Climate in Earth History: Studies in Geophysics (1982)

THEODORE C.MOORE, JR.

Exxon Production Research Company

NICKLAS G.PISIAS

Oregon State University

L.D.KEIGWIN, JR.

Woods Hole Oceanographic Institution

The long-term (greater than 10⁶ yr) character of the Earth’s climate appears to exhibit distinct shifts from one state to the next. The succession of such states can be thought of as an evolutionary process with the average characteristics of each successive climatic state fundamentally different from any previous state. Each state may differ in terms of mean condition, in the amount of oscillation around the mean condition, and in the distribution of the amplitude of oscillation as a function of frequency. This long-term evolution of climate appears to be associated with telluric changes (i.e., changes in the geography and topography that form the boundaries to the fluid spheres). The rate of climatic change depends on the nature of the telluric effects. For example, the opening of an ocean gateway to deep and surface flow (such as passage between Antarctica and Australia) may have had a sudden and dramatic effect on average oceanic conditions, whereas the gradual opening of the Atlantic or closing of the Tethyan seaway may have caused longer term shifts in climatic conditions.

The geologic record of the deep sea affords us the opportunity to study the character of global oceanographic conditions over the past 100 million years (Ma) to define the steps in the evolution that have lead from the rather equitable climates of the Cretaceous to the ice ages of the last few million years and to relate these evolutionary steps to the changes in the telluric boundary conditions that are likely to have caused them. Furthermore, we should be able to characterize different parts of the climate system during each stage of this evolution. By studying the way the surface ocean, the deep ocean, the atmosphere, and the cryosphere have changed during each evolutionary stage, we will gain insights into the mechanisms that give rise to long-term climatic change. In addition, the investigation of the climate system under a variety of boundary conditions should give us a better fundamental understanding of how the different elements of this system can, and do, interact.

Before such research can proceed, quantitative data on each element of the climate system must be acquired and studied. One of the most extensive quantitative data bases that now exists for the Cenozoic is the record of change in oxygen isotopes of benthic foraminifera (see Table 19.1). These data have been compiled by many investigators (Douglas and Savin, 1973,

1975; Savin et al., 1975; Shackleton and Kennett, 1975a, 1975b). More recent work by Cenozoic Paleo-Oceanography Research Project (CENOP) workers has greatly added to this data base, particularly in the Miocene.

Oxygen isotopes measured on the shells of benthic foraminifers have given us a good representation (Figure 19.1) of the long-term changes that have occurred in the mean conditions of the deep ocean. The deep ocean contains more than 90 percent of the water on the Earth’s surface and represents a part of the climate system that is important to the storage and transport of heat. It is a rather slow-moving part of the system, with a response time on the order of 10³ yr, intermediate between the rapidly responding atmosphere and surface ocean and the much more slowly changing cryosphere and lithosphere. Compared with that of the surface waters, the isotopic composition of the modern deep ocean is relatively homogenous (Craig and Gordon, 1965); thus, an isotopic record of change in the deep ocean from almost any location is likely to give a picture of change in a large and important part of the climate system.

If it is assumed that the shells of benthic foraminifera are deposited in isotopic equilibrium (or that any vital effects can be taken into account), then changes in the oxygen isotopic ratio in the carbonate tests of the deep benthic fauna indicate changes in either the temperature of the bottom waters (with each 1°C equivalent to roughly 0.26 ‰ change in the ¹⁸O/¹⁶O ratio) or in the isotopic composition of deep waters. Changes in the isotopic composition of the deep ocean would most likely be caused by the transfer of a large amount of isotopically light water from the oceans to continental glaciers (δ¹⁸O change of 0.1 ‰ is roughly equivalent to a 10-m glacial sealevel change); however, changes in the mode of formation of deep water that involved a significant change in their salinity would also affect their isotopic composition [with about a 0.1 ‰ change in δ¹⁸O for every 0.2 ‰ change in salinity (Craig and Gordon, 1965)].

Studies of the long-term record of the isotopic record of foraminifera, together with other geologic and geophysical studies, suggest that in the last 100 Ma there were two times when major continental ice caps were formed and extended into the sea: in the Middle Miocene, marking the buildup of the Antarctic ice cap about 14 Ma ago (Shackleton and Kennett, 1975a, 1975b), and in the late Pliocene, marking the formation of continental glaciers in the northern hemisphere about 3 Ma ago (Shackleton and Opdyke, 1977). If the estimates of the effect of ice volume on the isotopic composition of seawater are taken into account, and if it is assumed that changes in salinity have been small, then the record of the long-term changes in the oxygen isotopes can be interpreted as an oceanic temperature record (Figure 19.1). This record suggests that the deep waters have cooled by about 10–13°C in the last 60 Ma. Surface waters followed this trend to the mid-Miocene and then leveled off or warmed slightly {Douglas and Woodruff, 1981). This overall cooling of deep waters is neither monotonic nor gradual. Short reversals in the trend occur, and the major portion of the cooling appears to occur as distinct steps in the record (e.g., in the mid-Eocene, at the Eocene-Oligocene boundary, in the mid-Miocene, and in the Pliocene (Figure 19.1). These sharp drops in the isotopic record are thought to be associated with evolutionary changes in the climate system that give rise to shifts in the mean conditions. It remains to be seen whether other proxy records and oceanic and climatic changes exhibit the same shifts in their records and whether additional evolutionary steps can be identified in these records.

The most recent evolutionary stage as defined in Figure 19.1 is the Quaternary. An important characteristic of the Quaternary climate has been the large degree of variability around the mean climatic state. Most of the Quaternary variability in the benthic oxygen isotope signal is thought to be associated with changes in continental ice volume (Shackleton, 1967; Shackleton and Opdyke, 1973). The variability of both the oxygen isotopes (Shackleton and Opdyke, 1976; Pisias and Moore, 1981) and the planktonic fauna (Ruddiman, 1971; Briskin and Berggren, 1975) changed through the Quaternary and these changes appear to involve both the amplitude and frequency of oscillation. Spectral analyses of one 2-Ma-long record of oxygen isotopes (measured on a planktonic species of foraminifera) indicates that this record can be divided into at least three intervals, each having progressively more variance associated with progressively longer periods of oscillation (Pisias and Moore, 1981). These changes in the spectral character of oxygen isotope records may also indicate evolutionary changes in the climate system. In this example, these changes are thought to result from changing mechanisms of ice-cap growth and decay and may indicate the effects of extensive glacial erosion of continental areas (Pisias and Moore, 1981).

Similar spectral studies of Tertiary records have yet to be undertaken, primarily because (a) few long, relatively undisturbed marine sections have been recovered, and (b) establishing a sufficiently accurate time scale for a detailed spectral analysis is a difficult task. Although it may not be possible yet to investigate the spectral character (i.e., the distribution of variance as a function of frequency of oscillation) of Tertiary oxygen isotopic records, the total variance of such records and how this variance has changed with time and place can be studied.

It is sometimes assumed that the warmer climes of Tertiary and Cretaceous times were more equable than at present, that is, they were less variable. However, little work has been done on the short-term variability of Cenozoic climate. Certainly before the buildup of continental glaciers one might expect to see less variability in the oxygen isotope signal. But was there less oceanographic variability during the Eocene than during the Oligocene or Miocene? Did the variability of the benthic oxygen isotope record increase as the deep-ocean temperature cooled? How did variability in this record change with evolution of the climate system, and what degree of variability is associated with each evolutionary step? Such questions are addressed here in hopes of better defining the true nature of climatic and oceanographic variability through the Cenozoic.

An accurate estimate of the total variance in a data set does not require as long a record as a spectral analysis. It does not require that the time scale be known accurately, nor does it re-

quire that the mean values of two data sets be the same before a comparison of the variances can be made. Thus, virtually all of the short-time series of oxygen isotope data (Table 19.1), regardless of the particular species used to obtain the data, can be used in a comparison of the total variance in different data sets. These variances have units of (‰)²; however, this notation is omitted from the text discussion.

Most of the samples in the Tertiary data series are rather widely spaced within the recovered sections (usually 100 cm). Accumulation rates of 10–50 meters/million years (m/m.y.) are common in these pelagic sediments and thus would indicate a time spacing of the samples that is often greater than 10⁵ yr. Sample spacing in the Miocene isotopic data collected by the CENOP group (Table 19.1) was designed to be between 50,000– 100,000 yr. Although such sample spacings are much broader than commonly used in Quaternary studies, they do provide an estimate of the total variance in the long-term record. If the Pleistocene record is sampled at a 50,000-yr spacing, the variance estimate is the same as for a 5000-yr sample spacing. Table 19.1 lists data sources to be used in this analysis and Table 19.2 lists site locations. These data have the following characteristics: (1) Oxygen isotope measurements in each data set were carried out on samples of a single species, a species group, or a mixed benthic assemblage. Data derived from different species groups or size fractions were not combined; however, if their variances were not significantly different (F test), they

FIGURE 19.2 Variance in benthic oxygen isotopes as a function of age in the Cenozoic. Data sources are listed in Table 19.1. Horizontal lines indicate stratigraphic range represented by the data sets, vertical lines indicate 80 percent confidence limits of the variance estimates. Closed circles represent Pacific Ocean data sets; open circles, Atlantic Ocean data sets; squares, Southern Ocean data sets; triangles, Indian Ocean data sets. Shaded area encompasses most of the data from the Pacific and Southern Oceans; excluded are the very deep Pacific sites in the Early and Late Miocene and the highly variable sites of the Atlantic and Southern Oceans sites in the Late Miocene.

could be pooled. (2) Each data set comes from a single site. (3) Each group of measurements is associated with a known stratigraphic age.

Stratigraphic intervals over which the variance was calculated were kept as short as possible to lend more detail to the long-term record of variability. Data were selected to avoid any clear jumps or shifts in the record that might be associated with evolutionary changes in the system. Such shifts in the data would lead to an abnormally large estimate of the variance. To guard further against more gradual shifts in the data, trends were removed from each data set using a simple linear regression. The estimate of variance used in this study is the variance around this regression line.

For each stratigraphic age the variances of all data sets were compared using the M test (Thompson and Merrington, 1944) for homogeneity of variances. If the difference in the variance estimates were nonsignificant (p≥0.95), then the variances of all the data sets of the age could be pooled. This procedure allows the combination of several Quaternary time series. If variances were nonhomogeneous, they were subdivided according to ocean basin locations, and again tested for homogeneity. Data sets from the Atlantic, Pacific, and Indian Oceans were compared; when sufficient data were available, their variability as a function of water depth was contrasted.

In a few cases the variability in individual data sets is high, greatly exceeding that found in other sites at the same time and depth interval. Such data might result from unrecognized diagenic or stratigraphic problems. They are considered spurious here, and, although included in Table 19.1, they are not plotted in the figures. Sites having extremely low variability might result from severe disturbance and mixing of rotary-drilled sections; however, such sites appear to show some degree of spatial and temporal coherence and are not excluded from the figures.

The record of the variance in benthic oxygen isotopes is shown in Figure 19.2, with vertical lines giving the 80 percent confidence limits of the estimate and horizontal lines indicating the range of the stratigraphic age of the individual data sets. Each symbol is located at the midpoint of the stratigraphic range. Different symbols are used for Pacific, Atlantic, Antarctic, and Indian Ocean data sets.

Although the amount of data varies greatly through the Cenozoic (Figure 19.2), there appear to be several important changes in the character of isotopic variance as a function of time: (a) a decrease in the estimated variance from mid-Eocene to Oligocene time, (b) a slight Middle Miocene maximum in variance, (c) an increase in the variance of Atlantic sites relative to Pacific sites in post-Early Miocene times, and (d) a sharp increase in the variance at both Atlantic and Pacific sites beginning in the late Pliocene. The background variability above which the intervals of high variance rise is surprisingly consistent. It averages about 0.04 and is four times greater than the variance associated with laboratory error.

In the latest part of the record the high variance in benthic oxygen isotopes increases from values of 0.06–0.1 in the late Pliocene to approximately 0.2 in the late Quaternary. Most of this high degree of variability is related to changes in the isotopic composition of the oceans as continental glaciers waxed and waned (Shackleton, 1967; Shackleton and Opdyke, 1973).

There is some indication that the pooled variance in Atlantic sites is slightly higher than in Pacific sites. The difference is not significant in the Quaternary (M test, p≥0.95); however in the Pliocene and Late Miocene the Atlantic and Pacific do appear to show different degrees of variability. Shackleton and Cita (1979) noted the generally higher degree of varability in Atlantic sites during the latest Miocene. Such differences are

also seen in the Pliocene and Late Miocene data presented here (Figure 19.2), where several {but not all) Atlantic and Southern Ocean sites show a higher variance than most Pacific sites, with the highest variances measured in the earlier part of the Late Miocene. During the Early Miocene and Oligocene, the variance of benthic oxygen isotopes in Atlantic sites were also slightly higher than those of the Pacific; however, the differences are not statistically significant (F test, p≥0.9).

In the Pliocene and Miocene, measurements from deep sites (>4000 m) are included in the data set, These Pacific sites (DSDP 71 and DSDP 77 of the Deep Sea Drilling Project) show generally higher variance than others from the Pacific Ocean basin during both the Early and Late Miocene. During the Middle Miocene, however, detailed studies of a mid-depth site (DSDP 289—Woodruff et al., 1981) have an equally high degree of variability. This interval spans the time of glacial buildup in Antarctica (Shackleton and Kennett, 1975a) and appears as a major shift in the benthic isotopic values (Figure 19.1). The maximum in variance associated with this shift in isotopic values is clearly seen in the original data (Woodruff et al., 1981). To remove the effect of this shift in isotopic mean values on estimates of variance, the Middle Miocene data are subdivided into three groups (Table 19.1, Figure 19.1): pretransition, transition, and posttransition. Trends within these data subsets were removed using a linear regression.

There are few accurate estimates of oxygen isotopic variance in the Paleogene; however, the data that are available indicate a minimum in variability during Late Eocene through Oligocene times. Discounting the high mid-Eocene variance of DSDP 398 (which may be spurious), the variance in benthic oxygen isotopes of the early Paleogene was near 0.06, whereas those in the Late Eocene through Oligocene were significantly lower (pooled variance=0.027). This apparent decrease in variance parallels a general cooling trend through the mid- to late-Paleogene (Figure 19.1) when bottom-water temperatures are estimated to have dropped from almost 15°C in the mid-Eocene to 6°C in the Oligocene (Figure 19.1). This late Paleogene minimum in variability is also significantly less than that measured in the Miocene, when the variance was usually near 0.04.

Many of the changes in the variance of the oxygen isotopic data can be associated with major evolutionary transitions in the Cenozoic climate. The maxima in variance during the Quaternary and mid-Miocene are both correlated with the growth of major ice caps. It is presumed that like those of the Quaternary, the mid-Miocene intervals of high variance are associated with instability in the climate system and that most of the variation is due to changes in the isotopic composition of the oceans.

The next older major step in climatic evolution indicated by the oxygen isotopes (Figure 19.1) occurs at the Eocene-Oligocene boundary. Although the data are more sparse around and across this boundary, the existing data (Kennett and Shackleton, 1976; Keigwin, 1980) indicate a rapid, monotonic shift in isotopic values. This shift is observed in both planktonic and benthic species in high latitudes but only in the benthics at low-latitude Pacific sites (Keigwin, 1980). Thus this evolutionary step, which is thought to be associated with the opening of the Australian-Antarctic seaway (Kennett and Shackleton, 1976), appears to be related to a marked cooling of deep waters and high-latitude surface waters. However, there does not appear to have been a change in the variability of deep-ocean waters on either side of this boundary. Nor was there a marked instability associated with the transition from one climatic state to the next, as observed with the growth of continental ice in the Middle Miocene and Quaternary.

The increase in benthic isotopic variability that occurred in the mid-Eocene is based on sparse data (Table 19.1); but if further work supports its existence, it is the third maximum in variability associated with a marked shift in the mean ¹⁸O content of benthic tests (cf., Figure 19.1). The cause of the mid-Eocene shift in mean isotopic values is not certain. It could have been caused by a telluric change, such as the opening of a passage between South America and Antarctica (Norton and Sclater, 1979), which might have led to a marked cooling of the ocean waters. It might also have been caused by an early buildup of fairly large mountain glaciers in Antarctica (Matthews and Poore, 1980) or by some combination of temperature and ice volume effects.

Sufficient data have been gathered from the Miocene and Pliocene (Table 19.1) to allow a view of how benthic isotopic variability is distributed in space as well as time. In the Late Miocene there is an increased oxygen isotopic (temperature) contrast between deep and bottom waters (Douglas and Woodruff, 1981). This is also the time when marked differences between the isotopic variability of the Atlantic and Pacific Oceans are first noted. This divergence of Atlantic and Pacific estimates of isotopic variance in the Late Miocene suggests the development of different source regions for deep waters in the two basins.

The general pattern of change with depth seen in the Miocene and Pliocene of the Pacific Ocean is from low variance in shallower sites to higher variance in deeper sites. In both the Early and Late Miocene, the variance in the deepest sites (DSDP 71 and DSDP 77) is higher than in any other depth zone in the Pacific Ocean and is exceeded in magnitude only by the Quaternary data and the Late Miocene data from the Atlantic and Southern Ocean. In the Atlantic, Late Miocene variability of benthic isotopes at shallow depths is high (up to about 0.25). The variance decreases with depth, so that at 3000 m all oceans show approximately the same rather low degree of variability. The marked difference in the variance of Atlantic and Pacific benthic isotopic data has been noted previously (Shackleton and Cita, 1979); however, this difference does not appear to occur prior to the mid-Miocene.

The data presented here indicate that the variability of benthic oxygen isotopes have changed with time and that these changes have often been associated with major steps in the evolution of the oceans. Although the data from the Paleogene are sparse, they indicate that the variability in benthic oxygen isotopes of

the Oligocene and Late Eocene was significantly less than at any other time in the Cenozoic. The Oligocene was a time of relatively cool and equable climate (Kennett, 1978; Fischer and Arthur, 1977) and low eustatic sea levels (Vail et al., 1977). The low isotopic variability of this interval may be attributable to homogeneous deep waters derived from a single source region. The higher variability of the early Paleogene was associated with much warmer high-latitude and bottom-water temperatures. These conditions might have given rise to deep waters with a wider range of isotopic compositions. Different source regions and the initial buildup of sizable glaciers on Antarctica could also have served to introduce variability in the isotopic composition of the deep waters; however, the data are not sufficient to explore these possibilities.

The well-documented isotopic shift across the Eocene-Oligocene boundary is interpreted as cooling of the high-latitude ocean and deep waters by about 3°C (Keigwin, 1980). There is no maximum in isotopic variability associated with the buildup of continental ice during the mid-Miocene and Pliocene-Pleistocene. Rather the record indicates a sudden, monotonic shift from one relatively stable oceanic state to another.

In the Early Miocene the average variability increased again, but not up to the levels of the early Paleogene. This change is not readily associated with major oceanographic changes (see Figure 19.1). and the data are not sufficient to tell whether this shift to increased variance was relatively sudden (as in the mid-Miocene and Quaternary) or more gradual. The gradual shoaling of the calcite compensation depth and increase in the carbonate dissolution gradients (Heath et al., 1977) through the Oligocene and Early Miocene suggest a slow, long-term change in the character of the deep waters during this time interval.

An apparent maximum in variability is, associated with the growth of the Antarctic ice cap during the mid-Miocene and suggests that some degree of instability in this ice cap may have existed during its early growth phase (Woodruff et al., 1981). In the Pacific Ocean. variation in the benthic oxygen isotopes was approximately the same before and after growth of the Antarctic ice cap; however. variability in data from the Atlantic Ocean greatly increased by Late Miocene times.

The development of northern hemisphere ice sheets in the Late Pliocene is associated with an increase in oxygen isotopic variability that continues into the Quaternary and reaches a maximum in the late Pleistocene. This increase in variance is readily associated with the fluctuations in the isotopic composition of seawater caused by the growth and decay of continental ice sheets.

There are several keys to assessing the changes in the Cenozoic oceans that may have led to changes in the variability of benthic isotopes. Changes in the global ice volume (which caused a 1.6 ‰ change in the oxygen isotopic composition of the oceans during the late Quaternary) is clearly associated with two of the maxima noted in the historical record (Figure 19.2). Such compositional changes may also be associated with the high variance of the early Paleogene (Matthews and Poore, 1980).

By itself the range of oxygen isotopic compositions of modern deep waters (0.57 ‰) is close to the range of variation in the Cenozoic record of benthic oxygen isotopes; however, the temperatures of these modern watermasses are such that the isotopically heaviest waters [North Atlantic deep water (NADW) at +0.12 ‰] are also comparatively warm (about 4°C), and the ¹⁸O-depleted waters [Antarctic bottom water (AABW) at −0.45 ‰] are cold (about 2°C). Thus, the temperature-fractionation effects of calcite precipitation on oxygen isotopes tend to offset the compositional differences and result in benthic foraminiferal tests with similar isotopic compositions.

The temperature and isotopic differences between modern AABW (at about 5°C and −0.15 ‰) and NADW (at 4°C and +0.12 ‰) tend to enhance the differences measured in the benthic foraminifera. If over long periods of time a site were alternately bathed by NADW and AABW, the isotopic record measured on benthic foraminifera would have a variance close to that measured in most of the Cenozoic sites studied. This suggests that the range of isotopic compositions of modern deep waters is at least sufficient to account for most of the long-term Cenozoic variation in benthic oxygen isotopes (excluding the large compositional changes associated with development of the cryosphere).

How such a degree of long-term variation actually takes place remains unresolved. There are four mechanisms that seem plausible: (1) changes in the isotopic composition of the deep waters at their area of formation; (2) changes in the temperature of salinity characteristics of the deep water masses [which might also be associated with (1) above]; (3) changes in the number of source areas producing deep waters of dissimilar isotopic composition; and (4) changes in vertical structure of the oceans that would lead to fluctuations in hydrographic boundaries between waters of different physical and/or isotopic character.

To evaluate the likelihood of any of these mechanisms operating at a particular time, data are needed from sites that sample a wide depth range in several ocean basins. Such data are not available for the Paleogene but are now being produced for the Neogene. The most common pattern seen in the Miocene data is that of relatively low variability at shallow depths (<1500 m), moderate variability between about 1500 and 4000 m, and increased variability below 4000 m.

High variability in the deepest sites is not seen in Pliocene times. The Early Pliocene has relatively low variability (0.01– 0.03) at all depths except between 2000–3000 m. In the Late Pliocene, the data are rather homogeneous within the Atlantic and Pacific Oceans. There is some indication that variability increases slightly with depth and that the Atlantic is more variable than the Pacific:. These tendencies are statistically significant, but the data base is not large. During this time interval, northern hemisphere glaciations are thought to have begun (Shackleton and Opdyke, 1977), Although variance estimates for the Pacific Ocean are only slightly less than those of the mid-Miocene ice buildup in Antarctica, the amount of variability estimated for this interval of northern hemisphere glacial buildup is no greater than that found in the shallower waters of the Atlantic Ocean in the Late Miocene and is nearly the same as that estimated for the Pacific Ocean during the early Paleogene. Thus, relatively high isotopic variability may be closely

associated with times when major changes occur in the average oxygen isotopic composition of the oceans (such as during glacial buildup), but they may also occur in association with the creation of new, isotopically different types of deep and intermediate water masses. If the later mechanism applies, large differences in the variability of benthic oxygen isotopes may be found between different oceans and between different depths in the same ocean.

The oxygen isotopic data presented here are discussed in terms of only one simple statistical characteristic—its variability. Even with this simple tool, major changes in the deep waters of the oceans can be discerned. Clearly, the record of paleo-oceanographic change is dependent not only on its position in time but also on its geographic and depth location. A more thorough understanding of the variability of the oceans awaits an evaluation of the spectral character of oceanic oxygen isotopic variability that could take into account the relative importance of long-term and short-term oscillations. Questions concerning how such variance spectra have changed with time and the likely mechanisms of such change await the gathering of longer time series and the further refinement of the geologic time scale.

We wish to express our appreciation to the CENOP project scientists for their thoughtful comments and criticisms of this work. Discussions with Sam Savin, Michael Bender, Nick Shackleton, and David Graham were particularly helpful. We also thank Fay Woodruff, Edith Vincent, Robley Matthews, Mike Sommers, Michael Bender, and Sam Savin for making their unpublished data available to us. This research was supported by National Science Foundation grants to the CENOP project, including OCE 79–14594 at the University of Rhode Island.

Boersma, A., and N.J.Shackleton (1977). Tertiary oxygen and carbon isotopic stratigraphy, Site 357 (mid latitude South Atlantic), in Initial Reports of the Deep Sea Drilling Project 39, U.S. Government Printing Office, Washington D.C., pp. 911–924.

Boersma, A., N.Shackleton, M.Hall, and Q.Given (1979). Carbon and oxygen isotope records at DSDP site 384 (North Atlantic) and some Paleocene paleotemperatures and carbon isotope variations in the Atlantic Ocean, in Initial Reports of the Deep Sea Drilling Project 43, U.S. Government Printing Office, Washington, D.C., pp. 695–717.

Briskin, M., and W.A.Berggren (1975). Pleistocene stratigraphy and quantitative paleo-oceanography of tropical North Atlantic core V16–205, Late Neogene Epoch Boundaries, Micropaleontology, 167–198.

Craig, H., and L.I.Gordon (1965). Isotopic oceanography: Deuterium and oxygen-18 variations in the ocean and marine atmosphere, in Symposium on Marine Chemistry, D.R.Schinck and J.T.Corless, eds., Occas. Publ. No. 3 of Narragansett Mar. Lab., Graduate School of Oceanogr., U. of Rhode Island, pp. 277–374.

Douglas, R.G., and S.M.Savin (1973). Oxygen and carbon isotope analysis of Cretaceous and Tertiary foraminifera from the central North Pacific, in Initial Reports of the Deep Sea Drilling Project 17, U.S. Government Printing Office, Washington, D.C., pp. 591–605.

Douglas, R.G., and S.M.Savin (1975). Oxygen and carbon isotope analyses of Tertiary and Cretaceous microfossils from Shatsky Rise and other rise sites in the North Pacific, in Initial Reports of the Deep Sea Drilling Project 32, U.S. Government Printing Office, Washington, D.C., pp. 509–520.

Douglas, R.G., and F.Woodruff (1981). Deep sea benthic foraminifera, in The Sea, Vol. 7, C.Emiliani, ed., Wiley-Interscience, New York.

Fischer, A.G., and M.A.Arthur (1977), Secular variations in the pelagic realm, in Deep Water Carbonate Environment, H.E. Cook and P. Enos, eds., Soc. Econ. Paleontol. Mineral. Spec. Publ. No. 25, pp. 19–50.

Heath, G.R., T.C.Moore, and T.H.van Andel (1977). Carbonate accumulation and dissolution in the equatorial Pacific during the past 45 million years, in The Fate of Fossil Fuel CO₂in the Oceans, N.R.Anderson and A.Malahoff, eds., Plenum, New York, pp. 627– 640.

Keigwin, L.D., Jr. (1979a). Late Cenozoic stable isotope stratigraphy and paleontology of DSDP sites from the eastern equatorial and central Pacific, Earth Planet. Sci. Lett. 45, 361–382.

Keigwin, L.P., Jr. (1979b). Cenozoic stable isotope stratigraphy, biostratigraphy, and paleoceanography of deep-sea sedimentary sequences, Ph.D. dissertation, U. of Rhode Island, 188 pp.

Keigwin, L.D., Jr. (1980). Paleoceanographic change in the Pacific at the Eocene-Oligocene boundary, Nature 287, 722–725.

Keigwin, L.D._, Jr., M.L.Bender, and J.P.Kennett (1979). Thermal structure of the deep Pacific Ocean in Early Pliocene, Science 205, 1386–1388.

Kennett, J.P. (1978). The development of planktonic biogeography in the southern ocean during the Cenozoic, Mar. Micropaleontol. 3, 301–346.

Kennett, J.P., and N.J.Shackleton (1976). Oxygen isotopic evidence for the development of the psychrosphere 38 Myr ago, Nature 260, 513–515.

Kennett, J.P., N.J.Shackleton, S.V.Margolis, D.E.Goodney, W.C.Dudley, and P.M.Kroopnick (1979). Late Cenozoic oxygen and carbon isotopic history and volcanic ash stratigraphy: DSDP site 284, South Pacific, Am. J. Sci. 279, 52–69.

Matthews, R.K., and R.Z.Poore (1980). Tertiary δ¹⁸O record and glacial-eustatic sea-level fluctuations, Geology 8, 501–504.

Norton, I.O., and J.G.Sclater (1979). A model for the evolution of the Indian Ocean and the breakup of Gondwanaland, J. Geophys. Res. 84, 6803–6830.

Pisias, N.G., and T.C.Moore (1981). The evolution of Pleistocene climate: A time series approach, Earth Planet. Sci. Lett. 52, 450–458.

Ruddiman, W.F. (1971). Pleistocene sedimentation in the equatorial Atlantic: Stratigraphy and faunal paleoclimatology, Geol. Soc. Am. Bull. 81, 283–302.

Savin, S.M., and H.W.Yeh (1981). Stable isotopes in ocean sediments, in The Sea, Vol. 7, C.Emiliani, ed., Wiley-Interscience, New York.

Savin, S.M., R.G.Douglas, and F.G.Stehli (1975). Tertiary paleotemperatures, Geol. Soc. Am. Bull. 86, 1499–1510.

Shackleton, N.J. (1967). Oxygen isotope analyses and Pleistocene temperatures re-assessed, Nature 215, 15–17.

Shackleton, N.J. (1977). The oxygen isotope stratigraphic record of the Late Pleistocene, Phil. Trans. R. Soc. London B 280, 169–182.

Shackleton, N.J., and M.B.Cita (1979). Oxygen and carbon isotope stratigraphy of benthic foraminifers at site 397: Detailed history of

climatic change during the Late Neogene, in Initial Reports of the Deep Sea Drilling Project 47, U.S. Government Printing Office, Washington, D.C., pp. 433–445.

Shackleton, N.J., and J.P.Kennett (1975a). Paleotemperature history of the Cenozoic and the initiation of Antarctic glaciation: Oxygen and carbon isotope analyses of DSDP sites 277, 279, 281, in Initial Reports of the Deep Sea Drilling Project 29, U.S. Government Printing Office, Washington, D.C., pp. 743–755.

Shackleton, N.J., and J.P.Kennett (1975b). Late Cenozoic oxygen and carbon isotopic change at DSDP site 284: Implications for glacial history of the Northern Hemisphere and Antarctica, in Initial Reports of the Deep Sea Drilling Project 29, U.S. Government Printing Office, Washington, D.C., pp. 801–807.

Shackleton, N.J., and N.D.Opdyke (1973). Oxygen isotope and paleomagnetic stratigraphy of equatorial Pacific core V28–238: Oxygen isotope temperatures and ice volumes on a 10⁵ and 10⁶ year scale, Quat. Res. 3, 39–55.

Shackleton, N.J., and N.D.Opdyke (1976). Oxygen-isotope and paleomagnetic stratigraphy of Pacific core V28–239 late Pliocene to latest Pleistocene, in Investigations of Late Quaternary Paleoceanography and Paleoclimatology, R.M.Cline and J.D.Hays, eds., Geol. Soc. Am. Mem. 145, pp. 449–464.

Shackleton, N.J., and N.D.Opdyke (1977). Oxygen isotope and paleomagnetic evidence for early northern hemisphere glaciation, Nature 270, 216–219.

Thompson, C.M., and M.Merrington (1944). Tables for testing the homogeneity of a set of estimated variances, Biometrika 33, 296– 304.

Vail, P.R., R.N.Mitchum, Jr., and S.Thompson (1977). Global relative changes of sea-level, in Seismic Stratigraphy—Application to Hydrocarbon Exploration, C.E.Payton, ed., Am. Assoc. Petrol. Geol. Mem. 26, Part IV, pp. 83–98.

Vergnaud-Grazzini, C. (1979). Cenozoic paleotemperatures at site 398, eastern North Atlantic: Diagenetic effects on carbon and oxygen isotopic signal, in Initial Reports of the Deep Sea Drilling Project 47, U.S. Government Printing Office, Washington, D.C., pp. 507–512.

Vincent, E., J.S.Killingley, and W.H.Berger (1980). The magnetic Epoch 6 carbon shift: A change in the oceans ¹³C/¹²C ratio 6.2 million years ago, Mar. Micropaleontol. 5, 185–203.

Woodruff, F., S.Savin, and R.Douglas (1981). A detailed study of the Mid-Miocene isotopic record and its paleoclimatic implications, Science 212, 665–668.