the occasional similarity in deep and surface ¹⁸O reflects variability in the intensity of surface stratification, as the ¹⁸O gradient swings between c. 0.5 and nearly 2 during OAE 1b (c. 112 Ma). Hence, the isotopic data suggest that the western North Atlantic was probably filled with moderately high-salinity waters not unlike the modern Mediterranean Basin that were occasionally more strongly stratified by runoff from the adjacent continents or by inflow of low-salinity waters from adjacent ocean basins such as the Arctic or the Pacific. Indeed, tectonic reconstructions show that the North Atlantic could have been hydrographically restricted from the deep Indo-Pacific by shallow sills across the Central American Seaway and the myriad of elevated plateaux and tectonic terranes in the Tethys Seaway (Hay et al. 1999).

Our isotopic data (Erbacher et al. 2000) suggest that the western North Atlantic behaved like the Plio-Pleistocene Mediterranean basins. There, the modern vertical temperature gradient can be as low as 3-4^oC and salinities of c. 36 increase ¹⁸O of planktonic and benthic foraminifera by as much as +/- 1 over the open eastern North Atlantic (Miller et al. 1970). The potential for unusually high salinities in the North Atlantic during early Albian time makes it difficult to estimate absolute sea surface temperatures (SST). However, heavy ¹⁸O values from other records (Huber et al. 1995; Clarke & Jenkyns 1999) may suggest that temperatures during early Albian time were lower than during any other period in the mid-Cretaceous.

We have no foraminifer ¹⁸O data from mid-Albian time. Indeed, a survey of existing DSDP and ODP sites suggests that mid-Albian time is very poorly represented everywhere in the deep oceans, or where present, has little calcareous fossil material suitable for stable isotopic analysis. The situation changed by the late Albian, where isotopic results from ODP Hole 1052E show that both planktonic and benthic ¹⁸O were nearly the most negative ratios seen in the last 100 Ma. Only around the Cenomanian-Turonian boundary interval can lighter isotopic values be observed. The large vertical ¹⁸O gradients between planktonic and benthic foraminifera suggest that by late Albian time, the North Atlantic had ceased to be an extension of a silled Tethys Seaway with an estuarine circulation and had developed deeper marine connections to other deep ocean basins.

Norris & Wilson (1998) inferred that SST reached at least 30-31^oC during late Albian and earliest Cenomanian time (c. 98-102 Ma) in the western North Atlantic. These high temperatures were maintained or even raised further during parts of the Cenomanian. For example, planktonic foraminifera from DSDP 144 (9^oN) in the equatorial western North Atlantic have average ¹⁸O values of -3.9, equivalent to temperatures of c. 32-34^oC using estimates from standard palaeotemperature equations, modern salinity, and assumption of an ice-free world. Notably, benthic ¹⁸O also peaks in upper Cenomanian time, reaching ratios over 1 more negative than that seen at any point in the deep oceans during the Cenozoic. Intermediate waters may have been unusually warm near the C-T boundary (Huber et al. 1999). Alternatively, tectonic barriers to exchange with other ocean basins may have become sufficiently restrictive that the North Atlantic was filled with thermocline waters or upper intermediate waters flowing in over shallow sills.

Our data (Huber et al. unpubl. data) from ODP Site 1050 illustrate that the mid-Cretaceous thermal optimum was not uniformly warm or stable. A pronounced increase in both planktonic and benthic ¹⁸O occurred in the mid-Cenomanian during the Rotalipora reicheli Zone (c. 97 Ma). Stoll & Schrag (2000) recorded a similar event in ¹⁸O of bulk limestone and marl in outcrop sections from Spain and Italy. They suggested on the basis of the magnitude of the event and its abrupt onset, that it records glaciation and a shift in whole ocean ¹⁸O as a result of ice build-up. A similar event is present in our data from the middle Cenomanian sequence in ODP 1050 and the ¹⁸O analyses of bulk carbonates from the Indian Ocean (Clarke & Jenkyns 1999) and Tethys (Stoll & Schrag 2000). Although the widespread occurrence of the ¹⁸O maxima suggests a global shift in ¹⁸O , such as that produced by ice volume changes, intermediate water palaeotemperatures at this site and in the southern South Atlantic (Huber et al. 1995) were none the less higher (>11^oC) than would be expected if there were a significant volume of polar ice unless the North and South Atlantic basins were isolated from high-latitude sources of deep water.

The possible existence of large ice volume changes during the peak of mid-Cretaceous warmth raises questions about mechanisms that regulate ice growth and decay. Huber et al. (1995) showed that latitudinal thermal gradients were unusually low during intervals of the mid-Cretaceous thermal optimum. Therefore, it is hard to understand how ice growth could begin when high-latitude SSTs were nearly as high as equatorial temperatures. The abrupt ¹⁸O swings in Cenomanian time suggest that there may be threshold effects that can dramatically alter ocean circulation and temperature at the warm end of the climate spectrum, in much the same way that feedback systems such as moisture balance and runoff operate to abruptly change boundary conditions in Pleistocene time at the cool end of the climate spectrum.

Campanian-Maastrichtian refrigeration

We have limited stable isotope data for foraminifera from late Turonian to mid-Campanian time, an interval of c. 15 Ma. Data are contradictory for this interval. Huber et al. (1995) showed that DSDP Site 511 in the South Atlantic was bathed with warm waters from Cenomanian to Coniacian time and SST did not begin to fall appreciably until early Campanian time. Barrera (1994) presented a handful of measurements from North Pacific DSDP sites that also suggest peak temperatures were maintained until some time in the Santonian period. However, ¹⁸O data of bulk carbonates from Indian Ocean DSDP sites (Clarke & Jenkyns 1999) suggest that the decline in temperatures occurred during late Turonian time. The planktonic foraminifera show a tendency towards heavier isotopic values during the course of the Turonian as suggested by the dataset from ODP Site 1050 (Huber et al. unpubl. data), although some extreme light isotopic values can still be seen. Results from Site 1050 suggest that planktonic ¹⁸O was similar to or more positive than that in late Campanian time and suggest that the surface waters were significantly cooling during Turonian time, although highly variable. One could conclude from the planktonic isotope record that the mid-Cretaceous climatic optimum was over by c. 90 Ma if not earlier, although the benthic isotope data show that a persistent fall towards heavier values occurred in Campanian time. More benthic isotope data are needed to document the exact timing of the end of the mid-Cretaceous climatic optimum.

Barrera & Savin (1999) published Campanian and Maastrichtian data for planktonic and benthic ¹⁸O from a number of sites around the world. Most records show the long-term trend to more positive ¹⁸O in both surface and deep waters and an abrupt step in this trend about 71 Ma. Miller et al. (1999) interpreted the step increase in foraminifer ¹⁸O to reflect an increase in ice volume and glacioeustatic sea-level lowering. MacLeod & Huber (this volume) have shown that the positive ¹⁸O deflection in benthic foraminifera is not nearly so pronounced in benthic records from the North Atlantic. Their data suggest either that the ¹⁸O shift observed elsewhere is not a glacial step or that the North Atlantic record is biased by the introduction of an unusually warm deep water mass that overprints the glacial increase in ¹⁸O. Notably, the size of the mid-Maastrichtian increase in ¹⁸O in planktonic foraminifera is about a third to half the amplitude of the change in benthic foraminifera even in the Pacific. Therefore, it seems likely that at least half of the mid-Maastrichtian ¹⁸O shift is due to cooling or an increase in salinity of deep waters.

Cretaceous-Palaeogene boundary and Danian climate

The Cretaceous-Palaeogene (K-P) boundary is preceded by a rise in global ocean ¹³C and bottom-water temperatures starting in mid-Maastrichtian time (c. 71-72 Ma) that culminates in early Danian time (Fig. 9). The overall rise in benthic ¹³C and deep-water temperatures probably reflects carbon burial and CO2 sequestration (Zachos et al. 1989; Stott & Kennett 1990). Sea-level fall may have played a role in carbon burial through the formation of large coal swamps with the retreat of epicontinental seas in late Maastrichtian time. The deep-sea ¹³C record is also influenced by a general decrease in inter-basin ¹³C gradients near the end of Maastrichtian time that reduced the isotopic contrast between relatively young deep waters in the North Atlantic and relatively old deep waters in the Pacific and Indian Ocean (e.g. Corfield & Norris 1996, 1998; Barrera & Savin 1999). Frank & Arthur (1999) explained the reduction of interbasinal ¹³C gradients by suggesting that the opening of deep passages between the North and South Atlantic played a key role in ventilating the deep North Atlantic during mid-Maastrichtian time. None the less, the North Atlantic continued to maintain a distinctive young deep and intermediate water-mass through the mid-Palaeocene (Corfield & Norris 1996, 1998).

Bottom temperatures stayed the same or rose from mid-Maastrichtian time to the K-P boundary, whereas surface water temperatures fell until the last 200 ka of the Cretaceous period. The general decline in SST may be related to the withdrawal of epicontinental seas (Frank & Arthur 1999) or to the sequestration of carbon and CO₂ during the rise in global ¹³C in late Maastrichtian time. The deep North Atlantic and deep waters elsewhere in the oceans display distinctly different ¹⁸O throughout late Maastrichtian time. The North Atlantic was consistently c. 2^oC warmer or less saline than the other deep basins, a contrast that was maintained through early Palaeocene time (e.g. Corfield & Norris 1996). Relatively high deep-water temperatures in the North Atlantic may reflect the presence of young deep waters conditioned by overflow from Tethyan basins much like the conditioning of modern North Atlantic deep water by Mediterranean outflow waters.

The early Danian period (64-65 Ma) is best known as a time of tremendous turnover in marine pelagic ecosystems following the Cretaceous-Palaeogene mass extinction (D'Hondt & Keller 1991; Gerstel et al. 1987; Jablonski & Raup 1995; Keller 1988; MacLeod 1993; Olsson et al. 1992; Smit 1982). The extinction eliminated c. 95 % of planktonic foraminifer species and had a profound effect on other members of the plankton. The extent of the devastation of the pelagic ecosystem is reflected by the extended (c. 3 Ma) collapse of the carbon pump and the vertical ¹³C gradient in the oceans (e.g. D'Hondt et al. 1998).

The recovery was also associated with large-scale changes in marine climate and carbon-cycle dynamics. There is a general decrease in global ocean ¹³C that may reflect reduced productivity and carbon burial (Shackleton & Hall 1984). In addition, the vertical ¹⁸O gradient was greatly reduced for an interval of almost 3 Ma in early Danian time coincident with the reduction in the vertical ¹³C gradient. The declines in vertical ¹⁸O and ¹³C gradients are both likely to be partly related to reorganization of biotic communities brought on by the end-Cretaceous mass extinction. A reduction in the vertical ¹⁸O gradient is perhaps best explained by the widespread extinction of surface-dwelling species of planktonic foraminifera during the K-P mass extinction and a tendency for palaeoceanographers to analyse species that grew mostly in thermocline waters during early Danian time.

At the K-P boundary, the mass extinction resulted in a dramatic 1 decrease in surface ocean ¹³C with little or no change in deep-water ¹³C. The absence of any large negative shift in ¹³C of deep waters strongly suggests that changes in the vertical ¹³C gradient are due to the extinction of surface ocean biota rather than a change in the global ¹³C reservoir (Hsü et al. 1982; Zachos & Arthur 1986; Zachos et al. 1989; D'Hondt et al. 1998). Some records show an inversion of the surface-to-deep ¹³C gradient (Hsü & McKenzie 1985; Zachos & Arthur 1986) that has been attributed to biomass burning (Ivany & Salawitch 1993) but could also reflect measurement artifacts due to low carbonate content in the boundary interval (e.g. Shackleton 1986).

Following the collapse of planktonic and benthic ¹³C gradients during the mass extinction, global ¹³C continues to rise. Maximum benthic ¹³C of c. 2.2 is recorded less than 100 ka above the K-P boundary and is succeeded by a long-term decline in ¹³C over the next 4 Ma. Hence, it appears that the process of carbon burial that led to the Maastrichtian rise in benthic ¹³C was reversed just after the K-P mass extinction and set in motion a long-term interval of unroofing previously deposited carbon. We suggest that the change from net carbon burial to net erosion may reflect the final draining of epicontinental seas and the onset of weathering of coal and organic shales deposited during late Cretaceous time. By the end of the Danian (c. 61 Ma) benthic ¹³C had returned to a ratio very similar to that achieved in mid-Maastrichtian and late Albian time, and very close to that later reached during the early Eocene.

Palaeocene-Eocene climate trends

One of the most striking features of the early Cenozoic and Cretaceous stable isotope records is the dramatic positive shift in ¹³C of both planktonic and benthic foraminifera in late Palaeocene time. Benthic foraminifer ¹³C increased by c. 2 and planktonic foraminifer ¹³C increased by c. 3 between 61 and 58 Ma. Part of the increase in ¹³C of planktonic foraminifera is probably due to the re-evolution of photosymbiosis after the K-P mass extinction (e.g. Corfield & Norris 1998). However, the increase in ¹³C of benthic species is seen throughout the oceans (e.g Miller et al. 1987; Corfield & Cartlidge 1992), suggesting that it represents a reservoir effect of burying large quantities of organic carbon. It is not at all clear where all this carbon was deposited, as there are few large oil or coal reservoirs of late Palaeocene age. The relatively rapid decline of ¹³C of both benthic and planktonic foraminifera between 58 and 55 Ma suggests that much of the carbon buried during Palaeocene time was exhumed by the end of the Palaeocene epoch. Beck et al. (1998) have suggested that much of the carbon deposited during late Palaeocene time may have accumulated in Tethyan basins that were subsequently uplifted and eroded during the initial stages of the Himalayan Orogeny.

The peak of the Palaeocene ¹³C increase coincides with the most positive benthic foraminiferal ¹⁸O ratios in the Palaeogene period. A modest increase in ¹⁸O of planktonic foraminifera is also present, suggesting cooling of both surface and deep waters during the late Palaeocene 'carbon isotope maximum'. We speculate that CO2 drawdown associated with organic carbon burial may have been responsible for the fall in ocean temperatures. Evidence for snowmelt and cool interior climates in the Rocky Mountains (e.g. Dettman & Lohmann 2000) as well as palaeobotanical estimates of mean annual temperature (e.g. Wing 1998) suggest that cooling in late Palaeocene time occurred over the continental interiors as well as the oceans. The late Palaeocene cool phase was succeeded by a c. 3 Ma interval of increasingly warm conditions leading up to the Late Palaeocene Thermal Maximum (LPTM) at c. 55 Ma.

The LPTM occurred at the point where deep-water temperatures and SST had nearly reached the highest levels in Cenozoic time. Zachos & Dickens (1999) and Katz et al. (1999) have proposed that the LPTM occurred because deep-water temperatures exceeded a threshold level above which methane hydrates began to catastrophically destabilize and contribute to a runaway greenhouse effect. However, Bains et al. (1999) pointed out that the long-term warming trend in latest Palaeocene time is highly aliased and that detailed ¹⁸O records from both foraminifera and fine fraction carbonate display no significant warming trend for at least 200 ka before the LPTM. Hence, it is unclear whether the million-year scale drift to higher temperatures has anything to do with the LPTM. Other proposed mechanisms to initiate the LPTM include changes in deep-water circulation inspired by tectonics (Beck et al. 1998), volcanism (Eldholm & Thomas, 1993; Bralower et al. 1997b) or slope failure (Bains et al. 1999; Norris & Röhl 1999).

Dickens (1999) suggested that sedimentary carbon reservoirs act as 'capacitors' in the global carbon cycle that store and release greenhouse gases to modulate global climate. This theory supposes that methane dissociation events, as proposed for the LPTM, are common in the geological record (Dickens 2000b). Indeed, the example of the LPTM has spawned a resurgence of interest in the Cretaceous and Palaeogene climate record and has led to the discovery of other large perturbations in the carbon cycle. Large negative ¹³C anomalies have been identified with the onset of several Oceanic Anoxic Events (OAEs) in Cretaceous time (Jenkyns 1995; Wilson et al. 1999). Likewise, analysis of benthic foraminifer assemblages and isotopic anomalies provides strong hints that there may be other events like the LPTM in late early Eocene and mid-Palaeocene time (Thomas & Zachos 1999). Short, but intense, ¹³C anomalies are of great interest as 'natural experiments' in the biological and climatological effects of transient perturbations of the carbon cycle. The increasing evidence that there may be several LPTM-like events offers the exciting opportunity to compare and contrast these events to better evaluate the palaeoceanographic context, trigger, duration and biological effect of large-magnitude changes in carbon reservoirs of whatever cause. There is also the possibility that some of the ¹³C events represent large emissions of greenhouse gases and can provide natural analogues for modern greenhouse warming.

The early Eocene benthic ¹⁸O record shows that the Earth was essentially warmer than today. That is to say, the high-latitude oceans were much warmer than today, raising the global mean temperature (e.g. Zachos et al. 1994), and the Earth was characterized by the absence of large ice sheets. None the less, maximum sea surface temperatures in the tropics were not necessarily higher than modern temperatures. Most planktonic isotope records show that low-latitude early Eocene sea surface temperatures were lower than at present (Boersma et al. 1987; Zachos et al. 1994; Bralower et al. 1995). This is not well understood. Either heat diffused into the deep ocean or the stable isotope records of the planktonic foraminifera do not record the surface, or have been affected by diagenesis. Wade et al. (this volume) have shown that massive upwelling in late mid-Eocene time influenced low-latitude sea surface temperatures. Upwelling at Milankovitch periodicities may explain why sea surface temperature estimates were low in the low latitudes. Towards mid-Eocene time increases in ¹⁸O can be seen in both the planktonic and benthic foraminiferal records. These trends indicate cooling of the Earth (Kennett & Shackleton 1976) starting near the early Eocene-mid-Eocene transition and continuing in a series of steps to the Eocene-Oligocene boundary.