Past glaciation of temperate‐continental mountains: A model for a debris‐charged plateau icefield/cirque glacier landsystem in the Southern Carpathians, Romania

Reconstructing the extent, style, timing and drivers of past mountain glaciation is crucial in both understanding past atmospheric circulation and predicting future climate change. Unlike in high‐elevation mountains situated in maritime and continental climates, less is known of past glaciation in mid‐altitude mountains, located in transitional climates, such as the Southern Carpathians of Romania. Despite these mountains harbouring a rich glacial geomorphology, this has never been systematically mapped according to well‐established morphological criteria, nor confidently related to former styles of glaciation. Therefore, filling this gap is important for not only accurately identifying glacial extents but also for establishing past glaciation styles and relating them to past ice dynamics and climate. Here, we devise the first glacial landsystem model for the Southern Carpathians, to enhance our understanding of the glaciation style and dynamics of past mountain glaciers in temperate‐continental climates. We present a geomorphological map, from which we infer the spatial and temporal evolution of glacial, periglacial and paraglacial landforms. We then assess the links between the mountain geomorphology and glaciation style and dynamics to construct a glacial landsystem model for debris‐charged plateau icefields and cirque glaciers in temperate‐continental mountain settings.


| INTRODUCTION
Reconstructing the extent, style, timing and drivers of past mountain glaciation is crucial for both understanding past atmospheric circulation and predicting future climate change (Li et al., 2016).Unlike high-elevation mountains situated in maritime (e.g., Federici et al., 2017) and dry-continental climates (e.g., Xu et al., 2010), less is known of the style, extent and dynamics of past glaciation in midaltitude mountains located in transitional climates (e.g., temperatecontinental/subarctic type), such as the Carpathian Mountains, Romania (Figure 1a,b).This climate zone is characterised by long, cold winters and short, warm summers because of its continental location and distance from the influence of oceanic air masses (Beck et al., 2018).
In this paper, we compile the first glacial landsystem model for the Romanian Carpathians, in order to understand glaciation style and past glacier dynamics.Based on the Godeanu Mountains in the southwestern part of the Southern Carpathians, this is achieved through the following objectives: (1) identifying and interpreting the origin and spatial distribution of mountain landforms, based on a remote and ground-truthed geomorphological map; (2) determining the glacial debris transport pathways and genesis of selected glacial depositional landforms, using sedimentological analyses; (3) hypothesising on the relationship between the mountain topography and glaciation style and dynamics, based on the spatial relationships between the landforms (glacial landsystem) in the study area; and (4) assessing the value of the landsystem model as an analogue for similar glacierised uplands.

| STUDY AREA
The Romanian Carpathians occupy the centre of Romania and comprise three chains, including the Eastern Carpathians, the Southern Carpathians and the Apuseni Mountains (Figure 1a,b).They are composed of sedimentary, metamorphic and volcanic rocks (Urdea et al., 2011) and resulted from Mesozoic-Cenozoic continental collision and Pliocene-Quaternary erosion (15-2.6Ma), including glaciation during cold climate cycles (Popescu, Urdea, & Vespremeanu-Stroe, 2017;Sanders et al., 2002;Sanders, Andriessen, & Cloetingh, 1999).Their position in Central-Eastern Europe marks a transition between the temperate-maritime climate of Western Europe, the Mediterranean climate of Southern Europe and their continental-arid counterpart of Central Asia (Mîndrescu et al., 2010;Reuther et al., 2007).The mountains are not currently glacierised but have numerous peaks above 2000 m a.The Godeanu Mountains are a $100 km 2 , northeast-southwestorientated mountain ridge located in the southwestern part of the Southern Carpathians (45 15 0 N 22 42 0 E; Figure 1c).Part of the Retezat-Godeanu group, the mountains form flat/rounded plateau surfaces overlooking $70 cirques and short (3-4 km) U-shaped valleys with a range of orientations (Mîndrescu, 2016;Niculescu, 1965).The mountains are composed of gneiss, paragneiss, crystalline schist, amphibolite and pegmatite lithologies, with sparse limestone.With a maximum elevation of 2291 m a.s.l.(Gugu Peak; Figure 1c), these mountains are influenced by a sub-Mediterranean climate, with modern prevailing winds from the north, west and southwest (Dragomir et al., 2016;Niculescu, 1965).The climate is warmer (0-2 C average/year) and wetter (1000-1400 mm/year) than other areas of the Southern Carpathians (Dragomir et al., 2016;Niculescu, 1965).Therefore, although the geology and topography of the Godeanu Mountains are similar to other mountains in the eastern Southern Carpathians (e.g., F ag aras ¸and Iezer), the present climate characteristics are unique to it.
This assumption was challenged by Ignéczi and Nagy (2016), who tested the hypothesis of a plateau icefield style of glaciation by performing numerical modelling of palaeoglacier extents.Using a steadystate plastic numerical model, they reconstructed 40 km 2 of ice over the range, constrained by the geomorphological map of Niculescu (1965).The reconstruction included plateau icefields, with an average palaeo-ELA of 1855 m a.s.l. on the northern slopes and valley glaciers predominant in the south-facing cirques in the west.These yielded an average palaeo-ELA of 1700 m a.s.l.Ignéczi and Nagy (2016) suggested that the discrepancy between the ELAs is related to missing geomorphological evidence of palaeoglacier extents in the northern valleys, possibly caused by postglacial erosion.This demonstrates that re-mapping of the area is necessary before further modelling and dating of past ice extents is attempted to derive paleoglaciological and palaeoclimate reconstructions.

| Geomorphological mapping
Guided by protocols for remote and field-based mapping of former glaciers (Chandler et al., 2018), we produced a geomorphological map of the Godeanu Mountains, at a scale of 1:120,000 in ArcMap 10.7.1.This was undertaken remotely for the entire range and field-checked in the area of the Sc aris ¸oara plateau and four surrounding valleys.
Because of low-resolution digital elevation models (max = 30 m, ALOS DEM), mapping was undertaken on high-resolution (1 m) orthorectified aerial photographs from ANCPI, ArcGIS World Imagery, as well as on imagery viewed on Google Earth Pro.
The mapping focused on landforms and sediments of glacial depositional origin (moraines and till veneers/drift limits), glacial/ glacifluvial erosional origin (ice-moulded bedrock and ice-marginal meltwater channels), periglacial origin (pronival and protalus ramparts, relict rock glaciers, rock-glacierised moraines) and paraglacial landforms (RSFs).Plateau areas were also demarcated to assess the possibility of former plateau icefields.Landforms were initially classified using descriptive terms (e.g., ridges and channels), avoiding premature genetic classifications (Darvill et al., 2017;Evans, Dinnage, & Roberts, 2018).Genetic classifications were subsequently made based on morphological diagnostic criteria, as outlined in Table 1.Field investigations in the area of the Sc aris ¸oara Plateau and four surrounding valleys (Bulz, Vl asia, Sc aris ¸oara and Scurtele; Figure 1c) were employed as control sites for extrapolation to the remote mapping for the entire range.

| Sedimentological analyses
Landform origins and potential glacial debris transport pathways were assessed in the field via sedimentological analyses on natural exposures in depositional features in the ground-truthed area.Specifically, this entailed sediment logging and descriptions using a lithofacies approach, augmented by clast shape analyses (cf.Benn & Ballantyne, 1994;Lukas et al., 2013;Evans & Benn, 2021).Clast shape indices include clast roundness, quantified using the RA (percentage of A and VA clasts) and the average roundness (AvgR; cf.Evans, 2010), and clast form, quantified using the C 40 index as an indicator of 'blockiness', and hence subglacial modification (Benn & Lukas, 2021).
The degree of clast modification is then related to glacial debris transport pathways by employing the Type II RA-C 40 covariance plot (Lukas et al., 2013) for anisotropic (foliated) lithologies, similar to those of the crystalline schists of the Sc aris ¸oara plateau.

| Glacial landsystems reconstruction
Glacial landsystems are assemblages of sediments and landforms that are representative of specific glacial processes, styles and dynamics in the context of the topographic and climatic characteristics of the investigated terrain (Evans, 2003).These characteristics, and thus, the landsystems, may change spatio-temporally in response to climate drivers (e.g., Benn et al., 2003;Bickerdike, Evans, et al., 2018;Bickerdike, Ó Cofaigh, et al., 2018).The importance of glacial landsystems lies in their integration with geochronology and ice T A B L E 1 Inventory of glacial depositional, glacial erosional, periglacial depositional and paraglacial landforms in the Godeanu Mountains, with details on their associated morphology, sedimentology, significance and key references.modelling to produce a holistic understanding of glaciation style, dynamics and palaeoclimate through time.However, their compilation in ancient settings requires the recognition of geomorphological process-form regimes, informed by suitable modern analogues.Hence, we consulted modern analogue landsystems from similar climatic and topographic settings, such as those for plateaux (Rea et al., 1998;Rea et al., 1999;Rea & Evans, 2003, 2007) and glaciated valley (Benn et al., 2003;Benn & Owen, 2002) settings where paraglacial and periglacial process-form regimes interact with those of glacial systems (Knight et al., 2019;Leigh et al., 2021).

| Spatial distribution
Latero-frontal moraines occur in all glacial valleys and on the Borascu plateau (Figures 1c and 2). in palaeoglacier extents through time (e.g., Benn et al., 2003;Evans et al., 2010;Owen et al., 2001; Table 1).The predominance of subdued drift limits in north and northeast-facing valleys reflects lower glacier debris turnover, as well as relatively greater postglacial erosion and sediment reworking, promoted by the steeper, stepped bedrock valley profiles (Ignéczi & Nagy, 2016;Niculescu, 1965).The increased number of moraine ridges in proximity to cirques in valley complexes shows the separation of different ice sources and concomitant relative stabilization of the glaciers once they retreated to higher elevations (Glasser et al., 2006).Notwithstanding the differences in north-versus south-facing valley geometries, the difference in number, spacing and size of the latero-frontal and recessional moraines is possibly due to north-facing palaeoglaciers being larger (Ignéczi & Nagy, 2016) and less responsive to climate fluctuations.are the bedrock steps, which can act as pinning points, solar insolation, and lee-side aspects relative to southerly winds (e.g., Barr & Lovell, 2014;Benn & Ballantyne, 2005;Coleman et al., 2009;Mitchell, 1996;Oerlemans et al., 1998).In the C arbunele (Figure 3c) and Soarbele (Figure 3a) valleys, there is a preferential slope aspect for the occurrence of large lateral moraines.

Aspects of valley topography that are likely influential in this respect
For example, in the C arbunele valley, a prominent lateral moraine occurs on the northwest-facing slope, where RSFs are also common, whereas there are no lateral moraines on the debris-poor southeastfacing equivalent.The variable occurrences and sizes of lateral moraines reflect the availability of paraglacial/supraglacial debris, resulting in lateral moraine asymmetry (Benn, 1989;Evans, 1999;Matthews & Petch, 1982; Table 1).This refers to the preferentially larger sizes of lateral moraines situated on steep, freeface-dominated and failure-prone valley sides (cf.Jarman, 2009), particularly in association with RSFs and talus cones, as observed in the C arbunele and Soarbele valleys.
Despite their generally low preservation potential in glaciated terrains (Benn & Evans, 2010) and Table 1).These features occupy a specific elevation niche at 1630-1950 m a.s.l., where they grade down-valley into recessional latero-frontal moraines (Figure 3d).Medial moraine production appears to have taken place via two distinct modes, based upon the models of Eyles and Rogerson (1978) and Rogerson et al. (1986).
Firstly, 'ice stream interaction' (ISI)-type medial moraines are formed through the coalescence of lateral moraines of two tributary glaciers or ice flow units (Eyles & Rogerson, 1978).This model seems appropriate for the confluence between the two cirques comprising the Bulz valley, demarcated by three transverse ridges (Figures 2, 3d and 4b).Secondly, 'avalanche-type' medial moraines do not require the existence of coalescing ice flow units, but rather a supraglacial rockfall debris source, such as from a plateau or high cirque backwall nunatak (Rogerson et al., 1986).This model is most applicable to the eastern Bulz, Mocirlu, C arbunele and Soarbele valleys (Figures 1c,2 and 3a,d).In both cases, with ice retreat and thinning, the debris position of each flow unit narrows and shifts towards the valley sides, forming inset ridges that represent parallel medial moraines in the main valley.The highly lobate bouldery ridge in the Bulz valley (Figure 3d) is more complicated in that it resembles the teardrop-shaped looped medial moraines of surging mountain glaciers (e.g., Ellis & Calkin, 1979;Gusmeroli et al., 2014;Meier & Post, 1969;Paul, 2015).Consequently, this reflects the differential speeds of individual flow units within the former valley glaciers of the study area rather than surging, as other features indicative of surging are lacking: for example, crevasse squeeze ridges, thrust moraines and zigzag eskers (Evans & Rea, 2003).We attribute the high preservation potential of medial moraines to large volumes of debris sourced supra-, sub-, and englacially (e.g., Eyles & Rogerson, 1978;Goodsell, Hambrey, & Glasser, 2005;Goodsell, Hambrey, Glasser, Nienow, et al., 2005;Small, Clark, & Cawse, 1979).

| Morphology and genesis
Ice-moulded bedrock is represented by abraded and plucked outcrops that form roches moutonnées and whalebacks/rock drumlins.
These features are indicators of subglacial erosion and former ice flow directions, and are generally reflective of a warm-based glacial thermal regime (Glasser, 2002;Glasser & Bennett, 2004;Rea, 2013; Table 1).

| Spatial distribution
In where their genesis is incompatible with a fluvial origin (Dyke, 1993).
The ice-marginal meltwater channels are useful in identifying former ice limits, and they offer insight into glacial thermal regime, meltwater volumes and subglacial water pressures (Table 1).Previous research has used meltwater channels as a diagnostic criterion for cold-based ice margins, whereby glacial meltwater is diverted along frozen margins, carving channels corresponding to their retreat positions (Dyke, 1993;Evans, Hughes, et al., 2017;Greenwood et al., 2007;Livingstone et al., 2010;Maag, 1969).However, Dyke (1993) and Syverson and Mickelson (2009)

| Morphology and genesis
Pronival and protalus ramparts are single, largely continuous ridges with asymmetric cross profiles, formed by the accumulation of rockfall debris at the foot of perennial snowpatches (Shakesby, 1997;Shakesby et al., 1999;Matthews et al., 2011; Table 1).Although some debate has surrounded a snowpatch versus glacial (morainic) genesis for many mountainside ridges (e.g., Shakesby, 1997;Shakesby & Matthews, 1993), there are specific diagnostic criteria for palaeoprotalus rampart interpretations (Table 1).Typically, the distance between a protalus rampart and the talus foot that it formerly covered can only be 30-70 m, otherwise the snowpatch would have turned into glacier ice (Ballantyne & Benn, 1994).Pronival ramparts, which represent ridges accumulating on mid-slopes, would thus be related to even smaller snow accumulations.The sedimentology of a pronival rampart should also typically be predominantly composed of openwork clasts in places, including a fine matrix (Table 1).

| Spatial distribution
We mapped 48 pronival and protalus ramparts in the Godeanu Moun-  and 2).The spatial distribution of pronival and protalus ramparts offers potential insight into palaeoclimatic trends (Ballantyne & Kirkbride, 1986).Firstly, favourable conditions, such as shading and windward-facing basins, are reflected in preferential locations on northfacing slopes (Ballantyne & Kirkbride, 1986;Colucci et al., 2016).The lower elevations at which the ramparts are located in south-facing valleys might suggest these formed at a different time and climatic conditions than their north-facing equivalents.Secondly, the close association of pronival and protalus ramparts with recessional moraines reflects a spatio-temporal transition from small glaciers to snowpatches (Ballantyne & Benn, 1994;Niculescu, 1965).Indeed, Hedding et al.
(2007) noted pronival and protalus ramparts forming under decreasing precipitation at the inception/demise of regional glacier activity.
Because of their similarity to, and juxtaposition with glacial landforms, RSFs can be misidentified as former glacier ice limits; they can also obscure them in cases of postglacial failures (Table 1).Additionally, RSFs that pre-date ice advance can be overrun and streamlined by glacier flow (Ballantyne, 2018a; Figure 4h).

| Spatial distribution
We identified 97 RSFs in the Godeanu Mountains.1c and 2).Our detailed mapping approach differs from that of Gunnell et al. (2022), which recognised only 21 RSFs in the Godeanu Mountains amongst 215 in the Southern Carpathians.
The work of Gunnell et al. (2022) and that of Vasile et al. (2022) constitute the first inventory and application of 10 Be cosmogenic nuclide dating of RSFs in the Southern Carpathians, respectively.These pioneering studies suggest that Neogene tectonic uplift and climate warming in the Southern Carpathians drove RSF development and that periglacial, fluvial and parafluvial triggers were more important than glacial and paraglacial processes.The dominance of Early Holocene (pre-9 ka) dated boulders in RSF deposits in particular supported these proposals (Vasile et al., 2022).Indeed, the spatio-temporal relationships between glacial, periglacial and paraglacial landforms may indicate a deglacial/postglacial (Early Holocene) development period for RSFs.
Furthermore, the higher number of RSFs on the north-facing slopes and lower association with periglacial landforms on this aspect than on the south-facing ones re-inforces the hypothesis of a later deglaciation on the north-facing valleys, which may still be undergoing failure, whereas their southern counterparts have already completed the paraglacial cycle (Ballantyne, 2002).Over the longer timescales of multiple Quaternary glaciations, however, RSF operation likely initiates and/or contributes to cirque development through headwall erosion into plateau surfaces (cf.Evans, 2021;Turnbull & Davies, 2006), sourcing debris for the development of rock glaciers and moraines.

| GLACIAL LANDSYSTEMS
The spatio-temporal relationships between the mapped landforms and the sedimentological evidence provide a clear signature of past glaciation style, debris turnover and moraine construction in the Godeanu Mountains, thereby facilitating the compilation of a glacial landsystem model for the region of the Southern Carpathians.This landsystem model displays many of the characteristics of the conceptual plateau icefield (Rea & Evans, 2003), glaciated valley (Benn et al., 2003) and the cirque/niche (Bickerdike, Ó Cofaigh, et al., 2018) glacial landsystems, as well as elements of debris-charged geomorphology observed in a modern analogue in Norway (Leigh et al., 2021).Therefore, it is best classified as a debris-charged plateau icefield/cirque glacier landsystem.The most extensive phases of glaciation were characterised by simultaneous occupation of valleys by ice descending both from surrounding plateaux and from cirques, where narrow mountain ridges dominate.Conversely, phases of less extensive glacierisation were characterised by localised cirque and niche glaciers, detached snouts/reconstituted glaciers and perennial snowpatches (e.g., Benn & Lehmkuhl, 2000;Evans, Ewertowski, & Orton, 2016;Evans et al., 2006;Gellatly et al., 1986;Rea & Evans, 2003), whereas plateau ice may have been restricted to upland surfaces only or completely disappeared.The latter phases may have occurred initially on the more exposed south-facing slopes, with the more shaded north-facing counterparts possibly deglaciating later.

| Spatio-temporal change in the Godeanu Mountains debris-charged plateau icefield/cirque landsystem
Using the most extensive moraine limits mapped, the first phase (Figure 7a) depicts the plateau icefield at its maximum extent, with ice descending into two neighbouring valleys.The valleys are separated by an ice-free nunatak, their bedrock slopes feeding extraglacial debris via talus cones to lateral and medial moraines of the outlet glaciers.
Plateau icefields were developed on the Moraru, Branu, Sc aris ¸oara, Bor ascu, Paltina and St anuleți summits, where the lack of glacial landforms (except rare moraine ridges documenting ice recession from surrounding valley heads) indicates cold-based ice dispersal centres.
Cirque glaciers accumulating in the more alpine terrain flow into the main valleys, coalescing with the plateau outlet glaciers.This results in the deposition of further medial moraines; the medial moraines on the cirque glaciers, created predominantly as avalanche types fed by couloirs at headwalls (Eyles & Rogerson, 1978;Rogerson et al., 1986), are also added to the valley glacier surface downflow of the coalescence zone.The transition from predominantly cold-based ice on the plateaux to warmer ice in the outlet glaciers was likely driven initially by the strain heating at the base of icefalls on the plateau edges (Rea et al., 1999;Rea & Evans, 2003;Whalley et al., 1996).Together with the subglacial modification of clasts down-valley, this explains the contrast between plateau summits devoid of glacial landforms and the widespread occurrence of glacial erosional landforms at lower elevations.Over time, the enlargement of cirque headwalls, as well as trough head development at plateau edges, promotes not only further glacial erosion but also slope destabilisation, and the initial triggering of RSFs, sourcing large volumes of extraglacial debris to the glacial system.
The second phase (Figure 7b) depicts initial ice retreat from the maximum extent.During this phase, the lowermost set of laterofrontal moraines were abandoned, and medial moraines were lowered onto the subglacial footprint.Also prominent are the looped latero- (partially) coupled ice margin, with high ice and debris supplies, and the efficient meltwater transfer of glacial debris from the snout to proglacial outwash systems (Benn et al., 2003).Although icemarginal meltwater channels typify cold-based plateau icefields, their lower valley position, the association with depressions and occasional intercalation with moraine ridges (Figure 3f) suggests a warm-based or polythermal basal thermal regime (Dyke, 1993).This is supported by 10 Be cosmogenic dating-inferred erosion rates in the Retezat Mountains, which indicate warm-based conditions at the ice margins and colder ice at the valley head during initial deglaciation (Ruszkiczay-Rüdiger et al., 2021).Additionally, the reconstructed coupled margin at this stage is a defining characteristic for glaciated valley landsystem settings with high debris turnover (e.g., New Zealand), whereby debris is efficiently evacuated from the margin by high subglacial meltwater pressures and volumes (Benn et al., 2003;Dyke, 1993).Phase 2 moraines are located at $1500 ≥ 1750 m a.s.l. and may correspond to the M2b moraine system (1600-1750 m a.s.l.) in the Retezat Mountains, which yielded early Lateglacial ages of 18.6 +0.9/À0.8-16.9 ±0.9 ka (Ruszkiczay-Rüdiger et al., 2016, 2021).
The third phase (Figure 7c) was characterised by continued reces-  Rogerson, 1978;Jennings et al., 2014;Roberson, 2008).As medial moraine preservation potential is usually very low after complete deicing (Benn & Evans, 2010), the prominence of such features in the Godeanu Mountains (especially in increasing numbers up-valley) indicates significant debris provision via extraglacial pathways.This was presumably derived from RSF sources, which began dominating large, steep areas of uncovered trough heads surrounding the plateaux upon ice recession (Ballantyne & Dawson, 2019;Evans, Ewertowski, & Orton, 2016, 2017;Rea & Evans, 2003).Examples of lateral moraine asymmetry (e.g., Figure 3c) reflect the uneven distribution of freefaces delivering such material during deglaciation.Over time, the uncovering of increasingly larger areas of such steep bedrock outcrops during glacier recession and downwasting resulted in enhanced debris production that locally overwhelmed glacier surfaces.
This resulted in the formation of rock glaciers and rock-glacierised moraines and a transition towards an uncoupled snout (Benn et al., 2003).Unlike coupled ice margins, the meltwater produced in uncoupled systems is inefficient at removing sediment from the foreland, resulting in gradual ice burial by debris (Benn et al., 2003).This is specific to glaciated mountain valleys under drier climate conditions, with lower ice supplies (Benn et al., 2003).In summary, this phase underscores the evolving importance of topography and debris provision in this landsystem.Located between $1750 and 1900 m a.s.l., Phase 3 moraines may have been deposited between the construction of the M2b and M3 moraines (1900-2000 m a.s.l.) in the Retezat Mountains, and hence likely date to 15.8 +0.9/À0.6 -15.2 +0.7/À0.8ka (Ruszkiczay-Rüdiger et al., 2016, 2021).
Also evident during Phase 4 is the role of topography in creating a geomorphic contrast between north and south-facing valleys.Some valleys contain far fewer recessional moraines and are devoid of medial moraines (e.g., the north-facing Cârnea, Mâțu, Scurtele and Sc aris ¸oara valleys and the south-facing Sc arița and Micus ¸a valleys; Figures 1c and 2).This contrast is likely a product of the variable occurrence of bedrock steps, which may act as topographic pinning points.Bedrock steps provide marginal stability, limiting variations in glacier accumulation areas when ELAs are coincident with them (Barr & Lovell, 2014;Dugmore, 1989;Oerlemans, 1989).Such topographic pinning points favoured the formation of rare, large moraines in north-facing valleys, contrasting with the small, but more numerous moraines that accumulated in shallower and smoother south-facing valleys (Barr & Lovell, 2014; Figure 2).Situated at $1900-1950 m a.s.
l., Phase 4 moraines may be equivalent to the M3 moraines in the Retezat Mountains (Ruszkiczay-Rüdiger et al., 2016, 2021).A further phase of moraine deposition (M4 at 2100-2150 m a.s.l.) was dated at 14.4 ±0.5 -13.5 +0.5/À0.4ka in the Retezat Mountains, which have a higher elevation than the Godeanu Mountains (max = 2509 m a.s.l.; Ruszkiczay -Rüdiger et al., 2016-Rüdiger et al., , 2021)).This suggests that topography may have an inhibiting role in sustaining glaciers, lower elevations promote higher temperatures and lower precipitation, which are not compatible with glacier growth/persistence (DeBeer & Sharp, 2009).This is especially relevant for the cirques in the Godeanu Mountains, generally peaking at lower elevations ($1750-2000 m a.s.l.) than the plateau-fed valleys in the Godeanu Mountains ($2000-2250 m a.s.l.) (Figures 1c and 2) and the cirque/valley glaciers described for the Retezat Mountains (Reuther et al., 2007;Ruszkiczay-Rüdiger et al., 2016, 2021).Secondly, plateau-fed glaciers alter ice dynamics by either (a) depressing ELAs, leading to a lower elevation moraines than in cirque/valley settings (Aa, 1996;Dahl & Nesje, 1992;Rea et al., 1999), or (b) rapidly retreating to their plateau surfaces during a warming climate, leaving disconnected ice masses in the ablation area, which may or may not imprint geomorphological evidence (Rea et al., 1999;Whalley et al., 1996).Therefore, undated M4 moraines may be found at lower elevations in the plateau-fed valleys in the Godeanu mountains than in the Retezat, subject to confirmation by an absolute dating programme therein.

| CONCLUSIONS
Drawing on geomorphological mapping and sedimentological investi- s.l.(max: 2544 m a.s.l.-Moldoveanu Peak).In the present climate, the F I G U R E 1 Location maps of the study area.(a) The location of Romania within Europe; (b) the location of the Godeanu Mountains within the Southern Carpathians; (c) large scale map of the Godeanu Mountains, showing the highest elevation point (Gugu Peak, 2291 m a.s.l.), plateau surfaces (highlighted in yellow), the names of the main cirques and valleys in the range, as well as the neighbouring mountain ranges: Țarcu Mountains in the West and Retezat Mountains in the East.Cirque and valley toponymy derived or adapted from Niculescu (1965) and Mîndrescu (2016).Digital elevation model for Europe (TOPO-WMS and TOPO-OSM-WMS), sourced from Terrestris, and that of Romania, Southern Carpathians, and the Godeanu Mountains sourced from EORC, JAXA.[Color figure can be viewed at wileyonlinelibrary.com] 0 C isotherm is situated at 1650 m a.s.l. in the Eastern Carpathians and 2050 m a.s.l. in the Southern Carpathians, and a mean precipitation above 2000 m a.s.l. of 1400 mm/year in the Eastern Carpathians and 1000 mm/year in the Southern Carpathians (Popescu, Urdea, & Vespremeanu-Stroe, 2017).
continuous linear valley-side depositional ridges and associated valley-floor arcuate ridges, forming inset sequences in an up-valley direction, are clearly latero-frontal moraines (Figures 3a and Sedimentological investigations were undertaken on exposures through moraines in the Bulz, Vl asia, Sc aris ¸oara and Scurtele valleys, surrounding the Sc aris ¸oara plateau (Figure 5a), including the lateral moraines BLZ1, BLZ2, BLZ4 and SCT1, and the frontal/recessional moraines BLZ3, BLZ5, VLS1, VLS2 and SCS1.The moraines are composed of coarse, bouldery, sandy-silty matrix or clast-supported diamictons (Figure 6a,b).The RA-C 40 covariance plot (Figure 6c) reveals the relatively high C 40 values, typical of anisotropic lithologies, indicating that roundness indices such as RA and AvgR are more appropriate in diagnosing clast modification within the former glaciers of the area.Down-valley trends can be assessed only in one location, the Bulz Valley, where the most intensive sampling was undertaken.
Figures 1c and 2), recessional moraines can be traced back onto each cirque floor.With respect to valley aspects, latero-frontal and recessional moraines are smaller, more closely spaced (average = 189.5 m) and greater in number (average = 9.05 moraines/valley) in south-facing valleys than those in north-facing ones, which are larger, more widely spaced (average = 303.75m) and lower in number (average = 7.16 moraines/valley).Additionally,

F
I G U R E 2 The mountain geomorphology of the Godeanu Mountains, Southern Carpathians, Romania, focusing on the main glacial, periglacial, paraglacial and contextual landforms.The numbered dashed line rectangles reveal the location of landform examples featured in Figure 3. Produced in ArcGIS, using two hillshade models (transparency = 95%, illumination angles = 180 and 315 ), derived from the ALOS DEM (EORC, JAXA, resolution = 30 m).High-resolution version (without the numbered rectangles) is available to consult in the Supporting Information section in the online version of the article.[Color figure can be viewed at wileyonlinelibrary.com] Lateral moraines are uncommon, with more extensive examples occurring in the Olanu West, Vl asia, C arbunele, M aneasa, Scurtele, Mâțu and Branu North valleys.More fragmented examples occur in the Scurtele, Mâțu and Sc arița-Bulz valleys (Figures 1c, 2 and 3c).
, medial moraines are present in the four south-facing valleys of Mocirlu, Bulz, C arbunele and Soarbele and in F I G U R E 3 Orthophotograph-map pairs of mountain landforms in the Godeanu Mountains.Black arrows indicate examples of the relevant landform observed.The maps use the location names in Figure 1c, and the legend in Figure 2, and include specific examples for some of the statistics discussed in the text: (a) latero-frontal and recessional moraines in the Soarbele valley; (b) till veneers in the Scurtele valley; (c) lateral moraines in the C arbunele valley; (d) medial moraines in the Bulz valley; (e) ice-moulded bedrock in the Bonțica valley; (f) ice-marginal meltwater channels in the Vl asia valley; (g) pronival ramparts in the Bor ascu Mic valley; (h) rock glacier in the Sc aris ¸oara valley; (i) rock-glacierised moraine in the Vl asia valley; (j) rock slope failure (rockslide) in the Bor ascu Mare valley.[Color figure can be viewed at wileyonlinelibrary.com] the northeast-facing Galbena North valley (Figures 1c, 2, 3d and 4b

F
I G U R E 4 Field photograph example of mountain landforms in the Godeanu Mountains.The white arrows indicate examples in each landscape context: (a) recessional moraine in the Bulz valley; (b) 'ice-stream interaction'-type medial moraines in the Bulz valley; (c) ice-moulded bedrock in the Sc aris ¸oara valley; (d) ice-marginal meltwater channel in the Vl asia valley; (e) pronival ramparts in the Scurtele valley; (f) relict rock glacier in the Sc aris ¸oara valley; (g) rock-glacierised moraine in the Vl asia valley; (h) ice-moulded rock slope failure blocks in the Scurtele valley.[Color figure can be viewed at wileyonlinelibrary.com] the Godeanu Mountains, ice-moulded bedrock is commonly found at altitudes of 1860-1920 m a.s.l.In areas of paragneiss and amphibolite in particular, roches moutonnées are widespread.Ice-moulded bedrock is common and prominent in the north-facing Bonțica, Gropița, Mâțu, Scurtele and Sc aris ¸oara, as well as the southern-facing Balmos ¸East, Sc arița, Micus ¸a, Stâna, C arbunele and Soarbele valleys(Figures 1c, 2, 3e and 4c).Niculescu (1965) suggested that icemoulded bedrock outcrops in areas where more durable lithologies, such as paragneiss, amphibolite and pegmatite, resist glacial erosion better than the weaker lithologies, such as crystalline schist.The prevalence of ice-moulded bedrock forms in north-facing valleys has been explained byNiculescu (1965) as a result of valley orientation, and hence ice flow being oriented in the opposite direction to the dip of the underlying strata.4.3 | Ice-marginal meltwater channels4.3.1 | Morphology and genesisRelict channels indicative of former glacial meltwater incision appear as elongate depressions or interlinked channel networks, interpreted as ice-marginal meltwater channels.They are interspersed with bedrock and sometimes sediment ridges and are present in locations

F
I G U R E 5 Clast sedimentological data on moraines in the valleys surrounding the Sc aris ¸oara plateau: (a) Bulz, Scurtele, Sc aris ¸oara and Vl asia.Each sample is named after the valley of provenance and order of collection: (b) BLZ3, (c) BLZ2, (d) BLZ 1, (e) BLZ5, (f) BLZ4, (g) SCT1, (h) SCS1, (i) VLS1 and (j) VLS2.Various sedimentological statistics (clast form ternary diagrams, clast roundness, C 40 index, relative angularity (RA), and average roundness (AvgR) are shown in individual boxes.The map legend can be consulted in the bottom right corner.[Color figure can be viewed at wileyonlinelibrary.com] in the elevation range 1530-1890 m a.s.l., the meltwater channels range from 60 to 330 m long, with one exception in the Soarbele valley, of 823 m.Ground-truthed examples in the Bulz and Vl asia valleys are 2-5 m deep.We cannot exclude a cold-based thermal regime in the Godeanu Mountains under specific conditions (elevation range and slope location).However, the association of meltwater channels with other landforms indicative of active deposition (recessional moraines), as well as warm-based thermal regimes (ice-moulded bedrock), indicates that an active temperate or polythermal setting was operating for at least some of the time.

F
I G U R E 6 Key information for interpreting the glacial transport pathways in the study area.(a) Stream-cut exposure in moraine in the Bulz valley, where clast data for sample BLZ2 were collected.Note the poorly sorted, matrix-supported, predominantly A and VA clasts.(b) Close-up photograph into a stream-cut exposure in a till veneer in the Sc aris ¸oara valley, where the SCS1 sample was taken from.Notice the poorly sorted, clast-supported, sandy matrix, with predominantly SA and SR clasts.Trowel for scale.(c) Covariance plot, showing how the clast data samples in the study area plot in terms of transport pathways.Each sample can be correlated with the valley of provenance by a different symbology.The colour gradient indicates the samples' relative position in their associated valleys, from the lighter upper valley samples to the darker lower valley ones.[Color figure can be viewed at wileyonlinelibrary.com] tains (Figures 2, 3g and 4e).Pronival and protalus ramparts are dominant in north-facing valleys (n = 31) in contrast to south-facing valleys (n = 17).They are situated at lower elevations in the south-facing valleys (1620-2100 m a.s.l.; average = 1790 m a.s.l), compared with north-facing ones (1710-2100 m a.s.l.; average = 1945 m a.s.l.).They also lie upslope of moraine ridges in many locations (e.g., up-valley distance between moraine and pronival rampart of 14.5 m in the Olanu West, and 15.03 m in the Cârnea valleys, respectively) (Figures 1c Onaca, Urdea, and Vespremeanu-Stroe (2017) andȘerban et al. (2019)    noted a discrepancy in rock glacier and block stream elevation between north-and south-facing slopes in the Southern Carpathians, with the northern slope features located at higher elevations than their southern counterparts.Popescu, Onaca, Urdea, and Vespremeanu-Stroe (2017) andȘerban et al. (2019)  attributed this to south-facing valleys receiving higher solar insolation, and thereby earlier glacier retreat and slope debutressing than in north-facing valleys.Because of their shading, the north-facing valleys might have preferentially supported higher elevation rock glaciers more recently.Although geochronology might constrain landform age and associated climate change, the incrementally polygenetic origin (glacial-to-periglacial) of many features makes their employment as palaeoclimatic indicators complicated (cf.Harrison et al., 2008;Jarman et al., 2013;Knight et al., 2019).

4. 6 |
Rock slope failures 4.6.1 | Morphology and genesis Depending on the failure style, travel distance and degree of disintegration, RSFs are represented by a range of discrete debris accumulations or partially intact bedrock masses, situated on, or at the foot of, steep rock walls.They can be classified as follows: (1) rock avalanchesdiscrete debris assemblages on valley floors, whose transport path can be traced directly uphill; (2) rockslides-arrested rock masses showing little deformation due to their short-travel distance; (3) rock slope deformations-subtle ruptures, bulges, tension fault ridges and troughs frontal and medial moraine signatures of constricted cirque glacier ice, flowing along the valley edge.Ice-marginal meltwater channels were excavated along lower valley slopes and the valley floor.Glacier snout thinning during retreat promoted further slope destabilisation and development of RSFs.The occurrence of continuous latero-frontal moraine ridges and ice-marginal meltwater channels indicates a sion of valley and cirque-based glaciers until the different glacier bodies became disconnected.Inset sequences of cirque moraines then started to record recession into localised alpine uplands.Overall recession was under active, warm-based conditions, as indicated by widespread ice-moulded bedrock and multiple inset latero-frontal moraines.Large longitudinal concentrations of supraglacial debris are indicated by the development of significant medial moraines, linked to tight arcuate or looped frontal moraines, each loop demarcating a separate flow unit within the valley outlet glacier (Eyles &

F
I G U R E 7 Conceptual model for the debris-charged plateau icefield/cirque glacier landsystem based on the Godeanu Mountains.The model is based on amalgamated elements from different valleys.Four phases (a-d) are represented to show how the landsystem signature changes spatio-temporally during deglaciation.Key features and landforms are as follows: (1a) plateau icefield dispersal centre; (1b) plateau icefield; (2) cirque glacier; (3) medial moraines; (4) nunatak; (5) snowpatches with protalus lobes (a) and protalus/pronival ramparts (b); (6) rock slope failure; (7a) latero-frontal end moraine; (7b) recessional moraine loops; (8) proglacial streams, ice-marginal meltwater channels and outwash deposits; (9) debris covered glacier snout developing into a rock glacier; (10) protalus rock glacier; (11) lateral moraines; (12) rock-glacierised lateral moraines; (13) reconstituted or stagnant glacier; (14) blockfield; (15) glacially scoured bedrock outcrops (roches moutonnées and whalebacks).[Color figure can be viewed at wileyonlinelibrary.com] gations, we devise a glacial landsystem model for the Godeanu Mountains, Southern Carpathians, Romania, to understand past glaciation styles and dynamics of mid-altitude mountain glaciers located in temperate-continental climates.We derive the following conclusions: Firstly, we identified glacial depositional (latero-frontal, recessional, lateral, medial moraines, drift limits/till veneers), glacial erosional (ice-moulded bedrock, ice-marginal meltwater channels), periglacial (pronival and protalus ramparts, relict rock glaciers and rock-glacierised moraines) and paraglacial (RSFs) landforms.The spatial distribution and association of landforms are compatible with a debris-rich glacier retreat, sourced from plateau surfaces/cirques.Secondly, we performed sedimentological analyses on selected glacial landforms.These further support a debris-charged glacial environment, with transport pathways transitioning from passive to active with distance from valley heads and sides.Although active transport in the lower valleys relates to subglacial erosion from the bed mostly during the earlier stages of deglaciation, passive transport became more dominant in the latter stages, wherein abundant extraglacial sediment was sourced from cirque heads/valley slopes and transported via well-preserved medial moraines to the glacier foreland.Thirdly, we produce a four-phase, time-transgressive, debrischarged plateau icefield/cirque glacial landsystem model.Phase 1 corresponds to the maximum glacial extent, when plateau ice coalesced with cirque ice, highlighting a transition from cold-based uplands to the warm/polythermal regimes in the valleys.Alongside material sourced from active bedrock erosion at lower elevations, extraglacial debris from RSFs at the valley heads was transported passively via medial moraines and deposited as latero-frontal ridges at the glacier snout.Phase 2 follows ice retreat from the maximum extent.The association of ice-marginal meltwater channels with recessional moraines is evidence of a warm-based/polythermal coupled ice margin, with a high debris turnover.Phase 3 underpins continual active but increasingly debris-charged recession and disconnection between cirque glaciers and plateau-fed outlets.This is supported by recessional moraines and looped medial moraines pertaining to individual ice flow units and asymmetrical lateral moraines, all of which reflect additional sediment supply sourced from fresh RSFs and the gradually uncovered ice-moulded bedrock ridges.Phase 4 illustrates the final retreat of the cirque and plateau outlets to their upland sources, the increased confinement of cirque glaciers to their niches and the more important role of debris and topography in driving landscape evolution.The likely climatic amelioration and abundance of extraglacial debris from debuttressed slopes led to the burial and uncoupling of glacial snouts from their foreland, which then became inefficient at removing the sediment load.This is supported by the development of pronival and protalus ramparts, rock glaciers and rock-glacierised moraines, which became more dominant over recessional moraines.Ultimately, the role of bedrock steps as topographic pinning points, and the elevation and topographic contrasts between plateau and cirque-fed glaciers explain the contrasts in moraine size, spacing and elevation between different outlets of the Godeanu Mountains and higher alpine summits in other areas of the Southern Carpathians.Finally, the new landsystem model can be used as an analogue for other plateau/cirque-flanked mountains in Central, Eastern and Southern Europe that exhibit a debris-rich geomorphic signature on their former glacier beds but ideally in conjunction with absolute dating and numerical modelling of glacier extents.AUTHOR CONTRIBUTIONS Cristina I. Balaban co-designed the study, obtained funding for fieldwork, designed the methodology, conducted fieldwork and analysis and drafted the initial and final manuscript.David H. Roberts co-designed the study, supported the methodology and fieldwork, supervised CIB and helped draft the final manuscript.David J. A. Evans co-designed the study, supported the methodology, supervised CIB and helped draft the final manuscript.Stewart S. R. Jamieson codesigned the study, supported the methodology, supervised CIB and helped draft the final manuscript.