Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Quaternary Sea-Level Change ROBLEY K. MATTHEWS Brown University ABSTRACT Attempting to understand Quaternary sea-level history provides a vigorous intellectual workout. After negotiating a long path through data and concepts of mixed quality, one finds global ice volume has fluctuated by tens of meters sea-level equivalent at rates that are difficult to resolve. Satellite monitoring of global ice volume appears only prudent. INTRODUCTION From a geologist's point of view, the entire history of modern civilization is written within the context of an anomalously high stand of sea level. While 10,000 yr seems like an extremely long time in the history of man, it is an extremely brief time in the history of glacio-eustatic sea-level fluctuations. While late Quaternary sea-level events revealed by the geologic record are not likely to influence short-term decision making, they do serve as an awesome reminder of the dynamic geologic context of mankind. We will not spend vast sums planning for an event that is perhaps thousands of years into the future. However, it is equally certain that we shall not simply acknowledge the imminent demise of the human race. Knowledge must come first; planning can come later. Acquisition of knowledge in sedimentary geology is an interesting subject in itself. Historically, this field is driven by the inductive mode of investigation; more recently, the deductive mode is taking on new importance. In the in 88 ductive mode, one is driven by necessity or by curiosity. The task is begun with virtually zero understanding, but with faith that some new truth will emerge if one works at it hard enough. The emphasis is on data-gathering first, followed by ad hoc explanation of the data later. Con- versely, in the deductive mode one attempts to begin with a general a priori model and then add sparse amounts of carefully chosen new data to solve problems where the model is in obvious need of refinement. Intellectual conflict arises when the scientist in a de- ductive mode balks at "explaining" vast amounts of sub- stantially irrelevant information compiled by the scientist operating in an inductive mode. The inductive scientist does not like being told that his study area is too compli- cated to be fit to a general model; the deductive scientist does not wish to make his model unnecessarily compli- cated simply to fit every random scrap of data. By way of example, classic inductive investigation of late Cenozoic sedimentary geology is rooted in the winter vacations of wealthy Englishmen. Nevermind that the
QUARTERNARY SEA-LEVEL CHANGE Mediterranean region is an extremely complicated tec- tonic area; it was a great place to spend the cold winter months. The local fossiliferous strata constituted a very interesting hobby. Surely if one described enough strati- graphic sequences in enough detail, some new truth would emerge. To this day, a "classic locality syndrome" re- mains in force. Because much is known about these strata, some workers continue to return to them to learn still more, regardless of the fact that we do not understand their regional tectonic history. More recently, we find a lot of inductive geologic in- vestigation driven by government edict. The geologic mapping of a nation is substantially an inductive process. One does it because the government is paying one to do it. More to the point at hand, one cannot fault a government for inductive investigation of coastal stratigraphy and neotectonic history with regard to the siting of nuclear power stations, for example. However, the fact that these inductive investigations needed to be carried out by no means ensures that the resultant data will be worthy of incorporation into a priori models of Quaternary sea-level history. This chapter takes a deductive approach to the question of Quaternary sea-level change. The 6~80 record in deep- sea cores provides the best opportunity to obtain a con- tinuous record of Quaternary (perhaps even Cenozoic) sea-level change. There is elegant simplicity to the state- ment that 6~80 variation in deep-sea cores reflects changes in the isotopic composition of sea water and thereby re- flects changes in continental ice volume. The major focus of this chapter will be to add simple refinements to this a priori model. The prospects for future ice-volume vari- ation and the prospects for a more predictive approach to stratigraphy will then be examined. APPROPRIATE (AND INAPPROPRIATE) TECHNOLOGY There are several technologies that are appropriate to attacking problems in the late Quaternary. These are re- viewed below. Further, the inductive mode (and the need to come to some conclusion concerning a practical prob- lem) often leads the geologist to use inappropriate technol- ogy. One's mind becomes set on gathering all the data one possibly can; one loses sight of the fact that much of the data may be seriously flawed. These situations are like- wise reviewed below. Stratigraphic Context of Study Materials To obtain a local sea-level history on the 103- to 104-yr time scale, one must have a stratigraphy that can be related 89 to sea level and one must be able to date events within that stratigraphy. There are lots of good types of data to work with and there are lots of pitfalls. Morphostratigraphic Units To work with the strati- graphic record of sea-level events, it is highly desirable to establish formal morphostratigraphic units. Coral-reef terraces and former strand-line complexes in elastic sedi- ments are examples of morphostratigraphic units. They can be mapped as a physical topographic or bathymetric feature and exist as stratigraphic units regardless of age or age relationships among units. A reasonably formal defi- nition of morphostratigraphic units has been followed with regard to coral-reef terrace sequences on Barbados (Bender et al., 1979), New Guinea (Bloom et al., 1974), and Haiti (Dodge et al., 19833. To a lesser extent, a similar formal- ism has been applied to fossil strand-line deposits in clas- tic environments of the U.S. Gulf coast and east coast (see Cronin, 1983, for an excellent review). Layer-Cake Stratigraphy In the absence of morpho- stratigraphic units, geologists often fall back to physical stratigraphic relationships defined in vertical profile such as a core or borehole. In some cases, solid inference can be derived from age dating within such a physical strati- graphy. A good example of this is the dating of Holocene peat deposits that immediately overlie late Pleistocene subaerial exposure surfaces (e.g., Nelson and Bray, 1970~. Bad examples of this practice exist especially where ge- ologists correlate on the basis of questionable radiometric dates rather than on the basis of physical characteristics or stratigraphic continuity. Deep-sea cores from many portions of the world ocean sample an extremely reliable "layer-cake stratigraphy." Sedimentation rates commonly can be assumed to be nearly constant for long periods of time. Magnetic stratigraphy and 6'80 variation can be correlated globally (e.g., Imbrie et al., 1984). Peats Holocene brackish water peat deposits in close proximity to late Pleistocene subaerial exposure surfaces provide quite good stratigraphic indicators of paleosea level, which can be easily dated using the 14C method. Importantly, it is difficult to move peats around as sea level moves up and down. The peats would simply disin- tegrate if moved at all. Thus, to find peat resting on former subaerial exposure surfaces provides good indica- tion of the time at which sea level first inundated the former subaerial landscape. Care must be taken in interpreting the sea-level signifi- cance of peat dates taken from within thick sections of peat. Time lines within peat accumulation may be hard to define. Correction of present stratigraphic elevation of the
90 sample with regard to compaction of peat below may prove difficult. In the discussion that follows, only peat samples from close proximity to subaerial exposure surfaces below are used to establish sea-level history. Mollusc Shells Mollusc shells from dredge hauls are by far the most abused "datable material." If mollusc samples are to have stratigraphic integrity, one must be able to demonstrate that the shells have not been transported by rising sea level, subsequent current activity, seagulls, or man. Most commonly, this means recovering articulated pelecypod valves (e.g., Curray, 19601. Numerous dates in the literature undoubtedly represent shells that were trans- ported landward by rising sea level (see MacIntyre et al., 1978, for a good discussion). Still other mollusc shells dredged from bathymetric lineations are species that are known to live in tens of meters water depth. Thus, old shells are washed upwards to shallower depth than where they actually grew and younger shells represent deep subtidal environments where offshore molluscs thrive. A further problem, rampant in the '4C dating of mollusc shell, is the exposure of sea-bottom samples to Holocene diagenetic processes, such as boring and infilling by micro- organisms. A small amount of contamination by Holocene calcium carbonate can makeavery old mollusc sample look like 20,000- or 30,000-yr-old material when dated by ]4C. Corals Reef-crest corals (A. palmata in the Atlantic) can be outstanding indicators of paleosea level. Such corals live within a few meters of sea level and are not subject to extensive transport. They provide excellent material for }4C dating or for 230Th dating (e.g., Lighty et al., 1978, 1982; Mesolella et al., 1969~. Virtually all corals other than the reef-crest species have the obnoxious habit of living either in the shallow back-reef or the deep fore-reef environment. Thus, while they do contain dat- able calcium carbonate, their relationship to past sea level must be inferred from paleoenvironmental reconstruction. Clearly, the coral did not live above sea level; but beyond that, things get complicated. "Beach Rock" Some of those who work on submerged late Pleistocene and Holocene deposits take cementation of sandy sediments to be an indication of deposition and cementation in the beach environment. None of these studies offer sufficient documentation of the beach envi- ronment as the source of cementation in these materials. Submarine cementation is equally capable of producing a lithified sandstone (e.g., Land and Moore, 1980~. The claim that "beach rock" identifies a former intertidal envi- ronment must be held to close scrutiny. Further, all of the datable materials to be found in such sediments are subject to the problems noted above. ROBLEY K. MATTHEWS Geochronologyl Chronostratigraphy Given good stratigraphic context of study materials, there are several dating methods available for the study of Quaternary sea-level change. ]4C Dating Back to the last glacial maximum (approxi- mately 18,000 yr before present (18 kaBP)), the geologist can virtually "date at will" with regard to ice margin ad- vance on the land, with regard to sea-level fluctuations at the shoreline and with regard to downcore stratigraphy in deep-sea cores. In the best of worlds, one would hope that all the material ]4C dated would have met selection criteria noted above. In fact, this is not the case, and much of the existing data must simply be regarded as irrelevant to the questions under consideration. Uranium-Series Dating 230Th and 23iPa dating of arago- nite corals constitutes the backbone of late Pleistocene reef-terrace geochronology. Unfortunately, a large amount of literature provides only one or two dates per terrace. When one takes the time to make multiple analyses on multiple samples from the same terrace, one learns that there is a considerable spread to these data. Presumably, part of this spread is due to the complicated nature of the analysis (e.g., Harmon et al., 1979~. Undoubtedly, some of the spread is due to slight diagenetic alteration of sample in the subaerial environment (Bender et al., 1979~. Even more outrageous numbers are obtained when one tries to date materials from within the present-day phreatic lense. Uranium mobility is rampant, and such dates have no . . . ,%. geocnronolog~c s~gn~cance. The summary data of Harmon et al. (1979) with regard to three particular samples exemplify the problem faced with regard to geochronologic significance of 230Th dates within the time interval of the last interglacial. On the basis of impeccable geologic arguments, Rendezvous Hill (Barbados III) terrace, the Curacao +6 m terrace, and the Key Largo limestone of South Florida all represent deep- sea isotope stage Se (event 5.5 in the SPECMAP terminol- ogy; taken to be centered around 122 kaBP; Imbrie et al., 1984~. However, on the basis of 13 analyses of each sample, the 230Th coral-dating community came up with a pooled age estimate of 118 + 9, 124 + 12, and 139 + 19 kaBP for these samples, respectively (see Harmon et al., 1979~. Two points are worthy of note. First, there appear to be systematic differences among these three samples as re- flected by the respective means. Further, Barbados is significantly uplifted whereas Curacao and Key Largo limestone samples occur at an elevation presumed to rep- resent the actual high stand of the sea. If anything, the Barbados sample should appear somewhat older than the
QUARTERNARY SEA-LEVEL ClIANGE other two because an uplifting island would experience a null point with rising sea level earlier in the history of sea- level rise and subsequent high stand of the sea. In fact, the reverse is true; the Barbados date is the youngest of the three examples. Very likely, some small amount of postde- positional alteration has affected at least two of three of these samples. Indeed, x-ray data indicate considerable calcite in the Key Largo sample; but the other samples appear quite satisfactory. Next, consider precision of these age determinations. Even with 13 analyses of aliquots of the same sample, precision (1~) remains in the range of 10,000 yr. This is quite large in comparison with the chronostratigraphic precision attainable from deep-sea cores by stacking and spectral tuning of the INTO record (discussed below). Thus, even under the best of circumstances uranium- series dating in the range of 125 kaBP (an age range extremely important in the geologic record)-uranium- series dating affords precision and accuracy that are mar- ginal in comparison with the chronology available through the use of deep-sea cores. Where a sufficient number of analyses are available on individual terraces, the dates may serve as a crude constraint on correlations to deep-sea cores. However, discussion of whether "the 120-kaBP terrace" of one sequence is the same or a different glacio- eustatic event from "the 130-kaBP terrace" in another sequence must rest on geologic arguments, not on ura- nium-series dating. Further, one simply cannot trust dates on sparse samples from a single or a few terraces. Such reports are nothing more than a potentially interesting lead suggesting that more work might be desirable. Finally, one simply should not trust uranium-series dates on materials that were col- lected from within the modern freshwater phreatic tense or that can be demonstrated to have resided in a paleophreatic tense at some time in the history of the sample. Magnetic Stratigraphy Identification of the Brunhes/ Matuyama magnetic boundary in deep-sea cores is funda- mental for late Quaternary chronostratigraphic studies. The independent chronology provided by identification of the Brunhes/Matuyama boundary and the continuous nature of the deep-sea record are the prime reasons for attempting to correlate terrace and shoreline stratigraphy to the deep- sea record. One will never get the simple stratigraphic continuity of deep-sea cores in the complicated nearshore environment. Thus, it is important to develop a deductive research strategy in which the deep-sea record and chro- nology are the a priori model that one takes to the nearshore environment. The SPECMAP Time Scale The SPECMAP group has generated a continuous deep-sea chronology for the last 91 780 ka, which claims a precision on the order of 2 ka (95 percent confidence level) for major events in the deep-sea GINO ice-volume record. The technique utilizes a formal taxonomy for recognizable events in the INTO stratigraphy and utilizes orbital tuning to make local adjustments to sedimentation rate (Imbrie et al., 1984; Prell et al., 1986~. The fact that numerous individual records can be incor- porated into the final grand chronology promises continu- ing improvement of both chronology and the resultant INTO composite estimate of the glacio-eustatic signal. Further, application of the technology to the entire Ceno- zoic is technologically feasible. In my opinion, this is the chronostratigraphy that shall prevail. The ~80 Ice VolumelTemperature Relationship Calcium carbonate precipitated from sea water records some combination of information concerning the isotopic composition of the water and the temperature at which the calcium carbonate grew. Sizable variation in the isotopic composition of the sea water results from variation in global ice volume. If one can constrain global ice volume, one can read local temperature. Conversely, if one can con- strain local temperature, one can read global ice volume. This technology is applicable to planktonic foraminif- ers, benthic foraminifers, corals, and molluscs. In order to obtain useful results, one must further constrain strati- graphic variability by working with single taxa. Time- series data based on "mixed assemblages" of planktonic foraminifers, for example, may reflect variation in propor- tions of deep dwelling and shallow dwelling taxa with time. Similarly, there is considerable variation among benthic foraminifer taxa (Graham et al., 1981~. Records based on "mixed benthics" or on several different benthics strung together in time series must be considered suspect. Similarly, some taxa can be shown to have considerable variation in isotopic composition as function of size of the specimen (Curry and Matthews, 1981~. Whether this re- flects seasonal variation or variation in habitat among juveniles and adults remains to be unraveled. Finally, variation in degree of dissolution of deep-sea sediments as a function of proximity to the lysocline can vary the iso- topic composition of individual taxa by as much as 0.5 per mil (Peterson and Prell, 19851. The above-cited difficulties notwithstanding, many deep- sea cores yield a remarkably similar global INTO signal. This signal, taken in conjunction with magnetic stratigra- phy, provides the data base for stacking and spectral tun- ing, which produces a remarkably precise a priori model of the global INTO signal (Imbrie et al., 1984; Prell et al., 19861. The isotope-stage terminology of Imbrie et al. (1984) is adopted here. Whether this signal should be
92 interpreted as a global ice-volume signal or a temperature signal remains a main point for discussion below. d'80 and ~3C Relationships to Subaerial Exposure Surfaces Stable isotope data also provide incontrovertible identi- fication of subaerial exposure surfaces within stratigraphic sequences of bank margin carbonates (Allen and Mat- thews, 19821. Recrystallization of unstable aragonite and high-magnesium calcite in proximity to soil gas imparts a highly negative 6~30 signal to sediments immediately be- low subaerial exposure surfaces. Similiarly, recrystalliza- tion of various unconformity-bounded packages of car- bonate sediment generally occurs under slightly different conditions of global ice volume and/or climate. Thus, unconformity-bounded packages of carbonate sediment commonly display slightly different SILO values from those sediments above and below. A detailed stratigraphy of sedimentation at or near sea level, followed by subaerial exposure, makes the bank margin carbonate environment a veritable gold mine of information concerning glacio-eustacy (e.g., Major and Matthews, 19831. THE 18 kaBP TO PRESENT TIME INTERVAL This time interval encompasses the transition from the last glacial maximum to the present and is within the range of convenient ]4C dating. Latest Pleistocene Maximum Continental Ice Volume The 18-kaBP perimeter of continental ice sheets is reasonably well known. Calculations based on this pe- rimeter and assumptions concerning ice mechanics of equilibrium glaciers allow estimation of the volume of ice tied up on land during times of maximum glaciation (Demon and Hughes, 1981~. Estimates for ice-volume sea-level equivalent tied up in equilibrium ice sheets range from as high as 163 m to as low as 102 m. Importantly, all of these calculations presume the ice sheets were at equilibrium. It is possible for ice sheets to exist for long periods of time at less than these theoretical equilibrium ice volumes. Thus, the various estimates constitute formal statements of the maximum ice volume given different assumptions. It is important that confirmation be sought elsewhere for the amount of ice tied up in continental glaciers. Lowest, Low Sea-Level Estimates for the 18-kaBP Shoreline When we seek to confirm maximum sea-level lowering caused by the maximum extent of the 18-kaBP continental ROBLEY K. MATTHEWS ice sheets, we find few high-quality data points below 40 m. Two data points from the Texas Gulf Coast at -57 m and 12,900 + 400 yrBP for paired valves of the brackish water bay clam, Rangia cuneata, tell a very clear story (Curray, 1960~. Similarly, a freshwater peat off southern New England at-56.5 m and 12,320 + 350 yrBP appears trustworthy (Oldale and O'Hara, 1980~. Thus, we are left with a high-quality number for maxi- mum sea-level lowering at 18 kaBP that is both small and young in comparison with the numbers suggested by con- tinental ice-volume calculations cited above. There is a lot of bathymetry suggestive of a sea stand at around -120 m (e.g., Curray, 1960), but it is poorly dated. These features certainly look like good candidates for 18-kaBP shorelines, but this common assertion has not been satis- factorily proven. We definitely want to know maximum sea-level lower- ing during the time of maximum ice volume on the conti- nents. This question will be examined below with regard to the transition from isotope stage 6 to isotope stage 5.5 (Se). Submergence Curves for 12 kaBP to Present High-quality data are more readily available for the shallow end of the sea-level rise curve. Key papers (Gould and McFarland, 1959; Curray, 1960; Coleman and Smith, 1964; Scholl, 1964; Redfield, 1967; Scholl and Stuiver, 1967; Schnable and Goodell, 1968; Baltzer, 1970; Nelson and Bray, 1970; Thom and Chappell, 1975; Hopley et al., 1978; Lighty et al., 1978; Field et al., 1979) have been reviewed with regard to criteria described above under stratigraphic context of sedimentary materials. With re- gard to Bermuda, 12 data points meet these criteria and range back to -21 m at 9100 yrBP; Florida, 22 data points ranging back to -27 m at 9400 yrBP; Texas/Louisiana Gulf coast, 15 data points ranging back to -58 m at 13 kaBP; and South Pacific (Australia and New Caledonia), 11 data points ranging back to -37 m at 9700 yrBP. All of these curves show sea level to be within less than 5 m of present sea level by approximately 5000 yrBP. Differ- ences among these curves in this time range are taken to reflect continuing isostatic adjustment to unloading of late Pleistocene glacial ice (e.g., Peltier, Chapter 4, this vol- ume). Calibration of Deep-Sea ~80 Signal as Continental Ice Volume The deep-sea 3~8O record from 18,000 yr to the present is well known and well studied (see Mix and Ruddiman, 1985, for examples). The amplitude of the planktonic signal in high-sedimentation-rate cores averages 1.7 + 0.1 per milt Inasmuch as there is no well-dated confirmation
QUARTERNARY SEA-LEVEL CHANGE of the 18-kaBP shoreline (discussed above), estimates of calibration between the deep-sea 8~80 record and the shore- line submergence record must be based upon i4C correla- tion of the few good shoreline data (cited above) with the high-quality deep-sea isotopic stacked record of Mix and Ruddiman (1985, Table 41. The four independent esti- mates thus resulting yield a value of 0.025 per mil per meter for the calibration between deep-sea INTO fluctua- tion and shoreline fluctuation during the Holocene sea- level rise. If this change in 6~80 value as a function of changing sea level were entirely an ice-volume effect, it would require that glacial meltwater be returned to a well- mixed ocean with an isotopic composition on the order of -9S per milt This is a seemingly outrageously negative number; glaciers today range from -20 to -50 per milt To escape this dilemma, we might also pursue the possibility that (1) there is systematic offset in the two age models; (2) the shape of the geoid has changed, thus affecting the observed depth difference between paleoshorelines and present sea level; (3) sea-surface warming explains part of the isotopic curve from 13 kaBP to the present; or (4) that surface water of the world ocean (or specifically, the tropi- cal Atlantic study area of Mix and Ruddiman) was con- taminated by unusually negative precipitation at the time of glacial melting. THE 180- TO 18-kaBP TIME INTERVAL Interest in this time interval stems from two facts; one technological and one scientific. This is the time interval for application of the uranium-series dating techniques. This time interval also encompasses the second to last major glaciation (isotope stage 6) and the last interglacial (isotope stage S.S, formerly Se). In examining this time interval, we will pick up further information concerning maximum sea-level fluctuation from glacial to interglacial conditions and we will pick up important calibration con- cerning the relative importance of ice-volume variation and temperature variation in the observed 6~80 curve from deep-sea cores. Coral-Reef Terrace Sequences Absolutely essential to the calibration of the deep-sea 6~80 record as an ice-volume signal is the stratigraphy and geochronology of coral-reef terraces on tectonically emerg- ing islands. On the basis of the record within the range of ]4C dating, it appears that only the 6~80 signal in deep-sea cores covaries with the sea-level signal from submerged shorelines. Within the time interval from 180 to 70 kaBP, a number of coral-reef terraces are observed that are in substantial agreement with a well-defined 6~80 signal in deep-sea cores. 93 Well-Dated Morphostratigraphic Units At least three morphostratigraphic sequences have been relatively well dated using uranium-series methods. These are the up- lifted coral-reef terraces of Barbados (Mesolella et al., 1969; Bender et al., 1979), New Guinea (Bloom et al., 1974; Aharon, 1983), and Haiti (Dodge et al., 1983~. In each of these areas, distinct morphostratigraphic units can be recognized, which correspond to the Barbados terraces- Rendezvous Hill, Ventnor, and Worthing (approximately 125, 105, and 82 kaBP, respectively). In each of these study areas, the morphostratigraphic units stand indepen- dent of radiometric dating. Additionally, the New Guinea terrace sequence contains several younger terraces that are not represented on Barbados or Haiti because the uplift rate of Barbados and Haiti is considerably less than the uplift rate of the New Guinea terrace sequence. The Constant Uplift Rate Hypothesis Historically, a great deal of literature concerning relative sea level within the time interval 140 to 70 kaBP hinges on application of the "constant uplift rate hypothesis" (Broecker et al., 1968) to Barbados, New Guinea, Haiti, and elsewhere. The "125- kaBP terrace" has been considered to represent a +6-m high stand of the sea. With this te~Tace as the "known" datum, uplift rates for the various regions were calculated. Most such calculations indicate that sea level at 105 and 82 kaBP was considerably below present sea level. Mat- thews ~ 1973) showed that relative sea-level estimates can be made independent of the 125-kaBP sea-level assump- tion where terrace sequences with differing (but constant) uplift rates can be combined into a calculation. Unfortunately, the constant uplift rate hypothesis is substantially ad hoc. With the work of Stockmal (1983), this assumption seems reasonable for a relatively simple tectonic setting such as Barbados, but this assumption is not easily justified for such complex areas as New Guinea or Haiti. Here, scientific strategy bifurcates. Historically, I have argued to simply consider tectonic uplift a convenient mechanism for delivering coral-reef terraces into the subaerial environment where they may be easily sampled for geochronologic and isotopic studies. If we had to rely on submarine observations for our knowl- edge concerning sea-level stands at 105 and 82 kaBP, we would still be groping around in the dark ages of Pleisto- cene geochronology. We should be thankful for the uplift- ing of these terraces; but should not rely on the "constant uplift hypothesis" to tell us anything about relative sea level. Alternatively, some scientists argue that sea-level esti- mations based on the constant uplift hypothesis constitute an independent check on the isotopic estimates. In recent years, this approach has taken on still greater appeal be- cause of possible uncertainty in isotopic sea-level esti- mates introduced by the possible existence of thickened
94 Arctic ice pack (e.g., Broecker, 1975; Williams et al., 1981~. In deference to this latter uncertainty, I here give serious consideration to sea-level estimates based on the constant uplift hypothesis. Note, however, that this hy- pothesis remains substantially ad hoc and can be consid- ered valid at this time only on an empirical, statistical basis. Dodge et al. (1983) provided a detailed comparison between sea-level estimates derived from terrace elevation data in tectonically active areas and the isotopic record from deep-sea benthic foraminifers. They concluded that the deep-sea benthic 6~80 record is two-thirds a bottom- water temperature signal and only one-third a 6~80 ice- volume signal. There are two things wrong with this conclusion. First, it is misleading and erroneous to compare sea-level infor- mation to the benthic 6~80 record (e.g., Prentice and Mat- thews, 19881. Second, the isotope data of Fairbanks and Matthews (1978) for Barbados terraces (Worthing, Vent- nor, and Rendezvous Hill) are slightly inconsistent with data on new samples collected specifically to test the re- sult concerning paleosea-level elevation indicated for Barbados terraces Worthing and Ventnor. These two problems are examined in more detail below. The DO Record from Coral-Reef Terraces and Deep-Sea Cores Numerous deep-sea cores reveal a remarkably consistent 6~80 time series from planktonic and benthic foraminifers within this time interval. With the approximate chronol- ogy of the deep-sea record established from constant sedi- mentation rate assumptions and the Brunhes/Matuyama boundary, and with approximate chronology of coral-reef terraces established from uranium-series dating, it is pos- sible to undertake comparison of the INTO record from deep-sea cores with the 6~80 record from coral-reef ter- races. Such an effort has been quite successful on Barbados (Fairbanks and Matthews, 1978) and on New Guinea (Aharon, 1983~. Both attempts yield similar results; the Barbados record is stressed here because it contains more information concerning maximum glacio-eustatic sea-level fluctuation in the time interval of 170 to 130 kaBP. After discussion of the coral-reef terrace 6~80 record, I will return to integration of these data with the deep-sea 6~80 record and with sea-level estimates based on the coral-reef terrace constant-uplift hypothesis discussed above. The Barbados Coral-Reef Terrace ~ ~ SO Record Figure 5.1 presents a cross section of the Christ Church region of Barbados indicating the location of numerous cored bore- holes and the position of a subaerial exposure surface ROBLEY K. MATTHEWS encountered in these boreholes. Oxygen isotope data have been obtained on numerous samples of reef-crest coral (A. palmata) from terraces and from borehole materials above and below the prominent subaerial exposure surface indi- cated in Figure 5.1. Figure 5.2 presents a plot of INTO data versus actual present-day elevation of the A. palmata samples. We presume that uplift of the island of Barbados is relatively small in comparison to sea-level fluctuations that have occurred within the time interval under consid- eration. Thus, uplift rate, if present, may be considered as "noise" with reversed sign for rising versus falling sea level. Note that the slope of the 6'8O versus elevation relationship for Kendal Hill to borehole 20 (regressive) is quite similar to the slope for the borehole 20-Rendezvous Hill transgressive relationship. The data of Wagner (1983) confirm this relationship of 6~80 versus elevation for re- crystallization products of phreatic lenses related to vari- ous late Pleistocene stands of sea level. Figure 5.3 plots INTO variation versus age for Barbados A. palmata data and for data on planktonic foraminifers from deep-sea cores. The general agreement between these two records is unquestionable; there remains only some small points concerning detail. These are dealt with below. CHRIST CHURCH RIDGE C ROS S S E C T I ON KINGS. AND ( BARBADOS RENDE2V~ o > 120: h h J ;2 6 0 At: en 40 _ 100 J > 8 0 IS J ~ O ~ O J al ,_ -40 _ 2, ~ ,' PLEISTOCENE LIMESTONE TERTIARY LIMESTONE VERT EXAC. I5X FIGURE 5.1 Cross section of the southern flank of Christ Church Ridge, Barbados, indicating coral-reef terrace relationships, lo- cation of cored boreholes, and the existence of a buried caliche profile (dashed line) that records subaerial exposure during deep- sea isotope stage 6. Samples of the surf zone coral (A. palmata) were taken from immediately beneath the subaerial exposure surface in RKM 20, 22, and 23 and from immediately above the subaerial exposure surface in RKM 20. Isotopic data for these materials are plotted against elevation in Figure 5.2. (From Fairbanks and Matthews, 1978, with modification.)
QUARTERNARY SEA-LEVEr CHANGE -~ u -20 ~ T ~ I I ( BARBADOS Ill ) I R E ~ D E Z V O U S ~ I L L | (-0.012 ) KENDAL HILLY 1 1 1 (-0 010) R K M # 20 Tat- - - (- 0 022 ) ~M#20R (-0 048) 1/ fRKM # 22 R _ -I , , , , t , -40 -30 -20 -10 0 10 20 ELEVATION RELATIVE TO PRESENT SEA LEVEL (M) Frozen Bottom-Water Hypothesis The use of the benthic 6~80 record as a possible indicator of ice-volume fluctua- tions is rooted in the clever argument of Shackleton (19671. At a time when Emiliani (1966a, for example) was talking about a 4° to 6°C colder tropical sea-surface temperature (SST) during glacial times, and small ice-volume compo- nent to the bi80 signal, Shackleton took the trouble to analyze relatively scarce benthic foraminifers. The result was astounding. The benthic 6~80 curve was similar in shape and amplitude to the planktonic 6~80 curve of Emili- ani. Bottom water could not possibly have gotten 4° to 6°C colder during glacial conditions because this would put the water below the freezing point. The logical conclu- sion was that Emiliani had greatly overestimated the tem- perature effect and greatly underestimated the ice-volume effect in both of these signals. Shackleton proposed that the benthic bi80 signal was more like two-thirds ice-vol- ume signal and one-third temperature signal, although the precise relation could not be determined from the data at hand. Importantly, there is no a priori reason to assume con- stant bottom water temperature during the glacial/intergla- cial cycle. A priori, the bottom water shall not freeze, but that is about it. Beyond this, there are no "magic num- bers" for bottom-water production. Bottom water is sim- ply that combination of temperature and salinity that is the most dense water. With glacial/interglacial changes in geometry of high-latitude marginal seas, all bets are off concerning precisely what temperature/salinity combina- tion will end up being the glacial world's bottom water. Thus, the initial success of Shackleton (1967) must be viewed as substantially empirical and semiquantitative. If 95 30 40 50 ~ BARBADOS DATA ~ ECENT -3.0 ._ a, -2.0 Q o 0 a) n n FIGURE 5.2 Plot of mean isotopic com- position of Barbados A. palmata samples against elevation relative to present sea level. Assuming the rate of tectonic uplift to be small in relation to the rate of glacio- eustatic change, these data constitute a calibration of 6~80 as a function of glacio- eustatic sea-level fluctuation. A relation- ship of 0.011 per mil per meter is taken as the glacio-eustatic effect. More positive 6~80 values at RK 20 and 22 are satisfied by CLIMAP estimates that this region was approximately 2°C cooler than modern during full interglacial conditions. (After Fairbanks and Matthews, 1978.) RENDEZVOUS HILL KINGSLAND VENTNOR-t KENDAL HIL+BERDARE WORTHING: ~ / \ if, ' RKM 2)T ~RKM 23R ~ ~RKM 20R DEEP SEA RECORD ~RKM 22R THOUSANDS OF YEARS BEFORE PRESENT FIGURE 5.3 Comparison of Barbados coral isotope record with the stacked deep-sea planktonic isotope curve of Prell et al. (1986). Error bars on isotopic data indicate statistical certainty about the mean at the 95 percent confidence level. Coral isotope data for Worthing, Ventnor, and Rendezvous Hill are from Table 5.2; the remainder are from Fairbanks and Matthews (1978). Note peak-for-peak correlation; Aberdare equals 7.5, and so forward to Worthing equals 5.1. (Similar to Fairbanks and Matthews, 1978, Figure 7; modified to include new data and a more up-to-date deep-sea planktonic 6~80 record.)
96 ROBLEY K. MATTHEWS one wishes to constrain temperature to be constant, one lated by many working on materials within Tertiary time must hook up to a 6~80 recording system other than hen- intervals. thic foraminifers. There are a priori arguments to suggest that the low-latitude sea surface may provide such a re cording system. Low-Latitude Constant Sea-Surface Temperature Hy pothesis Sea-surface temperature estimates based on total faunal census data and core top regression equations (CLIMAP, 1976, 1984) provided the (seemingly) final death blow to the Emiliani concept of a substantially cooler global SST during glacial times. Application of this tech nology to the 18-kaBP maximum glacial reconstruction indicates that some regions of the low-latitude ocean were slightly warmer at 18-kaBP maximum glacial time than they are today. Other large areas were only slightly cooler. Only where major shifts in upwelling occurred are low latitude SST greatly cooler at 18 kaBP than today. Prell (1985) applied alternative methods and basically confirmed the CLIMAP (1984) results, suggesting, if anything, still less tropical cooling than CLIMAP. Further, there are physical climatologic/oceanographic reasons to predict relatively constant SST of low-latitude ocean surface. The balance between radiation and latent heat flux dictates a tropical SST of about 28°C (Done and Shaw, 1977; Newell et al., 1978; among others). A1 though mechanisms can be envisioned that might cause this number to increase slightly (increased atmospheric CO2, for example), such mechanisms generate negative feedback (increased frequency of hurricanes, for example) that is poorly understood. At the present state of the art, calculations of atmospheric circulation arrive at 28°C tropi cal SST; this temperature is in generally good agreement with the CLIMAP empirical estimates for 18 kaBP. In light of existing data and experience, it would be difficult to justify an hypothesis of runaway warmer tropical SST. The possibility of cooler tropical SST is a different matter. Variation in planetary albedo, atmospheric CO2, or perhaps even solar insolation could result in a generally cooler planet and with it generally cooler tropical SST. Indeed, Manabe and Broccoli (1985) simulated a glacial world with higher albedo and lower atmospheric CO2 and thereby calculated a tropical SST on the order of 1° to 2°C cooler than modern. Importantly, this model result is opposed by the empirical temperature estimates of CLI MAP (1984) and Prell (1985) discussed above. Regardless of the ultimate resolution of this discrep ancy between model and empirical results, all global atmospheric circulation model experiments to date sug gest that a colder SST in the overall tropical regions would be accompanied by still yet colder polar regions. While this relationship is generally proposed by most working on late Pleistocene materials, the relationship is grossly vio The validity of the constant tropical SST hypothesis is an ongoing research problem. There is room for change. However, tropical SST is a profoundly important number with truly global ramifications. Change should not be proposed lightly. If at some time in the future modeling results move forward to still higher SST under certain conditions, these deviations from constant temperature hypothesis will be based on a priori models. Likewise, if the disagreement between model and empirical data con- cerning 18-kaBP SST shall be resolved in favor of cooler temperatures, these estimates of cooler temperatures will likewise be based on a priori models and thoroughly well- reasoned discussion. These new numbers would then become the basis for ice-volume calculations, and so on toward a perfect understanding of Earth history. With regard to isotopic estimation of sea-level history, the point here is that the low-latitude tropical sea-surface environment offers the best opportunity of constraining temperature variation and thereby reading out the glacio- eustatic ice-volume signal. This water mass is sampled by reef-crest corals and by surface-dwelling planktonic fo- raminifers. Indeed, if Dodge et al. (1983) had chosen to compare their terrace elevation data to the planktonic INTO record instead of the benthic INTO record, they would have arrived at a much more favorable comparison (a point to which I shall return below). Maximum Glacio-Eustatic Sea-Level Lowering at Iso- tope Stage 6 The data presented in Figures 5.2 and 5.3 afford an estimation of the amplitude of the glacio-eustatic signal in late Pleistocene time. The deep-sea INTO record clearly demonstrates that isotope stage 2 and isotope stage 6 are comparable examples of maximum glaciation during late Pleistocene time. The Barbados borehole data capture 1.7 per mil 6~80 variation within an elevation variation of approximately 80 m. Allowing for the fact that the Kendal Hill INTO value is somewhat heavier than Recent and al- lowing for local faulting (Wagner, 1983), the amplitude of sea-level variation from full interglacial to data point "RKM 22 (regressive)" could be estimated at 108 m. Making further allowance for the fact that local INTO amplitude in nearby deep-sea cores is approximately 2.0 per mil, one has an additional 0.3 per mil to play with as either local temperature signal or global ice-volume sea-level signal. Note in Figure 5.2 that the isotope versus elevation variation increases dramatically below the data point "RKM 20 (regressive)." This reflects a still more negative INTO value for late-stage continental ice, or a larger proportion of floating ice, or a local temperature decline during ex- treme glaciated conditions. Assuming the effect from data points "RKM 20 (regressive)" to "RKM 22 (regressive)"
QUARTERNARY SEA-LEVEL CHANGE to be solely an isotopic effect, the full 2.0 per mil ampli- tude reflects 114-m sea-level lowering relative to present. Alternatively, CLIMAP-18 kaBP AT estimates for this area are approximately nil for summer and approximately -2°C for winter. Assumption of a-2°C temperature dif- ference from interglacial to glacial conditions and a 1 20-m sea-level difference from present to maximum glacial conditions satisfies the isotopic maximum amplitude data in this area. Assuming the 108-m amplitude figure cited above for actual Barbados data with some straightforward corrections, a temperature difference of-3.3°C from inter- glacial to glacial conditions is required. All three of these estimates are probably within the uncertainty of the data. Comparison of Isotope Stage 5 Data for Terraces and Deep-Sea Cores As noted above, Dodge et al. (1983) erroneously compared terrace elevation data to the benthic rather than the Planktonic 6~80 record. A second source of error in the Dodge et al. (1983) calculation concerns the isotopic data reported by Fairbanks and Matthews (1978) for Barbados terraces Worthing, Ventnor, and Rendezvous Hill. The isotopic data contained in Fairbanks and Mat- thews (1978) for Barbados surface samples were taken on scraps of sample left over from radiometric dating and petrographic studies. In many cases, these were relatively small samples. Relatively small samples of A. palmata present a special problem in that there is strong seasonal banding within this coral. Given a small sample and nondescript geometry of banding, it would be easy to take a nonrepresentative sample for oxygen isotope analysis. To evaluate this problem, an entirely new set of dia- mond-drilled, 2-inch-diameter core samples were collected 97 expressly for isotopic study. The middle columns of Table 5.1 present Fairbanks and Matthews data and new data (from Table 5.2) concerning the isotopic difference among Barbados terraces Rendezvous Hill (isotope stage 5.5) as compared to Worthing (5.1) and Ventnor (5.3~. The left- hand columns present similar information from Dodge et al. (1983) estimates of elevation differences converted to INTO values by the calibration of Fairbanks and Matthews (19781. The right-hand column presents observed isotopic differences among Planktonic foraminifers based on the average of low-latitude deep-sea cores. With regard to Barbados isotope data, note that the new data indicate slightly less difference between the terraces. At a glance, the difference between the Fairbanks and Matthews (1978) data and the new data is not that excit- ing. Both data sets put the younger terraces low relative to Rendezvous Hill; the differences among means are within the range of overlapping confidence intervals; in 1978, the situation looked well under control. However, over the years there has been harping criticism concerning the Fairbanks and Matthews estimate that sea level repre- sented by the younger Barbados terraces might have been as much as 45 m below present sea level. To me, the situation was "close enough; move on to other things." To those inclined to be more fastidious, the new data will be somewhat reassuring. When one compares the coral-reef terrace elevation data of Dodge (left-hand column of Table 5.1) with the Barbados terrace isotope data (middle columns) and with the deep-sea record (right-hand column), one notes re- markable agreement between the new Barbados isotope data and the deep-sea cores. Estimates based on the ad hoc TABLE 5.1 Comparison of Terrace Elevation Data to Isotope Data from Corals and Low-Latitude Planktonic Foraminifers Terrace Elevation Barbados Corals Low-Latitude Data, 6~80 Fairbanks and Planktonic Isotope Stage equivalent,a Matthews (1978), New Data,b Foraminifer,C Comparison 6~80 (per mil) 6~80 (per mil) 6~80 (per mil) 6~80 (per mil) 5.1-5.5 +0.24 + 0.1 (3) +0.54 + 0.4 (7) +0.35 + 0.2 (9) +0.34 + 0.2 (5) 5.3-5.5 +0.21 + 0.1 (3) +0.52 + 0.1 (15) +0.35 + 0 2 (9) +0.30 + 0.1 (5) aElevation data from Barbados, New Guinea, and Haiti converted to 6~80 via Fairbanks and Matthews (1978~. bSee Table 5.2. CPlanktonic 6~80 data from the following cores and their respective sources were used: Core 280 (Emiliani, 1958), Core P6304-9 (Emiliani, 1966b), Core P6408-9 (Emiliani, 1978), Core V28-238 (Shackleton and Opdyke, 1973', and Core V22-174 (Thierstein et al., 19771. NOTE: Data are reported as (mean /~) + (confidence about mean, 95% C.L.~; number of independent estimates are in parentheses.
98 TABLE 5.2 Isotope Data on Barbados Corals ROBLEY K. MATTHEWS Sample Number 6,80 613C Sample Number 6~80 613C and Description (per mil) PDB and Description (per mil) PDB Worthing (Barbados I) FS-SOA Duplicate I Ave. FS-50A -3.47 -3.31 -3.39 +0.19 +0.08 +0.13 Maxwell ("New" terrace between Ventnor and Rendezvous Hill; basically part of Rendezvous Hill) AEJ-20 -3.65 +0.01 FS-51-3.420 96 AEJ-21 -3.65 -0.58 Duplicate B3.16+0.92 Duplicate G -3.82 0.52 FoSC5520A-3 61-1 19 AEJ--2120 -3 821 +0 28 OC-51-3 22+020 AGP 1l _3 50 +0.49 OC-53-3 20+0 64 -3.46 Sandy Lane- 1-2 94+0.55 Dupllcate D -3.61 +0.38 Sandy Lane-2-3.27-1.14 Duplicate D -3.39 +0.55 Sandy Lane-4-3.64+0.53 Ave. AGP-11 -3.49 +0.50 Duplicate H-3.61+0.62 AGP-12 -3.63 +0.15 Ave. Sandy Lane-4-3.62+0.58 Mean-3.280 17 Mean _3 57 +g 19 c,0.230.81 0.25 0.65 C 95 O. 8 0.6 r = ~.40 (not significant) r = +0.45 (not significant) n = 6 n = 9 Rendezvous Hill (Barbados III) Ventnor (Barbados 11) AFM-20A -3.74 +0.38 ANM-20 _3 03 -0 60 AFM-22A -3.75 +0.20 ANM-22 -3.30 0 37 R-50 - Duplicate E -3.30 -0.37 Duplicate J 3.56 +0.29 Ave. ANM-22 _3 30 -0 37 3.56 +0.46 50 3 Z3 +070 R-5 -344 +0.32 Dopllca e A 3 33 +o 44 D pl ca e F 3 3 AFS- O -3 60 +0.77 Ave. FT-50 ~3 33 +0-49 AFS FT-51 -3.65 -1.30 AFS- 12 -3.76 +0.10 FT-53 -3.55 -1.88 Mean -3.63 +0.1 l Duplicate C -3.25 -1.57 ~0.11 0.6~ Ave. FT-53 -3.40 -1.73 CLgs 0.08 0.4E BAB-10 -3.46 +1.18 r = -0.04 (not significant: BAB-11 -3.00 _0.75 n 9 BAB-13 -3.24 -0.18 Mean -3.28 -0.40 ~0.21 0.87 CL,s 0.16 0.67 r = +0.025 (not significant) n = 9 NOTE: During the summer of 1977 A. palmata was recol- lected on Barbados Terraces I, II, and III expressly to confirm the results reported by Fairbanks and Matthews (1978~. Whereas their subsurface data were taken on high-quality core materials, the terrace materials were leftover scraps from other projects. As such, samples were rather small and cut in undetermined relation to growth lines. For this reason, it was highly desirable to resample. Each of the three terraces was sampled three times at each of the three localities. Samples were taken with a 2-inch-diameter diamond core drill. Each core was approximately one foot in length. Samples for isotopic study were taken perpendicular to growth bands to represent the average of the whole coral. Isotopic data were collected on the Benedum Stable Isotopes Laboratory VG 602D over four days from November 20, 1978 to February 20, 1979. Precision (1~) on standards (6) was +0.00 for 6~80 and +0.18 for 6~3C. Precision (half range) on random duplicates (9) was +0.06 for both 6~80 and 6~3C. Averages were used in calculation of terrace statistics.
QUARTERNARY SEA-LEVEL CHANGE constant uplift hypothesis for terrace elevation (Dodge et al., 1983) are only slightly smaller than the actual isotope data for Barbados terraces and the deep-sea cores. The difference is well within the confidence interval for the various data subsets. To bring the means into literal agree- ment requires tropical sea-surface cooling on the order of only 0.5°C. Thus within the uncertainty of the data, the low-latitude planktonic INTO record is precisely a measure of global ice-volume fluctuation within the sea-level fluctuation range represented among the three younger Barbados terraces. On the basis of isotope data from boreholes, this precise relationship of INTO variation to change in elevation of sea level may be extended to the depth of 50 to 70 m below present sea level. Below this depth, the scant data that are available (Figure 5.2) suggest that a local temperature effect may become important in the depths of the glacial portion of cycles. The Deep-Sea I;'80 Record as an Ice-Volume Signal In summary then, terrace elevation data, Barbados iso- tope data, and deep-sea isotope data are in close agreement within the time interval of Barbados terraces Worthing, Ventnor, and Rendezvous Hill (82 to 125 kaBP). Given the uncertainty within each data set, no temperature effect is needed to bring these data into agreement; likewise, no hidden floating ice is required. The calibration of Fair- banks and Matthews (1978) is sufficient to the task in and of itself. Further, the data on which that calibration is based (namely, data points "RKM 23 regressive" and "RKM 20 transgressive" of Figure 5.2) indicate that ice volume alone is a sufficient explanation of the deep-sea b'8O vari- ation to a depth of sea-level lowering of approximately 50 to 70 m below present sea level, accounting for some- where between 0.5 and 0.8 per mil of global ocean INTO enrichment from interglacial conditions toward glacial conditions. In the region of Barbados, the lower half of the INTO record in coral materials and in nearby deep-sea cores is satisfied by a combination of continued ice-volume buildup and a local temperature lowering of approximately 2°C (a INTO effect of 0.5 per mil). Thus, although this combina- tion of numbers arrives at the classical Shackleton (1967) estimate of one-third AT effect and two-thirds /\ice-vol- ume effect, it is important to note that the temperature effect (if valid at all) is concentrated in the glacial half of the signal. Importantly, to the extent that this temperature effect may represent global cooling, it appears to be decid- edly the result of ice-volume buildup (presumably an al- bedo feedback mechanism); strikingly, this is the reverse of the common wisdom scenario that global cooling initi- ates continental glaciation. 99 THE PAST AS THE KEY TO THE FUTURE The discussion above has dealt largely with geologic information concerning "the numbers" regarding late Quaternary sea-level history. What was the level of the sea at what date in the past; how good are the data; how shall we resolve apparent conflicts? Geologists also have insight and opinion concerning the workings of the Earth as a dynamical system. Two possible scenarios deserve consideration here with regard to the possible ice-volume effects of future climatic warming. West Antarctic Ice Surge Accompanying Warm Interglacial High Stand Among the world's major ice sheets, the West Antarc- tic ice sheet seems to be especially precariously situated today. It is substantially a marine-based ice sheet, and the present-day topography over vast areas lies below the theoretical equilibrium profile. It can be argued that this ice sheet is capable of disintegrating over the next few hundred years. If such a glacial surge should occur, world sea level would rise by approximately 6 m. The geologic question is whether or not such a surge event has occurred in the past. There is no doubt that sea level at isotope stage 5.5 stood at around present sea level or slightly higher. The question with regard to possible West Antarctic ice surge is whether or not we can be confident that it stood 6 m above present sea level. Moore (1982) provided a conven- ient summary of 230Th dating from both tectonically active and tectonically stable regions. Seemingly tectonically stable regions giving seemingly reliable estimates for iso- lope stage 5.5 sea level above present sea level include the Bahamas, Bermuda, western Australia, Aldabra Atoll, and the Yucatan Pennisula, Mexico. Estimates of sea level relative to present range from +2 to +6 m. Another approach to the relative position of sea level at isotope stage 5.5 would be to compare 6~80 data to Holo- cene values. However, the differences that we seek to quantify are relatively small in comparison to analytical and geologic precision of the isotopic data, the isotopic difference between modern and +6 m sea-level high stand being only approximately 0.07 per milt With regard to deep-sea cores, the problem is compounded by nagging problems concerning the reliability of core-top materials. CLIMAP (1984, p. 206) notes the lack of consistency in their data and refers the question to dated coral reefs. With regard to Barbados coral data and New Guinea mollusc data, modern and isotope stage 5.5 equivalent materials yield comparable values within the uncertainty of the esti- mates. Encouragingly, the New Guinea data do indicate brief number VIIa (isotope stage 5.5 equivalent) to be 0.03 per mil lighter than modern; this is in the right direction,
100 but a difference of only about half the uncertainty of the two estimates. The isotopic data concerning New Guinea VIIb (Aharon, 1983) are especially troublesome with regard to the surge hypothesis. On the basis of geologic correlation and on the basis of reef-crest 6~80 values, New Guinea reef VIIa is surely isotope stage Se equivalent. (The fact that its date is slightly older than isotope stage 5.5 must be considered subordinate to these other arguments; see discussion above concerning problems with dating technology.) Aharon (1983) continued to regard New Guinea reef VIIb as a surge event that follows the deposition of reef VIIa. The interesting problem is that reef-crest bi80 values for VIIb are 0.5 per mil heavier than the values for VIIa. Aharon explained these highly anomalous 6~80 values by calling on dramatic global cooling accompanying a Wilson hy- pothesis surge event (see Flohn, 1979, for a convenient review of this and similar topics). It is important to note that materials similar to New Guinea reef VIIb have not been observed in any other terrace sequence. Therefore, it is possible that these unusual data have a local explanation and do not require integration into a global scenario. Nevertheless, a scenario such as set forward by Aharon (1983) could be consistent with the deep-sea 6~80 record. Inasmuch as rapid cooling accompanies the surge of iso- topically light water into the world ocean, the proposed high stand of the sea is lost somewhere in the transition from isotope stage 5.5 to 5.4. Taking all of these data literally, one is left with the unsatisfactory proposition that isotope stage 5.5 (equiva- lent to New Guinea reef VIIa) achieved a level somewhere between +2 and +6 m relative to present sea level by removal of an ice sheet in a manner inconsistent with the Wilson hypothesis, whereas isotope stage 5.5 was fol- lowed closely by a glacial surge (and resultant New Guinea reef VIIb) that was consistent with the Wilson hypothesis. At this point, belief or disbelief becomes largely a matter of taste. I would prefer to see the New Guinea reef VII data set enlarged and the result reproduced within some other terrace sequence before I proceed to make choices. At this stage in the science, the simplest choice remains that isotope stage 5.5 sea level is the same or slightly higher than modern for reasons that are not fully under- stood. While the West Antarctic surge hypothesis remains exciting, the observation could equally be explained by slight deviations of East Antarctic ice sheet from equilib- rium conditions by either more rapid flow or insufficient . . . precipitation. Ice Growth Adjacent to a Warm Ocean An equally credible scenario for the end of an intergla- cial can be written around rapid growth of continental ice ROBLEY K. MATTHEWS fed from a nearby warm ocean moisture supply. This scenario is best documented with regard to growth of North American ice sheets at the end of the last intergla- cial (isotope stage 5.5/5.4 transition) (Ruddiman et al., 1980~. By generating both planktonic faunal data and planktonic isotopic data at close spaced sample intervals, it is clearly demonstrated that the isotopic values become relatively heavy while the fauna continues to indicate rela- tively warm water. The simplest explanation for these observations is that North American ice sheets grew to considerable size (hence, relatively positive oceanic INTO values) while the nearby North Atlantic remained rela- tively warm. A similar scenario can be envisioned for the Southern Ocean at times throughout the Tertiary. In this case, relatively warm water would be provided to the surface ocean by upwelling to the south of the polar front. Such a mechanism might be similar to the occasional occurrence of ice-free conditions among winter pack ice around Ant- arctica today (the Weddell polynya; see e.g., Gordon, 1982; Gordon and Huber, 19841. The fact that the polynya remains ice free throughout winter is ample demonstration that upwelling water is transferring large amounts of latent heat to the high-latitude southern atmosphere. At times in the past, this effect has almost certainly been much larger. For example, it is likely that the deep ocean was approximately 8°C in Oligocene time. Upwelling of such warm water around Antarctica may have provided the local moisture supply for substantial growth of the Antarctic ice sheet at approximately the Oligocene-Eo- cene boundary (Matthews and Poore, 1980~. Looking to the future, if man's warming of the planet tends to warm intermediate and deep water (Roemmich, Chapter 13, this volume), then one might script an Antarc- tic ice-growth scenario such as outlined above. However, modern Antarctica has the further complication of large ice shelves that serve to buttress the glacial ice (tingle, 1984, for example). If these ice shelves were removed by upwelling of warmer water, the initial effect would surely be to increase the flow rate of ice streams. The new equilibrium between these two competing effects (increased precipitation and increased flow rates) is conjectural. The fact that both effects are plausible emphasizes the need to observe Antarctic surface-elevation changes from polar- orbiting satellites. GLACIO-EUSTACY IN STRATIGRAPHIC PREDICTION Given the high integrity of the deep-sea, low-latitude planktonic INTO record as a glacio-eustatic signal, there exists a significant opportunity for stratigraphy to move beyond a descriptive mode to a predictive mode. The
QUARTERNARY SEA-LEVEL CHANGE PREDICTED STRATIGRAPHY E 2 SEA LEVEL HISTORY 1'' __ - ~~ 5~7e~ 93 1 100 200 300 400 500 THOUSANDS OF YEARS BEFORE PRESENT BASIN SUBSIDENCE 1 00m,'my SUBSIDENCE RATE ( ISOSTATICALLY COMPENSATED AT 2:1 1 ~ FIGURE 5.4 Stratigraphy predicted for bank-margin carbonates where a well-known (or presumed) sea-level history is superim- posed on basin subsidence. Sea-level rise deposits new bank- margin carbonate sediment. Sea-level fall produces subaerial exposure surfaces. Diagonal dashed lines from sea-level high stands indicate subsidence that occurs before the next high stand of the sea. This sort of forward modeling holds great promise for bringing a new level of understanding to stratigraphy and thereby petroleum exploration. opportunity is before us to model an empirically derived glacio-eustatic signal with mathematical formulations of basin subsidence and isostasy, for example. By way of example, Figure 5.4 presents a proposed stratigraphy of subaerial exposure surfaces within bank- margin carbonate rocks. The model is generated by inter- action of the late Pleistocene average INTO curve (Prell et al., 1986) with a regional driving subsidence of 100 m per million years. In such a model, deposition occurs only when rising sea level crosses the subaerial exposure sur- face of previously existing bank-margin carbonate sedi- ments. As the sea-level curve tops out and heads once again toward glacial low-stand, yet another subaerial ex- posure surface is formed. Such a diagram is, then, a prediction of the stratigraphy that one would find if one drilled a borehole through a subsiding passive margin (such as the modern reef tract of Belize, Central America) that was subsiding at approximately this rate. While such a prediction in the late Pleistocene is academic, such predic- tions in older rocks could have extreme importance in petroleum exploration. Indeed, I join Christie-Thick et al. (Chapter 7, this volume) in criticizing the alleged seismic stratigraphy record of sea-level changes. This famous curve is nothing more than an ad hoc explanation of the data growing out of enumerable inductive mode studies of seismic profiles. However, whereas Christie-Thick et al. propose still further inductive mode investigations to reevaluate the ages of sequence boundaries, the approach 101 outlined in this chapter would deductively compare inde- pendent models for eustatic history (also see Harrison, Chapter 8, this volume) with models for basin subsidence. Initial attempts at applying a similar research strategy to Tertiary materials have been at least partially success- ful. Major and Matthews (1983) suggested that the shape of the Vail et al. (1977) sea-level curve for middle Mio- cene is approximately correct, but that the amplitude of their proposed sea-level fluctuations is probably exagger- ated by a factor of three. CONCLUSIONS 1. The low-latitude planktonic INTO record surely of- fers the most promise for delivering a continuous record of glacio-eustatic sea-level fluctuations back through time. On the basis of existing data, it appears that ice-volume fluctuations are sufficient to account for the structure in approximately the more interglacial half of the deep-sea INTO signal. The claim that this relationship holds true to a sea-level lowering of approximately 20 or 30 m relative to present is confirmed by isotopic data on surficial samples from coral-reef terraces of Barbados, New Guinea, and elsewhere. The claim that this relationship extends to a sea-level lowering of 50 to 70 m relative to present rests on two data points obtained by Barbados core drilling. These are very important numbers; similar drilling needs to be carried out in other parts of the world to confirm or reject the Barbados data. 2. With regard to the more glacial half of the deep-sea INTO signal, we face still further problems with scarcity of high-quality information. Estimates concerning the gla- cio-eustatic amplitude of isotope stage 6 rest solely on a single cored interval in Barbados RKM 22. Even here, there is no claim that this data point represents maximum low stand of the sea. The amplitude of the glacio-eustatic signal is surely a fundamental property of the dynamic system. We really should know that number quite well; if it turns out to be 80 m as opposed to 130 m, the result would significantly narrow uncertainties concerning con- tinental ice sheets. Once again, Barbados style core drill- ing in other parts of the world would seem to be indicated. Similarly, renewed attempts to sample and date strati- graphically significant material representing isotope stage 2 from submerged continental margins would appear to be worth unusual effort. 3. With regard to future glacio-eustatic sea-level fluc- tuations, it is noteworthy that the entirety of the civiliza- tion of man has occurred within a single eustatic high stand of the sea that is indeed quite anomalous to the average late Pleistocene condition of the Earth. Without human intervention, surely sea level shall once again fall to intermediate levels. With inadvertent human interven
102 lion, the system might go either way; sea-level rise in accordance with the West Antarctic ice surge hypothesis; or sea-level lowering in accordance with the warm ocean ice growth hypothesis. The precise nature of the geologic record with regard to these two possibilities is substan- tially i~Televent to planning concerning possible sea-level consequences of future climatic warming. Both scenarios are conceivable; we really do not know the time scales involved; best we simply plan ahead to monitor the Earth's ice budget henceforth and forthwith. 4. The advent of predictive stratigraphy would appear to be at hand. It is technologically feasible to construct a detailed glacio-eustatic sea-level curve throughout the Cenozoic. Other long-term eustatic effects can easily be added to this curve. The interaction of a detailed eustatic sea-level curve with basin subsidence models awaits pri- marily the awakening of the stratigrapher and exploration- ist to the opportunity at hand. REFERENCES Aharon, P. (19831. 140,000-yr isotope climatic record from raised coral reefs in New Guinea, Nature 304, 720. Allan, J. R., and R. K. Matthews (1982). Isotopic signatures associated with early meteoric diagenesis, Sedimentology 29, 797-817. Baltzer, F. (19701. Datation absolue de la transgression Holo- cene sur la core guest de Nouvelle-Caledonie sur des enchan- tillons de tourbes a paletouiers; interpretation neotectonic, Acad. Sci. C. R. Ser. D271~25), 2251-2254. Bender, M. L., R. G. Fairbanks, F. W. Taylor, R. K. Matthews, J. G. Goddard, and W. S. Broecker (19791. Uranium-series dating of the Pleistocene reef tracts of Barbados, West Indies, Geol. Soc. Am. Bull. 90, 577-594. Bloom, A. L., W. S. Broecker, M. A. Chappell, R. K. Matthews, and K. J. Mesolella (19741. Quaternary sea level fluctuations on a tectonic coast: New 230Th,/234U dates from the Huon Pen- insula, New Guinea, Quat. Res. 4, 185-205. Broecker, W. S., D. L. Thurber, J. Goddard, Teh-lung Ku, R. K. Matthews, and K. J. Mesolella (19681. Milankovitch hypothe- sis supported by precise dating of coral reefs and deep-sea sediments, Science 159, 297-300. CLIMAP Project Members (19761. The surface of the ice-age Earth, Science 191, 1 131-1137. CLIMAP Project Members (19841. The last interglacial ocean, Quat. Res. 21, 123-224. Coleman, J. M., and W. G. Smith (1964~. Late Recent rise of sea level, Geol. Soc. Am. Bull. 75, &33-840. Cronin, T. M. (1983~. Rapid sea level and climate change: Evidence from continental and island margins, Quat. Sci. Rev. 1, 177-214. Curray, J. R. (1960~. Sediments and history of Holocene trans- gression, continental shelf, Northwest Coast Gulf of Mexico, in Recent Sediments Northwest Gulf of Mexico, F. Shepard, ea., American Association of Petroleum Geologists, Tulsa, Okla., pp. 221-266. ROBLEY K. MATTHEWS Curry, W. B., and R. K. Matthews (1981). Equilibrium oxygen- 18 fractionation in small size fraction planktic foraminifera: Evidence from Recent Indian Ocean sediments, Mar. Mi- cropaleont. 6, 327-337. Denton, G. H., and T. J. Hughes, eds. (1981). The Last Great Ice Sheets, Wiley, New York, 484 pp. Dodge, R. E., R. G. Fairbanks, and F. Maurrasse (1983). Pleis- tocene sea levels from raised coral reefs of Haiti, Science 219, 1423-1425. Donn, W. L., and D. M. Shaw (19771. Model of climate evolu- tion based on continental drift and polar wandering, Geol. Soc. Am. Bull. 88, 390-396. Emiliani, C. (19581. Paleotemperature analysis of core 280 and Pleistocene correlations, J. Geol. 66, 26~275. Emiliani, C. (1966a). Isotopic paleotemperatures, Science 154, 851-857. Emiliani, C. (1966b). Paleotemperature analysis of Caribbean cores P6304-8 and P6304-9 and a generalized temperature curve for the past 425,000 years, J. Geol. 74, 109-126. Emiliani, C. (1978~. The cause of the ice ages, Earth Planet. Sci. Lett. 37, 349-352. Fairbanks, R. G., and R. K. Matthews (19781. The marine oxygen isotope record in Pleistocene coral, Barbados, West Indies, Quat. Res. 10, 181-196. Field, M. E., E. P. Meisburger, E. A. Stanley, and S. J. Williams (19791. Upper Quaternary peat deposits on the Atlantic inner shelf of the United States, Geol. Soc. Am. Bull. 90, 618-628. Flohn, H. (1979~. On time scales and causes of abrupt paleocli- matic events, Quat. Res. 12, 135-149. Gordon, A. L. (1982~. Weddell deep water variability, J. Mar. Res. 40, 199-217. Gordon, A. L., and B. A. Huber (1984~. Thermohaline stratifica- tion below the Southern Ocean sea ice, J. Geophys. Res. 89, 641-648. Gould, H. R., and E. McFarland, Jr. (1959). Geologic history of Chenier Plain, South Western Louisiana, Trans. Gulf Coast Assoc. Geol. Soc. 9, 261-270. Graham, D. W., B. H. Corliss, M. L. Bender, and L. D. Keigwin, Jr. (19814. Carbon and oxygen isotopic disequilibria of recent deep-sea benthic foraminifera, Mar. Micropaleonol. 6, 483~97. Harmon, R. S., T. L. Ku, R. K. Matthews, and P. L. Smart (1979~. Limits of U-series analysis: Phase I: Results of the uranium-series intercomparison project, Geology 7, 405~09. Hopley, D., R. F. MacLean, J. Marshall, and A. S. Smith (1978~. Holocene-Pleistocene boundary in a fringing reef: Hayman Island North Queensland, Search 9~8-9), 323-325. Imbrie, J., J. D. Hays, D. G. Martinson, A. McIntyre, A. C. Mix, J. J. Moreley, N. G. Pisias, W. L. Prell, and N. J. Shackleton (19841. The orbital theory of Pleistocene climate: Support from a revised chronology of the marine 6'8O record, in Milankovitch and Climate, Part 1, A. L. Berger et al., eds., D. Reidel, Dordrecht, pp. 269-306. Land, L. S., and C. H. Moore (19801. Lithification, micritization and syndepositional diagenesis of biolithites of the Jamaican Island slope, J. Sediment. Petrol. 50, 357-370. Lighty, R. G., I. G. MacIntyre, and R. Stuckenrath (1978). Submerged early Holocene barrier reef South-East Florida Shelf, Nature 275, 59-60.
QUARTERNARY SEA-LEVEL CHANGE Lighty, R. G., I. G. MacIntyre, and R. Stuckenrath (19821. Acro- pora palmata reef framework: A reliable indicator of sea level in the western Atlantic for the past 10,000 years, Coral Reefs 1, 125-130. Lingle, C. S. (1984~. A numerical model of interactions between a polar ice stream and the ocean: Application to Ice Stream E, West Antarctica, J. Geophys. Res. 89, 3523-3549. MacIntyre, I. G., O. H. Pilkey, and R. Stuckenrath (1978~. Relict oysters on the United States Atlantic continental shelf: A re- consideration of their usefulness in understanding late Quater- nary sea-level history, Geol. Soc. Am. Bull. 89, 277-282. Major, R. P., and R. K. Matthews (1983). Isotopic composition of bank margin carbonates on Midway Atoll: Amplitude con- straint on post-early Miocene eustasy, Geology l l, 335-338. Manabe, S., and A. J. Broccoli (1985). A comparison of climate model sensitivity with data from the last glacial maximum, J. Atmos. Sci. 42, 2643-2651. Matthews, R. K. (1973~. Relative elevation of late Pleistocene high sea level stands: Barbados uplift rates and their implication, Quat. Res. 3, 147-153. Matthews, R. K., and R. Z. Poore (19801. Tertiary 6~80 record and glacio-eustatic sea-level fluctuations, Geology 8, 501-504. Mesolella, K. J., R. K. Matthews, W. S. Broecker, and D. L. Thurber (1969~. The astronomical theory of climatic change: Barbados data, J. Geol. 77, 250-274. Mix, A. C., and W. F. Ruddiman (19851. Structure and timing of the last deglaciation: Oxygen isotope evidence, Quat. Sci. Rev. 4, 59-108. Moore, W. S. (1982~. Late Pleistocene sea level history, in Uranium Series Disequilibrium: Applications to Environmental Problems, M. Ivanovich and and R. S. Harmon, eds., pp. 48 1~96. Nelson, H. F., and E. E. Bray (1970~. Stratigraphy and history of the Holocene sediments in the Sabine-High Island area, Gulf of Mexico, in Deltaic Sedimentation, Modern and Ancient, J. T. Morgan, ea., Spec. Publ. 15, Society of Economic Paleon- tologists and Mineralogists, pp. 48-77. Newell, R. E., A. R. Navato, and J. Hsiung (19781. Long-term global sea surface temperature fluctuations and their possible influence on atmospheric CO2 concentrations, J. Pure Appl. Geophys. 116, 351-371. Oldale, R. N., and C. J. O'Hara (19801. New radiocarbon data from the inner continental shelf off Southeastern Massachu- setts and a local sea-level-rise curve for the past 12,000 yr., Geology 8, 102-106. Peterson, L. C., and W. L. Prell (1985~. Carbonate dissolution in recent sediments of the eastern equatorial Indian Ocean: Pres- ervation patterns and carbonate loss above the lysocline, Mar. Geol. 64, 259-290. Prell, W. L. (1985~. The stability of low-latitude sea-surface 103 temperatures: An evaluation of the CLIMAP reconstruction with emphasis on the positive SST anomalies, prepared for U.S. Department of Energy, U.S. Government Printing Office, Washington, D.C., 60 pp. Prell, W. L., J. Imbrie, D. G. Martinson, N. G. Pisias, N. J. Shackleton, and H. F. Streeter (1986~. Graphic correlation of oxygen isotope stratigraphy application to the late Quaternary, Paleoceanography 1, 137-162. Prentice, M. L., and R. K. Matthews (1988~. Cenozoic ice- volume history: Development of a composite oxygen isotope record, Geology 16, 963. Redfield, A. C. (19671. Post glacial change in sea level in the Western North Atlantic Ocean, Science 157, 687-691. Ruddiman, W. F., A. McIntyre, V. Niebler-Hunt, and J. T. Du- razzi (19801. Oceanic evidence for the mechanism of rapid Northern Hemisphere glaciation, Quat. Res. 13,33-64. Schnable, J. E., and H. G. Goodell (1968~. Pleistocene-Recent stratigraphy evolution and development of the Apalachicolan Coast Florida, Geol. Soc. Am. Spec. Paper 112, 72 pp. Scholl, D. W. (1964~. Recent sedimentary record in mangrove swamps and its rise of sea level over the southwestern coast of Florida: Part 1, Mar. Geol. 1, 344-366. Scholl, D. W., and M. Stuiver (19671. Recent submergence of southern Florida: A comparison with adjacent coasts and other eustatic data, Geol. Soc. Am. t3ull. 78, 437-454. Shackleton, N. J. (1967~. Oxygen isotope analyses and Pleisto- cene temperatures reassessed, Nature 215, 15-17. Shackleton, N. J., and N. D. Opdyke (19731. 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. Stockmal, G. S. (19831. Modeling of large-scale accretionary wedge deformation, J. Geophys. Res. 88, 8271-8287. Thierstein, H. R., K. R. Geitzenauer, B. Molfino, and N. J. Shackleton (1977~. Geology 5, 500. Thom, B. G., and J. Chappell (1975~. Holocene sea levels rela- tive to Australia, Search 6, 90-93. Vail, P. R., R. M. Mitchum, Jr., and S. Thompson, III (1977J. Seismic stratigraphy and global changes of sea level, part 4: Global cycles of relative changes of sea level, in Seismic Stratigraphy-Applications to Hydrocarbon Exploration, Memoir 26, American Association of Petroleum Geologists, Tulsa, Okla., pp. 83-97. Wagner, P. D. (1983~. Geochemical Characterization of Mete- oric Diagenesis in Limestone: Development and Applications, Ph.D. dissertation, Brown University, Providence, R.I., 386 PP. Williams, D. F., W. S. Moore, and R. H. Fillon (19811. Role of glacial Arctic Ocean ice sheets in Pleistocene oxygen isotope and sea level records, Earth Planet. Sci. Lett. 56, 157-166.