The ∼52 km-wide Foundation Trough – an informal unofficial name for this valley – underlies the FPAS trunk, with the bed at the grounding zone extending to ∼2 km below sea level (Figs. 1d and 2) (Jeofry et al., 2018a, b; Morlighem et al., 2020). The bed of the FPAS trunk is smooth in the direction of ice flow, with little evidence, at least based on current observations, of across-valley bedrock highs or topographic narrowings that could encourage grounding-zone stabilisation through increasing basal and lateral drag and reducing mass flow across the grounding zone (Jeofry et al., 2018a, b; Morlighem et al., 2020) (Fig. 2f–g). In contrast, where Academy Glacier joins the FPAS trunk the topography rises inland and is rougher, with a complex of across-glacier ridges, presumed to be bedrock (Figs. 2b–c, h, and 4). However, even after full isostatic rebound under ice-free conditions (Paxman et al., 2022), parts of this area, and the ridges, would remain over 1 km below sea level (Figs. 2h and 4). The geomorphology of the bed here is critical for assessing the potential of the FPAS grounding zone to retreat inland. For example, will future (or past) grounding-zone retreat halt here, where the topography is narrow, rough and the slope prograding (though still well below sea level), and the ice flow of Academy Glacier turns sharply by 90°, or could the grounding zone retreat much further inland towards the South Pole? The distribution of ice thickness observations (Fig. 3a), and the differences between the bed topography products of Bedmap3 and BedMachine (v4) (Fig. 3c–d), demonstrate the remaining uncertainties associated with bed topography in this area, and other parts of the FPAS catchment.

The influence of the higher and rougher topography on ice flow has, so far, only been directly evaluated using 2D flowline models with Bedmap2 data (Huybers et al. 2017). Inverse modelling of ice flow velocities, using BEDMAP (Lythe and Vaughan, 2001), had previously suggested that the rough topography (implied by high values of basal shear stress) beneath Academy Glacier might continue 200 km further inland (Joughin et al., 2006). However, a derivation of basal roughness based on more recent regional RES surveying across the region (Fig. 4a) now shows relatively low roughness – which may present fewer pinning points during grounding-zone retreat – across the parts of Pensacola-Pole Basin closest to the FPAS grounding zone.

The bed of the FPAS is a complex mosaic of heterogeneous basal conditions (Fig. 4). In addition to inland Academy Glacier, relatively low RES-derived roughness values are characteristic of: (i) the lowermost 175 km of the FPAS trunk (Jeofry et al., 2018b), particularly the Foundation-Patuxent side; and (ii) the basin beneath Patuxent Ice Stream, some 225–300 km inland of the grounding zone (Fig. 4a–b). These two zones, modelled as areas of low basal shear stress (10–12 kPa) (Joughin et al., 2006), are both well below sea level (Behrendt et al., 1966, 1974) and would remain so even under fully ice-free conditions (Paxman et al., 2022) (Fig. 4b). In fact, both of these zones are neatly “enveloped” by the ice-free rebounded sea level contour, as is the relatively lower roughness zone beneath Academy Glacier (Figs. 2 and 4a–b).

There are several potential reasons for the lower roughness zones. For example: (i) a thin drape of soft fine-grained marine sediments deposited during periods of more limited glacial extent (Engelhardt and Kamb, 1998; Studinger et al., 2001); or (ii) a smooth bed surface associated with a thicker tectonically-controlled sedimentary basin (Aitken et al., 2023). Both scenarios are likely and not mutually exclusive beneath the FPAS, and warrant future investigation. Given proximity to the present-day grounding zone, the low roughness zone beneath the FPAS trunk (Fig. 4a) (Jeofry et al., 2018b) has the greatest potential to represent marine sediment. If this were to be the case, sediments recovered from this zone could be used to test ideas of Holocene retreat and readvance, or grounding-zone retreat during earlier Quaternary interglacials or even Pliocene retreat.

If the FPAS trunk is underlain by marine sediments, then there are potential modern-day ice-dynamical implications: deformable marine sediments can reduce resistance to ice flow at the ice stream bed (Joughin et al., 2006; Diez et al., 2018; Li et al., 2022a) and lead to flow instability (Joughin and Alley, 2011). Similar low roughness subglacial zones, associated with low basal shear stress, are recognised beneath the trunks of other major outlet glaciers in the Weddell Sea, including Institute Ice Stream (Bingham and Siegert, 2007; Ross et al., 2012; Siegert et al., 2016), Recovery Glacier (Bell et al., 2007; Diez et al., 2019) and Bailey Ice Stream (Diez et al., 2018). This may suggest a common mechanism of surface formation or emplacement of sediment.

The FPAS bed is characterised predominantly by higher and more variable roughness further inland, particularly up-ice of the fully rebounded 0 m (sea-level) contour (Fig. 4a and b) (Bingham and Siegert, 2009). Although this may suggest that soft, deformable sediments are more restricted or absent, modelling of gravity and magnetic data is consistent with the presence of an older sedimentary rock sequence, ∼2–3 km thick, beneath the Pensacola-Pole Basin (Paxman et al., 2019; Li et al., 2022a; Aitken et al., 2023). This sedimentary basin thins towards the coast (Paxman et al., 2019) (Fig. 5), tapering out close to where Academy Glacier joins the main FPAS trunk (Figs. 1, 2 and 5).

Across the Pensacola-Pole Basin (average elevation of 490 m b.s.l.), the relationship between present-day “enhanced” ice flow and subglacial topography is complex. For example, enhanced ice flow is not always spatially coincident with the deepest parts of the (topographic) basin (Figs. 1 and 4c–d). The Academy Glacier flows relatively fast (up to ∼100 m a⁻¹) across the ∼1000 m b.s.l. parts of the Pensacola-Pole Basin (Fig. 4c–d). In contrast, the flow of the adjacent Support Force Glacier is currently slower (≲50 m a⁻¹), despite overlying a system of geologically-controlled overdeepened valleys up to at least 1500 m b.s.l. (Paxman et al., 2019) (Fig. 4c–d). The upper Academy Glacier has two ice flow zones, which may be partly influenced by long-range buttressing from Berkner Island (Fig. 1a–b) (Reese et al., 2018), with the fastest flow (∼100–200 m a⁻¹) proximal to the Patuxent Range, despite a shallower (topographic) basin in this area (Paxman et al., 2019) (Figs. 1 and 4c–d). The geometry of the FPAS ice flow, and its relationship with subglacial topography and basal boundary conditions, is important because of the potential of the FPAS to dynamically interact with other parts of the ice sheet. The mismatch between fastest flow and deepest basins in this region may reflect relatively recent changes in flow regime, where recently accelerated ice flow has not had time to erode the expected deep basins.

The far-reaching catchment of the FPAS interacts glaciologically, and in some cases hydrologically, with ice stream catchments across East and West Antarctica, including Support Force Glacier, Whillans Ice Stream, Mercer Ice Stream, Möller Ice Stream, Nimrod Glacier and Beardmore Glacier; (Fig. 1) (Winter et al., 2018; Yang and Kang, 2022). Potential therefore exists for the FPAS to be sensitive to, and possibly able to drive, glaciological and hydrological change in other sectors of West and East Antarctica. Change in the Ross or Weddell sea sectors of WAIS, or in other large EAIS catchments, could impact the FPAS and vice versa (e.g. Pingree et al., 2011; Steig et al., 2015; Winter et al., 2018). One potential constraint on such change is the basal topography of the upper FPAS catchment. Here, the Foundation and Patuxent troughs, which are narrow and in places above sea level, extend contiguously from the Siple Coast to Filchner-Ronne Ice Shelf (Winter et al., 2018) (Fig. 1). These troughs, which are likely tectonic in origin, are fed by smaller tributary valleys (e.g., beneath Hercules Dome; Hoffman et al., 2023) and continue to steer ice flow through upper Foundation Ice Stream and Patuxent Ice Stream today (Winter et al., 2018). Flow steering means that the orientation of these troughs and the surrounding highlands (Whitmore, Thiel and Horlick mountain blocks) (Jankowski and Drewry, 1981) likely limits the potential for WAIS-driven drawdown of the EAIS in the `bottleneck' zone between East and West Antarctica (i.e. “the gap in the Transantarctic Mountains through which the EAIS flows into the WAIS”; Pingree et al., 2011; Winter et al., 2018). However, significant potential for dynamic glaciological or hydrological change does exist between Academy and Support Force glaciers (see Sects. 4.2 and 5.2).

Previous reconstructions of glacial history in the southern Weddell Sea presented two contrasting scenarios of LGM ice extent: a minimum scenario with a restricted grounding zone located east of Berkner Island (Fig. 8), and a maximum scenario with a grounding zone near the shelf break (cf. Hillenbrand et al., 2014; Siegert et al., 2019). Little is known about the detailed bathymetry of the Weddell Sea continental shelf seaward of FRIS owing to the nearly perennial sea ice in the southern Weddell Sea, which precludes systematic marine survey access (Hillenbrand et al., 2014). However, seafloor glacial landforms indicate that grounded ice extended to the continental shelf break along the eastern Antarctic Peninsula at the LGM, suggesting that this may also have been the case seaward of Ronne Ice Shelf (Ó Cofaigh et al., 2014; Lavoie et al., 2015). More recent cosmogenic nuclide analyses using in situ ¹⁴C analysis of rock samples from the Shackleton Range is consistent with the more advanced/thicker scenario rather than the minimum extent scenario (Nichols et al., 2019; cf. Scenario B of Hillenbrand et al., 2014). However, an embayment in the ice shelf, implying more restricted ice extent in Filchner Trough, is a persistent feature in some numerical models of ice sheet history in the Weddell Sea. Such models also exhibit significant variations in ice thickness between simulations (e.g., Pittard et al., 2022), suggesting that processes and/or boundary conditions are not well understood or represented. As a result, the detailed configuration of the maximum LGM extent and thickness of the FPAS remain an open question.

Knowledge of the post-LGM glacial history of the lower parts of the FPAS (Fig. 8b) has been improved by cosmogenic surface exposure dating constraints on ice thickness and the timing and rate of ice-surface thinning from mountains adjacent to the FPAS trunk (Balco et al., 2016; Bentley et al., 2017; Nichols et al., 2019). Initial ¹⁰Be cosmogenic nuclide exposure ages in the Pensacola Mountains and Patuxent Range were indicative of: (i) LGM ice sheet surface elevations up to 500 m above present (Balco et al., 2016); and (ii) progressive thinning of the FPAS between 10 and 2.5 ka BP, by which time ice surface elevations comparable to the present day were reached (Bentley et al., 2017). However, more recent analysis using the short-lived cosmogenic nuclide in situ ¹⁴C indicated that the FPAS trunk was at least 800 m thicker than present in the Schmidt Hills and Pensacola Mountains at the LGM (Nichols et al., 2019) (Fig. 8), and confirmed that there was major thinning of the FPAS in the Early Holocene. An ice core record from Skytrain Ice Rise at the southern edge of Ronne Ice Shelf records a decrease in ice surface elevation of ∼450 m from 8.2 to 8 ka and ice-shelf retreat between 7.7 and 7.3 ka (Grieman et al., 2024). This is consistent with both the magnitude and timing of ice thinning recorded by exposure ages adjacent to the FPAS.

The cosmogenic nuclide data indicate ice surface elevations were approaching present-day levels by ca. 2.5 ka (Fig. 8c), and therefore allow for continued thinning beyond present in the late Holocene followed by subsequent thickening (Bentley et al., 2017). Were this thinning and thickening to have occurred, then retreat (inland of present) and subsequent readvance of the grounding zone would also have occurred (Johnson et al., 2022). Indeed, this scenario has been proposed for ice stream outlets in this part of the southern Weddell Sea based on a range of glaciological, geological and solid earth deformation evidence (Siegert et al., 2013; Bradley et al., 2015; Bentley et al., 2017; Kingslake et al., 2018; Siegert et al., 2019). Existing above-ice cosmogenic nuclide data cannot, however, unequivocally demonstrate thinning and subsequent thickening of ice in response to grounding-zone migration (Whitehouse et al., 2017; Bentley et al., 2017; Siegert et al., 2019). Instead, measurements of cosmogenic nuclide concentrations (specifically in situ ¹⁴C, see Nichols, 2022) in subglacial bedrock, rather than in above-ice samples, are needed to reveal if the FPAS was thinner than present in the late-Holocene (see Balco et al., 2023). Modelling experiments have simulated Holocene retreat-readvance for the FPAS (e.g., Kingslake et al., 2018; Pittard et al., 2022), but generally produce a modern-day grounding zone more advanced than observations, potentially due to model grid resolution failing to accurately capture the narrow width of the FPAS trunk. The post-LGM, and particularly the late-Holocene, history of the FPAS has important implications for its future evolution. Specifically, a late-Holocene retreat would imply that the grounding zone could retreat landward of its current position under conditions similar to present-day.

Several alternative, observationally based approaches have been used to reconstruct other aspects of the glacial history of the FPAS, including changes in ice-flow velocity and surface accumulation. Analysis of RES-derived englacial layering and numerical modelling has suggested, for example, that major post-LGM changes in the pattern and rate of enhanced ice flow occurred in the upper Academy Glacier catchment around the South Pole and Titan Dome (Bingham et al., 2007; Beem et al., 2018, 2021). Analysis of englacial layering between Dome A and the South Pole has revealed widespread layer drawdown around the South Pole that could be attributable to past flow dynamics or higher surface accumulation rates (Sanderson et al., 2024). During the Holocene, the direction of ice flow near the South Pole and the spatial pattern of accumulation have remained unchanged, although an estimated 15 % increase in ice speed since 10.1 ka has been suggested (Lilien et al., 2018). This speed-up has been interpreted as reflecting a steepening or thickening of the ice-sheet interior at the head of Academy Glacier since the LGM (Lilien et al., 2018).

The FPAS has been subject to numerous numerical ice-sheet modelling experiments, across a range of spatial (i.e. continental, regional, FPAS-specific) and temporal scales (e.g. Pliocene, Last Interglacial, next 100–200 years). Whilst a detailed assessment of the various simulations that have resulted is beyond the scope of this review, we provide an overview of the categories of modelling efforts undertaken and a summary of the range of scenarios resulting from the model simulations.

Simulations of particularly warm (Pliocene) climates on the Antarctic Ice Sheet suggest that the grounding zone of Academy Glacier may have retreated as far inland as the South Pole (∼700 km inland from the current grounding zone), accompanied by extensive retreat of Patuxent Ice Stream (DeConto and Pollard, 2016; DeConto et al., 2021; Hanna et al., 2024) (Fig. 9d). However, such extensive retreat is only observed in simulations from one model that incorporates Marine Ice-Cliff Instability (MICI) and melt-driven hydrofracturing processes (Pollard et al., 2015; DeConto and Pollard, 2016; DeConto et al., 2021) (Fig. 9c and d). The likelihood and significance of MICI and related processes in contributing to rapid retreat in the “real world”, however, are still under debate (e.g., Morlighem et al., 2024). In contrast to models of extreme “Pliocene” warmth, simulations of last interglacial ice-sheet change (e.g. Sutter et al., 2016; DeConto and Pollard, 2016; DeConto et al., 2021; Golledge et al., 2021) are not characterised by retreat inland of the FPAS trunk (Hanna et al., 2024), even if they do incorporate MICI processes (DeConto et al., 2021).

Over shorter timescales, future ice-sheet model simulations suggest that the potential sea-level contribution of the FPAS over the next 200–500 years is likely to be more limited even than that of other large marine-based catchments elsewhere in the EAIS (e.g. Aurora, and Wilkes subglacial basins). For example, ice sheet-scale and regional (i.e. Weddell Sea sector) model experiments (e.g. Wright et al., 2014; Cornford et al., 2015; Ritz et al., 2015; Seroussi et al., 2020; DeConto et al., 2021; Hill et al., 2021; Johnson et al., 2023; Seroussi et al., 2024; Hill et al., 2024) suggest that the FPAS is unlikely to undergo substantive grounding-zone retreat over the next 100–300 years (e.g. Fig. 9a and b). In these models, grounding-zone retreat is predominantly limited to the FPAS trunk only (i.e. the zone before the Foundation-Patuxent and Academy tributaries become decoupled). However, model results and projected retreat rates could vary depending on specific model choices, for example the form of the basal sliding law, and the way in which sub-shelf melt rates are imposed in stand-alone ice sheet models, both of which are known sources of uncertainty in future projections (Ritz et al., 2015; Seroussi et al., 2020, 2024).

Future glaciological change may not be fully captured in existing numerical model simulations. The FPAS may evolve differently because: (i) current numerical ice sheet models are poorly constrained (e.g. Steig et al., 2015; Bradley and Hewitt, 2024); (ii) major potential retreat has been simulated for FPAS-adjacent ice streams between 2100–2200 (e.g. Wright et al., 2014; Cornford et al., 2015; Seroussi et al., 2020; DeConto et al., 2021; Hill et al., 2021); and (iii) thinning of FRIS may reduce overall buttressing of grounded ice (Reese et al., 2018) impacting the FPAS in unanticipated ways. The potential for the FRIS cavity to shift from a cold to warm regime, driven by the inflow of warm circumpolar ocean waters on the continental shelf (Hellmer et al., 2012), may drive the glaciological change to adjacent ice streams (Hill et al., 2024), with uncertain implications for the FPAS.

FPAS-specific factors may also increase uncertainties in model outputs. First, the steep, prograde basal topography up-ice from the confluence of Academy Glacier and Foundation-Patuxent Ice Stream in Foundation Trough (Fig. 2) (Huybers et al., 2017) has been poorly constrained in observations and poorly resolved in some model experiments, particularly those that simulate the whole ice sheet and are accordingly low-resolution. Secondly, the narrow FPAS trunk (Fig. 2), situated within the Foundation-Thiel Trough (Jeofry et al., 2018b), results in high lateral shear stresses that may not be fully resolved in lower resolution numerical models (Jamieson et al., 2012; Gudmundsson, 2013; Mas e Braga et al., 2023). Thirdly, the poorly constrained potential long-range buttressing effects of Berkner Island (Fürst et al., 2016; Reese et al., 2018), may not be characterised in all models (Fig. 1a–b). Finally, many of the existing simulations (e.g. DeConto et al., 2021) were run using the Bedmap2 basal topography (Fretwell et al., 2013), which did not include the more recently acquired PolarGAP, Operation IceBridge, FISS and COLDEX RES data (Paxman et al., 2019; Morlighem et al., 2020; Frémand et al., 2023; Young et al., 2025) and only coarsely resolved sub-ice shelf bathymetry. Indeed, we note substantive topographic differences between Bedmap3 and BedMachine for the FPAS catchment (Fig. 3).

In addition to other sources of uncertainty, numerical ice sheet model experiments vary depending on resolution and due to uncertainties in bed topography. For example, Antarctic-wide warm climate model runs (e.g. for the Pliocene) using 5–20 km resolution model experiments generate deglaciation into the Pensacola-Pole Basin, whilst model runs at 40 km resolution do not (Pollard et al., 2015), potentially because the topographic complexity of the FPAS (Fig. 2) is poorly resolved at low resolution.

Only two FPAS-specific models exist to date. These simulate 2D flowlines up either Foundation Ice Stream (Whitehouse et al., 2017) or Academy Glacier (Huybers et al., 2017) only, and are therefore naturally limited in diagnostic and prognostic capabilities. In an ideal scenario, all modelling experiments would be undertaken using the most up-to-date topography at high resolution, with sub-km adaptive-mesh resolution near the grounding zone (e.g. Hill et al., 2021, 2024) and other regions of high stress gradients. Recent advances in ice-sheet-wide models have partly addressed this issue, by using fully coupled atmosphere-ocean-ice approaches with model resolutions of 4–8 km and higher-resolution adaptive grids (e.g. Pelle et al., 2020; Naughten et al., 2021). Sensitivity studies on the specific FPAS area are required to evaluate and understand which factors are critical for accurately modelling this system. Such approaches (see Siegert et al., 2020), which have already been used for modelling experiments in other catchments, can overcome these issues. In particular, such experiments would benefit from integrating existing and/or new observations and measurements of the FPAS and the offshore ice shelf. As it is more robust for model grids to be constrained by direct observations of bed topography than interpolated data (see Pritchard et al., 2025), acquiring new measurements so that bed topography resolution better approximates the resolution of current-generation numerical models would further help to improve modelling of the FPAS response to change.

We do not currently know: (i) why the FPAS grounding zone is positioned where it is, and whether the system is “in balance”; (ii) what the current conditions are at the FPAS grounding zone; (iii) what the physical properties of the FPAS ice and subglacial materials (e.g. rheology, anisotropy etc.) are; and (iv) what the configuration and character of the FPAS's hydrological system is.

Addressing these unknowns, essential for evaluating how the FPAS may respond to future climate or oceanographic driven change (e.g. Hellmer et al., 2012), requires four targeted activities:

Improved characterisation of the basal topography of the FPAS. New airborne geophysical surveys of the FPAS catchment, deploying innovative approaches such as swath imaging radar (e.g. Holschuh et al., 2020; Hoffman et al., 2023, Carter et al., 2025), are required to obtain the 10s to 100s of meter resolution required for ice-sheet modelling of grounding zone change (e.g. Seroussi et al., 2024; Hill et al., 2024) (Fig. 9). There is an urgent need for specific RES surveys of the Academy-Support Force boundary (Fig. 7) and the Foundation-Patuxent catchment (Figs. 1 and 2) of the FPAS. The design of these new surveys must prioritise glaciological needs by considering, for example, the survey targets, survey line orientation, height above ice surface etc. In this regard, we highlight the statement made by the BedMachine consortium – “… we recommend flight tracks perpendicular to the flow direction to maximize constraints on ice flux, especially upstream of Academy and Support Force glaciers….” (Morlighem et al., 2020);

The upstream boundary of the FPAS catchment intersects numerous ice-stream catchments. This makes it sensitive to, and potentially capable of influencing, ice-sheet change in other sectors of West and East Antarctica, including FRIS (Pingree et al., 2011; Steig et al., 2015; Reese et al., 2018; Winter et al., 2018). However, we believe priority should go to the interaction between FPAS and Support Force Glacier, as their contact (Fig. 7) likely represents the most sensitive boundary of the FPAS catchment to change.

The pattern of ice flow, the morphology of the ice surface and bed topography, and the configuration of the subglacial hydrological system (Sect. 4.2) (Figs. 1, 2, 6 and 7) at the Academy-Support Force boundary make it is highly susceptible to both ice flow-switching (Conway et al., 2002; Dowdeswell et al., 2006; Siegert et al., 2013; Winter et al., 2015; Selley et al., 2025) and subglacial water piracy (Alley et al., 1994; Vaughan et al., 2008; McCormack et al., 2023). However, the potential for, evidence of, and glaciological consequences of flow-switching and water piracy across this hundreds-of-kilometres-long boundary (Fig. 7) have not yet been assessed. Such assessment requires improved determination of the current geometry of this boundary, the controls on boundary position, and whether the ice sheet records evidence of past changes to the ice or subglacial water catchments.

Although the contemporary ice flow boundary between these glaciers (Fig. 7a) has been defined from remote sensing-derived ice flow measurements (Mouginot et al., 2019), there are known limitations to the ice velocity dataset in this area. In addition, even small changes to ice surface geometry could cause flow direction change over short timescales (Fig. 7a). Improved constraints on the contemporary ice flow pattern will likely be realised through forthcoming SAR satellite observations (e.g. NISAR) that will result in new and improved ice velocity data products.

Characterising what controls the position of the Academy-Support Force margin, and assessment of whether the ice sheet englacial structure records evidence of past boundary migration, needs new RES data. This will provide improved bed topography (e.g. Franke et al., 2020; Carter et al., 2025) and the characterisation, analysis and deciphering of englacial layering and structure across the shear margin (Fig. 7), including ice crystal fabric from polarimetric radar (e.g. Ross et al., 2020; Young et al., 2021; Jordan et al., 2022; Franke et al., 2022; Gerber et al., 2023; Jansen et al., 2024). These data should be integrated with ground-based radar (ApRES) and seismic measurements for ice fabric anisotropy (Brisbourne et al., 2019). Before new data are acquired, analysis of the englacial structure and layers, and “radioglaciological artefacts” (Young et al., 2021) from existing RES data (e.g. PolarGAP, OiB, SPRI-NSF-TUD etc.) should also be performed. These existing datasets have not yet been used to evaluate the nature and stability of the current position of the Academy-Support Force boundary, and thus the possibility that it could migrate, or that the ice structure may record past change.

An outstanding uncertainty is whether the subglacial hydrological catchments of Support Force and Academy glaciers align directly with the ice flow catchments. Current datasets (see Fig. 7) suggest that the ice flow and subglacial water flow catchments may be non-aligned. For example, some reported “Academy Glacier” active subglacial lakes (i.e. A14 and A16) could be located beneath Support Force Glacier instead. A similar mismatch between ice flow and hydrological catchments has been proposed for Pine Island and Thwaites glaciers in West Antarctica (Napoleoni et al., 2020). We need to confirm whether this is the case for the FPAS, as any model investigation of the potential for future hydrological flow-switching requires that the current boundaries of the Support Force and Academy hydrological catchments are determined accurately. Any such offset between ice flow and hydrological catchments poses challenges for running coupled hydrology-ice dynamical models.

Five sub-questions arise from considering the sensitivity of the FPAS to changes in external forcing (i.e. oceanographic and/or atmospheric changes) and internal dynamic feedbacks: (1) Is the FPAS susceptible to future sustained retreat via ocean-driven melt, and over what timescale(s)? (2) Is the FPAS beginning to show evidence of change, or is it likely to change, on societally relevant timescales (i.e. next 100–500 years)? (3) Is there evidence for past ice-sheet retreat (or readvance) of the FPAS, and when did this retreat (or readvance) occur? (4) How might ocean conditions, ice-shelf thickness, hydrology, surface mass balance (SMB), and grounding zone geometry change over the coming centuries, impacting the FPAS? and (5) Could changes to other parts of the Antarctic Ice Sheet lead to glaciological change in the FPAS (and vice versa)? Addressing these questions requires the following strategy:

To determine the susceptibility of the FPAS to sustained retreat via ocean-driven melt in the foreseeable future (i.e. next 100–500 years), high-resolution, adaptive-grid, catchment-scale numerical modelling of the FPAS with up-to-date bed topography (e.g. Siegert et al., 2020; Hill et al., 2021, 2024) is required (Fig. 9). Similar model simulations have been performed for the Aurora Subglacial Basin (Sun et al., 2016; Pelle et al., 2020), with model resolutions in those experiments sufficiently high to capture the narrow, complex, and in places prograde, bed topography of the FPAS trunk up to 200 km up-ice of the current grounding zone (Huybers et al., 2017) (Figs. 2 and 10). In addition to the ability to model changes to ocean cavity circulation in a grounding-zone retreat scenario, such experiments will require assessment of the role that the basal boundary conditions beneath the FPAS (i.e. recent marine sediments vs. older sedimentary basins) (Fig. 4) could play in controlling ice flow (Diez et al., 2018) and grounding-zone retreat (Ritz et al., 2015), given their role in determining basal friction. Improved, FPAS catchment-specific, high-resolution modelling of Last Interglacial and Pliocene scenarios are also required, as has been performed for the Wilkes and Aurora basins (e.g. Mengel and Levermann, 2014; Aitken et al., 2016);