Palynomorph and dinocyst distribution for samples from the E-O transition from Holes 1170D, 1171D, and 1172A is depicted in Tables T1, T2, and T3, respectively. Extensive discussion on the diatom events used in the regional correlation below can be found in Stickley et al. (submitted [N1]). Recovery of palynomorphs is reasonably continuous, with dinocysts being the most prominent group throughout the studied samples. Sporomorphs are common, whereas skolochorate acritarchs (of various types), remains of chlorophyte algae, and foraminiferal inner linings mainly occur at background levels.

The absolute and semiquantitative distribution patterns of the palynomorph assemblages show remarkable resemblance between the studied intervals of Sites 1170-1172 (Tables T1, T2, T3). In general, the (top of the) pro-deltaic unit is characterized by abundant marine and terrestrial palynomorphs. Dinocysts occur in high concentrations of ~15,000 dinocysts per gram sediment on average. The overlying transitional unit contains increasingly less (~4000) dinocysts and terrestrial palynomorphs per gram of sediment. The composition of the dinocyst associations differs significantly from those from the underlying siliciclastic pro-deltaic unit (Fig. F2). The boundary between the pro-deltaic unit and the overlying transitional unit is locally marked by a short palynologically barren interval (e.g., at Site 1172). At Sites 1170 and 1171 this boundary is located in nonrecovered intervals (488.3-491.5 meters below sea floor (mbsf) and 271.3-276.2 mbsf, respectively). The top part of the transitional unit at Site 1172 and the overlying oceanic unit at Sites 1170-1172 are barren of palynomorphs, except for one sample in the oceanic unit at Site 1172 (Table T3). The transition to this barren interval is sudden at all sites.

From the shipboard and first follow-up palynological studies (Shipboard Scientific Party, 2001c, 2001d, 2001e; Brinkhuis, Sengers, et al., this volume), it is apparent that the early Paleogene dinocyst stratigraphic succession at Sites 1170-1172 matches that known from other sites around the region, particularly from New Zealand (e.g., Wilson, 1988, and references therein; Crouch, 2001) and Seymour Island (Wrenn and Hart, 1988). It was also noted that during the early middle Eocene, influence of the so-called Antarctic-endemic Transantarctic dinocyst flora (cf. Wrenn and Beckmann, 1982), constituted by species like Deflandrea antarctica, Octodinium askiniae, Enneadocysta partridgei, Vozzhennikovia spp., Spinidinium macmurdoense, Spinidinium luciae, and so on, increases. The E-O transition is marked at all sites by a turnover from transantarctic dominated to associations dominated by undescribed species of Brigantedinium? and Deflandrea and Spiniferites spp., with slightly later the influx of Stoveracysta kakanuiensis at Sites 1170 and 1172. This turnover is invariably associated with the boundary between the siliciclastic pro-deltaic and transitional unit.

Qualitative and Semiquantitative Results

The younger part of the middle Eocene from Southern Ocean localities is comparatively less well studied than the older part. Only few comparisons may be made to New Zealand (e.g., Wilson, 1985, 1988; Strong et al., 1995; Edbrooke et al., 1998), DSDP Sites 280 and 281 (Crouch and Hollis, 1996), Seymour Island (Wrenn and Hart, 1988), Scotia Sea (Mao and Mohr, 1995), and southern Argentinian successions (e.g., Guerstein et al., 2002), indicating broad similarity. The Southern Ocean middle Eocene appears to be characterized by several important last occurrences (LOs), including those of Membranophoridium perforatum, Hystrichosphaeridium truswelliae, Arachnodinium antarcticum, Hystrichokolpoma spinosum, and Hystrichokolpoma truncatum (Brinkhuis, Sengers, et al., this volume). The transantarctic and bipolar (restricted to polar regions on both the Northern and Southern Hemispheres) dinocysts continue to be the dominant component of the associations.

Even less is known from dinocysts from the E-O transition in the Southern Ocean. Some comparisons may be made to Southeast Australia (Browns Creek section; Cookson and Eisenack, 1965; Stover, 1975), to New Zealand (e.g., Clowes, 1985; Edbrooke et al., 1998), and to DSDP sites in the southernmost Atlantic (Goodman and Ford, 1983; Mohr, 1990). The early late Eocene dinocyst species composition at Sites 1170-1172 forms a continuation of the middle Eocene pattern. Transantarctic dinocysts predominate, and final acmes of Enneadocysta spp., the D. antarctica group, and S. macmurdoense are recorded.

Site 1172

So far, the E-O transition in Hole 1172A has been studied for palynology at the highest resolution of all the Leg 189 sites. Results from that hole, compared to literature, suggest that important first occurrences (FOs) across the E-O transition include those of Schematophora speciosa, Aireiana verrucosa, Hemiplacophora semilunifera, and Stoveracysta ornata. Toward the middle late Eocene the FOs of Achomosphaera alcicornu, Reticulatosphaera actinocoronata, and Alterbidinium distinctum and the LO of S. speciosa are recorded, events that appear important for interregional correlation. Vozzhennikovia spp. continue to be a common constituent of the associations. The LO of S. speciosa is very close to the LO of abundant D. antarctica group. The latter horizon marks the boundary between the pro-deltaic and transitional unit here, signifying a marked change in the associations, including a brief barren interval. Cosmopolitan taxa like Turbiosphaera filosa, Cleistosphaeridium spp., Spiniferites spp., and Lingulodinium machaerophorum become prominent while undescribed species of Deflandrea and Brigantedinium begin to dominate the succession. Slightly higher, in a succession dated as just after the Eocene/Oligocene boundary (sensu Global Stratotype Section and Point; GSSP, Chron C13n; Stickley et al., submitted [N1]), sediments conspicuously become barren of organic microfossils to only briefly reappear in the lower Oligocene (Table T3). In this single palynomorph-bearing sample thus far from the lower Oligocene from Hole 1172A, virtually all Transantarctic Paleogene dinocysts have disappeared (only a single, poorly preserved, probably reworked specimen of E. partridgei was recovered). The association in this sample (189-1172A-39X-2, 3-5 cm; 356.13 mbsf) is effectively characterized by the abundance of taxa more typical for Tethyan (i.e., lower latitude) waters, including the occurrence of Hystrichokolpoma sp. cf. Hystrichokolpoma oceanicum (cf, e.g., Brinkhuis and Biffi, 1993; Wilpshaar et al., 1996; Brinkhuis, Munsterman, et al., this volume, Site 1168).

Sites 1170 and 1171

Although the E-O transition at Sites 1170 and 1171 is as yet less densely sampled, a broadly similar signature to that recovered at Site 1172 is immediately apparent (Tables T1, T2, T3). Invariably, abundant Brigantedinium? sp. and Deflandrea sp. A characterize the interval just underlying the pelagic unit. Also, A. distinctum is abundant. At Site 1170, S. kakanuiensis is locally present in these assemblages as well. Although the basal parts of the considered succession at Sites 1170 and 1171 are similar throughout to those from the siliciclastic pro-deltaic phase at Site 1172, marked differences may be noted in the transitional unit (see below, and Fig. F3).

Global Dinocyst Biostratigraphic Considerations

A comparison between the recorded late Eocene-early Oligocene dinocyst events and literature data suggest that the following dinocyst events are potentially meaningful for recognizing and globally correlating upper Eocene strata and the E/O boundary (see overviews in, e.g., Wrenn and Hart, 1988; Wilson, 1988; Brinkhuis and Biffi, 1993; Brinkhuis and Visscher, 1995; Mao and Mohr, 1995; Wilpshaar et al., 1996; Truswell, 1997; Hannah and Raine, 1997; Jaramillio and Oboh-Ikunenobe, 1999; Williams et al., this volume):

The LO of H. semilunifera is frequently associated with strata of late Eocene age. In Italy, this event is closely associated with the E/O boundary sensu GSSP, located just below the base of Chron C13n (e.g., Brinkhuis and Biffi, 1993). In the present study it is only recorded at Site 1172 (besides an isolated occurrence at Site 1170; see Tables T1, T3). Correlation to Hole 1172A magnetostratigraphy indicates that this event takes place in mid-Chron C16r.1r. This is markedly earlier than, for example, in Italy. This aspect may be due to the overall global cooling trend during the latest Eocene (cf., e.g., Berggren and Prothero, 1992).

Similarly, S. speciosa is found to occur in late Eocene deposits around the world (e.g., Brinkhuis and Biffi, 1993; Williams et al., this volume). In the present study it is only recorded at Site 1172, where it ranges from deposits assigned to Subchron C16r.1r to the top of Subchron C16n.1n (Fig. F3). In Italy, its LO is associated with the younger Chron C15 (e.g., Brinkhuis and Biffi, 1993). Again, this points to an earlier extinction at higher latitudes, possibly due to the overall global cooling trend during the latest Eocene. The event was also recorded at Site 1168 (Brinkhuis, Munsterman, et al., this volume). Current magnetostratigraphic interpretation correlates the LO of S. speciosa with the termination of either Subchron C17n.1n or C16n.2n at Site 1168 (Fig. F3). However, sampling resolution is as yet insufficient across the E-O transition at Site 1168. S. speciosa was first described from the Browns Creek section in southern Victoria, Australia (Cookson and Eisenack, 1965) and later restudied from samples from that same locality by Stover (1975). Combining their records with later generated magnetostratigraphic and calcareous microplankton data (Shafik and Idnurm, 1997), it appears that (1) this taxon does not range above the upper Eocene glauconite-rich unit at that locality and (2) it becomes extinct in deposits assigned to either Subchron C16n.2n or Chron C15n. In view of the apparent regional unconformities associated with the Subchron C16n.1n to C15n interval in the wider Tasmanian area (Shipboard Scientific Party, 2001a; Stickley et al., submitted [N1]), the distribution of S. speciosa cannot at present be used for more precise correlation.

Perhaps the most significant bioevent to be associated with the globally reported mid-Chron C13n cooling (i.e., post-E/O boundary sensu GSSP) is the LO of Areosphaeridium diktyoplokum (see discussion in Brinkhuis and Visscher, 1995). Studies by Stover and Williams (1995) suggest that records of this taxon from the Southern Hemisphere represent a different species, namely the morphologically closely related E. partridgei. During our study of Leg 189 sites, in addition to E. partridgei, yet another morphologically closely related taxon was identified, here informally termed Enneadocysta sp. A. The latter may only be differentiated from A. diktyoplokum by being dorso-ventrally compressed (Areoligeracean style) and by having two, rather than one, antapical processes, which distally unite into a single distal perforated platform (see Pl. P1, fig. 9). This platform has the typical widened sexiform Gonyaulacean shape, a trademark of all Areoligeraceans. Locally, in Eocene assemblages from Sites 1170-1172, this taxon may dominate dinocyst associations (Shipboard Scientific Party, 2001c, 2001d, 2001e; Brinkhuis, Sengers, et al., this volume). In turn, it differs from E. partridgei by lacking cingular or sulcal processes and by having entire platform margins. The presence of two distally uniting antapical processes strongly suggests that the genus Enneadocysta should be considered to reflect a Gonyaulacoid sexiform (Areoligeracean) tabulation pattern rather than a Gonyaulacoid partiform one as suggested by Stover and Williams (1995).

Specimens of A. diktyoplokum sensu stricto were not recorded during the present study, despite earlier statements (Shipboard Scientific Party, 2001c) to the contrary. These specimens are herein assigned to Enneadocysta sp. A. Intriguingly, the recorded FOs of both E. partridgei and Enneadocysta sp. A are stratigraphically close (Brinkhuis, Sengers, et al., this volume) and closely match the timing of the FO of A. diktyoplokum from Northern Hemisphere records (Williams et al., this volume). The LOs of Enneadocysta sp. A and E. partridgei are less well cited than the LO of A. diktyoplokum, which is commonly associated with mid-Chron C13n.

Enneadocysta sp. A is not recorded above the onset of the pelagic unit, nor was it recorded at Site 1168. Its LO thus appears to be associated with the E/O boundary, like A. diktyoplokum. The LO of E. partridgei has been reported to be associated with the early/late Oligocene boundary in the Southern Ocean (e.g., Hannah and Raine, 1997). In the present study, at most localities, E. partridgei is a rare species in E-O transitional deposits, whereas it may dominate middle-early late Eocene assemblages (Brinkhuis, Sengers, et al., this volume). Considering (1) the abundances in the Eocene and (2) the glacial and associated ice-rafting erosive activities during the Oligocene in the Southern Ocean and (3) the clear decline in abundance over the E-O transition, early Oligocene occurrences should be regarded as reworked. In any event, the ranges of A. diktyplokum and allied species seem to be very similar and lead to the speculation that they might represent different phenotypic cyst manifestations of the same dinoflagellate species, or that the species are intimately related otherwise.

The FO of Hystrichokolpoma sp. cf. H. oceanicum is reported from mid-Chron C12r from Italy (e.g., Wilpshaar et al., 1996). This matches its occurrence in coeval Site 1172 deposits (Table T3). Not only does this aspect confirm an early Oligocene age for the basal pelagic unit at Site 1172, but it also suggest warm-water influence at this time.

The FO of the S. ornata was recorded at all studied sites in the late Eocene, including Site 1168 (Brinkhuis, Munsterman, et al., this volume), albeit with a rather scattered distribution. At Site 1172 it first occurs at the onset of Chron C15, whereas at the present low resolution studied Site 1168 its FO appears near the termination of Subchron C16n.1n or C16n.2n, depending on the magnetostratigraphic interpretation (Fig. F3). The species was described from the Browns Creek locality, Southeast Victoria, Australia, and was also reported from New Zealand (Cookson and Eisenack, 1965; Stover 1975; Clowes, 1985). Northern Hemisphere records comprise only those from the late Eocene and early Oligocene in Italy (see overview in Brinkhuis and Biffi, 1993). Available information thus indicates that it ranges from the latest middle Eocene to earliest Oligocene, with a more consistent occurrence in warmer waters.

The FO of R. actinocoronata was calibrated against mid-Chron C16 in central Italy (Coccioni et al., 2000). At investigated Leg 189 sites it is very rare in deposits of late Eocene age, but occurrences do suggest a Chron C16 base for this species. The same more or less holds for Gelatia inflata, a species first described from the late Eocene in the Bering Sea.

The other occurring species are either long ranging, not described, endemic to the Antarctic, or bipolar and therefore not useful for global correlation purposes. The above summary, although broadly confirming a late Eocene to earliest Oligocene age, does not lead to further refinements for (global) correlation. If anything, the Leg 189 records indicate earlier extinctions at this latitude than at coeval deposits at Tethyan sites.

Regional Dinocyst Biostratigraphic Considerations

A. verrucosa appears to occur in deposits calibrated against Chrons C17-C16 when combining the records of A. verrucosa of Cookson and Eisenack (1965) and Stover (1975) with the magneto- and biostratigraphic study of Shafik and Idnurm (1997) at Browns Creek. In the present study, A. verrucosa is to date only recorded in samples from Site 1172 and ranges in deposits assigned to mid-Subchron C16r.1r-mid-C16n.1n (Fig. F3). At Site 1168, depending on the magnetostratigraphic interpretation, the species occurs in deposits assigned to Subchron C16n.2n or C17n.1n (Brinkhuis, Munsterman, et al., this volume).

The FO of S. kakanuiensis is reported to range in the late Eocene-early Oligocene in New Zealand (Clowes 1985; G.J. Wilson, pers. comm., 2000). This event apparently also occurs in Browns Creek above the upper Eocene glauconite-rich unit there (i.e., of post-Chron C16 or C15 age, according to Shafik and Idnurm, 1997). Our results indicate the event calibrates against Chron C13r at Site 1172, and we use the event to identify that magnetochron at Sites 1170 and 1168 (Fig. F3). Future higher-resolution studies on the E-O transition at Sites 1170, 1171, and 1168 may yield more precise information. S. kakanuiensis potentially provides an excellent regional correlation, conspicuously appearing at the time of the deepening of the Tasmanian Gateway (e.g., Stickley et al., submitted [N1]).

The FO of Brigantedinium? sp. is recorded near the top of Subchron C16r.1r at Site 1172. Invariably, high abundances of this taxon mark the youngest deposits of the transitional unit. This is also recorded at Sites 1170 and 1171 and may therefore be useful for regional correlations (Fig. F2) and for paleoecological considerations discussed below. The species has not been recorded in any other E-O studies (including Site 1168), although it is most probably conspecific with forms depicted by Mohr (1990) as "Brigantedinium sp." from sediments with a similar age off Seymour Island (Site 696).

The FO of Deflandrea sp. A is recorded in Chron C15r at Site 1172, in the oldest palynomorph-bearing sample of the transitional unit (Table T3; Fig. F3). This event may be older at Sites 1170 and 1171, as there is a small hiatus at this level at Site 1172. This species was not recorded at Site 1168. Deflandrea sp. A may partly be conspecific with the as-yet only informally described species "Deflandrea prydzensis" (Truswell, 1997). If so, this taxon may reflect a truly high-latitude endemic signal, as records from Prydz Bay indicate an even earlier first appearance there (E.M. Truswell, M. MacPhail, pers. comm., 2002)

Quantitative Dinocyst Distribution Patterns

In terms of quantitative results, the E-O transition can be broadly characterized as a succession of three dinocyst associations, labeled 1 to 3 (Fig. F2, Tables T1, T2, T3), when data from Sites 1170-1172 are combined and correlated as proposed in Figure F3 (see below). These associations characterize the siliciclastic pro-deltaic unit (association 1), and the lower transitional unit (associations 2 and 3). The overlying upper transitional unit and the pelagic unit are virtually barren of organic matter, except for a single sample from Site 1172.

The D. antarctica group, Phthanoperidinium, Vozzhennikovia, and Spinidinium spp. dominate dinocyst association 1, whereas E. partridgei is common in its older parts. Dinocyst association 2 is characterized by high abundances of Phthanoperidinium and Vozzhennikovia spp., Deflandrea sp. A, Brigantedinium? sp. (and other Protoperidinaceans), A. distinctum, and T. filosa. Dinocyst association 3 (only recognized at Site 1172) is characterized by a dominance of Brigantedinium? sp., with locally A. distinctum, Deflandrea sp. A, Spiniferites spp., and T. filosa being common. Cleistosphaeridium spp., L. machaerophorum, and S. kakanuiensis may be common in this association as well.

Regional Correlation Model

Using the above qualitative and quantitative dinocyst aspects, a correlation among Sites 1170-1172 is proposed (Fig. F3). Despite the present low-resolution studies at Sites 1170 and 1171, the chronology of the events is virtually the same among Sites 1170-1172. Only the FO of Deflandrea sp. A appears to be older at Sites 1170 and 1171 than at Site 1172. The relatively young FO of S. ornata at Site 1170 is expected to be corrected in future high-resolution studies. The dinocyst events that co-occur at the boundary between the pro-deltaic and transitional units at Site 1172 are spread out over several meters at Sites 1170 and 1171. This implies a more expanded transitional unit at the latter sites, which also will be subject of future high-resolution studies.

The uppermost part of the transitional unit at Site 1172, representing earliest Oligocene strata, contains a volcanic ash layer and is palynologically barren (see also Shipboard Scientific Party, 2001e). At Sites 1170 and 1171 the volcanic ash layer was not recorded and the LO of palynomorphs co-occurs with the onset of the nannofossil ooze, implying that the top part of the transitional unit at Site 1172 was not deposited or eroded at Sites 1170 and 1171 or that oxidation and/or winnowing was/were not complete at the latter (Fig. F3). Moreover, the absence of association 3 at Sites 1170 and 1171 may be taken to indicate that the corresponding strata were eroded there. Alternatively, however, association 3 may reflect local Site 1172 conditions.

At Site 1172 a relatively complete magnetostratigraphic sequence was recovered across the E-O transition (Stickley et al., this volume, submitted [N1]), providing the first-ever opportunity to calibrate Southern Ocean dinocyst records to the GPTS. For example, the new data imply that the FO of A. distinctum correlates with Subchron C16n.2n, the LO of abundant D. antarctica group with the top of Subchron C16n.1n, and the FO of S. kakanuiensis is calibrated against Chron C13r (see also Williams et al., this volume).

The E-O transition at Site 1170 also carries a magnetostratigraphic signal (Stickley et al., this volume). When correlating Site 1170 dinocyst events to Site 1172, it appears that normal polarity intervals originally assigned to Chrons C17n, C16n.2n, and C13n at Site 1170 (Stickley et al., this volume) are more likely to represent Subchrons C16n.2n, C16n.1n, and C11n, respectively (Fig. F3). This correlation implies that Chrons C15n and C13n are located within the core gaps at Site 1170.

To complement the dinocyst-based correlation of Leg 189 sites, new results from ongoing diatom analysis are incorporated here (as outlined in the Eocene and earliest Oligocene age model of Stickley et al., this volume). Because at Site 1171 some of the diatoms were studied in Hole 1171C and the dinocysts in Hole 1171D, an independent correlation between the two holes is needed. Core 189-1171C-31X and 189-1171D-3R are here correlated using (corrected) magnetic susceptibility (MS) (from Shipboard Scientific Party, 2001c), which in these strata mostly reflects glauconite content (Fig. F4). Core photos and MS recalibration (Fig. F4) show that the onset of the nannofossil ooze is at ~273 and ~269.9 mbsf in Holes 1171C and 1171D, respectively (not the other way around as stated in Shipboard Scientific Party, 2001c). Next, the MS curve for Hole 1171C was correlated with that for Hole 1171D taking the decrease in MS at the onset of the nannofossil ooze as a calibration point (Fig. F4). Allowing for subtle local variations, the record of Hole 1171C overlaps the Hole 1171D record almost perfectly in the uppermost part of the glauconite-rich unit. As the offset in depth appeared to be 3.6 m, diatom event depths from Hole 1171C were transferred to Hole 1171D depth by subtracting 3.6 m.

Abundant and well-preserved diatoms, several of which are age significant, were recovered in sediments of the E-O transition at Sites 1170, 1171, and 1172. At Site 1172, where the transition is more complete than at the other two sites, well-known circum-Antarctic diatom events have proved particularly useful in defining the magnetostratigraphic record for the early Oligocene (see Stickley et al., submitted [N1]). These same events are also recognized at Sites 1170 and 1171, allowing a reasonable regional correlation between all three sites. The diatom events used in this paper are the LOs of Distephanosira architecturalis (~33.5 Ma) and Hemiaulus caracteristicus (~33.5 Ma) and the FOs of Cavitatus jouseanus (30.62 Ma) and Rocella vigilans (A) (30.24 Ma). R. vigilans (A) in this paper is the small form noted in Harwood and Maruyama (1992) as being morphologically and morphometrically distinct from the larger form therein, R. vigilans (B). At Sites 1170-1172, we take some of the diatom data from the age models presented in Stickley et al. (this volume). Sample resolution at these two southern sites is much poorer than that for Site 1172.

The LOs of H. caracteristicus and D. architecturalis co-occur within the transitional unit at 358.8 mbsf in Hole 1172A. It appears these datums occur within close succession of each other in the circum-Antarctic and can be correlated to magnetochron C13n (see Stickley et al., submitted [N1], for in-depth discussion). Both of these events are truncated near the top of their ranges at ~33.5 Ma. In Hole 1170D, these two datums occur in the large depth range 454.61-478.50 mbsf, with their true stratigraphic position most likely occurring in the transitional unit (below 472 mbsf). Such a large stratigraphic interval is caused by an interval barren of all biogenic silica in Hole 1170D. This barren interval (~462.70-476.99 mbsf) contains pyritized diatoms of no age significance. In Hole 1171D, they occur together at the corrected (see above; Fig. F4) depth of 270.9 mbsf.

The FO of C. jouseanus is strongly suggested to be associated with Chron C12n in the circum-Antarctic, whereas that for R. vigilans (A) is correlated to Chron C11r (see Stickley et al., submitted [N1], for details). In Hole 1172A, both of these datums co-occur at the very base of the carbonate-bearing top ~2 m of the transitional unit at 357.39 mbsf. The range of C. jouseanus is probably truncated near its base (op. cit.) allowing an age of ~30.24 Ma to be assigned to this horizon. At Site 1170, these datums occur within the 454.61- to 478.50-mbsf low-recovery and barren interval. It is highly likely the true occurrence of both these datums is at ~472 mbsf at the base of the carbonate ooze. At Site 1171, both of these datums occur at the corrected (see above; Fig. F4) depth of ~269.9 mbsf.

At Site 1172, the LOs of the diatoms H. caracteristicus and D. architecturalis, used to locate Chron C13n, co-occur with the onset of the palynologically barren interval. This is also the case at Site 1171 (and presumably also at Site 1170), implying synchroneity for these two events within the Tasman region within Chron C13n. Taking this (and above-mentioned arguments) into account, the upper (palynologically barren) part of the transitional unit was not deposited at Sites 1170 and 1171. This would explain why the FOs of R. vigilans (small) and C. jouseanus are recorded slightly later at the onset of the nannofossil oozes at Sites 1170 and 1171, whereas at Site 1172 they are recorded within the upper part of the transitional unit. It is therefore likely that the onset of the nannofossil ooze is synchronous among Sites 1170-1172.

Paleoecological Considerations of the Quantitative Dinocyst Distribution Patterns

Dinocyst association 1 essentially occurs throughout the middle to early late Eocene at Sites 1170-1172 (Shipboard Scientific Party, 2001c, 2001d, 2001e; Brinkhuis, Sengers, et al., this volume; Huber et al., submitted [N2]). Peridinioid cysts like Deflandrea and Phthanoperidinium spp. commonly dominate this association. Empirical evidence (summarized, e.g., in Brinkhuis et al., 1992, Brinkhuis, 1994; Firth, 1996; Stover et al., 1996) suggests that Deflandrea and Phthanoperidinium spp. represent marginal marine heterotrophic dinoflagellates closely tied to ancient deltaic settings and organic-rich facies. Other Leg 189 data (lithology, micropaleontology, and geochemistry; see Shipboard Scientific Party, 2001a, 2001c; Stickley et al, submitted [N1]) further support this aspect. Most representatives of Deflandrea are assigned to the D. antarctica group, whereas most representatives of Phthanoperidinium are assigned to the P. echinatum group. D. antarctica is a typical representative of the Antarctic-endemic Transantarctic Flora (cf. Wrenn and Beckmann, 1982), whereas most Phthanoperidinium species are common in high-latitude Eocene dinoflagellate cyst floras (Goodman and Ford, 1983; Mao and Mohr, 1995; Firth, 1996). In this association, typically Vozzhenikovia and Spinidinium (S. macmurdoense) spp. are abundant as well. These may be regarded to represent an overall similar signal: marginal marine, highly eutrophic, and high-latitude settings (siliciclastic pro-deltaic unit).

Of the Gonyaulacoid species (probably autotrophic) high abundances of Enneadocysta sp. A and E. partridgei typically alternate with dominances of the above peridinioids in association 1. A shipboard pilot study at Site 1170 indicated that these changes are cyclic (Shipboard Scientific Party, 2001c). It appeared that Enneadocysta maxima correlate to maximum nannoplankton occurrences, whereas Deflandrea maxima correlate to diatom maxima. This pattern prompted the suggestion that Enneadocysta maxima possibly reflect more offshore oligotrophic conditions, whereas Deflandrea maxima reflect more inshore eutrophic conditions.

Besides the Tasman region (e.g., Haskell and Wilson, 1975; Crouch and Hollis, 1996; Truswell, 1997) this Antarctic-endemic association 1 has also been found in the Eocene of McMurdo Sound (e.g., Hannah and Raine, 1997), the Ross Sea (Wrenn et al., 1998; Hannah et al., 2000), Seymour Island/Weddell Sea (e.g., Wrenn and Hart, 1988; Mohr, 1990), Bruce Bank/Scotia Sea (Mao and Mohr, 1995), Falkland Plateau (Goodman and Ford, 1983), Prydz Bay/Mac. Robertson Shelf (E.M. Truswell, pers. comm., 2002), and from Argentina and Chile (e.g., Guerstein et al., 2002), and is thus widespread throughout the Antarctic region to paleolatitudes of ~60S throughout the Eocene. Dinocysts of association 1 may hence be regarded to distinctly reflect endemic Antarctic high-latitude, marginal marine, and highly eutrophic settings. The dinoflagellate species that produce these cysts may have been adapted to high-latitude conditions while ocean circulation possibly precluded significant mixing of populations. Yet, typical cosmopolitan dinocysts like Spiniferites spp. and Thalassiphora pelagica co-occur with the endemic groups to a certain extent. Remarkably, association 1 dinocysts are virtually absent in coeval strata at Site 1168, off western Tasmania (Brinkhuis, Munsterman, et al., this volume). Instead, only largely cosmopolitan to typical lower-latitude species are abundant at Site 1168 across the E-O transition. This suggests that water masses influencing that locality were distinct from those influencing all other Leg 189 sites at this time. This aspect and possible paleoenvironmental and paleoceanographical consequences are further discussed elsewhere (Huber et al., submitted [N1]).

At ~35.5 Ma (early late Eocene), association 1 was abruptly or gradually replaced by dinocyst associations 2 and/or 3 at Sites 1170-1172. In modern oceans, representatives of Brigantedinium are characteristic for coastal and oceanic upwelling regions as well as sea ice conditions and possibly frontal regions (e.g., Wall et al., 1977; Rochon et al., 1999). Brigantedinium spp. represent heterotrophic dinoflagellates, typically feeding on diatoms. The FO and later abundance of Brigantedinium? sp. might indicate the installment of an (oceanic) upwelling system on a large scale. Combined Leg 189 data confirm this aspect, reflecting the deepening of the Tasmanian Gateway (see also Stickley et al., submitted [N1]). Abundant Spiniferites spp. implies more cosmopolitan waters. From this point of view, the endemic aspect of the associations declines during deposition of the transitional unit. Brigantedinium? sp. may be seen as endemic to the Antarctic, as it was never found in other E-O transitional associations worldwide, except from the Antarctic Site 696 (Mohr, 1990). This taxon is separated from Brigantedinium sensu stricto in view of it having a thin outer wall, not known from modern or fossil Brigantedinium. If this aspect, which may be related to preservation (also not always seen on specimens analyzed here), is discarded, it is difficult to separate representatives of this taxon from modern Brigantedinium (see also Brinkhuis, Sengers, et al., this volume). This would make them cosmopolitan, invariably associated with high nutrient oceanic settings.

At ~35.5 Ma at Sites 1170 and 1171, not only Brigantedinium? sp. but also other Protoperidinioid (heterotrophic) taxa like O. askiniae and Selenopemphix spp. became abundant (Tables T1, T2). It appears that the deepening event was synchronous among Sites 1170-1172 and even affected Site 1168 off western Tasmania, and possibly even localities such as Browns Creek (Southeast Australia; considering the unconformities in these strata) (Shafik and Idnurm, 1997) (see Fig. F3). If compared to similar data from off Seymour Island (ODP Site 696) (Mohr, 1990), a similar signal and timing are apparent on the other side of Antarctica.

At ~34 Ma, dinocyst association 2 was gradually replaced by association 3. This association, so far only recorded at Site 1172, is dominated by representatives of Brigantedinium? sp. Slightly younger deposits, from ~33.5 Ma, are virtually devoid of organic matter. This aspect is most probably related to increased winnowing and oxidation. This transition occurs close to the timing of the Oi-1 18O event of Zachos et al. (1992, 1994) at ~33.3 Ma. The latter event is widely seen to reflect the onset of major Antarctic ice sheet formation and concomitant cooling of surface and deep waters. Inception of cooler O2-rich bottom waters, as well as a stronger deepwater flow regime, may well have been responsible for the virtual disappearance of organic matter at Leg 189 sites at this time. Remarkably, however, in the single palynomorph-bearing sample thus far from the lowermost Oligocene (at ~32 Ma from Hole 1172A), virtually all Transantarctic Paleogene dinocysts have disappeared. This earliest Oligocene association is, perhaps surprisingly, characterized by the abundance of taxa more typical for Tethyan (i.e., lower latitude) waters. This assemblage markedly differs from age-equivalent ones recovered from the Ross Sea farther to the southeast (e.g., Hannah et al., 2000). There, apparently endemic protoperidinoid dinocyst communities continue to dominate the assemblages. The change in associations at Site 1172 signifies the introduction of completely different surface waters in early Oligocene times in this region. Unfortunately, no data from the coeval intervals of the Oligocene of Sites 1170 and 1171 are available at this stage, as samples are nonproductive (see Tables T1, T2). It may be speculated that somewhere between the Ross shelf and Site 1172 the ecological- or water mass-controlled boundary between these associations occurred, possibly even to the north of Sites 1170 and 1171. Further aspects and consequences of these findings are discussed in Stickley et al. (submitted) [N1] and Huber et al. (submitted) [N2].