In Svalbard, more than half of the lagoon area (56.3 %) is occupied by PMLs, covering over 82.9 km² (Table 1). In the 1930s, six of the 14 selected lagoons did not yet exist, and two were glacial lakes. The area of the six PMLs existing in the 1930s was approximately 27.7 km². Over the course of 88 years, the PML's area has nearly tripled. The largest Svalbard lagoon, Tjuvfjordlaguna (Fig. 2), is the best example of PML and in 2024 covers an area of more than 61 km² (73.7 % of the PML's total area). It is more than twice the total area of all 13 other lagoons combined (21.8 km²). In 1936, the lagoon measured approximately 3500 m east–west and 7000 m north–south; by 2024, these dimensions shifted to 11 000 and 4500 m, respectively. This eastward expansion aligns with a calculated glacier recession rate of 75.5 m yr⁻¹ (Table 2). Between 2010 and 2024 (during which time all lagoons already existed), the area of PMLs increased by more than 14 km². The Tjuvfjordlaguna area alone increased by as much as 10 km².

Further notable changes occurred in Snaddlaguna (Table 1 and Fig. 3) and Eidemlaguna (Table 1 and Fig. 4). Their areas increased by 1.751 and 0.837 km², respectively. It is worth noting that Eidemlaguna recorded the most spectacular relative increase. From a small lagoon with an area of 0.09 km² in the 1930s (the second smallest lagoon measured at that time), it has become a huge reservoir with an area of 6.1 km². This is an increase of over 6700 %, making it the most rapidly growing system in terms of growth rate. The dynamically changing Eidemlaguna environment is also highlighted by Šiaulys et al. (2026). It is also worth mentioning that the surface area of Eidemlaguna was as large as 6.6 km² in 2023 (Šiaulys et al., 2026) and 6.1 km² in 2024 (Table 1). This change may be due to the sediment flux from the Eidembreen, which built up the islands that were previously located there, thus reducing the lagoon surface area.

In most cases, PMLs are expanding, but in four cases, their area is decreasing. The largest area decrease was recorded in Charleslaguna (Fig. 5), from 0.991 km² to 0.867 km² (a 12.5 % area decrease over 14 years). Femtelaguna (Fig. 6) was second, with a 6 % change in area, losing 0.025 km². Overall, over the past almost 90 years, the area of PMLs in Svalbard has increased notably, with only a few lagoons showing a slight decrease.

The glaciers' retreat associated with PMLs exhibits notable spatial and temporal variability across Svalbard (Tables 2 and 3). Overall, every studied system is linked to a retreating glacier, with average recession rates ranging from 7.6 m yr⁻¹ to over 75 m yr⁻¹ (Table 2). Changes in PMLs are primarily a consequence of melting glaciers. In this case, the greatest change was observed in Deltabreen – the glacier connected to Tjuvfjordlaguna. Over the course of 88 years, the glacier retreated inland by more than 6.6 km (at a rate of approx. 75 m yr⁻¹). During the most recent interval (2010–2024), Deltabreen also exhibited the greatest recession, retreating by approximately 1.7 km, followed by Eidembreen at 0.85 km. Both glaciers demonstrated relatively consistent retreat rates throughout the entire study period. In contrast, Makarovbreen, Makarovlaguna's (Fig. 7) “parent” glacier, showed a notable deceleration. Despite previous retreat magnitudes ranging from 0.282 to 0.589 km per period (0.484 km between 1936 and 1960, 0.589 km between 1960 and 1990, and 0.282 km between 1990 and 2010), its margin remained nearly stationary between 2010 and 2024, with a negligible shift of only 0.002 km.

These rates are broadly consistent with previously published values for comparable glaciers, although glacier-specific retreat estimates for the individual systems studied here are largely absent from the literature, so comparisons are necessarily drawn at the scale of regionally and morphologically analogous glaciers. The closest analogue is the work of (Kavan et al., 2022), who examined 11 tidewater glaciers along the south-east coast of Spitsbergen using a 1970 DEM combined with recent ArcticDEM and Landsat/Sentinel-2 imagery, and reported an average frontal retreat rate of 48 m yr⁻¹ (range 10–150 m yr⁻¹), with four of the eleven glaciers becoming land-based as their termini retreated. Our values fall squarely within this published range: the lower end of our dataset (7.6 m yr⁻¹) corresponds to typical land-terminating behaviour, while Deltabreen's ∼75 m yr⁻¹ lies within the upper-middle of the tidewater range. This is also in line with the glacier-type typology of Rachlewicz et al. (2007), who distinguished land-terminating glaciers receding at roughly 5–15 m yr⁻¹, non-surging tidewater glaciers at about 15–70 m yr⁻¹, and surging tidewater glaciers reaching 100–220 m yr⁻¹. More broadly, our findings support the regional picture in which eastern Svalbard has experienced some of the most pronounced ice loss in the archipelago: the Edgeøya ice field has shown the largest retreat rates of all Svalbard glaciers over recent decades (Nuth et al., 2013), and since the Little Ice Age, glaciers on Barentsøya and Edgeøya have lost 16.7 % of their area – the highest proportional loss among the main islands of the archipelago (Martín-Moreno et al., 2017).

Differences between our rates and previously published values can be attributed to several factors. First, methodological and temporal differences are important: our long-term rates are centreline recession distances averaged over an 88-year, multi-epoch baseline (1936–2024), whereas satellite-era studies such as Li et al. (2025), who resolved calving-front change across the Barentsøya – Edgeøya sector, derive shorter-interval, calving-front rates that are more sensitive to seasonal and interannual variability; averaging over different epochs and reference baselines inevitably yields divergent values. Second, glacier dynamics play a role, as eastern Svalbard is one of the principal clusters of surge-type glaciers, and terminus advances during surges have historically punctuated the general post-LIA retreat in this region; the near-stationary margin of Makarovbreen between 2010 and 2024 is consistent with a post-surge quiescent phase rather than a simple climatic deceleration. Third, and most likely for the decelerating systems, the transition from a marine- to a land-terminating configuration removes the calving and submarine-melt component that drives rapid tidewater retreat; once a terminus grounds in shallower water or on land, retreat tends to slow, a mechanism invoked to explain the eventual stabilisation of retreating Svalbard tidewater glaciers (Li et al., 2025), and this paraglacial transition is itself the process generating the new coastal lagoons examined here (Kavan and Strzelecki, 2023). Finally, local controls – bed slope and overdeepening, water depth at the grounding line, aspect, and hypsometry – produce substantial inter-glacier variability superimposed on the regional climatic signal (Małecki, 2013; Rachlewicz et al., 2007).

Glacial dynamics analysis (Table 3) reveals a progressive loss of direct connectivity between glaciers and their respective lagoons. Currently, in five out of the fourteen studied cases, the glaciers have completely retreated onto land, losing their marine margins. This terrestrial transition is particularly evident in Charleslaguna, where the “parent” glacier no longer terminates in the lagoon waters but contributes via fluvial systems. In general, when a lagoon loses contact with a glacier, its dynamics change, most often stabilising the lagoon's development.

Quantitative shoreline-change metrics derived from DSAS (Table 4) confirm that moraine barriers associated with PMLs on Svalbard are highly dynamic and exhibit both rapid erosion and localised accretion, depending on sediment supply, glacier position, and hydrodynamic exposure.

The Tjuvfjordlaguna barrier-moraine system, totalling 9.5 km in length, shows substantial narrowing, with the northern arc decreasing in some points from 950 to 375 m and the southern arc from 741 to 260 m in width (Figs. 2 and 8A). This extreme retreat is reflected in the DSAS results, which show a minimum EPR of -7.91 m yr⁻¹ and a mean EPR of -1.50 m yr⁻¹, accompanied by a minimum NSM of -695.95 m and a mean NSM of -132.19 m. These values are an order of magnitude higher than those observed for other PMLs, indicating sustained, long-term erosion rather than short-lived fluctuations. Analysis of the inlet reveals a more than tenfold expansion: the narrowest point widened from 26.84 m in 1936 to 723.38 m in 2024, with the average width increasing from 72.8 to 805.6 m. The moraine has retreated more noticeably on the sea-facing side (west) than the lagoon-facing side (east). This asymmetric retreat reflects stronger erosion on the sea-facing side compared to the lagoon-facing side, with observed narrowing of the barrier and expansion of the inlet indicating ongoing morphological adjustment under marine forcing.

In contrast, Scheilaguna (Figs. 8E and 9) represents the opposite end-member of PML evolution, characterised by barrier growth and increasing isolation from the open sea. DSAS results show consistently positive shoreline-change rates (EPR mean 0.39 m yr⁻¹; NSM mean 34.73 m), indicating net accretion. The Kvasspiggbreen terminal moraine is largely submerged, emerging only during low tide. By 2024, this barrier had extended by 530 m compared to observations from the 2010s. Currently, the inlet at Scheilaguna narrows to 106 m at low tide. A similar case can be seen at Femtelaguna (Figs. 6 and 8G), where between 2010 and 2024 the barrier lengthened by 88 m, visibly narrowing the PML inlet. Despite locally positive EPR values (EPR max 2.45 m yr⁻¹), the mean EPR (-1.39 m yr⁻¹) and NSM (-19.45 m) indicate that accretion at the inlet coexists with erosion elsewhere along the barrier. In both cases, the “parent” glacier is located on land and no longer has a direct connection with the lagoon.

Eidemlaguna (Fig. 4) deserves particular attention due to its complex and non-linear evolutionary trajectory. Although DSAS indicates a predominantly erosional trend (EPR mean -1.18 m yr⁻¹; NSM mean -103.58 m) (Table 4 and Fig. 8H), this lagoon is the only one in the study area to have changed its openness type twice (Table 5). In 1936, Eidemlaguna had one inlet from the northern part of the barrier, and until 2010 it had another inlet from the southern part of the barrier. In 2010, the northern inlet was closed and remains so to this day. This demonstrates that shoreline retreat does not necessarily translate directly into increased openness and that local sediment redistribution and barrier reconfiguration can temporarily offset erosion-driven widening.

The remaining PMLs (Makarovlaguna, Sjulaguna, and Tromsølaguna) show comparatively low rates of change, with EPR mean values close to zero (from -0.13 to 0.05 m yr⁻¹) and modest NSM values, indicating limited net shoreline displacement (Appendix A). These systems appear to occupy an intermediate state in which erosional and depositional processes are broadly balanced. Taken together, the DSAS metrics and morphological observations demonstrate that PML barriers do not evolve uniformly but instead follow divergent pathways of either rapid erosion and fragmentation or short-term stabilisation and inlet closure. Snaddlaguna (Figs. 3 and 8C) exhibits a predominantly erosional trend, with DSAS results showing a mean EPR of -0.52 m yr⁻¹ (minimum -1.79 m yr⁻¹; maximum 1.47 m yr⁻¹) and a mean NSM of -46.06 m (minimum -157.91 m; maximum 128.96 m). This indicates that while some sections of the barrier experience accretion, overall retreat dominates, reflecting dynamic sediment redistribution and partial narrowing under variable hydrodynamic conditions.

Overall, these metrics demonstrate that PML barriers exhibit divergent pathways, ranging from rapid erosion and fragmentation to short-term stabilization and inlet closure, reflecting the range of observed morphodynamic behaviours in the study area.

The retreat of marine-terminating glaciers in Svalbard has exposed extensive new coastal sectors that undergo rapid geomorphological reorganization (Kavan et al., 2025; Kavan and Strzelecki, 2023). In this study, we define Paraglacial Moraine Lagoons (PMLs) as systems separated from the open sea by barriers composed of former terminal or lateral moraines. Often partially ice-cored, these landforms are a direct consequence of recent glacial recession (Owczarek, 2025). This moraine-based foundation distinguishes PMLs from deltaic or sandy barrier lagoons and fundamentally dictates their stability.

Over the past century, Svalbard's coastline has undergone substantial transformation, with 98 new lagoons forming and 29 losing marine connectivity (Owczarek, 2025). While open and leaky lagoons often transition into closed systems via spit progradation and barrier migration (Kavan and Strzelecki, 2023), glacio-isostatic uplift provides an additional control. Reaching up to 10 mm yr⁻¹ in certain regions, this uplift offers partial protection from wave energy but may also accelerate the eventual isolation of lagoons from the sea (Kierulf et al., 2022).

Shoreline migration quantified using DSAS indicates that most of the analysed PMLs are currently dominated by erosion. Six out of seven lagoons exhibit negative mean EPR values, reflecting net retreat. Erosion is particularly pronounced at Tjuvfjordlaguna, where the mean EPR reaches -1.50 m yr⁻¹ (minimum -7.91 m yr⁻¹) and the mean NSM is -132.19 m. Eidemlaguna and Femtelaguna also experience sustained retreat, with mean EPR values below -1 m yr⁻¹. These rates are generally higher than those reported for more sheltered Svalbard coasts, such as Isbjørnhamna (Zagórski et al., 2015), Calypsostranda (Zagórski et al., 2020), Adventfjorden (Nicu et al., 2021), or Kaffiøyra (Czarnecki and Sobota, 2025), where mean erosion rarely exceeds -0.4 m yr⁻¹. The variability in lagoon openness over time is also consistent with observations across Svalbard, where some lagoons have disappeared due to barrier erosion, glacial sediment infilling, or isolation from the sea (Owczarek, 2025; Wołoszyn et al., 2022; Ziaja et al., 2009, 2023), illustrating the sensitivity of these coastal systems to both marine and terrestrial processes.

Despite this, the retreat rates of the most dynamic PMLs remain notably lower than extreme Arctic erosion hotspots, such as ice-rich permafrost cliffs at Drew Point (-38.3 m yr⁻¹, Alaska, Wang et al., 2022) or Pelly Island (-5.5 m yr⁻¹, Canada, Malenfant et al., 2022). These comparisons are intended for contextual purposes rather than implying equivalent processes, as PML barriers differ substantially from permafrost-dominated thermoabrasional coasts. Nonetheless, the intermediate erosion magnitude highlights the high sensitivity of moraine-based lagoon barriers compared to typical Svalbard fjord shorelines. The wide range of NSM values within individual lagoons further indicates pronounced spatial heterogeneity, suggesting that erosion and progradation may occur simultaneously along different sectors of the same moraine barrier, reflecting the inherent instability of recently deglaciated paraglacial coasts. Rapid terrestrial events, such as glacial lake outburst floods (GLOFs), have also been documented (Wołoszyn et al., 2022) to dramatically reshape lagoon morphology within days to weeks, emphasizing the dynamic interplay between coastal and glacial processes in Svalbard.

A contrasting example is Recherchelaguna, one of the best-described lagoon systems in Svalbard. Although genetically paraglacial, its barriers were constructed from proglacial deltas in front of Recherchebreen in the early 20th century and lack a moraine core (Kavan et al., 2024; Zagórski et al., 2012). For this reason, Recherchelaguna does not meet the structural PML criteria. This distinction is important: while delta-built barriers are governed primarily by fluvial sediment supply, PML evolution is strongly controlled by the mechanical properties, ice content, and geometry of former moraines. Consequently, Recherchelaguna represents a unique delta-based paraglacial lagoon, whereas PMLs constitute a separate moraine-controlled category occurring in multiple Svalbard fjords. Furthermore, the formation and evolution of PMLs are influenced by ongoing post-glacial processes, including glacio-isostatic uplift (Kierulf et al., 2022) and sediment transport (Kavan and Strzelecki, 2023), which together determine whether lagoons remain connected to the sea, become isolated, or experience barrier fragmentation.

An important trend observed in this study is the relationship between lagoon surface reduction and glacier configuration. In 75 % of cases where the lagoon area decreased, the “parent” glacier had already retreated entirely onto land. Although this temporal correspondence does not establish direct causation, it strongly suggests a shift in sedimentary regime following glacier detachment from the marine margin. Once glaciers become land-terminating, calving and subaqueous scouring cease, and proglacial rivers deliver glaciogenic sediments directly into the lagoon basin. This enhanced terrestrial sediment flux likely accelerates infilling and barrier aggradation, progressively restricting marine connectivity. This stabilising–isolating trajectory is exemplified by Scheilaguna, where exclusively positive EPR values (mean 0.39 m yr⁻¹) and a calculated barrier extension rate of 2.86 m yr⁻¹ indicate systematic progradation (Table 4). If current trends continue, complete isolation from the sea may occur within several decades, leading to gradual terrestrialisation of the lagoon. These trends mirror broader coastal changes observed in Svalbard, where both natural sediment dynamics and the interaction of morainic barriers with post-glacial uplift govern lagoon evolution (Ziaja et al., 2009, 2023).

In contrast, Tjuvfjordlaguna illustrates an erosional–fragmenting trajectory. Over 88 years, the moraine width decreased by approximately 650 m (Fig. 8A, Table 4). Continued retreat at similar rates could result in breaching of the narrowest barrier sections within a few decades, potentially fragmenting the moraine and re-establishing an open marine embayment. This example demonstrates that moraine barriers may not stabilise post-deglaciation but can instead undergo substantial marine reworking and structural degradation.

Finally, Murraylaguna (Fig. 10) represents a configuration in which PML development appears possible but was not fully realised. While the former Murraybreen moraine is partly submerged and visible only in archival imagery, current sediment routing favours island formation near the glacier front rather than nourishment of the former moraine ridge. This evidence suggests a potential, though unrealised, pathway of PML development, likely constrained by local bathymetry and sediment distribution.

Overall, these results indicate that PMLs are short-lived geomorphological features confined to a limited paraglacial time window following glacier retreat. After detachment from the marine margin, lagoon systems diverge toward one of two dominant trajectories: (1) stabilisation, isolation, and infilling driven by enhanced terrestrial sediment flux, or (2) barrier erosion and fragmentation under dominant marine forcing. DSAS-derived shoreline metrics provide quantitative support for this dual behaviour, reinforcing the interpretation of PMLs as highly transitional elements within the evolving coastal landscapes of Svalbard. The broader historical record of Svalbard lagoons confirms these patterns (Owczarek, 2025; Ziaja et al., 2009, 2023), showing that coastal change operates over multiple scales and through a combination of gradual processes, episodic events, and post-glacial uplift, all of which critically shape the formation, evolution, and eventual fate of PMLs.

The evolution of PMLs is fundamentally dictated by terrestrial paraglacial processes, which redefine the traditional sediment flux of Arctic coastlines. While paraglacial coasts typically exhibit rapid sediment flushing, PMLs introduce a critical “sedimentary sink” – a temporal pause where coarse glaciogenic material is sequestered in the nearshore zone rather than being lost to the continental shelf or deep-sea basins. This mechanism disrupts the standard delivery of meltwater and sediment directly into the marine environment. The vulnerability of these systems to terrestrial events is exemplified by GLOFs, which can cause the instantaneous “extinction” of a lagoon via catastrophic infilling (Wołoszyn et al., 2022). Such events underscore the role of PMLs as geomorphological buffers. Furthermore, contemporary observations at Recherchelaguna (Kavan et al., 2024) and Charleslaguna demonstrate a shift toward terrestrialization; in the latter, the development of a proglacial delta (0.042 km²) between 2010 and 2024 (Fig. 11) has significantly reduced the lagoon's accommodation space. Neglecting PMLs in coastal models leads to a fundamental miscalculation of the sediment budgets defining glaciated Arctic margins.

The formation of PMLs introduces a fundamental shift in the physical energy of the nearshore environment. By establishing a moraine barrier, these systems transform high-energy, wave-dominated open coasts into sheltered, low-energy basins. This transition significantly alters the physical template available for colonization. In Svalbard, such sheltered conditions are typically associated with increased retention of fine-grained sediments and organic matter, which are otherwise flushed into the deeper fjord (Kavan and Strzelecki, 2023). While the temporal window of PMLs is transient, the immediate transition from a glacial terminus to a sheltered lagoon creates a unique brackish-water habitat. These geomorphological changes suggest that PMLs may function as temporary biodiversity centers, though empirical evidence regarding the specific rate of benthic species establishment or the role of these lagoons as stopover sites for migrating fauna remains to be gathered.

We propose that PMLs represent a distinct, previously unclassified genetic category of Arctic lagoons. In our conceptual model, PMLs are defined by a “glacial legacy” barrier – a terminal or lateral moraine system that decouples the glacier–coast interface from open-sea energy, a feature notably absent in typical paraglacial environments. We define the evolution of these systems through the following four key stages (Fig. 12).

Stage 1: Moraine Foundation (Little Ice Age Framework). During the glacial phase (Fig. 12A), terminal and lateral moraines deposited at the LIA maximum provide the structural foundation for the PML system. Here, the moraine is not a passive depositional feature but an active element in the feedback between glacial sediment supply and marine abrasion. It functions as a hydrodynamic buffer, reducing wave energy at the glacier front and potentially limiting calving intensity. The stability of this moraine barrier determines whether a lagoon can later form; without it, the system would not follow the PML trajectory.

Stage 2: Lagoon Genesis (Post-LIA Retreat). As climate-driven warming triggers the retreat of the glacier margin from its LIA maximum, the depression becomes flooded with seawater, forming a low-energy lagoon basin (Fig. 12B–C). This intermediate environment bridges the gap between classic glacial fjords and open paraglacial coasts.

The key transition is from a system directly exposed to the open sea (Fig. 12B) to a partially isolated lagoon controlled by the moraine (Fig. 12C). Even marine-terminating glaciers exhibit reduced calving efficiency in this environment. The moraine barrier not only facilitates lagoon formation but also modifies glacier retreat dynamics.

Stabilization of the lagoon basin promotes redeposition of glacial sediments, reshaping barrier morphology through sediment supply from the active glacier, lagoon water-level fluctuations, internal currents, and calving-generated waves. This process gradually transforms the moraine into a structure with increasingly coastal characteristics.

At this mature lagoon stage (Fig. 12D), the moraine still regulates energy exchange with the sea, but its morphology reflects both glacial deposition and paraglacial processes. Tjuvfjordlaguna, Makarovlaguna, Sjulaguna, and Snaddlaguna currently exemplify this stage.

Stage 3: Terrestrial Transition and Sediment Trap. As the glacier retreats onto land (Fig. 12E–F), the lagoon undergoes major sedimentary reorganization. Calving cessation and exposure of ground moraines promote proglacial drainage, increasing fluvio-glacial sediment supply and initiating delta formation. At this stage, PMLs serve as primary sediment traps within the paraglacial transport cascade, retaining coarse material that would otherwise reach the deep shelf. Their buffering role aligns with paraglacial slope readjustment and sediment redistribution after glacier retreat (Ballantyne, 2002; Forbes and Syvitski, 1994; Kavan et al., 2024; Strzelecki et al., 2020; Zagórski et al., 2012). Moraine stability becomes controlled by the balance between terrestrial sediment supply and marine erosive pressure. Two scenarios are possible now: the first, where erosion exceeds sediment supply (Fig. 12E^′), and the second, the opposite scenario, in which sediment supply exceeds the erosion rate (Fig. 12F^′). However, glacier re-advances (also glacier surges) can destroy lagoons, as likely occurred during LIA glacier expansions (Fig. 12G).

Stage 4: Marine-driven fragmentation versus terrestrial-driven isolation. The final stage is nonlinear, governed by the balance between sediment delivery and marine erosion. Proglacial delta growth can lead to increased sediment accumulation in the glacier's foreland. Limiting the sediment supply to the barrier-moraine while maintaining the same level of marine erosion may lead to the barrier's destruction. Murraylaguna can be considered an example of such a situation, even though Murraybreen is not yet completely on land (Fig. 12E^′′).

Despite the formation of a proglacial delta, the sediment supply is sufficient to maintain the current state of the barrier moraine or even to build it up (Fig. 12F^′′). As deltas expand, they reduce the lagoon's accommodation space, potentially bifurcating the basin or severing its connection to the sea; further driving the lagoon toward terrestrialization. Examples of such PMLs are Femtelaguna and Charleslaguna.

This stage highlights that PML evolution is controlled by glacial legacy but unfolds through paraglacial processes – slope, fluvio-glacial, and coastal – which collectively dictate the pace and direction of system transformation (Ballantyne, 2002).

The identification of Paraglacial Moraine Lagoons (PMLs) as a distinct landform category opens several new avenues for Arctic coastal research. Because these features represent a transient state in the paraglacial landscape cycle, their future evolution will likely serve as a high-resolution indicator of the pace of Arctic coastal reorganization. However, fully decoding the role of PMLs in the changing Arctic requires a shift toward integrated, interdisciplinary studies that bridge the gap between geomorphology, cryospheric science, and marine ecology:

A critical uncertainty remains regarding the internal structure of PML barriers, necessitating collaboration between geomorphologists and geophysicists. Since many terminal moraines in Svalbard are known to be ice-cored, rising Arctic temperatures and the impact of waves may trigger accelerated degradation through thermo-abrasion. This could lead to catastrophic barrier failure and breaching. Future research utilizing geophysical surveys with ground penetrating radars or electrical resistivity tomography is needed to quantify the presence and volume of ground ice within these barriers and predict their vulnerability to sudden structural collapse.