Northumbria Research Link across the Drake Passage region linked to late Eocene cooling and glacial disturbance after the Eocene– Oligocene

. The role and climatic impact of the opening of the Drake Passage and how it affected both marine and terrestrial 15 environments across the Eocene-Oligocene Transition (EOT ~34 Ma) period remains poorly understood. Here we present new terrestrial palynomorph data compared with recently compiled lipid biomarker ( n -alkane) data from Ocean Drilling Program (ODP) Leg 113 Site 696 drilled on the margin of the South Orkney Microcontinent in the Weddell Sea, to investigate changes in terrestrial environments and paleoclimate across the late Eocene and early Oligocene (~37.6-32.2 Ma). Early late Eocene floras and sporomorph-based climate estimates reveal Nothofagus -dominated forests growing under wet temperate conditions, 20 with mean annual temperature (MAT) and precipitation (MAP) around 13°C and 1660 mm, respectively. A phase of latest Eocene terrestrial cooling at 35.5 Ma reveals a decrease in MAT by around 2°C possibly linked to the opening of the Powell Basin. This is followed by an increase in Mesozoic sporomorphs together with a shift in terrestrial biomarkers and sedimentological evidence indicating ice expansion to coastal and shelf areas approximately 34.1 million years ago. However, major changes to the terrestrial vegetation at Site 696 did not take place until the early Oligocene, where there is a distinct 25 expansion of gymnosperms and cryptogams accompanied by a rapid increase in taxa diversity following 33.5 Ma. This unusual expansion of gymnosperms and cryptogams is suggested to be linked to environmental disturbance caused by repeat glacial expansion and retreat, which facilitated the expansion of conifer and ferns. We conclude that the timing of glacial onset rather suggests that the event at site 696 is linked to the global cooling at the EOT and that latest Eocene regional cooling cannot directly be linked. Therefore, confirming that the opening of ocean gateways alone did not trigger Antarctic glaciation, even 30 if ocean gateways may have played a role in stepwise cooling before the EOT. e al., 2013, 2019). Vegetation composition, structure and diversity patterns are reconstructed along with sporomorph-based quantitative climate estimates in order to explore the timing and nature of vegetation and climate change across the northern Antarctic Peninsula region and South Orkney Microcontinent (SOM). The results are compared with recently compiled lipid biomarker ( n -alkane) data (López-Quirós et al., in review), and dinoflagellate cyst data (Houben et al., 2013) to better understand shifts in marine as well as terrestrial environments and the source of terrestrial versus aquatic organic 55 matter. Our results reveal a possible link between changing oceanographic conditions and latest Eocene terrestrial cooling around 35.5 Ma, before glacial onset in the region. resulted in loss of paratropical taxa and slight decrease in taxa diversity; (C) EOT vegetation during glacial onset (~34.1 Ma), environmental disturbance caused by ice expansion and retreat resulted in the reduction of Nothofagus -dominated forested areas and increase in tundra-like vegetation; (D) early Oligocene (~33.5-32.2 Ma), during glacial retreat and the development of milder climates disturbance and reduction of Nothofagus -dominated forests facilitated the expansion of more competitive Podocarpaceae and pioneer cryptogam taxa.


Introduction
The Cenozoic progression from greenhouse to icehouse climate conditions was accompanied by the establishment of the Antarctic ice sheet around the Eocene-Oligocene Transition  Ma; e.g., Hutchinson et al., 2021). This change in Earth climate state is evidenced by a prominent excursion in oxygen isotope ratios from marine biogenic calcite (e.g., Zachos 35 et al., 2001Zachos 35 et al., , 2008Westerhold et al., 2020) during the Earliest Oligocene Oxygen Isotope Step (EOIS ~33.65Ma ;Hutchinson et al., 2021). The possible causes of the onset of Antarctic glaciation are poorly understood and ambiguity remains as to whether a single or combination of factors and feedbacks drove the Cenozoic climate transition (e.g., DeConto and Pollard, 2003;Coxall and Pearson, 2007). Uncertainties particularly remain over the role of the opening and deepening of the Drake Passage on the development of the Antarctic Circumpolar Current (ACC), and how this affected both marine and terrestrial 40 environments (Scher and Martin, 2008;Houben et al., 2019). Today the ocean currents and the ACC exerts a strong influence on the Earth's climate system, playing an important role in the global distribution of heat, nutrients, salt, carbon, as well as in the gas exchange between the atmosphere and the ocean. Given that unabated anthropogenic warming is expected to cause a poleward shift of the ACC and potentially weaken thermohaline circulation (Zhang and Delworth, 2005) this study forms part of a wider need to fully understand the Earth climate system in order to better predict future stability of the Antarctic ice sheet. 45 A major obstacle in understanding the role of the opening Drake Passage and ocean currents in Cenozoic climate change has been the lack of well dated continuous records spanning the EOT from the region. Here we present new terrestrial palynomorph data from Ocean Drilling Program (ODP) Leg 113 Site 696 Hole B (herein referred to as Site 696), containing a well recovered EOT section, depicting changes in terrestrial environments and paleoclimate from the late Eocene to early Oligocene (~37.6-50 32.2 Ma; Houben e al., 2013Houben e al., , 2019. Vegetation composition, structure and diversity patterns are reconstructed along with sporomorph-based quantitative climate estimates in order to explore the timing and nature of vegetation and climate change across the northern Antarctic Peninsula region and South Orkney Microcontinent (SOM). The results are compared with recently compiled lipid biomarker (n-alkane) data (López-Quirós et al., in review), and dinoflagellate cyst data (Houben et al., 2013) to better understand shifts in marine as well as terrestrial environments and the source of terrestrial versus aquatic organic 55 matter. Our results reveal a possible link between changing oceanographic conditions and latest Eocene terrestrial cooling around 35.5 Ma, before glacial onset in the region.

70
A total of 35 samples from the late-middle Eocene to earliest Oligocene (643.73-520.88 mbsf) were analysed for their pollen and spore content. Raw data collected is available from the PANGEA database (Thompson et al., awaiting validation). All palynological slides were prepared using standard chemical palynological processing techniques following the protocols at the University of Northumbria, Department of Geography and Environmental Sciences and the Laboratory of Palaeobotany and the Laboratory of Palaeobotany and Palynology of Utrecht University. Slides were analysed using a Leica DM500 and Leica 75 DM2000 transmitted light microscopes at 200x and 1000x magnification. Where possible, counts of 300 (excluding reworked grains) sporomorphs were made. Only samples containing 50 or more in situ sporomorphs were used for further analysis and evaluation.
Botanic and taxonomic affinities used to identify the Nearest Living Relatives (NLR) of fossil species were established mainly after Truswell and Macphail (2009) and Raine et al. (2011) and references therein (Table 2.). Identification of reworked grains are mainly based on the age-restriction of the species, with species older than Eocene or Oligocene (e.g., Mesozoic species) being easily recognised as reworked. Consideration was also given to whether a grain was reworked based on the level of 85 thermal maturity and its state of preservation. All palynomorphs identified as in situ are regarded as being penecontemporaneous with deposition and are included in the final calculation of sporomorph percentages. Pollen percentages were plotted using riojaPlot, based on the R package rioja (Juggins, 2020) and local zones were established using the CONISS (Constrained Incremental Sum-of-Squares: Grimm, 1987) cluster analysis function. Sporomorph diversity was measured using both the Shannon-Wiener index and the observed number of taxa. A rarefaction method for sums of ≥50 and ≥100 grains was 90 applied, so that the effect caused by differences in the sample size may be removed allowing the estimation of the number of sporomorph species at a constant sample size (Raup, 1975;Birks and Line, 1992). The Shannon-Wiener Index was also carried out as the second measure of sporomorph diversity accounting for species richness and evenness (Shannon, 1948;Magurran, 2013;Morris et al., 2014). Samples containing less than 50 grains were omitted from this analysis. Detrended Correspondence Analysis (DCA) was performed, with downweighting of rare species by removing pollen types whose representation is <5%. 95 This ordination technique is used in order to evaluate ecological patterns within the data, using knowledge of the distribution of NLR and their modern environmental gradients (Correa-Metrio, 2014 performed using the software R for statistical computing (R Development Core Team, 2013) and the package Vegan (Oksanen et al., 2013).

Bioclimatic Analysis 100
Estimates for terrestrial mean annual temperature (MAT), mean annual precipitation (MAP), warmest month mean temperature (WMMT) and coldest month mean temperature (CMMT) were obtained using the NLR approach in conjunction with the Probability Density Function (PDF) method. Fossil taxa used and their NLR are shown in Table 2.
Climate estimates based on the NLR approach use presence or absence data and are independent of the relative abundance of 105 individual taxa. This makes this method ideal for sporomorph based climate estimates from marine sediments, where hydrodynamic sorting of grains may cause variations in the percentages of individual taxa (Arias, 2015), and also helps reduce taphonomic biases (Klages et al., 2020). However, the assumption that modern species and their climate requirements are an analogy for the geological past is one of the biggest weaknesses of the NLR approach. This uncertainty inevitably increases the further back in the geological record (Hollis et al., 2019). It should also be noted that the modern distribution of species 110 may be a function of either its past climate or biogeographic history (Reichgelt et al., 2016;Willard et al., 2019). Nevertheless, temperature estimates derived from the NLR approach are often in agreement with those from other botanical methods and geochemical proxies, providing a certain level of confidence (Klages et al., 2020;Pross et al. 2012).
The PDF method is used to statistically constrain the most likely climate co-occurrence window for an assemblage (Harbert 115 and Nixon, 2015;Willard et al., 2019;Klages et al., 2020). The bioclimatic envelope for each NLR was identified by cross plotting the modern distribution from the Global Biodiversity Information Facility (GBIF; GBIF, 2021) with the gridded WorldCLIM (Fick and Hijmans, 2017) climate surface data using the dismo package (Hijmans et al., 2017) in R. Some taxa were grouped at the family level because of their potentially ambiguous climatic affinity. This includes (1) taxa belonging to the genus Nothofagus, a group that is one of the most abundant in the Antarctic fossil record, due to the ambiguous climatic 120 affinity of the subgenus Brassospora, now endemic to New Caledonia; (2) pollen taxa affiliated with the modern-day genus Microcachrys, of which Microcachrys tetragona is the sole species, on the basis that M. tetragona is only found in specific location in Tasmania under narrow climatic and environmental conditions which are likely not representative of this once widespread genus; and (3) the pollen taxa Peninsulapollis gillii, which has links to the modern genus Beauprea now also endemic to New Caledonia. In these cases, Nothofagaceae, Podocarpaceae and Proteaceae were used, respectively, rather than 125 the genus or species as the NLR.

Geochemical analyses
The following section will focus on the interpretation of lipid biomarker (n-alkane) and stable isotope data from Site 696. For a full description of geochemical methods see López-Quirós et al., in review.

Lipid biomarkers (n-alkanes) 130
The distribution of n-alkanes in sediments can be assessed on the basis of carbon chain length in order to determine potential biological sources (Cranwell, 1973;Rieley et al., 1991;Bi et al., 2005;Duncan et al., 2019;López-Quirós et al., in review).
Variations in the ACL index through time can be used as a proxy of terrestrial organic matter inputs and can also provide 150 information on changes in climate (Collister et al., 1994;Rommerskirchen et al., 2006;Mahiques et al., 2017;Duncan et al., 2019;López-Quirós et al., in review). Plants produce higher ACLs in warmer, tropical regions, whilst lower ACLs are generally observed from cooler climates (Poynter et al., 1989;Sicre and Peltzer, 2004;Jeng, 2006;Vogts et al., 2009;Bush and McInerney, 2015;Duncan et al., 2019;López-Quirós et al., in review). Studies have also suggested plants synthesise longer n-alkanes in more arid environments providing plants with a more efficient wax coating to restrict water loss (e.g., 155 Kolattukudy et al., 1976;Schefuß et al., 2003;Calvo et al., 2004;Zhou et al., 2005;Moossen et al., 2015;Jalali et al., 2017Jalali et al., , 2018, indicating aridity has a strong control on ACL and that ACL index values decrease under wetter conditions (Duncan et al., 2019;López-Quirós et al., in review). The Paq index provides an approximate measure of the relative sedimentary contribution of submerged and floating aquatic macrophytes relative to emergent and terrestrial vegetation (Ficken et al., 2000;López-Quirós et al., in review) and is calculated following Eq. (2): (2) 165 where relationships between n-alkane indices were studied using Pearson's correlation coefficients and assesses as statistically significant when p < 0.05.
Sphagnum mosses also have a molecular distribution similar to submerged and floating macrophytes, showing enhanced production of C23 and/or C25 (Baas et al., 2000;Nott et al. 2000;Nichols et al., 2006;Duncan et al., 2019). Therefore, the Paq  The TI index is based on the assumption that inputs from photosynthetic algae and bacteria are characterised by short-chain nalkanes (C12-C22; Clark and Blumer, 1967;Han and Calvin, 1969;Cranwell et al., 1987;Grimalt and Albaigés, 1987;Duncan et al., 2019) compared to higher land plants rich in C27, C29 and C31 (Bourbonniere and Meyers, 1996;Mahiques et al., 2017;López-Quirós et al., in review). The TI index is calculated as a ratio over the Total Organic Carbon (TOC; Mahiques et al., 180 2017) following Eq. (3): This parameter therefore assumes that higher values of TI characterize a greater input of terrestrial plant-derived organic matter (Mahiques et al., 2017; in review).

TOC (Total Organic Carbon) 185
TOC represents the organic fraction preserved in sediments and can be used to help distinguish between marine and terrestrial sources of organic matter, depositional conditions and organic matter production (Calvert and Pedersen, 1993;Meyers and Ishiwatari, 1993;Avramidis et al., 2014Avramidis et al., , 2015. Organic matter in marine sediments is mainly derived from the decomposition of plants, animals and most importantly plankton (Avramidis et al., 2015). High planktonic primary production and zooplankton grazing causing an increased export of organic matter through the water column to the sea floor support increased 190 preservation of organic carbon in sediments. Therefore, TOC may be used as an indicator for palaeoproductivity even though it also depends on degradation and thus the residence time in the water column (Sarnthein et al. 1988;Lyle et al. 1988;Berger and Herguera 1992;Freudenthal et al. 2002;Jahn et al. 2003;Luo et al., 2013;Frihmat et al., 2015). Coastal sediments with a shorter water column contain higher amounts of TOC than open sea sediments. Organic carbon burial is also affected by redox conditions, and terrigenous detrital matter influx also exert a control and should be taken into account when interpreting TOC 195 in terms of palaeoproductivity (Luo et al., 2013). At Site 696 increased levels of TOC coincide with higher abundance of heterotrophic dinoflagellate cysts (Houben et al., 2013) and may be used to support the notion of high marine palaeoproductivity.

Results
The recovery of palynomorphs is good throughout the section. Of the 34 samples analysed 5 do not contain a sufficient amount 200 of sporomorphs and were discarded from further analysis. In total 74 pollen taxa (58 angiosperms and 16 gymnosperms), 24 spores and 1 sporomorph of unknown affiliation were identified (excluding reworked and unidentified sporomorphs), containing 54 genera. The stratigraphic distribution and relative abundance of major taxa groups is shown in Fig. 3. Major pollen and spore taxa, in order of decreasing abundance, include Nothofagidites spp., Podocarpidites spp., undifferentiated bisaccates, Cyathidites spp., Phyllocladidites spp. and Retitriletes, which occur commonly throughout the Eocene and 205 Oligocene sections.
Based on the results of CONISS ordination the succession is divided into 2 main zones (I and II; Fig. 3). In addition, Zone I is further subdivided (Ia Ib), based on the abundance and frequency of taxa. The results of rarefaction and DCA analysis along with the diversity indices results also show a good distinction between Zones I and II. are more abundant than the rest of Zone I. Whereas, Coptospora (Conostomum) and Liliacidites are less abundant than the remainder of Zone I. Of the entire studied section, Proteacidites cf. Scabratriporites greatly increase in abundance and frequency from Subzone Ib. Other rare taxa also only occur in Subzone Ia or Ib of Zone I but are represented by one or two 245 specimens. Sporomorph-based climate reconstructions reveal significantly higher temperatures within Subzone Ia compared to Subzone Ib, with an interval of latest Eocene cooling occurring around 35.5 Ma. MAT ranges from 10.9°C to 16.6°C and MAP ranges from 1418mm to around 1850mm, with an average of 12.8°C and 1661mm respectively for Subzone Ia. For Subzone Ib MAT ranges from 10.3°C to 11.4°C and MAP is between 1362mm and 1858mm, with an average of 10.8°C and 1591mm respectively (Fig. 4). The 11 samples of Zone II (53R 1W 80-82 to 54R 3W 38-41) are assigned an Oligocene age. Zone II records a strong increase in gymnosperms and cryptogams, accompanied by a rapid rise in taxa diversity between ca. 33.5 and 32 Ma and a contemporaneous increase in reworked Mesozoic sporomorphs (Fig. 3). Based on the results of rarefaction analysis the average number of sporomorph species for a count size of 50 individuals is 19.63 ± 2.00. The results of the Shannon diversity index 255 are between 1.97 and 2.12, with an average of 2.06 ± 0.05.
The Zone II sporomorph assemblage shows a significant decrease in Nothofagus compared to Zone I. Nothofagus pollen make up 51.2% of all non-reworked taxa and 89.9% of all angiosperm taxa in Zone II. Pollen taxa belonging to the Nothofagidites lachlaniae complex (subgenus: Fuscospora) remain the most abundant, followed by taxa belonging to the N. 260 rocanensis/saraensis complex (subgenus: Nothofagus), with undifferentiated Nothofagidites spp. sporomorphs close behind.
However, other Nothofagus taxa are less abundant. Although a slight increase in other angiosperms (non-Nothofagus) occurs in Zone II they remain the smallest botanical group, representing just 5.8% of all non-reworked sporomorphs. In order of abundance, from least to most abundant, significant non-Nothofagus angiosperm taxa include Proteacidites, Tricolpites, Myricipites harrisii (Casuarinaceae) and Peninsulapollis gillii (Proteaceae). Additional angiosperm taxa are typically 265 represented by one or two occurrences. Gymnosperms remain the second most abundant botanical group, but their abundance has increased strongly, representing 28.3% of all non-reworked sporomorphs in Zone II. The gymnosperm assemblage remains dominated by Podocarpidites and Phyllocladidites, which are the two most common gymnosperm taxa respectively. However, other changes in the gymnosperm pollen assemblage are taking place with Dilwynites (Wollemia) particularly increasing in frequency and abundance, along with Alisporites cf. australis (Gymnospermopsida), Microcachryidites antarcticus, 270 Podosporites and Trichotomosulcites subgranulatus (all Podocarpaceae), among others. Undifferentiated bisaccate grains, that likely belong to Podocarpidites, also continue make up a large proportion of the assemblage but are unable to be properly identified. Sporomorph-based climate estimates provide no evidence for abrupt cooling at the Eocene/Oligocene boundary.
Within the early Oligocene Zone II MATs are between ~10.3°C to 13.7°C and MAP ranges from 1423mm to 1860mm a year, with an average of 11.5°C and 1626mm respectively (Fig. 4). These results indicate a slight increase in both temperature and 275 precipitation compared to the latest Eocene Subzone Ib.

Sediment Transport and Provenance
The late Eocene terrestrial vegetation assemblage from Site 696 shares a number of similarities with Antarctic Peninsula palaeofloras of similar ages (e.g., Warny and Askin 2011b;Warny et al., 2019). Both Site 696 and Antarctic Peninsula late 280 Eocene assemblages are dominated by Nothofagidites pollen, predominantly those related to the modern subgenus Fuscospora, with secondary gymnosperms including high frequencies of podocarp conifer pollen. Similar angiosperm and cryptogam assemblages are also seen between the two, with angiosperms such as Proteaceae and Liliaceae and cryptogams such as Cyatheaceae and Sphagnum. This finding may therefore suggest that pollen from the Antarctic Peninsula region could have been transported to the SOM during this time. Furthermore, similarities between nearby Seymour Island sediments, both in 285 composition and paleogeographic setting, suggest a related source, and that the SOM was proximal enough to receive some detritus from the Antarctic Peninsula López-Quirós et al., in review). However, despite these similarities' significant differences in the palaeoflora occurs between the two regions indicating the Antarctic Peninsula may not have been the primary sediment source. In agreement with previous observations by Mohr (1990) the sporomorph assemblage from Site 696 contains a greater diversity of angiosperm pollen compared to late Eocene Antarctic Peninsula palaeofloras (e.g., Anderson 290 et al., 2011;Warny and Askin 2011b;Warny et al., 2019). This higher diversity has also been reported in southern South American Paleogene sporomorph floras (e.g., Romero and Zamaloa, 1985;Romero and Castro, 1986). In addition, the late radiatus (Davallia) not recorded in coeval Antarctic Peninsula assemblages, possibly due to the more northern latitude of the 295 SOM resulting in milder climatic conditions. In addition, the occurrence of moderate to well-preserved in situ benthic foraminifera, with predominantly angular to subangular terrigenous particles, does not support the notion of long-distance transport of sediments from adjacent sources (e.g., Seymour Island; López-Quirós et al., in review). These observations, together with an expansion of gymnosperm conifers and cryptogams recorded during the early Oligocene (33.5-32.2 Ma) at Site 696, but absent from Antarctic Peninsula floras (e.g., Askin et al., 1992;, suggest that the vegetation 300 of the SOM was unique in character and that a significant proportion of detrital material was likely of local origin (e.g., exposed parts of the SOM).
Furthermore, the SOM and the northern Antarctic Peninsula underwent significant rifting during the late Eocene and early Oligocene (~37-30 Ma; King and Barker, 1988;Eagles and Livermore, 2002;van de Lagemaat et al., 2021), forming what 305 would become the Powell Basin (Eagles and Livermore, 2002;Eagles and Jokat, 2014;van de Lagemaat et al., 2021;López-Quirós et al., in review). Throughout the latest Eocene (~35.5-34.1Ma), a decrease in the delivery of coarse terrigenous sediments and a drop in sedimentation rates by almost half is observed as the SOM became more distal from the Antarctic Peninsula due to the opening of the proto-Powell Basin (Eagles and Livermore, 2002;López-Quirós et al., in review).
Deposition of moderately to intensely bioturbated silty mudstones across the EOT (~34.1-33.6Ma) indicate continued 310 subsidence-related marine transgression at Site 696 (López-Quirós et al., in review). This subsequent and continued isolation of the SOM would have caused Site 696 to receive a greater proportion of localised sediments from exposed parts of the SOM.
This supports the results of this study that the majority of sediments supplied to Site 696 at this time were of local origin, perhaps still with some contribution from the northern Antarctic Peninsula.

315
Conversely however, Carter et al. (2017) suggested the majority of the late Eocene (∼36.5-33.6 Ma) sediments deposited at Site 696 are not of local origin. Using detrital zircon U-Pb and apatite thermochronometry analysis these authors concluded that sand grains from the late Eocene Site 696 best matched sources within the Ellsworth-Whitmore Mountains in West Antarctica and that sediments therefore must have been transported to the SOM by ice. However, the presence of in situ paratropical taxa within the early-late Eocene of Site 696 (37.6-35.5 Ma) suggests mild and even ice-free conditions during 320 this overlapping time period. Furthermore, palaeo-sea-surface temperature reconstructions (Douglas et al., 2014;Hoem and Bijl, unpublished) indicate relatively warm conditions (~14°C), and fossil dinoflagellate cyst (Houben et al., 2013(Houben et al., , 2019, calcareous nannofossils (Wei and Wise, 1990) and smectite-dominated clay mineralogy (Fig. 2: Robert and Maillot, 1990) support temperate depositional conditions (López-Quirós et al., in review) not favourable for transport by ice. Sediments deposited within two coarse-grained mudstone intervals within a fine-grained transgressive sequence are likely sourced from 325 coastal/shelf areas of the SOM (López-Quirós et al., in review), attesting to the continued supply of detritus from the SOM. Therefore, these results and those of this study indicate that transportation by ice was unlikely and that a majority of sediments transported to Site 696 are likely of local origin from exposed parts of the SOM.

Late Eocene Palaeoenvironment 330
Sediments from Site 696 record two distinct palaeofloras from the late Eocene Zone I to the early Oligocene Zone II assemblage that evolved in response to an increase in environmental disturbance beginning around 34.1 Ma. Throughout the Zone I assemblage (~37.6-33.6 Ma) abundant Nothofagus with secondary Podocarpaceae, minor angiosperm and cryptogam elements indicate the presence of a relatively humid Nothofagus-dominated temperate rainforest, growing under MATs between ~10.3°C and 16.6°C, and MAP of 1362mm and 1858mm (Fig. 4). Comparison with lipid biomarker n-alkane results 335  (Houben et al., 2013) and calcareous nannofossil (Wei and Wise, 1990) assemblages attest to temperate marine depositional conditions further suggesting temperate conditions prevailed during this time. Nothofagus (predominantly Fuscospora-type), together with less common Podocarpaceae, formed the forest canopy across much of the mid-to higher-altitude areas, with tracts perhaps dominated by one or the other due to natural differences in shade tolerance 340 (Poole, 1987;Veblen et al., 1996;Gallagher et al., 2008;Bowman et al., 2014). Microcachrys along with Araucariaceae, Ericaceae, Liliaceae, Chenopodiaceae and low growing proteaceous shrubs, also reflect better drained higher-altitude habitats as well as coastal and marginal forest environments (Kühl et al., 2002;MacPhail et al., 1999;Kershaw and Wagstaff, 2001;Bowman et al., 2014). Today, similar cool temperate Nothofagus-dominated mixed-podocarp forests occur in the temperate Valdivian region of southern Chile, between 37°45' and 43°20'S (Veblen et al., 1983(Veblen et al., , 1996Poole et al., 2001Poole et al., , 2003Cantrill 345 and Poole, 2012a;Bowman et al., 2014) across elevations greater than 2000m to lowland areas (Kershaw, 1988;Punyasena et al., 2011;Arias, 2015), where westerly trade winds from the Pacific result in high precipitation.
Pollen taxa representing vegetation communities with very different temperature requirements exist within the early-late Eocene (~35.5-37.6 Ma) Subzone Ia. The presence of the paratropical taxa Arecaceae (palms), Beauprea, Myrtaceae and 350 Dacrydium, each occurring intermittently throughout thus subzone (643.73-597.66 mbsf), indicates the existence of a temperate-paratropical vegetation community. These communities are not recorded in coeval Antarctic Peninsula assemblages perhaps due to the Antarctic peninsulas high mountainous palaeotopography (Wilson et al., 2012) and/or the lower latitude of the SOM. Paratropical taxa likely occupied sheltered lowland areas and favourable coastal margins and would have required mild temperatures and the absence of winter frosts, owing to the frost sensitivity of extant palms (Larcher and Winter, 1981, 355 Tomlinson, 2006, Eiserhardt et al., 2011, Reichgelt et al., 2018. Sporomorph based climate estimates reveal Coldest Month Mean Temperatures (CMMT) between 7.5°C and 13.5°C, well above freezing (Fig. 4). In addition, warmth-loving ferns including Gleicheniaceae and rare Davalliaceae also occur together with moisture-loving conifers such as Dacrydium, which only occur in Subzone Ia, and Phyllocladus, further indicating warm wet temperate conditions throughout this subzone. Today these taxa occur in subtropical to temperate regions in lowland sheltered environments, often thriving in wet humid conditions 360 and severely disturbed or pioneer habitats at the margins of rainforests and waterways (Specht et al., 1992;Chinnock and Bell, 1998;Bowman et al., 2014;Arias, 2015).
The co-occurrence of prominent vegetation communities, each with very different temperature and moisture requirements therefore suggests that late Eocene forests across the northern Antarctic Peninsula and SOM were subject to climatic gradients 365 related to differences in elevation and proximity to the coastline. Furthermore, the presence of paratropical and warm taxa within Subzone Ia and the lack of cold temperature taxa reveal conditions were warmer, by around 2°C between 37.6 and 35.5  Askin, 2000;Askin, 2011a, 2011b). Furthermore, an upwards-increase in illite clay minerals (Robert and Maillot, 1990) between approximately 36.4 to 33.9 Ma, signifying a shift in weathering regime from chemical to physical, supports the 375 idea of latest Eocene climate cooling.
An initial spike in reworked Mesozoic sporomorphs at around 34.1 Ma at the onset of the EOT indicates an increase in reworking. This coincides with increasing Eocene dinoflagellate cyst taxa percentages over Protoperidiniaceae (Houben et al., 2013). Sediments within this EOT interval also exhibit two coarsening-upward packages, within an otherwise fine-grained 380 sequence. Furthermore, these sedimentary packages contain the first evidence for ice-rafted debris (IRDs; Barker et al., 1988;López-Quirós et al., in review) in conjunction with a high percentage of illite clay minerals (Robert and Maillot, 1990). Based on these observations, significant ice build-up around the northern Antarctic Peninsula and SOM is inferred, with a period of continental ice expansion to the coast or beyond (López-Quirós et al., in review). This is supported by the presence of glacial surface textures on sand grains  and rare drop stones  from late Eocene 385 (34-37 Ma;Bohaty et al., 2011) sediments offshore James Ross Basin, as well as other sedimentological and geochemical evidence indicating late Eocene and early Oligocene cooling and ice expansion on the northern Antarctic Peninsula (e.g., Robert and Maillot, 1990;Ivany et al., 2008). Furthermore, topographic reconstructions indicate the northern Antarctic Peninsula at the EOT was comparable in elevation to the Trans Antarctic Mountains and Dronning Maud Land (Wilson et al., 2012). In model simulations these are suggested nucleation points for late Eocene and Oligocene glaciation (DeConto and 390 Pollard 2003;DeConto et al., 2007), suggesting a glacial presence in the Antarctic Peninsula region is reasonable during this time (Carter et al., 2017;Lepp, 2018). However, the pollen and spore assemblage from site 696 as well as other late Eocene and Oligocene sporomorph assemblages from the Antarctic Peninsula (e.g., Askin and Warny 2011a), indicate the region still retained some vegetation and therefore was never fully glaciated.

Early Oligocene Palaeoenvironment 395
Despite the evidence for a cooling of terrestrial climate between 35.5 Ma and 35 Ma, and latest Eocene glacial onset around 34.1 Ma the terrestrial palynomorph assemblage from Site 696 indicates that Nothofagus-dominated forests did not change dramatically in composition until the early Oligocene, after the EOIS. An expansion of conifer trees and cryptogams accompanied by a rapid increase in taxa diversity is recorded between approximately 33.5 and 32 Ma. This significant transformation of Antarctic flora in the early Oligocene is quantitatively reflected by the results of DCA analysis, as well as 400 by a decrease in n-alkane ACL, which in turn suggest herbaceous plants and/or conifer dominance (Fig. 5;in review). The diversity patterns derived from the Shannon diversity index and rarefaction analysis show that early Oligocene forests contained a significantly higher number of taxa compared with the late Eocene forest communities. Previous studies (e.g., Cantrill, 2001;Rain and Askin, 2001;Prebble et al., 2006;Griener and Warny, 2015) suggest that significant cooling and drying of the climate across the EOT led to decreasing diversity and a lowering of the forest canopy, with low 405 stature forests formed of scrubby dwarf Nothofagus and podocarps in sheltered regions and low altitude coastal zones.
However, based on the relatively high diversity and frequency of cryptogam taxa and non-Nothofagus angiosperms within the Zone II assemblage, these Nothofagus-podocarp forests would have been at least intermediate in stature and relatively open, allowing the development of fern, shrub and bryophyte communities (e.g., Macphail and Truswell, 2004). When compared to n-alkane results an increase in the abundance of cryptogams, specifically Sphagnum moss, is also supported by an increase in 410 the aquatic plant n-alkane index (Paq) and n-C23/n-C29 ratios throughout the early Oligocene (López-Quirós et al., in review).
Within the early Oligocene (~33.5-32.2 Ma) Zone II assemblage, the increase in Coptospora and Stereisporites (Sphagnum), along with some angiosperms such as Liliacidites and possibly Myricipites, suggest the progression towards colder environments. Coptospora and Liliacidites have been found in Oligocene and Miocene assemblages across Antarctica, 415 including the Ross Sea region (e.g., Askin and Rain, 2000;Prebble et al., 2006), Meyer Desert Formation (e.g., Lewis et al., 2008), Wilkes Land (e.g., Sangiorgi et al., 2018) and the Antarctic Peninsula (e.g., Warny and Askin 2011a). These Oligocene/Miocene assemblages have been inferred to represent tundra mosaic vegetation in cold, possibly glacial, landscapes (Francis and Hill, 1996;Macphail and Truswell, 2004;Prebble et al., 2006). In addition, the presence of common millimetre sized IRDs between approximately 564 and 560 mbsf suggests phases of continental ice expansion to coastal and possibly 420 shelf areas López-Quirós et al., in review). Within this same interval sporomorph-based climate reconstructions also reveal a cooling and drying step, with a decrease in MAT and MAP from around 13.4°C to 10.7°C and 1815mm to 1423mm, between ~33.5 Ma and 33.4 Ma. The initially relatively high temperatures during the earliest Oligocene may be associated with the reported return to near-Eocene climate soon after the EOT (Liu et al., 2009;Houben et al., 2012;Wilson et al., 2013) and is possibly also reflected in a shift in some organic matter indices across the EOT and after the EOIS 425 (Fig. 5;in review). However, further interpretation of this sediment section is hampered by a gap in core recovery. The increase in typical tundra taxa together with common IRD indicates a potential phase of cooling and glacial expansion during the earliest Oligocene, possibly punctuated by the development of milder climates (e.g., Liu et al., 2009;Houben et al., 2012;Wilson et al., 2013).

430
Warming and cooling phases with episodes of ice growth and retreat would have caused environmental disturbance, likely reducing the extent of Nothofagus-dominated forested areas that were established throughout the Eocene. In New Zealand today Nothofagus seedlings are able to capitalise more readily on small canopy openings enabling them to out-compete podocarps in old-growth stands (Lusk et al., 2015). However, unlike podocarps juvenile Nothofagus also require shelter from frost and desiccation, finding it hard to establish themselves amongst other vegetation in open and marginal forest 435 environments until this other vegetation has been partially suppressed through overtopping by Nothofagus (Wardle, 1964;Lusk et al., 2015Rawlence et al., 2020. Thus, conifers are probably favoured by exogenous disturbance, providing a shortlived reprieve from angiosperm competition (Enright & Hill 1995;Lusk et al., 2015). Across the EOT and earliest Oligocene glacial related environmental disturbance and the development of milder climates, possibly after the EOT (e.g., Liu et al., 2009;Houben et al., 2012;Wilson et al., 2013), would have therefore facilitated the expansion of different vegetation types 440 previously suppressed by the dominance of Nothofagus, due to the limited dispersal ability of Nothofagus. Furthermore, a study by Galeotti et al. (2016), suggested that until ca. 32.2 Ma any Antarctic ice sheet would have been extremely sensitive to orbitally paced, local insolation forcing and would have be prone to large fluctuations. Variability in ice volume during the early Oligocene are also reflected by greatly varying benthic 18 O confirming large fluxes in Antarctic ice during this time.
Therefore, the unusual expansion of gymnosperms and cryptogams seen at Site 696 is suggested to be related to an 445 increase in environmental disturbance caused by repeated glacial expansion and retreat, with the first major glacial expansion around 34.1 Ma, together with the competitive dominance of podocarps on exposed disturbed sites (Fig. 6).
A lack of evidence for marine reworking after the EOT and good agreement between terrestrial biomarkers (López-Quirós et al., in review) and the fossil sporomorph record suggest that the changes in the terrestrial palynomorph assemblage during the early Oligocene (~33.5 Ma) reflect true climate signals and increased environmental 450 disturbance caused by glacial onset. Moreover, the results of this study reveal that major changes in terrestrial vegetation took place after the onset of glaciation rather than after terrestrial climate cooling that took place during the latest Eocene after 35.5Ma.

Paleoceanography
At the same time as terrestrial cooling at 35.5 Ma, indicated by the loss of paratropical taxa and a decrease in sporomorph-455 based MAT estimates, sharp changes to marine environments at site 696 are signalled by the appearance of glauconitic packstone (~588.8 to 577.9 mbsf; López-Quirós et al., 2019). The formation of this mature glaucony-bearing facies is suggested to be related to a decrease in the delivery of terrigenous sediments to Site 696 and suboxic reducing conditions at the sediment water interface (López-Quirós et al., 2019, in review). This change in oceanic environmental conditions may be explained by the opening of the proto-Powell Basin and changes to ocean currents, with several studies (e.g., Lawver and 460 Gahagan, 1998;Eagles and Livermore, 2002;Livermore et al., 2007)  Condensed glauconitic sections on outer shelf-upper slope setting commonly occur beneath upwelling areas with high productivity (e.g., Cook and Marshall, 1981;Wigley and Compton, 2006;Banerjee et al., 2016). Importantly these change in oceanographic conditions associated with the opening of the Powell Basin occur synchronously with terrestrial cooling at 35.5 Ma, within the resolution of this study. Furthermore, large-scale changes in vegetation 480 composition and decreasing diversity from Antarctic Peninsula (e.g., Askin, 2000;Askin, 2011a, 2011b) also occur at this time. Therefore, this may suggest a link between marine and terrestrial environments and that the opening of the Powell Basin and the establishment of oceanic upwelling may have driven a large-scale regional cooling step at 35.5 Ma. However, the regional change in oceanography and marine environments cannot be directly linked with the terrestrial vegetation change and glacial onset in the region, that took place about one million years later, at 34.1 Ma. The 485 timing of the second cooling rather suggest that the event at site 696 is linked to the global cooling at the onset of the EOT which is documented by a combination of deep-ocean cooling and global ice sheet growth marking the step from a largely ice-free greenhouse world to an icehouse climate (Hutchinson et al. 2020).

Conclusion 490
The terrestrial palynomorph assemblage from ODP Site 696 in the Weddell Sea records two distinct palaeofloras that evolved in response to increased environmental disturbance and provide insight into late Eocene and early Oligocene terrestrial climate and cryosphere evolution. Late Eocene pollen and spore assemblages reveal a terrestrial climate cooling at 35.5 Ma with a decrease in MAT by an average of 2°C, associated with a shift from temperate-paratropical Nothofagus-dominated forests to cool temperate Nothofagus-dominated forests. This cooling of terrestrial climate after 35.5 Ma coincides with changes in floral 495 diversity and composition in palaeoflora records from the Antarctic Peninsula (e.g., Askin, 2000;Askin, 2011a, 2011b), which have been interpreted to reflect the onset of prolonged cooling in the region. Despite evidence for terrestrial cooling and ice expansion, Nothofagus-dominated forests did not change dramatically in composition until the early Oligocene, when there was distinct expansion of gymnosperms and cryptogams accompanied by a rapid increase in taxa diversity between approximately 33.5 and 32 Ma. We suggest that glacial related environmental disturbance, starting around 500 34.1 Ma, reflected by an increase in cold climate taxa and sedimentological evidence for ice transport and erosion (e.g., Robert and Maillot, 1990;López-Quirós et al., 2019, in  Basin and reorganisation of ocean currents triggered a regional cooling step at 35.5 Ma prior to glacial onset at 34.1 Ma. However, the large temporal gap (~1.4 Ma) between oceanographic changes and glacial onset suggests that the opening of ocean gateways did not alone trigger glaciation, even if ocean gateways may have played a role in stepwise cooling.