Background: Diatom biostratigraphy

Sedimentary particles, including diatoms, are mixed and transported in glacial till. Consequently, biostratigraphic analysis of diatoms in glacial sediments must be applied with unusual caution. Although Antarctic diatom biostratigraphic information is always qualitative, the age ranges of many Antarctic diatoms are well documented and biostratigraphic results are generally quite reliable. The latest composite biostratigraphic scheme for Antarctic sediments (Reference Harwood, and Maruyama,Harwood and Maruyama, 1992) includes 184 defined diatom data (“first” and “last” stratigraphie occurrences), divided into 33 diatom zones spanning the last 36 Ma, suggesting an average zonal resolution of 1.1 Ma. Faunal turnover was unusually high during the Pliocene (5.2–1.6 Ma ago), where zonal resolution is 0.4 Ma. The use of individual diatom ranges and sub-zones can yield higher chronostratigraphic resolution. This biostratigraphic scheme was developed from more than 20 years of study of sediment records recovered by stratigraphie drilling in the Antarctic and sub-Antarctic zones by the Ocean Drilling Program (ODP) and the Deep Sea Drilling Project (DSDP).

Stratigraphie records south of about 60° S tend to be more discontinuous due to oceanographic factors and the influence of glacial activity. Due to limited high-resolution stratigraphie reference sections near the continent, the accuracy of Antarctic diatom ranges in near-shore deposits at higher latitudes has been questioned (Reference Clapperton, and Sugden,Clapperton and Sugden, 1990; Reference Burckle, and Pokras,Burckle and Pokras, 1991; Reference Burckle,Burckle, 1993). These authors suggested that the stratigraphie ranges of certain diatoms may be different in the near-shore zone of Antarctica than in the Southern Ocean, where nearly continuous biostratigraphic records are preserved. This effect is called diachrony.

Limited diachrony has been documented in sediments near the Antarctic polar-front zone, the boundary between Antarctic and sub-Antarctic water masses (Reference Burckle, and Abrams,Burckle and Abrams, 1987; Reference Fenner,Fenner, 1991). These water masses have distinct diatom assemblages. The diachrony exhibited in this region, which is generally limited to glacial-interglacial time-scales, is due to migration of the polar front with oceanographic changes. However, significant diachrony of biostratigraphically important diatom species wholly within the Antarctic water mass (entirely south of the polar front) has not been conclusively documented (Reference Webb, and Harwood,Webb and Harwood, 1991; Reference SchererScherer, 1992). In fact, the accuracy of Pliocene diatom ages in inshore waters of the Antarctic continent has recently been independently confirmed (Reference Barrett,, Adams,, Mcintosh, Swisher and WilsonBarrett and others, 1992).

This is not the appropriate forum for a detailed debate of this issue, so I will limit further discussion to the specific biostratigraphic issues raised by Reference Burckle,Burckle (1993). Burckle presented five specific challenges to the data and interpretations of Reference Scherer,Scherer (1991). His comments refer to possible diachrony, the apparent absence of certain taxa and the limitations of biostratigraphy in glacial sediments. I discuss each of these issues in the order presented in his paper.

The validity of Actinocyclus ingens a an isochronous datum

Reference Scherer,Scherer (1991) suggested that the youngest diatom assemblage identified at UpB is more similar to a late Pleistocene Antarctic assemblage than a Pliocene or early Pleistocene assemblage. In support of this interpretation, I cited the apparent absence of well-documented Pliocene species (such as those present in Sirius Group tills) and the apparent lack of the diatom Actinocyclus ingens ( Fig 1a), a heavily silicified form with an unusually high preservation potential in glacial sediments (Reference Abelmann,, Gersonde,, Spiess,, Bleil and ThiedeAbelmann and others, 1990). I suggested that the assemblage may represent sedimentation in the West Antarctic interior, subsequent to the extinction of A. ingens from Antarctic waters, ∼600 000 years ago (the datum is now known to be closer to 620 000 years ago (Reference Harwood, and Maruyama,Harwood and Maruyama, 1992)).

Fig. 1.

Micrographs of key diatoms referred to in the text. Examples from the Southern Ocean. Magnification 1500 times. Scale bar at upper left equals 20 µm, equivalent to the smallest available sieve size. a. Centric diatom Actinocyclus ingens. b. Pennate diatom Nitzschia kerguelensis. c. Pennate diatom Nitzschia curta.

Burckle questioned this interpretation and made the Statement “Α. ingens disappeared from the record during the late Quaternary in sections recovered north of about 61° S but to the south of this latitude they disappeared much earlier in time”. This statement is not substantiated by available stratigraphic data. The extinction of A. ingens is unusually well documented. The coherence of this circum-Antarctic datum is demonstrated by the fact that the extinction of A. ingens, 620 000 years ago, has defined the top of the A. ingens zone in all of the ten high-latitude Southern Ocean diatom biostratigraphic schemes proposed since 1970 (Reference Donahue,Donahue, 1970), despite many other refinements made in Antarctic diatom biostratigraphy after nearly 25 years of study (see Reference Gersonde, and Burckle,Gersonde and Burckle, 1990; Reference Baldauf, and Barron,Baldauf and Barron, 1991; Reference Harwood, and Maruyama,Harwood and Maruyama, 1992).

The extinction of this species is known to be diachronous across the global ocean, but contrary to what Burckle reported, Antarctic waters were the last refuge of this species. The extinction of A. ingens was progressive, beginning in tropical latitudes 11.5 Ma ago, and 4.4 Ma ago in the mid and high latitudes of the North Pacific (Reference Barron,, Bolli, Saunders and Perch-NielsenBarron, 1985; Reference Spencer-Cervato,, Sarnthein and ZahnSpencer-Cervato and others, 1992). However, this species persisted, indeed thrived, in the Southern Ocean until 620 000 years ago. This datum is particularly easy to identify because the heavily silicified A. ingens preserves particularly well and reaches an abundance acme in lower Pleistocene sediments of the Southern Ocean, prior to its final disappearance. Antarctic sediments deposited during the interval 1.60.7 Ma typically contain 40–60% A. ingens relative to other diatoms (Reference Abelmann,, Gersonde,, Spiess,, Bleil and ThiedeAbelmann and others, 1990).

To support his assertion that A. ingens disappeared “much earlier” than 620 000 years ago south of 61°S, Burckle cited his own work, holes 690B (65° S), 695A (62° S) and 697B (62° S) of Ocean Drilling Program leg 113 in the Weddell Sea (Reference Gersonde, and Burckle,Gersonde and Burckle, 1990). Referring to hole 690 B, Reference Burckle,Burckle (1993) stated “the last occurrence of this species (A. ingens) is in upper Pliocene sediments”. But a critical fact was not reported. This site has an hiatus spanning the interval from 2 Ma ago to about 0.75 Ma ago (see figure 7 and table 6 of Reference Gersonde, and Burckle,Gersonde and Burckle, 1990). The entire interval characterized by unusually abundant A. ingens is missing from this core! The top of the A. ingens zone is identified at the appropriate level in hole 690B, above the unconformity, though this interval is rather diatom-poor (Reference Gersonde, and Burckle,Gersonde and Burckle, 1990). Similarly, hole 695A, which Burckle also cited, has a highly condensed upper Pliocene and Pleistocene section, with disconformities and poor diatom preservation (see figure 10 of Reference Gersonde, and Burckle,Gersonde and Burckle, 1990).

Referring to hole 697B, Burckle stated “ (A. ingens) occurs up to the middle Quaternary, but is about 300 ka short of the 600 ka mark”. He went on to state “All samples above the last occurrence of this species in this hole are barren of diatoms”. It cannot be concluded that A. ingens became extinct 300 000 years too early at Site 697 if no diatoms are preserved in sediments that are younger than 900 000 years old! In fact, the top of the A. ingens zone could not be identified in this core, due to locally poor preservation and stratigraphic disturbance (see figure 11 of Reference Gersonde, and Burckle,Gersonde and Burckle, 1990). These examples, from Burckle’s own work, offer no evidence that Actinocyclus ingens disappeared earlier than 620 000 years ago south of 61° S.

Contrary to what Burckle wrote, there is good evidence that this species was an important part of the plankton south of 61^e S during the early to mid-Pleistocene. Few continuous stratigraphic reference sections exist south of this latitude (note problems at sites 690, 695 and 697), but the better records all show significant numbers of A. ingens in the Pliocene, and up to the early late Pleistocene. For example, this datum is demarcated in site 274 at 69° S at the mouth of the Ross Sea, off Cape Adare. At site 273, at 74°32′ S in the Ross Sea, the top of the A. ingens zone may be suggested, although glacial reworking and drilling disturbance complicates the interpretation of these sites (Reference McCollum,McCollum, 1975; Reference Savage,, Ciesielski,, Oliver, James, and Jago,Savage and Ciesielski, 1983). The A. ingens datum may be present in the CIROS-2 drill hole, at 77°41′S, in McMurdo Sound, though stratigraphic problems are apparent in this core as well (Reference HarwoodHarwood, 1986). This datum is recognized at site 324 at 69°03′s in the Bellingshausen Sea (Reference Schrader,Schrader, 1976), as well as at virtually all Plio-Pleistocene sections in the circum-Antarctic siliceous ooze belt.

In referring to the Pleistocene record of core 1458 of Reference Abelmann,, Gersonde,, Spiess,, Bleil and ThiedeAbelmann and others (1990), from the Weddell Sea at 64°59′S, Burckle stated “(this core) has few diatoms in that interval”. However, in discussing core 1458, Reference Abelmann,, Gersonde,, Spiess,, Bleil and ThiedeAbelmann and others (1990) stated “A distinct change in diatom assemblage occurs at about 1.6 Ma, when Actinocyclus ingens increases considerably in numbers (toward the top of the core)”. Reference Abelmann,, Gersonde,, Spiess,, Bleil and ThiedeAbelmann and others (1990) did not specifically report the level where A. ingens disappears, but the identification of the top of the A. ingens zone (620 ka) is clearly implied in the text and in their figure 5.

As outlined above, Burckle’s statement that “ A. ingens disappeared from the record during the late Quaternary in sections recovered north of about 61° S but to the south of this latitude they disappeared much earlier in time” is not only unsubstantiated, it is almost certainly incorrect. Clearly, more and better near-shore reference sections are needed, but at present there is no justification for assuming that the A. ingens datum is not valid for West Antarctic interior basins.

Lack of well-demarcated biostratigraphic first occurrences

Gaps in our knowledge of Neogene Antarctic diatom biostratigraphy exist; however, a great deal is well established. More than ten of the species reported from Upstream Β have never been reported from Miocene sediments. Many of these have well-established first occurrences in either the early or late Pliocene and the details of first occurrences of others still need to be established. What these taxa do have in common is that they are common in Antarctic waters today. Furthermore, there is a conspicuous absence of well-established Pliocene taxa in association with the extant (living) species that have been found. This strongly implies, though does not directly indicate, a post-Pliocene assemblage. The absence of A. ingens provides further Suggestion of a late Pleistocene age, as discussed above. I have never suggested that these data provide “proof” of late Pleistocene collapse of the West Antarctic ice sheet, though the diatom data certainly are supportive of such an hypothesis.

A complete, undisturbed and well-preserved stratigraphic record of Pliocene and Pleistocene biosiliceous sedimentation on the Antarctic continental shelf is not yet available. Burckle used this fact to suggest that we know almost nothing of the stratigraphic ranges of common living Antarctic diatoms. For example, Burckle stated “The geologic ranges of Thalassiosira gracilis var. expecta and T. antarctica have never been investigated so it is not known how far back in time they ranged”. These diatoms are common and well-known in Antarctic waters today, yet they have never been reported from Miocene or Pliocene sediments from the Southern Ocean or from Antarctic continental shelf deposits. If they were a common component of sediments of these ages, then their occurrences would have been reported. The fact that their first occurrences have not been specifically documented may imply, though does not prove, a young age (i.e. post-Pliocene).

Burckle pointed out uncertainties with regard to the stratigraphic record of the small pennate diatom Nitzschia separanda and the rare centric diatom Tkalassiosira trifulta. Biostratigraphic uncertainty with regard to these taxa is clearly acknowledged in figure 3 of Reference Scherer,Scherer (1991). It is unlikely that the degree of uncertainy associated with these species would alter the basic interpretation of the data in a significant way. I do not dispute that there remain ambiguities with regard to the documented ranges of individual diatom taxa. It is my belief that many of the uncertainties are due to limited reference sections and a lack of high-resolution studies, rather than problems inherent in the diatom record.

Apparent absence of Nitzschia kerguelensis and Nitzschia curta

Burckle’s third argument centered around the apparent absence of the pennate diatoms Nitzschia kerguelensis (Fig. lb) and Nitzschia curta (Fig. lc). Both of these taxa are abundant in Antarctic waters today. N. curta dominates the diatom assemblage of much of the modern Ross Sea. N. kerguelensis is uncommon in the Ross Sea today but it dominates sediments of the circum-Antarctic ocean. It is true that these diatoms have not been unequivocally identified at Upstream B. I must point out that lack of observation does not necessarily constitute absence, as Burckle suggested. This is particularly true with regard to these taxa, given their morphology and the laboratory methods used, as discussed below.

Micropaleontological analysis of glacial diamictons is fundamentally different from micropaleontological analysis of in-situ deep-sea sediments. In essence, analysis of diatom assemblages in subglacial till is glacial geology, not paleo-oceanography. Consequently, the methods employed are quite different. Diatoms in till are rare, mixed and often mechanically degraded, whereas Antarctic marine sediment generally includes abundant diatoms which are relatively unmixed. In order to recover identifiable diatoms from a clay-rich till, it is necessary to use methods that are inappropriate for paleo-ocean-ographic studies; specifically, the use of fine sieves (20 µm mesh) and other methods to concentrate larger diatoms (Reference SchererScherer, 1992). Nearly all of the diatoms I consider Pleistocene are relatively robust centric forms (large and well silicified, as is A. ingens). In contrast, Nitzschia curta and Nitzschia kerguelensis are pennate (long and thin) forms ( Fig 1).

There is a notable paucity of pennate diatoms of any age in data reported from Upstream Β 1988–89 matrix sediments. This is due largely to the bias against pennates and diatom fragments imparted by the use of sieves, and possibly by abrasion in active till. Pennate diatoms pass through the screen, effectively eliminating them from most of the slides examined. Slides of the unsieved fraction of matrix sediments are also examined, but considerably more time has been spent on sieved preparations, due to a higher rate of data generation. The unsieved fraction examined contains diatom fragments, including small fragments of pennate diatoms, but identifiable diatom debris in these preparations of UpB 1988–89 sediment is quite rare.

Nizschia curta ( Fig 1c) is abundant in the Ross Sea today, but it is a small and rather delicate form, and may be more susceptible to dissolution and breakage than the larger and more robust cèntrics. I point out that the Miocene and Pliocene ancestors of N. curta (including N. praecurta) are also apparendy absent from the UpB 198889 matrix sediments, despite the fact that upper Miocene diatoms strongly dominate the diatom assemblage. N. praecurta is similar to N. curta in dimensions and construction, and is present (up to 5%) in upper Miocene diatomaceous sediment clasts that were recovered from Upstream Β during the 1991–92 field season (Scherer, in press; see discussion below). The apparent absence of this taxon is comparable to the apparent absence of N. curta. Small fragments of diatoms in this lineage were encountered in examination of unsieved preparations, but tentative identification of small fragments was not reported. It would be impossible to distinguish N. curta from N. praecurta based on analysis of fragments.

Nitzschia kerguelensis ( Fig 1b) is present in low abundance in the modern Ross Sea but is present in mid-Holocene Ross Sea sediments (Reference Kellogg, and Truesdale,Kellogg and Trues-dale, 1979). Why has it not been identified at Upstream B? First, I reiterate that this is a pennate diatom, thus it is not expected to be represented in sieved material. Secondly, I was reluctant to report the occurrence of fragments of diatoms that are obviously akin to N. kerguelensis. Some fragments are likely to be from N. kerguelensis, but I cannot rule out the possibility that they are from, for example, a coarse specimen of the closely related, but much older, N. miocenica. Positive identification of small diatom fragments is a dangerous practice.

Burckle argued that N. curta and N. kerguelensis have “dimensions similar to those pennate forms which were recovered from Ice Stream Β samples”, but all pennates reported, mostly from unsieved samples, represent less than 5% of all the marine diatoms counted, whereas typical counts of in-situ marine sediment typically contain abundant pennate forms. Small clasts of upper Miocene diatom-rich sediment were recovered from the matrix sediment from UpB 1991–92. These clasts contain more than 20% pennate diatoms when prepared without the use of sieves (Reference SchererScherer, 1992). No similar diatomaceous clasts have been recovered from UpB 1988–89 matrix sediments (Reference Scherer,Scherer, 1991).

The bias toward centric diatoms in sieved preparations has always been acknowledged and discussed (Reference Scherer,Scherer, 1991, Reference Scherer1992). Bearing in mind that this is a till, with many natural biases already present, the methodological bias is an unfortunate but necessary evil.

Finally, there are no “mysterious absences” among the centric diatoms I report. Nearly all well-silicified centric diatoms ( >20 µm) common in modern Antarctic sediments from south of the polar front are represented in my data, in roughly appropriate proportions (Reference Scherer,Scherer, 1991, table 1).

Sediment accumulation rates during the Pliocene

Burckle abstracted the following sentence from Reference Scherer,Scherer (1991) “The apparent lack of Pliocene and early Pleistocene diatoms beneath Ice Stream Β is probably a result of low net biogenic accumulation during the Plio-Pleistocene, coupled with the erosional effects of repeated grounding-line advances across the sea floor”. He then drew a comparison with high sediment-accumulation rates in the northern Weddell Sea region during the Pliocene. His argument is unclear, because the Weddell Sea and the continental shelf of the Ross Embayment sector of the West Antarctic interior are very different settings. Deposition during marine phases in the West Antarctic interior and non-deposition and erosion during glacial phases cumulatively result in low net accumulation, regardless of sedimentation rate at a given time. This is readily apparent in the southern Ross Sea where a major unconformity leaves middle and lower Miocene sediments exposed near the sediment surface (Reference Savage,, Ciesielski,, Oliver, James, and Jago,Savage and Ciesielski, 1983).

I suggested that the youngest diatoms from UpB may represent the most recent deglacial (Scherer, 1991). Most sedimentary evidence of earlier deglacial intervals has been removed from this region by glacial advances. The Pleistocene diatoms beneath Ice Stream Β are currently being transported northward, toward the grounding line. I have not suggested that the Ross Embayment was unglaciated throughout the Pliocene and Pleistocene, nor have I suggested that West Antarctica was continuously glaciated throughout that interval. During deglacial phases, biogenic sediment-accumulation rates in the West Antarctic interior were quite high (Reference Savage,, Ciesielski,, Oliver, James, and Jago,Savage and Cielsielski, 1983; Reference Harwood,, Scherer, and Webb,Harwood and others, 1989; Reference Scherer,Scherer, 1991, Reference Scherer1992). Subsequent erosion, presumably during glacial advances, removed most of this material.

The entire central Ross Embayment has been subjected to extensive glacial erosion. Massive volumes of Plio-Pleistocene sediment have been eroded from the West Antarctic interior. Glacial-bed processes carved the Ross Sea Unconformity (Reference Savage,, Ciesielski,, Oliver, James, and Jago,Savage and Cielsielski, 1983; Reference Karl,, Reimnitz,, Edwards,, Cooper, and DaveyKarl and others, 1987), transporting eroded sediments and depositing them as thick glacial sediment packets that reached up to several kilometers thick at the seaward edge of the Ross Sea continental shelf (Reference Bartek,, Vail,, Anderson, Emmet and WuBartek and others, 1991; Reference Cooper,, Barrett,, Hinz, Traube, Leitchenkov and StaggCooper and others, 1991). The ODP sites discussed by Burckle are from oceanic depths of the northern Weddell Sea, which experienced relatively continuous sedimentation during much of that interval, resulting in high net sediment accumulation during certain Pliocene intervals.

Interpretation of non-marine diatoms

Arguments about lakes in West Antarctica that existed more than 20 Ma ago have no bearing on the debate regarding possible Pleistocene West Antarctic ice-sheet collapse, so I will only respond to Burckle’s comments in brief. In Reference Scherer,Scherer (1991), I speculated that the surprising discovery of common small sediment clasts composed of planktonic (floating) fresh-water diatoms in Upstream Β sediments suggests that large lake systems may have been present in West Antarctica’s past. I suggested that the lakes might have been fairly deep, based on the complete absence of benthic (bottom-dwelling) diatoms co-occurring with the planktonics in these clasts. (Diatoms need light to survive.) Burckle argued that the lakes might have been small or shallow, rather than large and deep, citing an extensive but shallow-lake system in Africa. I would expect that the African lakes discussed by Burckle contain benthic as well as planktonic diatoms, but this is an assumption on my part, based on my knowledge of diatoms. The two papers Reference Burckle,Burckle (1993) cited (Reference Flohn,Flohn, 1965 [incorrectly cited by Burckle as Flohn, 1966] and Reference Burckle,Flohn and Nicholson, 1980) contain no diatom data whatsoever!

I believe that it is reasonable to speculate that an absence of benthic diatoms in lake sediments that contain abundant planktonic diatoms may suggest deposition in relatively deep water. The relatively high abundance of these clasts in the UpB 1988–89 till and the recent discovery of the same fresh-water diatom (Alaucosiera italica var. A; Reference Scherer,Scherer, 1991) in sediments collected at Upstream Β during the 1991–92 season (Scherer, unpublished data) can be more easily explained with large lake systems than smaller ones.

Discussion

Burckle suggested that sediments from beneath a current ice sheet “cannot be used to establish ice-sheet history”, due to re-working and sediment mixing, which he called “smearing”. He proposed instead, that deep-sea sediment cores contain the needed evidence for past ice-sheet reconstruction. I do not argue that deep-sea sediments can provide a relatively complete record and generally the best available chronology for interpreting océanographie changes around Antarctica, and that these changes may be related to changes in the ice sheets. Analysis of deep-sea sediments provides an indispensible approach to paleoenvironmental reconstruction. However, as Burckle pointed out, proxy records such as these do not necessarily indicate linkage with any specific ice mass. This is particularly true with regard to relatively small ice sheets. Consequently, the history of the Wes Antarctic ice sheet cannot be written based only on analysis of deep-sea sediments. Diatom data from beneath the ice sheet provided a valuable independent line of evidence for comparison with better-constrained time-series records from the deep sea.

There is a fundamental difference between glacial-geologic evidence, which includes diatoms in subglacial till and micro-fossil evidence from deep-sea sediment cores. Whereas marine cores may contain excellent time-series data, glacial-geologic records provide evidence of specific glacial (or deglacial) events. A complete ice-sheet history could not be written using glacial–geologic methods, because the record is, by nature, episodic. Depositional events are followed by erosional events. Glacial-geologic evidence and deep-sea proxy evidence both have specific limitations but, taken together, they can provide critical checks and balances, which may lead to consensus.

Burckle asked “Is there direct evidence for late Pleistocene collapse of the West Antarctic ice sheet?”. This question contains two distinct issues: (1) Do the diatoms in question date from the late Pleistocene, and (2) Do they represent direct evidence of marine deposition in the West Antarctic interior following a collapse of the West Antarctic ice sheet?

I have interpreted the age of the youngest diatoms, based on all biostratigraphic information available, as most likely late Pleistocene. Although the data are sparse, they may, nevertheless, be very significant. Burckle challenged the age assignment, suggesting that the diatoms might be older, though he offered no specific age alternative. The broadest interpretation possible is that the diatoms could be as much as about 2 Ma older than I have suggested. However, a Pliocene age is unlikely, considering the fact that the entire (centric) diatom assemblage is indistinguishable from late-Pleistocene (centric diatom) assemblages, and that it differs greatly from well-dated mid-Pliocene assemblages of the Ross Sea continental shelf (Reference HarwoodHarwood, 1986; Reference Barrett,, Adams,, Mcintosh, Swisher and WilsonBarrett and others, 1992).

Regarding the mechanism of emplacement of these diatoms in UpB sediment, I have considered alternative hypotheses, including (1) possible aeolian deposition of diatoms to the surface of the ice sheet, (2) introduction by advection of diatoms by bottom currents beneath the ice shelf, or (3) as sample contamination. These hypotheses are considered less likely than the ice-sheet collapse hypothesis, as discussed elsewhere (Reference SchererScherer, 1992, unpublished data). Reference Burckle,Burckle (1993) did not disagree with this interpretation. He wrote that, if the diatoms are late Quaternary, “this may be taken as direct evidence that collapse occurred sometime during this interval”. Even if the diatoms reported from Upstream Β are late Pliocene, rather than late Pleistocene, the significance of the observation would not be diminished. They would still seem to represent the first direct evidence of late Neogene disintegration of the West Antarctic ice sheet. In my judgement, based on a wide variety of data, the most reasonable interpretation for these diatoms is that they were deposited in the West Antarctic interior, south of Upstream B, during a marine interval (implying ice-sheet disintegration), and that the most likely time for the most recent event was within the last 620 000 years. During a subsequent glacial advance, these diatoms were eroded and transported to the site where they were eventually recovered.