Cirques in the Transantarctic Mountains reveal controls on glacier formation and landscape evolution

In this study, we analyse the morphometry of cirques in the Transantarctic Mountains (TAM) to understand regional glacier formation and landscape evolution since the onset of Cenozoic glaciations. We find that, unlike most glacierised regions worldwide, aspect bias for cirques in the TAM is not particularly strong, indicating that glaciers were able to form and cirques develop on slopes with a variety of aspects. This is perhaps unsurprising, given that Antarctica ’ s climate has been conducive to long-lived and extensive glaciation for many millions of years. Surprisingly, where cirques in the TAM show an aspect bias, this is typically towards the North, NW and/or NE, rather than favouring South-facing slopes where direct solar radiation is comparatively limited. This lack of a poleward aspect bias is unlike most cirque populations globally and indicates that total solar insolation was not a key control on where former glaciers in the TAM were able to initiate. Instead, we argue that prevailing wind directions played a dominant role in controlling the slopes on which past glacier development was favoured. Specifically, South-facing slopes in the TAM are directly exposed to katabatic winds which originate from the interior of the East Antarctic Ice Sheet (EAIS). These slopes are therefore susceptible to wind deflation, with snow and ice being redistributed to lee-side slopes where it can accumulate due to protection from the wind. For this reason, North, NE and/or NW facing slopes may have favoured glacier development, and therefore resulted in a concentration of cirques with these aspects. This evidence suggests that most cirques in the TAM are no older 34 Ma, as outward radiating winds from the continent ’ s interior could only have prevailed when the EAIS was present (in some form). By contrast, the very highest (and likely oldest) cirques in the TAM have more varied aspects, indicating that they may have formed before katabatic winds came to dominate, and by extension, before widespread growth of the EAIS at 34 Ma, and could perhaps have formed as far back as 60 Ma. In general, we find that cirques in the TAM have similar dimensions to those in other regions globally, despite having been occupied by glacial ice for far longer. Thus, our findings support a growing body of evidence which suggests that cirque size and glacier occupation times are not directly coupled, though there is some evidence of spatial variability in cirque size which might relate to the differences in the dynamics of former glaciers.


Introduction
Cirques are topographic depressions formed (and largely shaped) by subglacial erosion.They reflect sites of former mountain glaciation and are ubiquitous in glaciated regions globally (Evans, 1977).Given their association with past mountain glaciers, their distribution and dimensions can provide palaeoenvironmental information, and they have been used for this purpose in many regions globally (e.g., Evans and Cox, 1995;Wallick and Principato, 2020;Zhang et al., 2020;Evans et al., 2021).Despite this, cirques in Antarctica (the most extensively glaciated continent on Earth) have received only limited research focus, partly because most are currently submerged beneath ice sheets (see Bo et al., 2009;Rose et al., 2013), and due to the historical paucity of highresolution satellite imagery (particularly south of ~83 • S) and digital elevation models (DEMs).Exceptions are Selby and Wilson (1971) who considered the possible age of a small number of cirques (n ~15) in the Wright Valley region of the Dry Valleys (part of the TAM), Andrews and LeMasurier (1973) who mapped and analysed a small number (n = 8) of very large cirques on the flanks of volcanoes in Marie Byrd Land (West Antarctica), Aniya and Welch (1981) who investigated the morphometry of 56 (mostly ice free) cirques in the Victoria Valley system, part of the Dry Valleys, and Holmlund and Näslund (1994) who mapped and measured the dimensions of a small number (n = 12) of subglacial cirques and U-shaped valleys in Dronning Maud Land, East Antarctica.
Fortunately, over the past decade, various digital datasets of Antarctica's subglacial (e.g., Fretwell et al., 2013) and surface (Howat et al., 2019) topography have become available and fundamentally changed the way the continent's land and ice are investigated (e.g., Jamieson et al., 2014;Small et al., 2021).The subglacial data are excellent for analysing large features (e.g., Rose et al., 2013) but are currently of insufficient spatial resolution to map and analyse comparatively small features such as cirques (typically <1 km in diameter, Barr and Spagnolo, 2015;Evans and Cox, 2017).By contrast, topographic models of the ice-free landscape are now of sufficient spatial resolution (<10 m) that features such as cirques can be mapped and analysed in unprecedented detail (Barr et al., 2022).In this study we use ice-free topographic data to analyse the morphology (aspect, size, and shape) of glacial cirques in the Transantarctic Mountains (TAM).This is the first study to analyse cirque morphology across the TAM to reveal information about Antarctica's glacial and landscape history during the Cenozoic.

Study area
The TAM extend >3000 km across the Antarctic continent, forming a geographic divide between the East and West (Elliot, 2013).Given their extent, the TAM cover ~30 degrees of latitude, and reach up to ~4500 m in elevation.They have existed for hundreds of millions of years (since the Neoproterozoic), but their current form is primarily Cenozoic in origin (Goodge, 2020).The most recent phase of the range's evolution involved Cenozoic uplift and volcanism alongside glaciation (Goodge,  (Barr et al., 2022).During this period, both warm-and coldbased mountain glaciers were present, before much of the range became shrouded by ice sheets at the Eocene-Oligocene boundary (~34 Ma, DeConto and Pollard, 2003).Thus, the mountains have a history of longterm glacier occupation, and are currently extensively glacierised.Despite this, there are large areas of exposed bedrock, not currently submerged beneath glacial ice (Goodge, 2020).
Cirques in the TAM are presumed to have formed during Cenozoic glaciations (Barr et al., 2022), though as with other cirque populations globally (Barr and Spagnolo, 2015), assigning a specific time of formation to these cirques is very difficult.It is possible that some date back to the onset of Cenozoic mountain glaciation, but others may be far younger (perhaps forming in the last few million years).Irrespective of their time of formation, it is likely that all cirques have experienced periods of active erosion and growth beneath warm-based ice masses, and periods of minimal growth beneath cold-based and therefore minimally erosive ice masses (Barr et al., 2019).
At present, climate in the TAM varies as a function of altitude, though cryo-arid conditions prevail, with mean monthly temperatures consistently below zero, even in coastal areas during the Austral Summer (Dec, Jan, Feb) (Wang et al., 2023).The region's weather is also dominated by katabatic winds which originate from the continent's interior and are funneled by ice topography through the valleys of the TAM (Parish andBromwich, 1987, 2007).Though the strength of these winds reduces during summer, they are some of the strongest (up to 90 m/s) and most persistent low atmosphere winds on Earth (Parish and Bromwich, 2007).

Methods
The cirques analysed in this study were first mapped by Barr et al. (2022) using the Reference Elevation Model of Antarctica (REMA) Digital Surface Model (DSM), which has an 8 m spatial resolution and vertical error of <1 m (Howat et al., 2019).Where possible, this mapping was cross validated with existing cirque maps, though these are few and only cover small regions (see Selby and Wilson, 1971;Andrews and LeMasurier, 1973;Aniya and Welch, 1981;Holmlund and Näslund, 1994).Barr et al. (2022) mapped all cirques identifiable in the TAM, irrespective of their degree of glacier-coverage, i.e. since the mapping was performed from DSM data, all features with cirque-like topography were recorded, even if partially ice-covered (see examples in Fig. 1).Barr et al. (2022) subsequently divided this dataset of 14,060 cirques into 'glacier-occupied' (n = 12,768) and 'glacier-free' (n = 1292) populations (Fig. 2a).In the present study we use both populations (which can be downloaded as shapefiles from Barr et al., 2022).For the analysis of cirque aspect (i.e., orientation) all cirques in the dataset were considered, irrespective of their degree of ice cover, and aspect was measured as the orientation of the median axis, following Barr and Spagnolo (2013).Maximum altitude (Z_max) was also calculated for all cirques.For size and shape analysis, only 'glacier-free' cirques were considered, since cirques occupied by ice of unknown thickness are unlikely to yield robust information.For 'glacier-free' cirques, length (L), width (W), depth (H) minimum altitude (Z_min) and mean altitude (Z_mean) were calculated using an automated GIS tool, ACME, developed by Spagnolo et al. (2017).Additional calculated metrics for glacierfree cirques include the shortest distance to the modern coastline (Dist), following Oien et al. (2022), and the dominant lithology (accounting for greatest surface area of each cirque) using geological data from Cox et al. (2023).

Results
The 14,060 cirques are distributed throughout the TAM (Fig. 2a) but are most abundant and most densely spaced in areas immediately south and west of the Ross Sea and Ross Ice Shelf (Fig. 2b).

Cirque aspect
The entire cirque population shows a weak Northern aspect bias, with a vector mean (VM) aspect of 002 • (Fig. 3a).Vector strength (VS), which measures the degree of asymmetry in aspects, is only 11 % which suggests very weak aspect asymmetry, i.e., cirques do not show clearly preferred orientations.When cirques are divided into sub-populations (Fig. 3a), aspect shows a NW, North, or NE bias, but again this is typically very weak (Table 1).When the aspect of only glacier-free cirques are considered (Fig. 3b) very similar aspect trends are observed (Table 1).

Cirque size and shape
Glacier-free cirques in the TAM have L ranging from 199 m to 2445 m, with a mean of 779 m and median of 698 m; W ranging from 138 m to 3170 m, with a mean of 833 m and median of 746 m; and H ranging from 79 m to 1271 m, with a mean of 413 m and median of 380 m (Table 2).Frequency distributions for L, W and H are all unimodal, with positive skew (Fig. 4a).All size metrics (L, W, H) are positively correlated (0.59 < r < 0.82) (Table 3) indicating that long cirques also tend to be wide and deep.To investigate how shape varies with growth, cirque L, W, and H are plotted against size ( 3 √LWH, see Evans, 2006a) (Fig. 4b).Based

Table 1
Aspect Vector Mean (VM) and Vector Strength (VS) for cirques in the TAM and for different sub-regions of the TAM (for region acronyms see Fig. 2).on these plots, power exponents for L, W and H are 1.04, 1.01, and 0.95 respectively.This indicates that over time, cirques have grown roughly isometrically, though lengthening has slightly outpaced widening, which has slightly outpaced deepening.This comparatively isometric growth is atypical of cirque populations globally, which often show deepening substantially outpaced by lengthening and widening (see Barr and Spagnolo, 2015), and may reflect the varied lithology and structure of the TAM, as found for some large cirque populations elsewhere globally (e.g., Evans, 2006b;Barr and Spagnolo, 2013).In considering other relationships with cirque dimensions (Table 3), it is apparent that all size metrics (L, W, H) are positively correlated with Z_max (0.16 < r < 0.41, Fig. 5).W and H show a statistically significant positive correlation with Z_mean (r = 0.08 and 0.23, respectively).Only H shows a statistically significant correlation with Z_min (r = 0.10).These correlations suggest some relationship between cirque size and altitude, with the highest cirques also tending to be the largest and deepest.

Cirque size and shape with aspect
When the glacier-free cirques are divided into 8 aspect bins (N, NE, E, SE, S, SW, W, NW), one-way analysis of variance (ANOVA) reveals statistically significant variations in L (F-ratio = 8.6, F-crit = 2.0) and W (F-ratio = 3.7, F-crit = 2.0), and overall size (F-ratio = 4.2, F-crit = 2.0), but not H (F-ratio = 1.4,F-crit = 2.0).The NE-facing cirques show the largest mean L and W, and the South-facing group has the smallest (Table 4).The shallowest cirques (lowest H) are also found in the Southfacing group, with the deepest in the East-facing group (Table 4).Despite these patterns, Fourier (harmonic) regression suggests that aspect related differences in cirque dimensions (L, W, and H) are not statistically significant.

Cirque lithology
When the glacier-free cirques are divided into 5 separate geological classes (Fig. 6), one-way ANOVA indicates statistically significant relationships for L (F-ratio = 9.0, F-crit = 2.4), W (F-ratio = 9.8, F-crit = 2.4), and H (F-ratio = 5.0, F-crit = 2.4).However, there are no consistent relationships to suggest clear geological controls on cirque size.For example, mean L and W are greatest for mixed sedimentary rocks and lowest for intrusive igneous rocks and volcanic igneous rocks (Fig. 6a  and b), whereas mean H is greatest for igneous intrusive rocks and lowest for metamorphic rocks (Fig. 6c).

Controls on glacier formation
Since cirques are formed in locations which favour the accumulation and preservation of snow and ice, the two key topoclimatic factors that determine cirque aspect are total solar insolation and prevailing wind directions (Evans, 1977), although topography and geological structure play a role (Gordon, 2001;Bathrellos et al., 2014).In general, total solar insolation promotes cirque formation on South-and North-facing slopes, in the Southern and Northern hemispheres, respectively.In contrast, prevailing winds promote cirque development on lee slopes, where snow and ice redistributed from up-wind slopes accumulates, and is protected from deflation (Evans, 1977;Evans, 1990).These impacts on cirque aspect are often most apparent where glaciation is marginal (i.e., where the topography extends just above the regional equilibrium line altitude, ELA), meaning that glaciers are only able to form on climatically favourable slopes (Barr and Spagnolo, 2015).In such cases, the aspect bias in resulting cirque populations is often particularly strong, and reflected by a high VS (see Evans, 1977).By contrast, in the TAM, it is notable that VS is very low (Fig. 7).This might reflect the fact that Cenozoic glaciation in Antarctica has been extensive, often covering most of the continent (Anderson et al., 2002), meaning that cirqueforming glaciers were able to develop on slopes with a range of aspects.It is also apparent that none of the cirque populations in the TAM show an overall aspect bias (as reflected by the VM) towards the south (Fig. 3a & b).The lack of a southerly aspect bias likely indicates that total solar insolation was not a key control on where glaciers were able to initiate in the TAM.This might reflect the region's high latitude, where Sun angles during the ablation season (i.e., Austral summer) are such that aspect-related contrast in the receipt of solar insolation on slopes are minimal when compared to lower latitudes.Similarly weak aspect bias is observed for populations of high latitude modern glaciers, e.g., glacier populations in Novaya Zemlya (~74 • N), SE Ellesmere Island (~77 • N), Svalbard (~78 • N), Axel Heiberg Island (~79 • N), Severnaya Zemlya (78-81 • N), and Franz Jozef Land (80-83 • N) have relatively low VS of 23 %, 23 %, 21 %, 7 %, 33 %, and 31 % respectively (Evans, 2006a).In addition, Evans and Cox (2005) demonstrate that for modern glaciers, little N-S altitude asymmetry is expected between 70 • (N or S) and the Pole, indicating the negligible role that aspect-related contrasts in the receipt of solar insolation play in regulating glacier locations at such latitudes.
Though there is no southerly cirque aspect bias in the TAM, the population as a whole and each of the sub-populations show some aspect bias towards the North, NW, or NE (Fig. 3a & b), suggesting that prevailing wind directions may have played the dominant role in regulating on which slopes glaciers and therefore cirques were most readily able to form.At present, the prevailing winds in the TAM are katabatic (Parish and Bromwich, 1987), and flow radially towards the coast (with winds directions partly governed by the topography of the underlying ice) (Fig. 3c) from the centre of the East Antarctic Ice Sheet (EAIS) where they are generated by radiative cooling of air masses (Parish and

Table 5
Cirque aspect vector strength (VS) data for different populations globally.These data are presented in Fig. 7. Data from the present study (i.e., 1,292 glacier-free cirques in the TAM) are displayed in red.Citations for each study:  48) Evans (1999); (49) Clough, 1974, Clough, 1977;(50) Evans (1994); (51) Derbyshire and Evans (1976); ( 52) Gordon (1977); (53) Unwin (1973); (54) Sugden (1969); (55) Evans (2006a); (56) Oberbeck (1964); (57) Evans (1974); (58) Evans and Cox (1995); (59) Evans (1974); (60) Ergenzinger (1967); (61) Kornilov (1964) , 1987, 2007;Parish and Cassano, 2003;van Lipzig et al., 2004).The aspect data for many of the cirques in each of the sub-regions of the TAM suggests that similar wind patterns prevailed when they were being formedi.e., in each sub-region the cirque VM is roughly aligned (lies 180 • ) to the prevailing wind direction (Table 1, and compare Figs. 3 a & b with Fig. 3c).This suggests that the EAIS was present (in some form) at the time of formation of many, though not necessarily all, cirques.The EAIS formed ~34 Ma (DeConto and Pollard, 2003) and was likely centred over the Gamburtsev Mountains (Bo et al., 2009), >1000 km from the TAM (Fig. 2a).However, some of the highest, and therefore oldest (i.e.those that were the first to be occupied by glaciers), cirques in the TAM may have formed 60 Ma ago (Barr et al., 2022).This implies that cirque formation occurred in different regions at different times, in some cases perhaps separated by many millions of years.This is potentially demonstrated when cirque aspect is considered by altitude, i.e., by separating the entire cirque population into ten discrete altitudinal groups (each containing 1406 cirques) data show that as population altitude increases, VS generally decreases, and there is typically more deviation from North-facing aspects (Fig. 8).Specifically, VS is strongest (i.e., >20 %) for the lowest two altitudinal groups of cirques, and < 15 % for the reminder.VM is N in the lowest three altitudinal groups; NE is the next three; and more varied at higher elevations.This might indicate that the highest cirques were generated first, prior to the development of the EAIS (and associated katabatic winds), while the lower altitude cirques formed once the EAIS was in place (and therefore favour North-facing slopes in the lee of katabatic winds).

Landscape evolution
There is ongoing uncertainty about specifically how and when cirques grow (i.e. are enlarged) and therefore how glaciated mountain landscapes evolve (Crest et al., 2017;Barr et al., 2019;Ruszkiczay-Rüdiger et al., 2021;Salcher et al., 2021).For example, it is possible that cirques are enlarged continuously throughout occupation by glacial ice, or growth may be episodic, with active erosion largely occurring during the initiation and termination of glaciations, when glaciers are small and subglacial erosion is focused within the cirques and promoted by steep and mostly warm/wet-based ice (Barr andSpagnolo, 2013, 2015;Crest et al., 2017).It is also possible that cirques are mostly enlarged during the very early stages of glaciation, and that their dimensions are then altered little during subsequent occupation by glacial ice (Barr et al., 2019;Ruszkiczay-Rüdiger et al., 2021).Results from the present study show that glacier-free cirques in Antarctica are not notably larger than in other regions globally (Fig. 9).Given that some of these Antarctic cirques have been occupied by glacier-ice for millions of years (DeConto and Pollard, 2003;Barr et al., 2022), far longer than in many of these other regions globally (e.g.<140 thousand years for some cirques in British Columbia, see Barr et al., 2019), this seems to indicate that cirque size is not directly linked to the duration of glacier occupationi.e.Pre-1981 data from Evans (1977).p.c. is personal communication in Evans (1977).
I.D. Barr et al. cirques don't continuously grow when occupied by glacial ice.There are several reasons to suspect this.First, when glaciation is extensive (as in Antarctica) cirques may become occupied by cold-based, minimally erosive ice (Näslund, 1997;Hassinen, 1998).Second, as cirques overdeepen, sediment may become trapped subglacially, protecting the underlying bed from erosion (Hooke, 1991;Gądek et al., 2015).Third, cirques may grow until they reach a 'least resistance' shape after which bedrock erosion is minimal (Barr et al., 2019).However, our data do show some spatial variability in cirque size across the TAM, a general increase in cirque size with altitude (Table 3), and a tendency for NEfacing cirques to be larger than those facing South (Table 4).A possible explanation for these size differences might be related to ice occupation time, including by the ice sheet, and/or reflect differences in the dynamics of former glaciers (e.g., Delmas et al., 2015).This is particularly true of size variability with aspect, which might relate to differences in the dynamics of former glaciers.For example, on NEfacing slopes the small mountain glaciers that are often presumed to drive cirque development may have been more dynamic (with higher mass-turnover) and therefore more erosive than those on South-facing slopes.The rationale for expecting more dynamic glaciers on these slopes is that the total annual receipt of solar insolation is greatest on North-facing slopes (in the TAM), and glaciers may therefore have experienced more melting than those facing South.In addition, as outlined in Section 4.1.,NE-facing glaciers may have experienced more accumulation than South-facing examples since katabatic winds (Fig. 3c) are likely to have re-distributed snow and ice from South-facing slopes and deposited this material on those facing North.This combination of comparatively high melt rates and high accumulation rates on North-facing slopes in the TAM may have resulted in particularly dynamic and erosive glaciers (with steep mass balance gradients) which were able to develop larger cirques.

Conclusions
In this study, we conduct the first systematic morphometric analysis of cirques in the Transantarctic Mountains to yield information about past glacier formation and landscape evolution.The main study findings are:

Table 6
Cirque size data for different populations globally.These data are presented in Fig. 9.In each case, values refer to minimum, mean, and maximum for cirque length (L), width (W) and depth (H), all recorded in metres.Data from the present study (i.e., 1,292 glacierfree cirques in the TAM) are displayed in red.Citations for each study: (1) Bathrellos et al. (2014) 28) Marinescu (2007); (29) Aniya and Welch (1981).1. Across the TAM, when glacier-free and glacier-occupied cirque populations are compared, VM and VS are similar, however neither shows a particularly strong aspect bias (as reflected by VS).This likely reflects Antarctica's climate which has been conducive to extensive and widespread glaciation for many millions of years (Miller et al., 2005;Barr et al., 2022), allowing glaciers to form and thereby generate cirques on slopes with a variety of aspects.2. The lack of a southerly cirque aspect bias indicates that total solar insolation was not a key control on where former glaciers in the TAM were able to initiate, perhaps because of the region's high latitude, where the aspect-related contrasts in the receipt of solar insolation during the ablation season are minimal when compared to lower latitudes.3.Where cirques in the TAM show some aspect bias, this is typically towards the North, NW and/or NE.This likely reflects the dominant role of prevailing wind directions in controlling the slopes on which past glacier development was favoured.If so, this suggests that during the formation of some cirques, prevailing wind directions in the TAM were similar to present, i.e., dominated by katabatic winds which flow from the interior of the EAIS.This implies that the EAIS was present (in some form) at the time, meaning that many of these cirques are no older than 34 Ma.By contrast, the very highest (and likely oldest) cirques in the TAM have more varied aspects, indicating that they may have formed before katabatic winds came to dominate (therefore sometime prior to 34 Ma, and perhaps as far back as 60 Ma, Barr et al., 2022).4. In general, we find that cirques in the TAM have similar dimensions to those in other regions globally, despite having been occupied by glacial ice for far longer.Thus, our findings support a growing body   5).Data from the present study (i.e., all 14.060 cirques in the TAM) are displayed in red (circle with hollow centre).
of evidence which suggests that cirque size and glacier occupation times are not directly coupled. 5.There is some evidence of variability in cirque size which might reflect differences in ice occupation time and/or relate to the differences in the dynamics of former glaciers, i.e., glaciers on slopes with North, NW and NE aspects (with high mass turnover) were likely more erosive and therefore able to generate larger cirques than minimally erosive glaciers (with comparatively low mass turnover) on other aspects.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability
Data will be made available on request.

Fig. 3 .
Fig. 3. Aspect frequency for (a) all cirques (n = 14,060) and (b) all glacier-free cirques (n = 1292) in the Transantarctic Mountains and for separate sub-regions.PR = The Pensacola Range; QMM = Queen Maud Mountains; QER = Queen Elizabeth Range; SVL = South Victoria Land; NVL = North Victoria Land.The vector strength (VS) and vector mean (VM) are also shown (the arrow in each rose diagram corresponds to the vector mean).Rose Diagrams are oriented according to each subregion's approximate orientation.(c) Model representation of wind vectors across the TAM (from Parish and Cassano, 2003).Length of wind vectors corresponds to wind magnitude.

Fig. 4 .
Fig. 4. (a) Frequency distributions for the Length (L), Width (W) and depth (H) of glacier-free cirques in the Transantarctic Mountains.(b) Allometric (double logarithmic) plot of L, W, and H against size ( 3 √LWH) for each of these cirques.

Fig. 6 .
Fig. 6.Variations in the Length (L), Width (W) and depth (H) of glacier-free cirques in the Transantarctic Mountains according to differences in dominant lithology (i.e., the geological unit which accounts for most of a cirque's surface area).Each boxplot shows the median (horizontal line), mean (cross), 1st and 3rd quartiles, and outliers (i.e.values >1.5 box lengths beyond the interquartile range).

Fig. 7 .
Fig. 7. Cirque aspect vector strength (VS) data for different populations globally.Numbers refer to different studies (details are provided in Table5).Data from the present study (i.e., all 14.060 cirques in the TAM) are displayed in red (circle with hollow centre).

Fig. 8 .
Fig.8.Aspect data for all 14,060 cirques in the TAM grouped according to their altitude (Z_max).Groups range from (a) the lowest 1406 cirques to (j) the highest.In each image, the arrow corresponds to the Vector Mean (VM).

Fig. 9 .
Fig. 9. Cirque size data for different populations globally.Numbers refer to different studies (details are provided in Table 6).Data from the present study (i.e., 1292 glacier-free cirques in the TAM) are displayed in red (circle with hollow centre).(a) Length.(b) Width.(c) depth.Note that the y-axis is logarithmic.Mean, maximum and minimum values are shown here.

Table 2
Size and shape metrics for glacier-free cirques in the Transantarctic Mountains.

Table 3
Pearson product moment correlation among attributes of glacier-free cirques in the Transantarctic Mountains.

Table 4
Variation in size metrics with aspect for glacier-free cirques in the Transantarctic Mountains.