Atmospheric conditions and temperatures are a first order control on environmental systems across the Peninsula, impacting cryospheric processes and ecosystem environments (Carrasco et al., 2021). The Antarctic Peninsula has been warming since at least the 1950s, when most regional instrumental records began (e.g., Turner et al., 2020). However, an earlier starting record, at Orcadas Station in the South Orkney Islands, demonstrates a warming beginning in the 1930s (Zitto et al., 2015), while a proxy temperature record from an ice core from James Ross Island suggests a sustained warming began ∼600 years ago (Mulvaney et al., 2012). Based on a quality controlled meteorological dataset of monthly and hourly data from year-around stations (http://legacy.bas.ac.uk/met/READER/; Turner et al., 2004; Carrasco et al., 2021), Vernadsky Station, on the west coast of the mainland Peninsula has seen the largest warming, by 0.45 °C ± 0.19 °C per decade (or 3.26 °C ± 1.40 °C over 73 years) during 1951–2024, with the temperature increase in winter twice this rate, predominantly driven by sea ice loss in the Bellingshausen Sea (Turner et al., 2020).

The warming at meteorological stations in the north-east Peninsula is smaller, but nevertheless still significant: for example, Esperanza Station warmed by 0.31 °C ± 0.11 °C per decade between 1946 and 2024 CE (or 2.43 °C ± 0.83 °C over 78 years), and this area has also seen a marked increase in extreme temperatures, as discussed later. A shorter-term cooling trend was also evident in the Peninsula region from 1999 to 2016 CE, particularly in austral summer, consistent with the region’s large natural variability and the influence of the tropics (Turner et al., 2016).

Some of the warming has been due to a positive trend in the SAM during summer, in response to the Antarctic ozone hole, which began to develop from the mid-20th century onwards (Thompson and Solomon, 2002; Polvani et al., 2011). This caused summer westerlies to strengthen over the Peninsula, resulting in increased maritime air on the western side, and contributing to the warming trend here (Orr et al., 2004). The strengthened winds also resulted in increased deflection of the flow over the Peninsula and the enhanced transport of relatively warm air masses to its eastern side, which is further enhanced by the formation of warm foehn winds (Marshall et al., 2006; Orr et al., 2008). The recent warming has coincided with marked increases in regional precipitation on the windward side of the Peninsula (e.g., Goodwin et al., 2016; Medley and Thomas, 2019).

The rapid Antarctic Peninsula warming described above can be compared with the globally averaged warming of 1.34 °C–1.41 °C for the years 2014–2025 relative to 1850–1900 CE (World Meteorological Organisation, 2025). Globally, the average trend since the late 1970s has been 0.2 °C per decade (The Copernicus Programme, 2025), indicating that the warming of the Antarctic Peninsula is occurring at a faster rate than the global average, particularly on the northern Antarctic Peninsula and Scotia Arc archipelagos.

Alongside this background of multi-decadal period of warming, the Antarctic Peninsula currently experiences extreme warm-temperature events that can last for a few days and cause surface melting of snow and ice (Orr et al., 2023), which have been linked to atmospheric rivers and localised foehn-induced warming, especially when they occur in combination (King et al., 2017a; King et al., 2017b; Gorodetskaya et al., 2023; Lu et al., 2023; Bozkurt et al., 2024; Gilbert et al., 2025). This warming has been accompanied by reduced sea ice extent (Gilbert and Holmes, 2024), increased surface melt (Gorodetskaya et al., 2023) and increased upwelling of warm mid-depth currents such as Circumpolar Deep Water. Antarctica, like other regions of the world, is now subject to extreme climate and weather alongside global heating and disruption to natural systems (Siegert et al., 2023).

To explore future climate, we calculate the multi-model mean of an ensemble of CMIP6 models (listed in Supplementary Table S1) to provide projections of atmospheric variables over the Peninsula (delineated in Supplementary Figure S1). Relative to 1850–1900 CE (to be in line with IPCC projections), this multi-model mean projects global near-surface (2-m height) temperature increases of 1.99 °C, 4.07 °C and 4.84 °C for the period 2081–2100 CE under SSP 1–2.6, SSP 3–7.0 and SSP 5–8.5, respectively. These values are not directly comparable with the IPCC best estimates given in Table 2 which are based on both model and observational evidence but do fall well within the very likely range.

The same multi-model mean projects a Peninsula near-surface temperature change of +2.3 °C, +5.2 °C and +6.1 °C for each scenario respectively for the decade 2090–2099 CE compared with pre-industrial (Table 4), showing enhanced warming across the Peninsula compared with the global mean. This is due to a series of feedbacks that amplify global warming, including changes in albedo, sea ice, and large scale circulation. These circulation changes bring mid-latitude air masses towards the Peninsula, and increase the regional foehn effect, warming the eastern Peninsula.

Comparing the end of the 21st Century (decade 2090–2099 CE) with the present period (2015–2024 CE), this multi-model mean predicts an annual mean increase in near-surface temperature over the Antarctic Peninsula domain that varies from 0.55 °C in SSP 1–2.6 °C–4.23 °C in SSP 5–8.5 (Tables 3, 4). The largest seasonal warming is in winter and the smallest in spring or summer, depending on the SSP. This would result in a longer summer melt season. Daily minimum near-surface temperature increases generally surpass those for daily maximum, with the difference exceeding 1 °C in spring and summer in SSP 5–8.5. The annual number of days across the Peninsula with mean temperatures greater than 0 °C increases in all season-SSP combinations other than spring-SSP 1–2.6. This annual count increases from 19.7 to 47.6 days in SSP 5–8.5, with, unsurprisingly, the largest proportional change in summer: under SSP 5–8.5 the proportion more than doubles from 13.8% to 30.5%.

Antarctic Peninsula precipitation increases with SSP severity, with decreases only projected in spring and summer in SSP 1–2.6. Annual precipitation increases by ∼29%, from 1.91 mm day- 1 -2.46 mm/day in SSP 5–8.5, with the largest seasonal change in autumn. Projected change in annual mean precipitation scales approximately in line with a Clausius–Clapeyron thermodynamic benchmark using α = 0.07 K⁻¹ (≈7% per K) as in Dalaiden et al. (2020), except for SSP1-2.6 where the change is well below this scaling, indicating dynamic or humidity changes could be important. Note that unlike Dalaiden et al. (2020) we analyse precipitation rather than net accumulation.

There is marked seasonal variation in the sign of changes in the snow precipitation percentage. Increases are projected in winter and spring for every SSP with the converse in autumn. In summer, there are decreases in SSP 3–7.0 and SSP 5–8.5, with the latter demonstrating a reduction from ∼76% to ∼62%, signifying that more than a third of precipitation will fall as rain. Supplementary Figure S2 shows the spatial pattern of precipitation change from the multimodel mean for each of the three emissions scenarios considered. Precipitation change exhibits an east-west pattern, which becomes even more pronounced under higher emissions. The majority of the precipitation change is projected to occur over ocean regions just west of the main Peninsula.

Increased greenhouse gases under the higher emissions scenarios are associated with a positive trend in the SAM, with strengthened and poleward-shifted westerlies over the Peninsula. However, some of these circulation changes are likely to be cancelled or offset if the anticipated recovery of the ozone hole by the end of the current century takes place, which is expected to reverse the positive SAM trend associated with the ozone hole (Son et al., 2010).

Extreme events can disrupt ecosystems, lead to extensive surface melt, and impact sea ice, driving step-changes in multiple systems (Siegert et al., 2023). For example, ice-shelf collapse can have long-lasting significant impacts on land ice, and extreme sea ice minima can stress the marine environment, impact climate through changes in ocean albedo, influence glacier calving, and lead to self-perpetuating processes as ocean heat uptake increases (Abram et al., 2025).

Evidence from multiple realms point to increasing intensity and frequency of extreme events, which have the potential to adversely affect wider systems across the Peninsula. The highest temperature recorded on the Antarctic mainland occurred in February 2020 near the northern tip of the Antarctic Peninsula, at 18.6 °C (González-Herrero et al., 2022). In this same 2019/2020 austral summer, George VI Ice Shelf, southwest Antarctic Peninsula, experienced a 32-year record high surface melt and ponding event (Banwell et al., 2021) (Figure 2C). In February 2022, the Antarctic Peninsula again underwent record surface melt during an extreme warm event, associated with an atmospheric river (Gorodetskaya et al., 2023). Warming of this kind, if occurring in high summer, has the unprecedented ability to melt snow and ice in regions not observed to melt previously.

Atmospheric rivers can transport vast amounts of moisture from mid- and subtropical latitudes to Antarctica (Clem et al., 2022). These long (>2000 km) and narrow (<1000 km) bands travel through the troposphere and are associated with extreme precipitation events. When they intersect mountainous topography, they can produce intense windward precipitation and cloud formation, with strong foehn warming on the leeward side (Turner et al., 2021; Wille et al., 2025). The Peninsula has a trend of increasing occurrence of atmospheric rivers (+0.89 atmospheric rivers per decade) from 1979 to 2022. Atmospheric rivers are associated with most extreme melt events in West Antarctica and the Antarctic Peninsula (Wille et al., 2025).

The ocean is also subject to warming events, resulting in the repeated breaking of minimum sea-ice records since 2017 (Doddridge et al., 2025; World Meteorological Organisation, 2025). The years 2022–2024 saw the three lowest Antarctic sea ice extents in the satellite era (World Meteorological Organisation, 2025). Extreme sea ice lows like these can negatively impact Antarctic fauna such as emperor penguins that rely on sea ice for breeding (e.g., Fretwell et al., 2023). Marine heat waves are intensifying in frequency and magnitude, a trend that will continue in the Southern Ocean under future projected warming (Gurumoorthi and Luis, 2025). A marine heat wave was recorded in early January 2020 in the Drake Passage, with sea surface temperature anomalies of +3 °C (Rigual-Hernández et al., 2025). These marine heat waves can lead to substantial impacts on ecosystems and sea ice. Marine heat waves are projected to increase in frequency and become more intense under climate warming scenarios, with the trend more intense at higher warming levels (Frölicher et al., 2018).

The multi-model mean from CMIP5 simulations under RCP 4.5 and RCP 8.5 from 15 global climate models (GCMs) provide projections for Antarctic heat waves. Heat waves (three or more consecutive summer days with very high temperatures) are expected to become more frequent and last longer in the coming century (Feron et al., 2021). Under RCP4.5, the number of heat waves is projected to double by 2070–2099 CE compared with 1961–1990 CE in West Antarctica. This could result in ∼15 days per season by 2070–2099 under RCP 4.5 and 25 days under RCP 8.5, rising from 5 days per season in 1961–1990 CE (Feron et al., 2021). Temperature anomalies of 6 °C under RCP 4.5 °C and 8 °C under RCP 8.5 could result in increased surface melt on the Antarctic Peninsula.

Extreme weather under a higher emissions scenario could also include more frequent rain events (Vignon et al., 2021). Atmospheric rivers are projected to increase in frequency globally as a result of increased moisture availability (Wille et al., 2025), though when the mean increase integrated water vapour is accounted for, atmospheric river frequencies are comparable to the present day, with a small increase in precipitation (Maclennan et al., 2025). Atmospheric rivers can accelerate glacier mass loss by washing or melting snow from glaciers and land surfaces. More frequent and more intense rainfall events generally coincide with positive surface temperatures, which will become more frequent on the Peninsula in the summer under all scenarios, but particularly under SSP 5–8.5 (Table 3e).

Increasing extreme events are likely to have a large impact on Antarctic Peninsula systems, including through accelerating frontal ablation from glaciers (Gorodetskaya et al., 2023). Increasing surface melt on ice shelves has the potential to destabilise them by infiltrating into the firn and refreezing, reducing the firn air content (Dunmire et al., 2024). This promotes meltwater pond formation on impermeable ice surfaces. In turn, meltwater ponds (e.g., Figures 2C,F) melting downwards through the ice can reduce the structural integrity of these huge floating ice slabs by inducing hydrofracture (Scambos et al., 2009). Marine heat waves and sea ice minima can impact Antarctic fauna such as emperor penguins (Fretwell et al., 2023).

The Southern Ocean plays an important role in regulating Earth’s climate, absorbing heat and carbon dioxide (Brown et al., 2019; Williams et al., 2023). The area around the western Antarctic Peninsula is warming rapidly, particularly at intermediate depths (200–1000 m). Sea surface temperatures of the Bellingshausen Sea warmed by > 1 °C from 1955 to 1994 (Meredith and King, 2005). This trend has continued, with a warming trend on the western Antarctic Peninsula continental shelf from 1993 to 2021 below 100 m depth (Davison et al., 2024). Increased wind strength and Antarctic Circumpolar Current transport is driving increased upwelling of the relatively warm Circumpolar Deep Water ocean current onto the continental shelf in the Bellingshausen Sea (Dinniman et al., 2012; Davison et al., 2024; Moffat and Meredith, 2018; Cai et al., 2023), bringing increased heat fluxes onto the continental shelf.

The increased flow of Circumpolar Deep Water onto the continental shelf on the western Antarctic Peninsula has contributed to the rapid recession of tidewater glaciers (Cook et al., 2016; Davison et al., 2024; Wallis et al., 2023). At Cadman Glacier, south of Anvers Island (65.6°S), rapid grounding line retreat has been linked to significant warm temperature anomalies from 50 to 500 m water depth from 2017 onwards (Wallis et al., 2023), related to increased incursion of Circumpolar Deep Water.

CMIP6 projections for Southern Ocean sea surface temperatures suggest that warming trends are likely under all future climate scenarios. The multi-model mean sea surface temperature change is ∼0.3 °C for SSP1-2.6 (Table 4). Increases in the northwestern Antarctic Peninsula are unlikely before 2080 CE, and not before 2100 in the northwestern Weddell Sea (Tonelli et al., 2021). Critically, the low emissions scenarios are associated only with pulses of warm water onto the continental shelf in the Weddell Sea (Teske et al., 2024).

The higher emissions scenarios will drive continued poleward intensification of atmospheric and oceanic circulation (Cai et al., 2023), impacting glaciers, sea ice, ice shelves and biodiversity across the western Antarctic Peninsula. For SSP 3–7.0, the approximate sea surface temperature change is ∼0.7 °C. For SSP 5–8.5, the mean sea surface temperature increase in the Southern Ocean is ∼1.6 °C (Table 4). For both SSPs 3–7.0 and 5–8.5, the timing of the emergence of the anthropogenic climate signal is earlier. The northwestern Antarctic Peninsula will warm more intensively and rapidly in these scenarios, as the Antarctic Circumpolar Current will drive warmer waters closer to the continent (Tonelli et al., 2021). Under these higher emissions scenarios, pulses of warm water onto the continental shelf in the Weddell Sea are replaced with continuous warm water flow (Hellmer et al., 2012; Naughten et al., 2021; Teske et al., 2024), which would negatively impact tidewater glaciers and ice shelves on the eastern Antarctic Peninsula. Marine heat waves will increase in frequency and severity in the Southern Ocean, placing additional stress on biological and cryospheric systems (Gurumoorthi and Luis, 2025).

Sea ice is a crucial component of the Antarctic cryosphere, acting as a climate amplifier through changing albedo (Riihelä et al., 2021), influencing ocean heat uptake (Brown et al., 2019), and impacting multiple glacier, oceanic and ecological systems (Abram et al., 2025). Its importance is recognised by the World Meteorological Organisation as a headline Global Climate Indicator, key to describing the changing climate (Trewin et al., 2021).

Antarctic sea ice (Figure 2) has responded more variably to warming than that in the Arctic. Instead of showing broad decreases over the satellite data record, Antarctic sea ice has exhibited both positive and negative trends depending on season, region, and time-period (e.g., Parkinson and Cavalieri, 2012; Matear et al., 2015; Doddridge et al., 2025). Regional dependency in these changes is partly explained by the interaction between stratospheric ozone depletion and rising greenhouse gas emissions. While these forcings can exert competing effects in East Antarctica (e.g., Marshall et al., 2014; Polvani et al., 2011), they act additively over the Antarctic Peninsula, strengthening westerly winds and contributing to the pronounced changes observed in this sector (Thompson et al., 2011; Turner et al., 2016). Consequently, sea ice patterns around the Antarctic Peninsula show distinct variability, with the Bellingshausen Sea just to the west (Figure 1) experiencing large declines in winter sea ice cover and overall duration since the late 1970s. In contrast, to its east, the Weddell Sea saw relatively stable or even slightly increasing sea ice cover until 2015. However, since 2015, the Weddell Sea has begun to experience more frequent periods of low sea ice cover, with conditions in this region in February 2023 contributing approximately 33% of the overall record low sea ice extent anomaly that year (Jena et al., 2024). Since 2023, sea ice conditions have remained lower than average, leading to some studies suggesting Antarctic sea ice has entered a new state (Purich and Doddridge, 2023; Abram et al., 2025). However, as in the Arctic (England et al., 2025), these changes could also be attributed to natural variability. Further loss of Antarctic sea ice risks self-perpetuating processes as sea ice thins and ocean heat uptake increases, potentially leading to irreversible future declines (Abram et al., 2025). The recent extreme sea ice lows are challenging for the skill of current climate models to accurately predict future sea ice trends and abrupt changes (Abram et al., 2025; Holmes et al., 2024; Diamond et al., 2024).

While there have been opposing sea ice trends around the Antarctic Peninsula, archived CMIP6 model projections (Figure 3) suggest continued reductions in both sea ice area and duration. Around the Antarctic Peninsula, this trend is particularly pronounced, with models projecting greater reductions in sea ice cover compared to other parts of Antarctica. Under the SSP 1–2.6 emission scenario the changes are, however, modest, with overall seasonal reductions in sea ice concentration of between 1% and 2%, largest in winter (Figure 3; Table 4). However, given the current sea ice minima, this may be an underestimate. Under the SSP 3–7.0 scenario, summer sea ice reductions of 10% are predicted, with winter reductions of 13%. Under the SSP 5–8.5 emission scenario, winter reductions increase to 20% (as averaged for all sea ice areas shown), with summer reductions of around 12% (Table 5). However, these predicted changes show strong regional dependence, with large winter and spring reductions in sea ice concentration in the Bellingshausen Sea, while the largest reductions in the Weddell Sea occur in summer and autumn (Figure 3; Table 5).

Summers with extreme low Antarctic sea-ice cover impact glacier calving, change the formation rate of Antarctic Intermediate Water, affecting heat and carbon uptake, and change the timing and size of surface phytoplankton blooms. These changes to ocean conditions, both at the surface and at depth, impact marine ecosystems and food chains, placing stress on marine mammals, birds and other fauna. Significant decreases in sea ice extent will likely provide positive feedback to the polar amplification of climate change, as the lower albedo of open ocean water results in increased ocean warming (Lenton and Armstrong, 2023).

Ice shelves (e.g., Figure 2H; Figure 4) fringe ∼75% of Antarctica’s coastline and regulate ice-sheet stability by buttressing the flow of inland ice (Furst et al., 2016). Recent oceanic and atmospheric warming have together driven considerable ice shelf melt and retreat (Cook et al., 2016; Rignot et al., 2019), and collapse events since 1995 (Cook and Vaughan, 2010). Ice-shelf collapses have resulted from multiple factors, primarily increased basal melting that drives long-term thinning (Glasser and Scambos, 2008; Etourneau et al., 2019), especially once ice shelves recede beyond key structural thresholds such as basal pinning points (Doake et al., 1998; Gilbert and Kittel, 2021) (Figure 4). Regional atmospheric and oceanic warming has contributed to ice shelf thinning and collapse, driving increased surface melt, ponding and subsequent hydrofracturing (van den Broeke, 2005; Banwell et al., 2013; 2024; Robel and Banwell, 2019). Changing sea ice concentrations have also played a role (Christie et al., 2022). Rapid ice shelf collapses have been followed by abrupt glacier acceleration and thinning, due to a loss of buttressing (De Rydt et al., 2015; Gudmundsson et al., 2019; Scambos et al., 2014; Parsons et al., 2024) (Figure 4). The most substantial grounding line retreat on the Peninsula was recorded at Hektoria Glacier, a tributary glacier of the former Larsen B Ice Shelf (Figure 1,Figure 2E). Hektoria receded 16.3 ± 0.5 km from 1996 to 2020 CE following that shelf’s collapse in 2002 (Wallis et al., 2024).

Currently, basal melting accounts for most ice shelf mass loss (Garbe et al., 2023; Davison et al., 2023). However, relative to surface melting, basal melting is less understood and more difficult to predict under future warmer climates (Fricker et al., 2025), because it is influenced by cavity geometry, basal channels and crevasses, influent subglacial meltwater, and warm seawater intrusions, all of which are difficult to simulate. In general, projected increases in ocean warming will increase ice shelf basal melting and hence thinning. Ocean warming does not necessarily lead to more ice shelf calving, and Coulon et al. (2024) show relatively unchanged calving fluxes for many Antarctic sectors, despite significant increases in basal melt and surface runoff for RCP 8.5. Coupled ocean-sea ice modelling indicates that basal melting driven by warm water intrusion, notably that of Circumpolar Deep Water, increases to 2140 CE under all emissions scenarios, but is particularly sensitive under high warming scenarios (Kusahara et al., 2023). Based on projections from the UK Earth System Model (UKESM1), the Ross Gyre may expand into the Amundsen and Bellingshausen seas, bringing ∼1 °C warming due to increased Circumpolar Deep Water transport (Gómez-Valdivia et al., 2023).

Although ice shelf surface melt is currently lower than basal melt at most ice shelves (e.g., Smith et al., 2020; Garbe et al., 2023), it is projected to increase non-linearly with future atmospheric warming (Trusel et al., 2015; Kittel et al., 2022). This is particularly the case above 0 °C (Banwell et al., 2023; Orr et al., 2023), largely driven by a positive melt–albedo feedback from surface ponding and slush (Dell et al., 2024). Additionally, a substantial near-surface layer of densified snow, or firn, on most Antarctic ice shelves (Dunmire et al., 2024) buffers against instability by storing surface meltwater (Holland et al., 2011; Kuipers Munneke et al., 2014). Under increased surface melt rates, this porous layer saturates and refreezes (Hubbard et al., 2016), depleting its firn air content (FAC), enabling meltwater to pond on the now impermeable ice (e.g., Figure 2E) and increasing the likelihood of hydrofracture and collapse (Lai et al., 2020; Jourdain et al., 2025).

Peninsula-wide FAC projections range from minimal change under a low-emissions scenario, with only Wilkins Ice Shelf expected to see a >10% decrease, to substantial depletion under a high-emissions scenario (Dunmire et al., 2024) (Table 4). Under the highest emission scenario, Kittel et al. (2022) projected 2500–6500 Gt of cumulative surface melt across all Peninsula ice shelves by 2100, with the greatest increase (>2000 mm w.e. yr⁻¹) expected in the northern and western Peninsula, including George VI Ice Shelf (Garbe et al., 2023). Under this very high emissions case, only Stange Ice Shelf retains >10 m of FAC, with FAC of Scar Inlet, Wilkins and Larsen C ice shelves amongst the lowest across Antarctica. Future intense atmospheric rivers could further destabilise these ice shelves (Wille et al., 2022).

Gilbert and Kittel (2021) projected extensive surface runoff, indicative of substantial FAC reductions, on Wilkins, George VI, and Larsen C ice shelves at 2 °C and 4 °C global mean warming above preindustrial temperatures, thresholds that approximately align with SSP 1–2.6 and SSP 5–8.5, respectively (cf. Tables 1, 2). Of these three ice shelves, several studies suggest that Larsen C and Wilkins ice shelves are likely to collapse by 2100 CE under very high emissions (Bell et al., 2018; Van Wessem et al., 2023; Dunmire et al., 2024), partly driven by their projected increased hydrofracture potential (Jourdain et al., 2025). Conversely, George VI Ice Shelf’s compressive flow regime makes it relatively resistant to hydrofracture (Lai et al., 2020; Gilbert and Kittel, 2021; Jourdain et al., 2025) despite exceptionally high present-day surface melt rates and pond density (Banwell et al., 2021; Banwell et al., 2024).

While the physical processes driving ice shelf collapse are poorly represented in numerical models, projections that include collapse result in substantially higher ice mass loss (Fricker et al., 2025; Seroussi et al., 2024; Edwards et al., 2021). Using an ensemble of 16 models, Seroussi et al. (2024) concluded that collapse would, on average, result in 1.1 m higher sea level from the Antarctic Ice Sheet by 2300 under a high emission scenario, with the Peninsula notably being amongst the most sensitive regions of Antarctica, already responding by 2100 CE. Predicting this future stability is further complicated by ice shelf 3D structural complexity, including stratification, foliation and fractures inherited from feeder glaciers (e.g., Miles et al., 2025), and subsequent stratification, rifts, suture zones and englacial and basal marine ice formed near and downstream of the grounding zone (e.g., Glasser et al., 2009; Miles et al., 2025).

Melting and recession of grounded land ice in the Antarctic Peninsula contributes to global sea level rise (Otosaka et al., 2023). Peninsula outlet glacier recession will increase habitat connectivity and make new ecological habitats available for colonisation, impacting terrestrial fauna, and ocean freshening and changes in sedimentological load impacts marine ecosystems (Griffiths et al., 2024). Land ice on the Peninsula influences wider ocean circulation (Meredith et al., 2013) and impacts regional albedo. Antarctic Peninsula land ice is also a sentinel for climate change, due to its rapid thinning and recession.

Otosaka et al. (2023) reported an average annual mass loss of 13 ± 5 Gt yr^-1 across the Peninsula for 1992–2020, rising to 21 ± 12 Gt yr^-1 from 2017 to 2020 CE. The mass losses in the earlier part of the 2000s were largely associated with ice shelf collapse and glacier acceleration, and from 2012 to 2016 were partially offset by an extreme snowfall event in 2016. Since 2016, mass loss has increased (Otosaka et al., 2023), driven by sustained retreat of marine-terminating glaciers (Hanna et al., 2024). This was forced by oceanographic processes, including ocean warming and disintegrating landfast sea ice in embayments in front of former Larsen B Ice Shelf areas, which removes buttressing and results in rapid glacier terminus recession (Ochwat et al., 2024; Loebel et al., 2025). Furthermore, grounded glaciers across the Peninsula are accelerating (Boxall et al., 2022; 2024; Wallis et al., 2023; Davison et al., 2024). Seasonal glacier acceleration across the Peninsula has been attributed to a mixture of ocean and atmospheric drivers (Fricker et al., 2025). This includes enhanced basal lubrication following increased surface melt (Boxall et al., 2024), warmer oceans (Cook et al., 2016; Davison et al., 2024) and reduced sea ice (Rott et al., 2020). These Peninsula outlet glaciers, terminating in narrow fjords directly into the ocean, are analogous to the outlet glaciers of the Greenland Ice Sheet, and processes long observed in Greenland are increasingly observed in the Antarctic Peninsula (Mottram et al., 2025).

Constraining future land ice changes on the Antarctic Peninsula is challenging, as there remain severe limitations in the predictive capability of ice sheet models (cf. Bamber et al., 2019). The current state of the art models and circum-Antarctic ISMIP6 multi-model ensembles (Nowicki et al., 2020; Seroussi et al., 2020; Seroussi et al., 2024; Figure 5) are hindered over the Peninsula, since these continent-wide simulations are challenged by the (often poorly observed) complex topography and low spatial resolution relative to ice thickness and valley width in this region. Very few studies have investigated in detail the mass balance sensitivities of Peninsula glaciers. It is, therefore, challenging to quantify with any degree of confidence the key drivers of ice mass loss and gain under different climate scenarios. This results in large uncertainty in the future behaviour of land ice on the Antarctic Peninsula, with a large spread of projections (Seroussi et al., 2024; Table 3) that are largely independent of scenario until well after 2100 CE. Major differences between projections only tend to emerge after 2150 CE (e.g., Lowry et al., 2021).

Under lower emissions scenarios, it is likely that current trends will continue, with some mass loss potentially offset by increased snowfall (e.g., Golledge et al., 2015; Pattyn et al., 2018; O'Neill et al., 2025). A mid to high-range scenario could result in increased runoff, decreased firn air content, and increased surface meltwater (which is already seen on land ice in the northern Peninsula, e.g., Figure 2F). This could potentially impact the basal hydrology of Peninsula glaciers, exacerbating current trends (e.g., Boxall et al., 2024).

Grounding line retreat and further ice shelf collapse is increasingly likely under SSP 5–8.5 (Table 4). Key areas of land-ice vulnerability include glaciers feeding George VI Ice Shelf and those within the Bellingshausen Sea sector (Seroussi et al., 2024; O'Neill et al., 2025); both are regions where grounded glaciers are currently accelerating (e.g., Boxall et al., 2024), and where substantial land ice loss under a higher emissions scenario could result in glacier ungrounding (Figure 5). Loss of Larsen C Ice Shelf, likely by 2100 CE under a very high emissions scenario (Dunmire et al., 2024), could result in increased grounding line discharge (Mitcham et al., 2022) and grounding line retreat in parts of the eastern Peninsula. Due to cold temperatures and the relatively low height of the Peninsula’s bedrock above sea level, even under the most extreme scenarios, glaciers will likely keep a marine-terminating front under all scenarios for multiple centuries (Figure 5). Loss of ice shelves and sea ice would likely exacerbate current trends by decreasing buttressing and increasing calving from marine terminating glaciers.

There is wide uncertainty in sea level contributions from the Antarctic Peninsula. Under a lower emissions scenario, the ISMIP6 multi-model ensemble mean predicts growth of land ice, with a limited sea level contribution: −7.8 ± 11.8 mm by 2100 CE and −19.7 ± 35.5 mm by 2300 CE (Table 3; Supplementary Table S3; Figure 6). Alternatively, under a very high emissions scenario where ice shelf collapses occur, total sea level contributions could reach 7.5 ± 14.1 mm by 2100 CE and 116.3 ± 66.9 mm by 2300 CE (Table 4). This indicates that very high emissions scenarios, in synergy with decreased sea ice and ice shelf collapse, would substantially impact the Antarctic Peninsula, with extensive grounded glacier recession and considerable new exposure of bare rock and ground (Seroussi et al., 2024).

Key physical-chemical drivers of change within marine ecosystems around the Antarctic Peninsula include warming atmospheric and oceanic temperatures, declines in ice, aragonite undersaturation and increasing extreme events such as storms and marine heatwaves (Schofield et al., 2024; Griffiths et al., 2024). For example, the retreat of glacier fronts in the western Antarctic Peninsula has resulted in a change from an ascidian-dominated fauna to a mixed fauna, driven by increased sedimentation rates (Sahade et al., 2015). These stressors impact on biota both directly (for instance, increasing adverse weather and flooding of penguin nest sites; Cimino et al., 2014) and indirectly, via changing light and nutrient supply which impact on the phytoplankton supply to the base of the food web (Hayward et al., 2025; Atkinson et al., 2025).

These impacts are frequently modelled or speculated upon, but the detail of projections in marine ecosystems is highly uncertain for two reasons. First, uncertainties in modelling future Antarctic Peninsula climate and sea ice propagate the uncertainties in food web dynamics (Ding et al., 2025). Second, species show variable degrees of resilience to change. Some species show a very limited ability to cope with even a 1 °C increase in water temperature, while others can tolerate greater increases (Peck, 2018; Clark et al., 2019). For instance, some species appear to have shifted their range southwards in response to warming (e.g., Pakhomov et al., 2002), while others have not (Tarling et al., 2018) or their shifts are decoupled from the pace of warming (Atkinson et al., 2022). However, there is considerable uncertainty due to experimental regimes that do not closely relate to the real world or the key life stages of marine fauna, and a broad lack of studies involving either multiple variables or multiple interacting species (Suckling et al., 2015; Peck, 2018; Morley et al., 2019).

Notwithstanding these caveats, CMIP6 projections under a very high emissions scenario project increases in phytoplankton around the Antarctic Peninsula (Fisher, 2024), with warming seabed temperatures projected to increase species richness east of the Antarctic Peninsula and decrease it to its west (Griffiths et al., 2017). Increased storminess in combination with warming is projected to enhance colonisation by currently non-native species such as those associated with drifting kelp rafts and floating plastic debris from further north, as well as from species commonly found in biofouling communities on ships (Barnes and Fraser, 2003; Fraser et al., 2018; McCarthy et al., 2019; Avila et al., 2020). At the same time, ranges of key pelagic species such as krill (Euphausia superba) and salps are projected to contract to the south (Chen et al., 2025), with the Antarctic continent and fringing shelves acting as a physical barrier to constrict available living space. Krill, which provide an important food source for whales, seals, and penguins, are dependent on winter sea ice as a nursery for their young. Range-expanding taxa and non-indigenous, invasive, species could become established in the west Antarctic Peninsula if factors that currently inhibit them (such as sea ice scour) are removed (Griffiths et al., 2024). Increased ice-free coastline due to loss of land ice in the Antarctic Peninsula could provide new habitats for colonisation by intertidal and shallow subtidal communities (ibid.).

The mechanisms by which invertebrates and their vertebrate predators respond to warming will reflect both their contrasting degrees of stenothermy and their trophic position. The more stenothermic invertebrates may need only modest degrees of warming to push them beyond their thermal limits, albeit with recent work describing a wide range of thermal sensitivity (Clark et al., 2019; Morley et al., 2024). Long-term (multi-year) field deployments of heated and non-heated benthic settlement plates have demonstrated variable influences on growth of different species and on interspecific competition (Ashton et al., 2017; Barnes et al., 2021), illustrating the multiple levels of complexity involved in predicting outcomes.

The warm-blooded predators, by contrast, are expected to be more resistant to these water temperature changes, but more sensitive to direct climate factors such as flooding of breeding colonies or reduction of sea ice substrate, coupled to indirect effects of climate mediated through changes in their prey (see Schofield et al., 2024). This interplay makes it challenging to project the futures of the suite of iconic species including penguins, albatrosses, seals, and whales, that vary both in their prey specialism and in their requirement for sea ice. Thus, for instance, a partial replacement of an ice- and krill dependent penguin species (Adélie penguin) with the non-ice-dependent gentoo penguin on the western Antarctic Peninsula (Herman et al., 2020), but whether this will continue depends also on the trajectory of krill, a major food source for these species, and the relative resistance of the species to variations in its supply. Finally, the 2022 CE sea ice record low in the Bellingshausen Sea region resulted in the regional breeding failure of emperor penguin colonies (Fretwell et al., 2023). It is therefore likely that changes in sea ice will negatively impact these iconic faunas, and sea ice loss could result in reduced survival capacity of some Antarctic species that depend on it (Abram et al., 2025).

Due to its warmer climate, the ice-free ground of the maritime Antarctic biological region, which includes the western Antarctic Peninsula, its offshore islands and the Scotia Arc archipelagos, is biologically richer and more diverse than the rest of the continent (Convey and Biersma, 2024). There are no resident terrestrial vertebrates other than the snowy sheathbill, a scavenging migratory bird species closely associated with marine vertebrate aggregations on land. The terrestrial fauna therefore consists entirely of invertebrates and is dominated by the microarthropod groups of mites and springtails and microinvertebrate groups of nematodes, tardigrades, and rotifers. Many of these are now known to be endemic to Antarctica, and often to specific smaller regions within the continent, and illustrate the long history of evolution in isolation and regionalisation of Antarctica’s terrestrial fauna (Convey et al., 2020).

Antarctic vegetation is dominated by lichens and mosses, the former similarly showing considerable levels of endemism (Øvstedal and Smith, 2001), while there is now also increasing evidence of microbial group endemism in Antarctica (e.g., diatoms, Verleyen et al., 2021). Thus, Antarctic terrestrial ecosystems are of considerable conservation importance in terms of the values recognised in the United Nations’ Convention on Biodiversity. The Peninsula region currently contains 32 Antarctic Specially Protected Areas (with further proposals in the process of assessment), of which three are entirely marine, with 16 on the Antarctic Peninsula, 13 on the South Shetland Islands and three on the South Orkney Islands (see https://www.ats.aq/devphBackEnd/StatusReport/pdf?lang=e). These are designated under the formal processes defined in the Protocol on Environmental Protection to the Antarctic Treaty, although it is recognised that the current area protection system in Antarctica is inadequate and unrepresentative (Coetzee et al., 2017) as well as having very little resilience in the face of climate change (Hughes et al., 2021).

Antarctic Peninsula terrestrial ecosystems face three primary contemporary challenges or threats, these being climate change (primarily related to warming and changes in precipitation form or amount), the anthropogenic introduction of non-native invasive species and the direct impacts of the expanding human footprint (e.g., station and facility construction, increased visitor numbers and visit locations, physical damage by vehicles and footfall) as a result of competition for use of the very limited total ice-free area available (Convey and Peck, 2019; Hughes et al., 2020; Hughes et al., 2025; Lee et al., 2022). If considered in isolation, many Antarctic terrestrial and freshwater species are likely to ‘benefit’ from current levels of warming (Convey, 2011) through the relaxation of developmental constraints imposed by low temperature and lack of liquid water availability, increases in the length and cumulative thermal energy receipt (‘day degrees’) of the summer biologically active season, and in the extent of available ice-free habitat. However, recent studies have highlighted that both microbial and invertebrate species may already be at increased risk of reaching their upper thermal limits, with increasingly frequent and more intense extreme climatic events (Misiak et al., 2020; Bahrndorff et al., 2025).

There is widespread expectation that vegetation extent will expand as a result of warming and increased water availability. However, this has only been explicitly demonstrated at local scales at two locations in the Peninsula region over the last 60 years, for the two native flowering plants (Parnikoza et al., 2009; Cannone et al., 2022) and a single functional group of bank-forming mosses (Cannone et al., 2017). Warming is predicted to lead to an important change in reproductive strategy in some members of Antarctica’s dominant vegetation type, mosses, where currently the contribution of sexual reproduction (inherently creating increased genetic diversity) and the production of highly dispersive spores is generally limited compared to lower latitudes (Smith and Convey, 2002). A recent study on King George Island provides one of the first instances of observation of this change in the regionally widely distributed moss, Sanionia uncinata (De Vargas et al., 2024). Similarly, in the mid-1990s it was suggested that the frequency of successful seed maturation in the native grass, Deschampsia antarctica, had already increased, supporting the local expansion of populations of the species (Convey, 1996), although no more recent studies monitoring this appear to be available.

The potential for remote sensing approaches to assess Antarctic vegetation extent has recently led to the first attempt to provide a vegetation map of the entire continent (Walshaw et al., 2024), which it is accepted will develop further in future. However, recent claims based on satellite data alone since the mid-1980s (Roland et al., 2024) have proved contentious, claiming greater than an order of magnitude (x14) increase in extent over the entire Antarctic Peninsula and South Shetland Islands, claims which lack biological realism (Colesie et al., 2025). The latter authors conclude that, at present, any overall vegetation expansion in this region (referred to as ‘greening’ in the publication and media reports, although this term has been taken from the Arctic literature, where it precisely refers to a different process in detail) is, at most, a tiny fraction of that claimed and, at present, may not even be demonstrable.

The change in precipitation from snow to predominantly rain in an extending proportion of the summer, while directly promoting biological activity, also brings risks of flooding or washing out of terrestrial ecosystems (comparable to the negative impacts of flooding on penguin rookeries noted above). It can also accelerate snow loss from glacial surfaces, changing the extent and timing of availability of habitat for snow algae, which are important Peninsula primary producers (Gray et al., 2021). As significant rainfall events extend later into the autumn and earlier in the spring period, the phenomenon of ‘rain-on-snow’ events, previously not considered in Antarctic biological research, becomes an increasingly likely expectation. In the High Arctic (e.g., Svalbard), these events have become increasingly frequent during the winter months, resulting in the formation of thick and impermeable ice layers either within the snow pack or on the ground surface (Pedersen et al., 2022). These events have wide ranging deleterious consequences, including direct damage to or loss of vegetation and contained invertebrate fauna through ice encasement, anoxia in underlying substrates, and restricted access to winter forage for grazing vertebrates. Although the latter is not pertinent to the Antarctic Peninsula region, most of these consequences are likely to apply and provide a new and unquantified threat to the region’s terrestrial ecosystems.

A recent continental North American study has identified a related issue with potential Antarctic relevance: Thompson et al. (2025) highlight the negative consequences for insects of the loss of the ‘subniveum’, a thermally buffered habitat between the surface snow layer and the underlying ground, which some species rely on for survival overwinter. This sub-snow habitat certainly exists in Antarctica and is documented to promote microbial activity in the spring shoulder season as melt commences (Cockell et al., 2002). Its loss, which could result from various reasons including winter thaws, rain-on-snow events or failure to form due to changed distribution or timing of snow accumulation in the early stages of winter (see Coulson et al. (2023) for an example from High Arctic Svalbard), exposes the soil surface habitat and its contained invertebrates to much more extreme and variable temperatures than in snow-buffered habitats, in some cases breaching their thermal tolerance limits.

The entire maritime Antarctic is an important region for breeding penguins and other seabirds, as well as seals. While these are considered under marine ecosystems above, they provide an important mechanism of transfer of marine-derived nutrients to terrestrial ecosystems, where they are recognised as a key driver leading to “hotspots” of terrestrial diversity; thus, future changes in distribution or numbers of these marine species are expected to have knock-on consequences for neighbouring terrestrial ecosystems and their biodiversity (Bokhorst et al., 2019; Convey and Biersma, 2024).