Geomorphology 327 (2019) 572–584 Contents lists available at ScienceDirect Geomorphology journal homepage: www.elsevier.com/locate/geomorph Multiphase breakdown sequence of collapse doline morphogenesis: An example from Quaternary aeolianites in Western Australia Matej Lipar a,⁎, Uroš Stepišnik b, Mateja Ferk a a b Anton Melik Geographical Institute, Scientific Research Centre of the Slovenian Academy of Sciences and Arts, Gosposka ulica 13, SI-1000 Ljubljana, Slovenia Department for Geography, University of Ljubljana, Aškerčeva 2, SI-1000 Ljubljana, Slovenia a r t i c l e i n f o Article history: Received 27 September 2018 Received in revised form 29 November 2018 Accepted 29 November 2018 Available online 3 December 2018 Keywords: Sinkhole Karst Geomorphology Calcrete a b s t r a c t The morphology of collapse dolines varies according to their maturity and effectiveness of the removal of collapsed material. In addition, the variable balance between various geomorphic processes due to local geologic, hydrologic and climatic settings results in a diverse morphology of collapse dolines and dynamics of their morphogenesis. A single generalised proposed sequence of collapse dolines morphogenesis has therefore limited value and more detailed study is needed in terrains which differ in terms of geology, hydrology and/or climate. There is a particularly well exposed karst in Stockyard Gully National Park in the southwestern coastal part of Western Australia, formed in Quaternary aeolianites, consisting of a dense field of collapse dolines up to 100 m in diameter and on average 10 m deep. Extensive field work combined with available data of local rock stratigraphy and a comparison to collapse dolines worldwide revealed that the principal processes of collapse doline formation in Stockyard Gully National Park are similar to processes responsible for collapse doline formation worldwide (i.e., collapses above underground chambers and removal of the collapsed material). However, rock characteristics in the area influence their morphometry due to the mechanical weakness of aeolianite compared to well-cemented limestones, and a surface calcrete layer with stronger mechanical resistance than underlying aeolianite. Consequently, we propose a new 4-stage multiphase breakdown sequence of collapse doline morphogenesis in aeolianites, divided into cave dome, calcrete caprock dome, young collapse doline, and mature collapse doline. Calcrete caprock dome is stabilised by the uppermost well cemented calcrete and represents a distinctive phase just before the final breakdown to form an actual collapse doline. © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). 1. Introduction Collapse dolines (also referred to as collapse sinkholes, collapse sinks; see Section 3.1) are one of the most prominent surface features in a karst landscape. They are closed karst depressions formed by sudden collapse or gradual lowering of a surface above a mechanically unstable underground cavity (Jennings, 1985; Ford and Williams, 2007). They are usually circular to subcircular in ground plan, measuring from a few metres to several hundred metres in diameter and up to several hundreds of metres deep. Commonly they have vertical to steep slopes with collapsed boulders or scree covering the floor (Waltham et al., 2005). They typically (but not exclusively) occur in porous soluble rocks (e.g., limestone, gypsum), where water can percolate through the rock and consequently dissolve it to form voids and cavities as well as to remove collapsed material through dissolution and erosion (Gabrovšek and Stepišnik, 2011; Hiller et al., 2014; Kaufmann and Romanov, 2016). Although they are particularly frequent and well developed in ⁎ Corresponding author. E-mail address: matej.lipar@gmail.com (M. Lipar). karst areas with concentrated shallow subsurface flow, they are found in all karst environments, including arid areas where the dynamics of their development is constrained due to the small amount of subsurface drainage (Jennings, 1985; Gunn, 2004; Abotalib et al., 2016). Rock characteristics significantly influence collapse doline' dynamics of formation, resulting in complexity and heterogeneity of their morphology (Šušteršič, 2000; Šušteršič, 2003; Doğan and Özel, 2005; Šušteršič, 2006; Gabrovšek and Stepišnik, 2011; Santos et al., 2012; Kaufmann, 2014; Kaufmann and Romanov, 2016). Thus, even though they attract substantial scientific attention, the general chronological stages of their formation cannot be applied to collapse dolines in environmental settings with complex rock characteristics. The study of collapse dolines is important from various aspects of karst geomorphology, hydrology and speleology. Monitoring the development of collapses contributes to understanding dynamics of the development of surface karst features. In particular, it supports an understanding of the development of slopes in karst (Stepišnik and Kosec, 2011; Kaufmann, 2014; Kaufmann and Romanov, 2016), and through rock exposures offers an insight into the karst vadose zone, which is important for paleoclimate reconstruction (Hartmann and Baker, 2017). Size, morphology and spatial distribution of collapses are https://doi.org/10.1016/j.geomorph.2018.11.031 0169-555X/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). M. Lipar et al. / Geomorphology 327 (2019) 572–584 direct indicators of the past and present hydrologic settings of the area, and therefore help the understanding of vadose and underground water, its flow, level, change, and human impact on it (Mihevc, 2001; Šušteršič et al., 2003; Waltham et al., 2005; Šušteršič, 2006; Mihevc, 2010; Gabrovšek and Stepišnik, 2011; Ferk, 2011; Ferk, 2016; Ferk and Stepišnik, 2011). Understanding collapse dolines, which therefore includes understanding of karst environments and subsurface drainage, has also a direct connection for better planned environmental and engineering procedures in such environments. When these developments expand in karst areas, so too do impacts and hazards associated with karst (Gutiérrez et al., 2014). The Stockyard Gully Cave System in the southwest part of Western Australia offers an excellent insight into formation of collapse dolines in Quaternary aeolianites (i.e., aeolian calcarenites) which have been subjected to a multiphase breakdown sequence. The overall rock characteristics, which include matrix porosity, weak cementation and the development of erosional resistant calcrete layers, differ from the rock characteristics of a diagenetically mature limestone. The latter rock characteristics are typical for a “classic karst” (a term used for the Dinaric karst in Slovenia and Italy where the first scientific investigations were carried out; Ford and Williams, 2007) where collapse dolines were first defined by Cvijić (1893). The comparison between those two different lithological environments can therefore increase understanding of the common processes responsible for collapse doline formation, including lithological influence. The aim of this paper is to present a multiphase breakdown sequence of collapse doline formation in a geological setting dominated by aeolianite and calcrete. This data, alongside a comparison with the 573 worldwide occurrence of collapse dolines, contributes to the overall understanding of the complex behaviour of karst. 2. Regional setting and geology The research area along the Stockyard Gully Cave System in Stockyard Gully National Park (Fig. 1) lies around 250 km north of Perth in southwestern Western Australia, and is characterised by a Mediterranean climate with hot and dry summers, and mild and wet winters. The average annual rainfall in the nearby town of Eneabba is ~490 mm and mean annual temperature is ~21 °C (Bureau Of Meteorology, 2018) (note Eneabba weather station is now closed and the last data is for March 2017). Low heath rich in Fabaceae, Myrtaceae and Acacias dominates the landscape, with Tuart trees in the gullies and valleys (Conservation Commission of Western Australia, 2014). The research area stretches over a landscape influenced by the cyclic deposition of marine carbonate dunes during the Quaternary, termed the Spearwood Dune System. This carbonate sand is now mostly lithified to form the Tamala Limestone (Teichert, 1947; Teichert, 1950; Passmore, 1970; Playford et al., 1976; Kendrick et al., 1991). The Tamala Limestone is predominantly a cemented Pleistocene calcareous coastal dune system, defining it as aeolianite, and stretches for N1000 km along the southwest Western Australian coast. Lipar and Webb (2014) described several members of the Tamala Limestone in the research area, all fine to coarse-grained well-sorted aeolianites with often visible aeolian cross bedding or horizontal planar lamination, high matrix porosity and relatively weak cementation (Table 1). The members are separated by mostly weakly cemented or loose reddish palaeosol horizons Fig. 1. Locality map of the research area in southwestern Western Australia. The satellite map is based on GoogleEarth imagery, Digital Elevation Model data is downloaded from Shuttle Radar Topographic Mission website, and the Tamala Limestone distribution is based on Geoscience Australia 1:1,000,000 scale Surface Geology of Australia (digital dataset, 2008). 574 M. Lipar et al. / Geomorphology 327 (2019) 572–584 Table 1 The stratigraphy of the Tamala Limestone Formation in the research area after (Lipar and Webb (2017)). Time of deposition Stratigraphy Karstification Palaeosol Calcrete MIS 7 Pinnacles Desert Member Recent soil MIS 9 Stockyard Gully Member Very karstified, abundance of pinnacles and solution pipes Absent MIS 11 Nambung Member Very karstified MIS ≥13 White Desert Member Slightly karstified (visible solutional voids) Some amount of case-hardened rock and abundant microbialitic calcrete Extensive up to 40 cm thick laminar calcrete and case-hardened rock Minor amount of case-hardened rock and abundant microbialitic calcrete Some amount of case-hardened rock and microbialitic calcrete with higher proportions of quartz grains and clay than the surrounding aeolianite. To the east, the Tamala Limestone area borders the Late Tertiary to Middle Pleistocene siliceous Bassendean Dune System (also termed Bassendean Sand) (Playford and Low, 1971; Mory, 1995), whilst to the west, the area borders the loose Holocene carbonate Quindalup Dune System (Safety Bay Sand) (Passmore, 1970; Semeniuk et al., 1989; Mory, 1995). The entire research area has a rolling topography of denuded coastal dunes with well-developed syngenetic karst. This type of karst is formed more or less simultaneously with the sediment lithification, with caves, solution pipes (tubular vertical voids in epikarstic zone (Lipar et al., 2015)) and pinnacles (limestone pillars (Lipar and Webb, 2015)) as the most prominent features (Bastian, 1964; Jennings, 1968; Grimes, 2006). The term syngenetic karst strongly overlaps with the term eogenetic karst, both related to karst in weakly consolidated limestones during their early stage of diagenesis (Grimes, 2006): it is widespread along shorelines (Vacher and Mylroie, 2002), and also occurs in continental settings (Florea et al., 2007; Lipar and Ferk, 2011). 3. Material and methods 3.1. Definition and terminology of collapse dolines Collapse dolines were first defined by Cvijić (1893) as a type of doline that genetically differs from more common solution dolines (collapse versus dissolution). The term doline is derived from dolina in Slovenian language meaning literally valley. In classic karst area solution dolines are enclosed depressions with diameters approximately between 10 and 100 m, where the main process of formation is dissolution (Gams, 2004; Resnik Planinc, 2016). Collapse dolines, on the other hand, are formed mainly by mechanical processes and are on average bigger than solution dolines, with steeper slopes and collapsed material on the bottom (Stepišnik, 2010). The other term for collapse dolines used mostly in American English is collapse sinkhole, whilst sinkhole itself is used as a general expression for all natural or anthropogenic closed depressions regardless of their size, genesis or material they have formed in (e.g., Benito et al., 1995; Kohl, 2001; Doğan and Çiçek, 2002; Gutiérrez et al., 2007; Nisio et al., 2007; Doğan and Yılmaz, 2011; Oosthuizen and Richardson, 2011; Kobal et al., 2015; Ozdemir, 2015; Vincent et al., 2015). Depending on the region, size, lithology and/or hydrology, there are several subclasses of collapse dolines that have local names. For Mostly weakly cemented reddish up to 1 m thick Mostly loose reddish up to 0.5 m thick Cemented reddish up to 0.5 m thick example, some of the largest collapse dolines in the world, exceeding 500 m in diameter and depth, are recorded in Wulong Karst in China, locally referred to as tiankeng (Senior, 1995; Waltham et al., 2005; Xuewen and Waltham, 2006). This term has been used for other large collapse dolines worldwide (Waltham, 2006). Tiankeng-size collapse dolines are, for example, Crveno Jezero in Croatia, a 528 m deep collapse doline, 385 m × 440 m in diameter (Kaufmann, 2009; Kaufmann and Romanov, 2016), and Velika Dolina in Slovenia, 129 m deep and in average 128 in diameter (Habič et al., 1989; Stepišnik, 2008, 2010). In the Yucatán karst in Mexico there are several hundreds roughly circular collapse dolines extending below the water table level, locally termed cenotes (Marker, 1976; Beddows, 2006; Coke, 2012). This has become the term for all permanently inundated collapse dolines (Kranjc, 2013). Flooded collapse dolines, cenotes, are found worldwide, e.g. in Australia (Grimes, 2006; Webb et al., 2010), Otaco karst in Namibia (Waltham et al., 2005), Guam (Mylroie et al., 2001) and Turkey (Jennings, 1985). Collapse dolines can occur in mixed lithological settings, for example, where non‑carbonate rocks (i.e., caprock) overlie karstified bedrock. The collapses can then be transmitted from interstratified carbonate karst rocks through the non-karst caprock to the surface. Such are termed caprock dolines, subjacent collapse dolines or interstratal collapse dolines (Jennings, 1966; Davey and White, 1992; Gunn, 2004; Waltham et al., 2005). Similar settings but with unconsolidated or only weakly consolidated non‑carbonate cover above the karstified bedrock can lead to formation of cover-collapse dolines or dropout dolines (Gunn, 2004; Panno and Luman, 2018). 3.2. Methods This study is based on (1) remote sensing technology and (2) field work. The overall distribution of collapse dolines was determined by analysing Google Earth satellite images in combination with digital Surface Geology of Australia (Geoscience Australia 1:1,000,000, digital dataset, 2008) for distribution of the aeolianites, and 3 arcsecond (90 m) digital elevation model (DEM) downloaded from Shuttle Radar Topographic Mission website for determining the overall topography of the area. In addition, we used the digital database of cave locations from the Western Australian Speleological Group (Inc.), including hard copies of cave maps to determine the relationship of the underground passages to the surface above. Due to increased vandalism and therefore in terms of cave protection, only the necessary location data is published within this study (Fig. 1). Fig. 2. Karst features in the research area. (A) Stockyard Gully Tunnel with sand on the floor (a), Nambung Member aeolianite (b) and Stockyard Gully Member aeolianite (c). (B) Beta Doline; note a small shelter cave on the left side of the photograph and solution pipe within the calcrete in the middle part of the photograph. (C) Delta Doline; note the circled person for scale. (D) Multiple entrance pits of the Wait Cave. (E) Epsilon Doline. (F) The collapse doline that represents the entrance to the Stockyard Cave. (G) Beekeepers Cave with exposed Stockyard Gully Member aeolianite (a), red palaeosol (b) and Pinnacles Desert Member aeolianite (c). (H) Aerial view of the Beekeepers Cave and Aiyennu Cave; note the circled car for scale. Photo courtesy (E–H) of Danny Wilkinson. M. Lipar et al. / Geomorphology 327 (2019) 572–584 Additional data is available directly from the Western Australian Speleological Group (Inc.). ESRI ArcGIS software was used to operate with and combine all the digital datasets. 575 Detailed field investigations were conducted after the remote sensing analyses, as application of the latter was limited by resolution and subterranean components of karst features. Field investigations were 576 M. Lipar et al. / Geomorphology 327 (2019) 572–584 conducted during wet and dry parts of the year to determine the occurrence and level of the water table. Detailed field mapping of the extent and characteristics of the collapse dolines in Stockyard Gully National Park was carried out using electronic surveying devices (GPS Garmin Oregon 550, laser distance meter Leica Disto, photo camera and drone). The collapse dolines` morphology was documented and analysed, and the same method was then extended to caves and other karst features nearby for comparison. For details of the lithology in the study area including the laboratory analyses used, the reader is referred to Lipar and Webb (2014, 2015, 2017) and Lipar et al. (2017); we used the lithological data published within these papers to inform the interpretation of the morphometric breakdown sequence model of collapse doline formation. To strengthen our findings, we included a review of collapse dolines and focused on different sequences of their formation. 4. Results 4.1. Characteristics of the cave system The Stockyard Gully Cave System is comprised of five main caves with separate entrances (Fig. 1); (1) Stockyard Tunnel Cave, (2) Stockyard Cave, (3) Wait Cave, (4) Aiyennu Cave, and (5) Beekeepers Cave. They are fragments of a former uniform cave passage, now separated by collapse dolines and an unroofed cave. Collapse dolines appear in an elongated cluster along the shallow Stockyard Gully Cave System and represent its surface expression in an area of 1.2 km 2 (Fig. 1). The cave system commences in the eastern part where nonkarstic allogenic water (Stockyard River) flows ephemerally into the karst landscape, forming an ~200 m long blind valley. The valley terminates at the cliffs with the entrance into the (1) Stockyard Tunnel Cave formed within the Nambung and Stockyard Gully Members of the Tamala Limestone (Fig. 2A). About 200 m to 500 m from the entrance the tunnel is fragmented by a ~300 m long unroofed section of the cave including a ~50 m long natural bridge, the Stockyard Bridge (Fig. 2F). The system then continues underground for N500 m and is known as the (2) Stockyard Cave. This cave mainly consists of a single meandering tunnel that divides at the very end into two passages. The western one terminates in the White Room cave chamber, with a pile of collapsed rocks blocking the potential continuation of the channel (Fig. 3B). Collapsed rocks and surrounding walls show no extensive weathering, indicating that collapses occurred relatively recently. The rock stratigraphy on the walls shows dune crossbedding, and some solution pipes filled with soil can be observed on the ceiling. Based on the detailed map of the cave and the location of the White Room from the surface point of view, there are no signs on the surface that would indicate a large underlying cavity, such as washing of the soil underground or slight depression in the relief. The eastern channel continues to the Alpha Doline. Collapsed rocks in the collapse doline block the continuation of the man-passable cave system, however, the cave system is accessible further in a northwest direction through vertical shafts and collapse dolines. Collapsed blocks throughout the Stockyard Cave indicate that collapses affected most of the original solutional passages. However, cave walls also contain mud sediments (including some larger logs jammed within the cave), which indicate occasional flooding and that most of the passages are at present in the epiphreatic zone. Further north to northwest, numerous small vertical voids (0.1–1 m in diameter) in two places lead into cave chambers (domes) with piles of collapsed rock on the floor. (3) Wait Cave entrance chamber (Fig. 2D) is small with only a metre or less between the ceiling and underlying collapsed rubble. (4) Aiyennu Cave has larger entrance chamber (up to 30 m deep; Fig. 3D). The rock stratigraphy on the walls indicate that the chamber of the Aiyennu Cave was originally formed within the Nambung Member or an even older member of the Tamala Limestone, and collapses occurred within the younger Stockyard Gully and Pinnacles Desert Members, separated by a thick palaeosol horizon. The uppermost Pinnacles Desert Member is relatively thin, karstified (numerous solution pipes, often enlarged to the point that pinnacle topography occurs; Fig. 3C), and case-hardened by calcrete. This very hard layer of calcrete, which is the case-hardened (palaeo)surface of the aeolianites, represents a strong contrast to the underlying softer aeolianite. The last entrance into the Stockyard Gully Cave System is through the (5) Beekeepers Cave (also known as Uniwa Cave), where explored cave passages continue for N1.400 m before becoming too narrow for further exploration (Fig. 2G). 4.2. Characteristics of collapse dolines There are six major collapse dolines which follow the main course of the subsurface discharge in the southeast to northwest direction towards the Indian Ocean and are clustered at the beginning of the allogenic inflow into the karst aquifer. The exposed walls of the collapse dolines offer excellent exposures of the limestone stratigraphy, indicating they are all formed in aeolianite capped by the calcrete layer on the surface, which correlates with the overall aeolianite deposition of the Tamala Limestone elsewhere north of Perth (Lipar and Webb, 2014; Lipar et al., 2017). None of the collapsed dolines extends below the water table level. The collapse dolines will now be described in turn, following the discharge direction of the Stockyard Gully Cave System (Fig. 1). a. Alpha Doline (Fig. 4C, D) has minor evidence of fresh collapses. Surrounding walls have partially been denuded and more than half of the slope is scree and colluvium accompanied by old collapsed rocks - most of the slopes are therefore relatively stabilised although rocky walls and scree also occur, including a small overhung shelter cave. The floor has a gully shape of collapsed blocks following the longest diameter of the doline. Its form is elliptical (NNW-SSE direction), its size is approximately 100 m × 60 m wide at the upper rim and about 10 m deep. Recently, a connection from Alpha Doline to Stockyard Gully Cave was discovered (Wilkinson, 2015) and thus confirmed its direct relation to the Stockyard Gully Cave System. The overall vegetation is abundant. b. Beta Doline (Fig. 2B) is expressed as an about 5 m deep depression with rather gentle and partly vegetated slopes. A shelter cave occurs on one side between the calcreted surface (of the Pinnacles Desert Member) and collapsed blocks beneath it. The floor is flattened and vegetated. It is slightly elliptical (NW-SE direction), approximately 40 m × 30 m wide. c. The most apparent collapse doline is the Gamma Doline (Fig. 4A, B) with recently exposed walls that belong to the Pinnacles Desert or Stockyard Gully Members and have well expressed dune cross bedding. The walls become overhung at the top where aeolianite becomes hardened due to calcrete formation. The slopes beneath the wall and overhangs are scree of collapsed blocks and sand. Collapsed rocks cover the floor of the doline with scarce moss and grass vegetation. The Gamma Doline is circular, approximately 25 m in diameter and about 12 m deep. The scarce vegetation and fresh rock exposures indicate recent and ongoing collapses. d. Delta Doline (Fig. 2C) is expressed as a gentle depression with cracked and collapsed top layer of the strongly cemented calcrete part of the Pinnacles Desert Member. The collapsed blocks are partially tumbled on the western side, whilst on the eastern side, the cracked surface layer was only lowered to form a slope towards the centre of the collapse doline. On the edge of the steep slope side with tumbled rocks is a calcrete overhang, known as Seismic M. Lipar et al. / Geomorphology 327 (2019) 572–584 577 Fig. 3. Aerial and inside view of typical examples for the first and second stage of collapse doline formation with relatively strong removal of collapsed material (Fig. 5). (A & B) The White Room – cave dome; note dune cross-bedding on the wall. (C & D) Aiyennu Cave – calcrete caprock dome; note the circled person on the rope for a scale (a) Stockyard Gully Member aeolianite, (b) loose palaeosol, (c) Pinnacles Desert Member aeolianite. Photo courtesy (A, C) of Danny Wilkinson. Cave or Facts-of-Life Cave, which represents a small rock shelter, leading some metres further down the collapse slope. Vegetation is mostly absent. It is generally circular, approximately 20 m in diameter and up to 3 m deep. e. Epsilon Doline (Fig. 2E) has partially to non-vegetated walls with some blocks of collapsed calcrete overhang beneath them on one side and a vegetated scree slope on the other. The floor is mostly flattened and vegetated. It is a slightly elliptical depression (N–S direction), approximately 30 m × 40 m wide and 6 m deep. There are no known connections accessible for a caver to reach the underground cave. f. The Beekeepers Cave Collapse Doline is the entrance to the eponymous cave (Fig. 2G, H). The collapse doline has up to 6 m high vertical to nearly vertical walls with vegetated colluvium and scree slopes beneath it, and gully-shaped elongated floor. It is elliptical (NNW to SSE), approximately 60 m × 45 m wide and up to 10 m deep. It has one apparent overhang with relatively freshly exposed limestone, however, it is believed that this part has been artificially affected by a blast to eliminate the bee hives. Active water flow was observed during extreme high-stands approximately four metres beneath the cave opening at the bottom of the doline. 5. Discussion 5.1. Formation of collapse dolines Syngenetic karst (Jennings, 1968) (see Section 2) is generally defined as a karst in a porous rock with matrix porosity, lack of joint/bedding control and slow water flow, resulting in horizontal cave systems of low, wide, irregular, and interconnected chambers and passages (i.e., syngenetic maze caves) (Grimes, 2006). However, the Stockyard Gully Cave System resembles linear stream cave morphology. Such morphology may be a result of concentrated water flow due to steep gradients (Grimes, 2006), less permeable and/or non‑carbonate 578 M. Lipar et al. / Geomorphology 327 (2019) 572–584 Fig. 4. Aerial and inside view of classic examples for the third and fourth stage of collapse doline formation with relatively strong removal of collapsed material (Fig. 5). (A & B) The Gamma Doline – young collapse doline. (C & D) The Alpha Doline – mature collapse doline. Photo courtesy (A, C) of Danny Wilkinson. palaeo-topography (Mylroie et al., 2001; Bastian, 2003; Grimes, 2006), or secondary porosity, triggered by fissures (Whitaker and Smart, 1997; Stafford et al., 2005). Although the water flow may to some degree be controlled by impervious stratum (Bastian, 1964), the absence of a steep gradient and the absence of shallow palaeo-topography favour the possible causes of linearity in the Stockyard Gully Cave System to be mainly fissure control. This was confirmed by field observations in the caves and collapse dolines, which are formed along evident fractures in the bedrock and mostly oriented NNW-SSE. Evident fissures and fractures within collapsed chambers and collapse dolines in otherwise high porous aeolianites were described also by Hill (1984). Neotectonics could have contributed to their occurrence as it has been noted in this area (Playford et al., 1976; Mory, 1994; Clark, 2010; Clark et al., 2012), including earthquakes recorded in South-west Seismic Zone of Western Australia (McCue, 1990; Clark, 2010). Although the relationship between seismicity, faults and present-day tectonics are still poorly understood (Clark, 2010), the observed fissures indicate a combination of weak cementation, matrix porosity and strong permeability on one hand, and the fissure control on the other. As for collapse dolines in carbonate bedrock within Dinaric karst (Stepišnik, 2010) and other collapse dolines worldwide (Waltham et al., 2005), the collapse doline formation in Stockyard Gully Cave System is therefore clearly related to caves formed by high discharge channelled groundwater flow through the aquifer, controlled by fissures (i.e., by rock characteristics). The occurrence of collapse dolines in syngenetic karst is not unusual and most of the accessible caves in syngenetic karst have at least partially been affected by underground collapses (Bastian, 1964; White, 1994; Grimes, 2006). Mylroie and Vacher (1999) postulated that collapse dolines seem more common in coastal karst and composite islands than in continental karst developed on well-cemented limestone. Two components are crucial for this fact: (1) a relatively shallow epiphreatic zone of cave formation (e.g. similarly described in Šušteršič (2002)), and (2) mechanical weakness of the rock in comparison to wellcemented limestones in classic karst. The mechanical weakness is exacerbated in the study area by palaeosol layers. As described before (Table 1), palaeosols can be weakly cemented or loose, so the overall mechanical strength of the rock is reduced (see Figs. 2G & 4D where a loose palaeosol “cuts” horizontally through the Tamala Limestone). M. Lipar et al. / Geomorphology 327 (2019) 572–584 Removal of the weakly cemented or loose palaeosol material forms notches in the walls of the collapse dolines. In addition, the palaeosol horizons appear to represent the level to which the upward stoping progresses relatively rapidly with a narrow parabolic profile. Further stoping above the palaeosol horizon is slower with a wider parabolic profile, as the overlying material is better cemented and mechanically more resistant (Fig. 3D). The volumes of large collapse dolines along the Stockyard Gully Cave System generally exceed the volumes of the passages. Therefore, formation of large collapse dolines cannot be related solely to cave collapses, but also to gradual removal of breakdown material within hydrologically active cave passages below collapse doline floors (Habič, 1963; Mihevc, 2001; Stepišnik, 2004, 2007). The collapse dolines occur in greater numbers and are more apparent in the south-eastern part of the research area, where concentrated water flow enters the karst. Ponor or spring areas with channelled groundwater flow are generally favourable for collapse doline formation (Waltham et al., 2005) due to the higher potential of solution and mechanical erosion to remove collapsed material, causing gradual undermining beneath fractured zones (Stepišnik, 2007). Larger cave chambers and collapse dolines were formed where the removal was efficient (e.g., Figs. 3B & 5B), and only minor depressions were formed where the removal was absent or not efficient (e.g., Fig. 2C). This could have happened due to blocked underground water flow that diverted and left the collapse section. All collapse dolines in the research area are strongly influenced by the calcrete layer capping the aeolianites. As stated before, it is well cemented and more resistant to mechanical weathering than underlying weakly cemented rock. This results in a delay of collapse of the 579 uppermost layer (i.e., calcrete) into the underlying dome. The best example is the entrance chamber of the Aiyennu Cave (Fig. 3C, D). When the final collapse of the surface calcrete layer occurs, the calcrete layer remains overhung in regards to the underlying rock strata (clearly seen within the Gamma Doline (Fig. 4A, B)). This is similar to collapse dolines where calcrete is present, including Pleistocene calcarenites of the southern Italian coast (e.g., Grotte della Poesia; Delle Rose and Parise, 2004b), and banana holes, karst features of (but not limited to) the Bahamas (Harris et al. (1995). The latter are small circular to oval voids with common vertical or overhung walls due to the calcrete, mostly b12 m in diameter and b4 m deep (Waltham et al., 2005). The initial shallow phreatic dissolution was followed by a direct collapse of the thin roof, which differentiates them from classic collapse dolines formed by progradational collapses from depth. The difference in hardness of the rock layers results in a form that resembles caprock dolines (Ford and Williams, 2007), but caprock dolines form by collapse breaching consolidated non-karst caprock. Thus, caprock dolines are surface depressions in insoluble rocks due to collapse into a cavity in underlying soluble rocks, e.g. in intrastratal karst settings (Klimchouk, 2013). However, calcrete caprock is calcareous and consequently karstic; it is a part of the epikarstic vadose zone, hydrologically connected and part of the overall karst system from the beginning of karstification, as indicated by soil-filled solution pipes on the ceiling of a dome. Besides the Stockyard Gully area, collapse dolines appear elsewhere in Western Australia within aeolianites and calcarenites, including in the wetter Margaret River area to the south, where they are even more common than in the north (Jennings, 1968). They also occur in southeastern Australia (Grimes, 2006), and on Eyre Peninsula, where Fig. 5. A diagram illustrating four stages of collapse doline formation in Stockyard Gully with relatively strong removal of collapsed material (see detailed descriptions of each stage in Section 5.2). 580 M. Lipar et al. / Geomorphology 327 (2019) 572–584 their main genesis (beside a presence of minor recent collapses) is suggested to be due to solution from below guided by passive fractures in the underlying basement rocks (underprinting) (Twidale and Bourne, 2000). They have also been reported from southern Italy (Delle Rose and Parise, 2004a), Guam (Mylroie et al., 2001) and Cyprus (Harrison et al., 2002). In all these cases, the collapse dolines are not big enough to define them as tiankengs (Waltham, 2006). However, some of the collapse dolines reach below the water table (or below the sea) (Mylroie et al., 2001; Delle Rose and Parise, 2004b) and can be in this case termed cenotes. 5.2. Multiphase breakdown sequence The most common subdivision amongst various published morphographic sequences of collapse doline formation includes three basic stages: young, mature and degraded (Habič, 1963; Šušteršič, 1984; Summerfield, 1996; Waltham et al., 2005). Young collapse dolines have almost vertical walls and debris within their floors, which might have cave opening(s). During the mature stage, the slopes degrade and are consequently less steep, whilst the floor is covered by weathered colluvium. Cave entrances of this stage are uncommon. The final or degraded stage of collapse doline formation is a rounded planshape hollow with no rock walls. The floor is usually covered by fine grained material of local weathered material or by allogenic alluvium (Waltham et al., 2005; Stepišnik, 2011). However, collapse doline morphology and the dynamics of morphogenesis is a result of the balance between various geomorphic processes that vary due to local geologic, hydrologic and climatic settings. Kaufmann (2014), for example, showed a sequence of evolution of a collapse doline in the setting where karstic rock is overlain by a non-karstic layer. It begins with the formation of a large cave void along the fractures and bedding within karstic rocks. Due to weaker strength along fault zones, the cave passage starts to collapse. Collapsing progresses upwards, whilst the collapsed rubble is removed by dissolution or mechanical erosion. The breakdown eventually reaches the insoluble layer, and the collapsed rubble starts accumulating as it cannot be dissolved and can only be partially removed by mechanical erosion. The progressive collapses finally cause a surface breakdown. In addition, different characteristics of the carbonate strata itself can influence the collapse doline formation, for example, the differences in solubility between two different layers may emphasise the dimensions of collapse dolines. Szczygieł et al. (2018) showed four stages of the development of tiankengs, where less soluble Ordovician strata on the surface result in concentrated and high-volume inflow, which in turn results in large volume caves in the lower more soluble Cambrian strata. Another example of the sequence of collapse doline formation was shown by Doğan and Özel (2005) in gypsum karst area east of Hafik, Turkey. They focused on processes that reshape the collapse doline once it is formed. The first stage is a fresh collapse of a cavity and the initial formation of the collapse doline, where morphometrics are mostly dependent on the size of the underlying cavity. The rim is generally not circular and the floors are covered with piles of collapsed blocks. The second stage is characterised by undercutting of doline slopes and further enlargement of a collapse doline rim. The rim is still not circular and at least one side of the collapse doline remains steep. The circular rim is reached in the third stage, followed by further retreat of the sides in the last stage, causing their diameter to increase and slope steepness to decrease. The most relevant example for the formation of the collapse dolines studied in this paper was discussed by Hill (1984). He noticed differences in morphology of collapsed cave passages between aeolianites (studied in the coastal area of Kangaroo Island, South Australia) and “more usual, dense, bedded” limestones, and described the mechanics of upward propagation of collapses. However, the collapse dolines differ from those in Stockyard Gully in the lack of calcrete caprock and the absence of strong removal of the collapsed material, resulting in the formation of inclined fissure caves on Kangaroo Island (Bastian, 1964; Jennings, 1968; Grimes, 2006). 5.2.1. Multiphase breakdown sequence in Stockyard Gully A single proposed sequence is therefore usually limited to certain climatic, geologic or hydrologic settings. The variable morphology of collapse dolines in Stockyard Gully reflects their maturity and consequently their different stages of formation, together with the influence of the lithology (weakly cemented aeolianite overlain by well cemented calcrete). The sequence of doline formation can be best presented as a 4stage multiple breakdown sequence: cave dome, calcrete caprock dome, young collapse doline, and mature collapse doline. We present two examples of collapse doline formation that occur in the Stockyard Gully: a sequence where collapses and removal of the collapse material are both active (Fig. 5), and a sequence where consequent removal of the material was minimal which led to different, less apparent forms (Fig. 6). (1) Cave dome: The formation of a cave dome. Once the collapse is initiated, the dynamics of cave dome formation is relatively rapid due to the weak cementation of the rock. This process continues until the dome reaches a shape that is structurally stable (Hill, 1984). It is not visible on surface (Fig. 3A), however, hydrological connection may be noticeable in the form of soil-filled solution pipes on the ceiling of a dome. Soil-filled solution pipes are generally not visible on the surface as they are overgrown by vegetation. Although a dome itself may develop a stable arch and cease enlarging (as described by Šušteršič (2000) and Ford and Williams (2007)), the relatively shallow cave system at Stockyard Gully favours further development towards the surface. If active removal of the collapsed material is present, a large spacious dome can form such as the White Room in Stockyard Gully Cave (Fig. 3B). Due to lower mechanical strength of the rock, these domes have higher parabolic profiles than domes in harder well-cemented limestones which have lower arches (Šušteršič, 1994; Koritnik and Šušteršič, 2000; Waltham et al., 2005). However, if the removal of the collapsed material is not active, only narrow inclined fissure type caves develop between the collapsed material and the dome ceiling. (2) Calcrete caprock dome: The formation of a dome is often stabilised by the uppermost well cemented and generally 0.5 m to 2 m thick layer of calcrete. At this stage, the upward stoping almost reaches the surface and it represents a stage before the final collapse of a surface layer to form an actual collapse doline. However, the shallow underlying chamber allows wash-off of the soil through epikarstic surface karst voids (e.g. solution pipes) into the cavern, resulting in exposed rocks and stripped vegetation (Fig. 3C). The wash-off is more apparent above large domes where removal of the collapse material is active. The best example is Aiyennu Cave (Fig. 3C, D), which is basically a dome just before becoming a collapse doline; the uppermost (surface) rock layer is at present still in place, but is already connected through emptied solution pipes and other vertical voids into the underlying collapse chamber of the cave. However, if the removal of the collapsed material is less active, the wash-off of the soil may be reduced as well (but not absent), and relatively smaller caverns (in volume, not in length) can be reached through the emptied solution pipes from the surface. One such example is Wait Cave (Fig. 2D) that appears ~150 m east from Aiyennu Cave where underground water flow had less impact on the dissolution and erosion of the collapsed material. (3) Young collapse doline: The span of the calcrete caprock roof weakens by progressive lateral undercutting from beneath as well as by dissolution from above. A final collapse of the uppermost layer (i.e., calcrete) results in a relatively large surface depression – a collapse doline. Gamma Doline (Fig. 4A, B) is the M. Lipar et al. / Geomorphology 327 (2019) 572–584 581 Fig. 6. A diagram illustrating four stages of collapse doline formation in Stockyard Gully with relatively weak removal of collapsed material (see detailed descriptions of each stage in Section 5.2). best example of a young collapse doline in the area (also noted by Jennings (1968)). The collapsed blocks of the calcrete containing abundant solution pipes are visible on the floor of the doline, whilst the walls are, due to the calcrete around the rim, still overhanging – a result of different mechanical strength of the rock within the strata. Vegetation is mostly absent due to ongoing recession of the walls. The rocky floor with a conical shaped depression in accumulated talus (Fig. 4B) indicates continuous removal of material above active cave passages. Where the removal of material is negligible or absent, only a shallow collapse or subsidence of the calcrete layer occurs, as best portrayed in the Delta Doline (Fig. 2C). (4) Mature collapse doline: Alpha Doline (Fig. 4C, D) would best describe a mature collapse doline with a past of active removal of the collapsed material. It has rare evidence of fresh collapses and abundant scree and colluvium as well as abundant vegetation. These collapse dolines have the steepest part at the top of the rim, which is not typical for mature collapse dolines in a classic karst (Stepišnik, 2010). At that stage, the slopes become stabilised, the floor becomes filled with weathered colluvium – similar to other collapse dolines worldwide. On the other hand, if the removal of the collapsed material was not active and only minimal collapses or subsidence occurred in a young collapse doline stage, the mature collapse doline becomes rapidly overgrown and is often only a minor noticeable depression. 5.3. Timing of collapse doline formation As discussed in Section 5.2.1, the collapse dolines in Stockyard gully are in different stages of their formation, so they did not all form at the same time. Examples of collapse dolines with increased diameter and depth due to active removal of the collapsed material as well as collapse dolines where the removal of collapsed material has been minimal, indicate that underground water flow has been diverted in certain collapsed places. In this case, the underground water began forming additional voids in the epiphreatic zone, which later became a subject of “secondary” (i.e., younger) collapses. This is, for example, evident in the Stockyard Gully Cave System section between Alpha Doline (collapse doline) and White Room (cave dome). Alpha Doline formed above the previous main cave passage, which was hydrologically active enough to remove a substantial amount of collapsed material and form a relatively large collapse doline. However, the collapsed rubble eventually blocked the main phreatic/epiphreatic void, so the main water flow diverted eastwards (see Fig. 1 for locations of the dome and the collapse doline), causing a new situation where solutional enlargement of voids by diverted water caused much younger collapses to occur, forming the White Room. A similar processes of underground water migration were described by Šušteršič (2002). On-going diversion of main underground water flow therefore produces collapsed dolines of different ages. To discuss timing of collapse doline formation in Stockyard Gully National Park, we refer to the overall process of initial solution followed by collapses. The final collapse which causes a cave dome to become a collapse doline is an instantaneous event, but understanding of previous processes is important, especially for modelling and future prediction of possible collapses (Kaufmann and Romanov, 2016; Xu et al., 2017). The commencement of solution (i.e., cave formation) can be constrained by aeolianite ages. Deposition of aeolianites in the Stockyard Gully area occurred during Quaternary interglacial cycles (Lipar et al., 2017). Examination of exposed walls of collapse dolines and caves in Stockyard Gully showed that three units are present: Pinnacles 582 M. Lipar et al. / Geomorphology 327 (2019) 572–584 Desert Member (deposited in MIS 7; ages of the marine isotope stages (MIS) are from Lisiecki and Raymo (2005)), overlying the Stockyard Gully Member (MIS 9) and Nambung Member (MIS 11). The initial karst voids of the Stockyard Gully Cave System would therefore have formed after the Nambung Member had become sufficiently lithified to support the voids, probably in MIS 11. The solution process has since occurred periodically with interruptions during glacial periods when the climate was drier (Lipar et al., 2017). During the lower sea level during glacial periods, the water table in Stockyard Gully probably dropped and the resultant loss of buoyancy (alongside the weakly cemented aeolianite) would cause the first prominent roof collapses. Collapses during sea-level regressions which resulted in the loss of buoyant support for the rock were also reported by Mylroie and Vacher (1999), Waltham et al. (2005) and Ford and Williams (2007). The wetter climate during interglacial and especially transitional periods (Lipar et al., 2017) would contribute to effective removal of the collapsed material, so the cave chambers were able to grow in size. Nevertheless, present-day surface exposure of collapses happened after the deposition of the uppermost surface Pinnacles Desert Member aeolianite in MIS 7. Furthermore, collapsed blocks of this member on the floors of collapse dolines indicate their collapse after the rock experienced calcretisation and karstification to form solution pipes and pinnacles (the main period of pinnacle formation occurred in MIS 5; Lipar et al., 2017). It is therefore probable that the most mature collapse dolines in the research area were formed around 100 ka or later. Four speleothems were dated in caves within the same geological setting between 70 km and 200 km south of Stockyard Gully, yielding ages from 13.8 to 8.4 ka. Two speleothems were also dated in a cave system ~15 km to the north of Stockyard Gully; one yielded an age of 3.1 ± 0.2 ka (U/Th; 2σ uncertainty), the other, collected at the inner edge of the mature collapse doline entrance, could not be dated due to a very low U/Th ratio (Lipar et al., 2017). If the speleothem in the collapse doline entrance is the same age as other speleothems in the area, the collapse doline probably already existed around 14 ka, and is certainly older than 3 ka. We can therefore assume that the most mature collapse dolines in the Stockyard Gully research area are certainly older than 3 ka. Nevertheless, the youngest possible ages for other (active, early) collapse dolines are irrelevant as they are still forming (enlarging) at present, evident in the active sediment removal at the bottom of the dolines or underlying caves. Weak cementation of the rock and consequently accelerated time for cavern development causes their formation to be relatively faster than a collapse doline of an 200 m × 200 m area in classic karst, where lowering of the surface by 100 m was demonstrated by a numerical model in a crushed zone (i.e., combination of solution and collapse) to last around 1 million years (Gabrovšek and Stepišnik, 2011), but still slower than, for example, in salt, where the formation of large cavities and subsequent surface collapses may appear within only months or years (Waltham et al., 2005). 5.4. Applied knowledge Collapse dolines commonly represent a hazard in urban areas built on porous soluble rocks prone to karstification (Karimi and Taheri, 2010; Gutiérrez et al., 2014). We discussed the formation of a calcrete caprock dome, stating that it represents a momentarily stabilised dome by the uppermost well cemented calcrete, yet only one more final sudden collapse is needed to become a surface opening, i.e., a collapse doline. This indicates that inadequate spatial planning (as described by Zorn and Komac (2015)) may be hazardous in areas where surface calcrete covers underlying karst rock, and tools with ground penetrating technology should be used as part of urban planning, as upward propagation of underground domes could already have reached the final stage before the collapse of the calcrete layer (as seen in the case of Aiyennu Cave (Fig. 3D)). In addition, the stability of underground passages is greater when they are filled with water due to the greater density of water compared to air. We discussed the beginning of accelerated formation of collapse dolines during glacial periods, when the sea surface was lower and consequently the water table within the coastal aeolianites dropped. Collapse therefore happens, when a flooded passage that has been enlarged beyond the mechanical stability limit in air, is suddenly drained. In accordance with buoyancy support, we can assume that such hazard becomes greater in cases where water table gradually drops (Legchenko et al., 2008), which may be the important case in densely urbanised areas built on karst rocks in a relatively dry climate where water is scarce and pressure on groundwater is high. On the other hand, collapse dolines in Stockyard Gully National Park demonstrate an excellent example of how useful they are in terms of deciphering the local evolution of aeolianite landscape and palaeoclimate (Lipar, 2018): the exposed aeolianite walls on their sides including associated features (e.g., calcrete and solution pipes) are affected by building agents (deposition, lithification and diagenesis) and erosional agents (physical, chemical and biological weathering), which are driven and dependent on climate (rainfall, temperature and wind) and can therefore serve as palaeoclimate indicators. In addition, collapse dolines also demonstrate how fragile and prone to changes they are in terms of climatic change; and how vulnerable the groundwater is to pollution as it can be accessed by so many enlarged karst conduits and/or collapsed zones without proper filtering (see also Kačaroğlu (1999), Boulton et al. (2003), van Beynen and van Beynen (2011), and Eberhard and Davies (2011) for further reading). Protecting the natural values of collapse dolines as well as karst landscapes in general is therefore of great importance for sustainable (living) environment. 6. Conclusion Karst in Quaternary aeolianites is particularly well exposed in Stockyard Gully National Park, southwestern Western Australia. The main Stockyard Gully Cave System is largely collapsed, and consequently numerous collapse dolines are present with variety of forms representing their different stages of development. The main difference to classical collapse dolines formed in diagenetically mature limestones is the presence of a mechanically resistant layer of calcrete, which influences collapse doline morphogenesis. The variable morphology of collapse dolines in Stockyard Gully reflects their maturity and consequently their different stages of formation. We present a 4-stage sequence of collapse doline formation with two possible variations based on effectiveness of the removal of collapsed material. The sequences embrace overall processes from initial underground collapses (stage 1 — cave dome, and stage 2 — calcrete caprock dome) to surface exposure as enclosed depressions (stage 3 — young collapse doline, and stage 4 — mature collapse doline). The age of different aeolianite members and speleothems enable us to constrain the genesis of mature collapsed dolines to between 100 and 3 ka, which contrasts with much longer time of collapse doline genesis on classic karst. Acknowledgements We wish to thank members of the Western Australian Speleological Group (Inc.) for the agreement to use their reference sources and for organising and leading cave trips required for this research. A permit to conduct research in national parks as well as to enter certain caves or cave passages was obtained from the Western Australia Department of Parks and Wildlife. The field work was financially supported by Australian Speleological Federation Karst Conservation Fund. Part of the research was carried out whilst M. Lipar and M. Ferk were supported by the European Regional Development Fund: European Union & Republic of Slovenia, Ministry of Education, Science and Sport M. Lipar et al. / Geomorphology 327 (2019) 572–584 (2017–2020; research programme OP20.01261) and Slovenian Research Agency (research programme P6-0101). The Golden Open Access of this paper is funded by the Research Centre of the Slovenian Academy of Sciences and Arts (ZRC SAZU). The authors thank Danny Wilkinson for permission to use his collection of aerial drone photos for the purpose of this research, Lex Bastian, Robert Susac and Blaž Komac for comments on the early draft of the text, and two anonymous reviewers for comments and suggestions, which substantially improved the manuscript. References Abotalib, A.Z., Sultan, M., Elkadiri, R., 2016. Groundwater processes in Saharan Africa: implications for landscape evolution in arid environments. Earth Sci. Rev. 156, 108–136. Bastian, L., 1964. Morphology and development of caves in the southwest of Western Australia. Helictite 2 (4), 105–119. Bastian, L., 2003. Hydrogeology and speleogenesis update, the Yanchep cave area, Western Australia. Under way 2003. 24th Biennial Conference of the ASF. Karst Studies Presentations, pp. 36–44. Beddows, P.A., 2006. Yucatan Phreas, Mexico. Encyclopedia of Caves and Karst Science. Fitzroy Dearborn, New York, pp. 1677–1682. Benito, G., del Campo, P.P., Gutiérrez-Elorza, M., Sancho, C., 1995. Natural and humaninduced sinkholes in gypsum terrain and associated environmental problems in NE Spain. Environ. Geol. 25 (3), 156–164. Boulton, A.J., Humphreys, W.F., Eberhard, S.M., 2003. Imperilled subsurface waters in Australia: biodiversity, threatening processes and conservation. Aquat. Ecosyst. Health Manag. 6 (1), 41–54. Bureau Of Meteorology, 2018. Climate Statistics for Australian Locations (Australian Government). Clark, D., 2010. Identification of Quaternary scarps in southwest and central west Western Australia using DEM-based hill shading: application to seismic hazard assessment and neotectonics. Int. J. Remote Sens. 31 (23), 6297–6325. Clark, D., McPherson, A., Van Dissen, R., 2012. Long-term behaviour of Australian stable continental region (SCR) faults. Tectonophysics 566–567, 1–30. Coke, J.G., 2012. Underwater caves of the Yucatán Peninsula. In: White, W.B., Culver, D.C. (Eds.), Encyclopedia of Caves, Second edition Academic Press, Amsterdam, pp. 833–838. Conservation Commission of Western Australia, 2014. Performance assessment – Nambung National Park management plan 1998–2008, Crawley. Cvijić, J., 1893. Das Karstphänomen. Geographischen Abhandlungen Wien 5 (3), 218–329. Davey, A.G., White, S., 1992. An unusual subjacent karst doline at east Buchan, Victoria. Helictite 30 (2), 41–43. Delle Rose, M., Parise, M., 2004a. Slope instability along the Adriatic coast of Salento, southern Italy. In: Lacerda, W.A., Ehrlich, M., Fontoura, S.A.B., Sayão, A.S.F. (Eds.), Proceedings of IX International Symposium on Landslides, Rio de Janeiro (Brasil), pp. 399–404. Delle Rose, M., Parise, M., 2004b. Speleogenesi e geomorfologia del sistema carsico delle grotte della Poesia nell'ambito dell'evoluzione quaternaria della costa adriatica salentina. Atti e memorie della Commissione grotte Eugenio Boegan. 40, pp. 153–173. Doğan, U., Çiçek, İ., 2002. Occurrence of cover-collapse sinkholes [cover-collapse dolines] in the May Dam reservoir area (Konya, Turkey). Cave and Karst Science 29 (3), 111–116. Doğan, U., Özel, S., 2005. Gypsum karst and its evolution east of Hafik (Sivas, Turkey). Geomorphology 71 (3), 373–388. Doğan, U., Yılmaz, M., 2011. Natural and induced sinkholes of the Obruk Plateau and Karapınar-Hotamış Plain, Turkey. J. Asian Earth Sci. 40 (2), 496–508. Eberhard, S.M., Davies, S., 2011. Impacts of drying climate on aquatic cave fauna in Jewel Cave and other caves in southwest Western Australia. Journal of the Australasian Cave & Karst Management Association 83, 6–13. Ferk, M., 2011. Morphogenesis of Rakov [kocjan Karst Basin. Geogr. Bull. 83 (1), 11–25. Ferk, M., 2016. Palaeofloods in karstic Ljubljanica River Catchment. Geografija Slovenije. 33. Založba ZRC, Ljubljana, p. 187. Ferk, M., Stepišnik, U., 2011. Geomorphological Characteristics of Rakov Škocjan. Georitem. 17. Založba ZRC, Ljubljana, p. 70. Florea, L.J., Vacher, H.L., Donahue, B., Naar, D., 2007. Quaternary cave levels in peninsular Florida. Quat. Sci. Rev. 26 (9–10), 1344–1361. Ford, D., Williams, P., 2007. Karst Hydrogeology and Geomorphology. John Wiley & Sons Ltd, Chichester. Gabrovšek, F., Stepišnik, U., 2011. On the formation of collapse dolines: a modelling perspective. Geomorphology 134 (1–2), 23–31. Gams, I., 2004. Kras v Sloveniji - v prostoru in času. Založba ZRC, Ljubljana. Grimes, K.G., 2006. Syngenetic karst in Australia: a review. Helictite 39 (2), 27–38. Gunn, J., 2004. Encyclopedia of Caves and Karst Science. Fitzroy Dearborn, New York, London. Gutiérrez, F., Guerrero, J., Lucha, P., 2007. A genetic classification of sinkholes illustrated from evaporite paleokarst exposures in Spain. Environ. Geol. 53 (5), 993–1006. Gutiérrez, F., Parise, M., De Waele, J., Jourde, H., 2014. A review on natural and humaninduced geohazards and impacts in karst. Earth Sci. Rev. 138, 61–88. Habič, P., 1963. Udarne vrtače koliševke in podzemski tokovi. 3rd Yugoslavian Speleological Congress, Sarajevo, pp. 125–129. 583 Habič, P., Knez, M., Kogovšek, J., Kranjc, A., Mihevc, A., Slabe, T., Šebela, S., Zupan, N., 1989. Škocjanske Jame speleological revue. Int. J. Speleol. 18, 1–42. Harris, J.G., Mylroie, J.E., Carew, J.L., 1995. Banana holes - unique karst features of the Bahamas. Carbonates Evaporites 10 (2), 215–224. Harrison, R.W., Newell, W.L., Necdet, M., 2002. Structure and genesis: karstification along an active fault zone in Cyprus. In: Kuniansky, E.L. (Ed.), U.S. Geological Survey Karst Interest Group Proceedings. Water-Resources Investigations Report 02-4174. U.S. Geological Survey, Shepherdstown, West Virginia, pp. 45–48. Hartmann, A., Baker, A., 2017. Modelling karst vadose zone hydrology and its relevance for paleoclimate reconstruction. Earth Sci. Rev. 172, 178–192. Hill, A.L., 1984. The origin of the Kelly Hill Caves, Kangaroo Island, S.A. Helictite 22 (1), 6–10. Hiller, T., Romanov, D., Gabrovsek, F., Kaufmann, G., 2014. The creation of collapse dolines: a 3D modeling approach. Acta Carsologica 43 (2–3), 241–255. Jennings, J.N., 1966. Solution doline and subjacent karst doline. Aust. Geogr. 10 (2), 132–133. Jennings, J.N., 1968. Syngenetic karst in Australia. In: Williams, P.W., Jennings, J.N. (Eds.), Contributions to the Study of Karst. The Australian National University, pp. 41–110. Jennings, J.N., 1985. Karst Geomorphology. Basil Blackwell, Oxford. Kačaroğlu, F., 1999. Review of groundwater pollution and protection in karst areas. Water Air Soil Pollut. 113 (1/4), 337–356. Karimi, H., Taheri, K., 2010. Hazards and mechanism of sinkholes on Kabudar Ahang and Famenin plains of Hamadan, Iran. Nat. Hazards 55 (2), 481–499. Kaufmann, G., 2009. Modelling karst geomorphology on different time scales. Geomorphology 106 (1), 62–77. Kaufmann, G., 2014. Geophysical mapping of solution and collapse sinkholes. J. Appl. Geophys. 111, 271–288. Kaufmann, G., Romanov, D., 2016. Structure and evolution of collapse sinkholes: combined interpretation from physico-chemical modelling and geophysical field work. J. Hydrol. 540, 688–698. Kendrick, G.W., Wyrwoll, K.-H., Szabo, b.J., 1991. Pliocene-Pleistocene coastal events and history along the western margin of Australia. Quat. Sci. Rev. 10, 419–439. Klimchouk, A.B., 2013. Evolution of intrastratal karst and caves in gypsum. In: Frumkin, A. (Ed.), Karst Geomorphology. Treatise on Geomorphology, pp. 438–450. Kobal, M., Bertoncelj, I., Pirotti, F., Dakskobler, I., Kutnar, L., 2015. Using lidar data to analyse sinkhole characteristics relevant for understory vegetation under forest cover-case study of a high karst area in the dinaric mountains. PLoS One 10 (3), e0122070. Kohl, M.S., 2001. Subsidence and Sinkholes in East Tennessee: A Field Guide to Holes in the Ground. Tennessee Division of Geology. Koritnik, J., Šušteršič, F., 2000. Modelling the stability of a very large cave chamber; case study: Brezno pri Medvedovi konti. Acta Carsologica 29 (2), 149–160. Kranjc, A., 2013. Classification of closed depressions in carbonate karst. In: Shroder, J.F. (Ed.), Treatise on Geomorphology. Elsevier, pp. 104–111. Legchenko, A., Ezersky, M., Boucher, M., Camerlynck, C., Al-Zoubi, A., Chalikakis, K., 2008. Pre-existing caverns in salt formations could be the major cause of sinkhole hazards along the coast of the Dead Sea. Geophys. Res. Lett. 35 (19), 1–5. Lipar, M., 2018. A combined approach to the study of Quaternary aeolian calcarenites for understanding the palaeoclimate. Geogr. Bull. 90 (1). https://doi.org/10.3986/ GV90103. Lipar, M., Ferk, M., 2011. Eogenetic caves in conglomerate: an example from Udin Boršt, Slovenia. Int. J. Speleol. 40 (1), 53–64. Lipar, M., Webb, J.A., 2014. Middle–Late Pleistocene and Holocene chronostratigraphy and climate history of the Tamala Limestone, Cooloongup and Safety Bay Sands, Nambung National Park, southwestern Western Australia. Aust. J. Earth Sci. 61 (8), 1023–1039. Lipar, M., Webb, J.A., 2015. The formation of the pinnacle karst in Pleistocene aeolian calcarenites (Tamala Limestone) in southwestern Australia. Earth Sci. Rev. 140, 182–202. Lipar, M., Webb, J.A., 2017. Speleogenesis in Stockyard Gully National Park and Beekeepers Nature Reserve, Western Australia. In: Moore, K., White, S. (Eds.), 17th International Congress of Speleology. Australian Speleological Federation Inc & Speleo2017 & International Union of Speleology, Sydney, NSW, Australia, pp. 226–229. Lipar, M., Webb, J.A., White, S.Q., Grimes, K.G., 2015. The genesis of solution pipes: evidence from the Middle–Late Pleistocene Bridgewater Formation calcarenite, southeastern Australia. Geomorphology 246, 90–103. Lipar, M., Webb, J.A., Cupper, M.L., Wang, N., 2017. Aeolianite, calcrete/microbialite and karst in southwestern Australia as indicators of Middle to Late Quaternary palaeoclimates. Palaeogeogr. Palaeoclimatol. Palaeoecol. 470, 11–29. Lisiecki, L.E., Raymo, M.E., 2005. A Pliocene–Pleistocene stack of 57 globally distributed benthic δ18O records. Paleoceanography 20, 1–17. Marker, M.E., 1976. Cenotes: a class of enclosed karst hollows. Z. Geomorphol. 26, 104–123. McCue, K., 1990. Australia's large earthquakes and recent fault scarps. J. Struct. Geol. 12 (5–6), 761–766. Mihevc, A., 2001. Speleogeneza Divaškega krasa. Založba ZRC, ZRC SAZU, Ljubljana. Mihevc, A., 2010. Geomorphology. In: Mihevc, A., Prelovšek, M., Zupan Hajna, N. (Eds.), Introduction to Dinaric Karst. IZRK ZRC SAZU, Postojna, pp. 30–43. Mory, A.J., 1994. Geology of the Hill River - Green Head 1:100000 Sheet. Geological Survey of Western Australia, Perth. Mory, A.J., 1995. Geology of the Wedge Island 1:100000 Sheet. Geological Survey of Western Australia, Perth. Mylroie, J.E., Vacher, H.L., 1999. A conceptual view of carbonate island karst. In: Palmer, A.N., Palmer, M.V., Sasowsky, I.D. (Eds.), Karst Modelling. Karst Waters Institute, Charles Town, WV, pp. 48–57. 584 M. Lipar et al. / Geomorphology 327 (2019) 572–584 Mylroie, J.E., Jenson, J.W., Taboroši, D., Jocson, J.M.U., Vann, D.T., Wexel, C., 2001. Karst features of Guam in terms of a general model of carbonate island karst. J. Cave Karst Stud. 63 (1), 9–22. Nisio, S., Caramanna, G., Ciotoli, G., 2007. Sinkholes in Italy: first results on the inventory and analysis. Geol. Soc. Lond., Spec. Publ. 279 (1), 23–45. Oosthuizen, A.C., Richardson, S., 2011. Sinkholes and Subsidence in South Africa. Council for Geoscience, Cape Town. Ozdemir, A., 2015. Investigation of sinkholes spatial distribution using the weights of evidence method and GIS in the vicinity of Karapinar (Konya, Turkey). Geomorphology 245, 40–50. Panno, S.V., Luman, D.E., 2018. Characterization of cover-collapse sinkhole morphology on a groundwater basin-wide scale using lidar elevation data: a new conceptual model for sinkhole evolution. Geomorphology 318, 1–17. Passmore, J.R., 1970. Shallow coastal aquifers in the Rockingham District, Western Australia. 18. Water Research Foundation Australia Bulletin. Playford, P.E., Low, G.H., 1971. Definitions of some new and revised rock units in the Perth Basin, Perth. Playford, P.E., Cockbain, A.E., Low, G.H., 1976. Geology of the Perth Basin Western Australia. Geological Survey of Western Australia Bulletin. Perth, Geological Survey of Western Australia 124. Resnik Planinc, T., 2016. The new paradigm of solution dolines. Geogr. Bull. 88 (1), 65–78. Santos, E.S.M.d., Silva, R.W.S., Sampaio, E.E.S., 2012. Analysis of the risk of karst collapse in Lapão, Bahia, Brazil. Explor. Geophys. 43 (3), 198–212. Semeniuk, V., Cresswell, I.D., Wurm, P.A.S., 1989. The Quindalup Dunes: the regional system, physical framework and vegetation habitats. J. R. Soc. West. Aust. 71, 23–47. Senior, K.J., 1995. The Yangtze Gorges Expedition: China Caves Project 1994. Cave and Karst Science 22 (2), 53–89. Stafford, K., Mylroie, J.E., Taboroši, D., Jenson, J., Mylroie, J.R., 2005. Karst development on Tinian, Commonwealth of the Northern Mariana Islands: controls on dissolution in relation to the carbonate island karst model. Journal of Cave and Karst Studies 67 (1), 14–27. Stepišnik, U., 2004. The origin of sediments inside the collapse dolines of Postojna karst (Slovenia). Acta Carsologica 33 (1), 237–244. Stepišnik, U., 2007. Loamy sediment fills in collapse dolines near the Ljubljanica River springs, Dinaric Karst, Slovenia. Cave and Karst Science 33 (1), 23–32. Stepišnik, U., 2008. The application of electrical resistivity imaging in collapse doline floors: Divača karst, Slovenia. Studia Geomorphologica Carpatho-Balcanica 42, 41–56. Stepišnik, U., 2010. Collapse Dolines in Slovenia (Udornice v Sloveniji). E-GeograFF, 1. Department of Geography, University of Ljubljana, Ljubljana. Stepišnik, U., 2011. Sediments in collapse dolines on the Kras plateau, Slovenia. Acta Geogr. Slov. 51 (1), 233–252. Stepišnik, U., Kosec, G., 2011. Modelling of slope processes on karst. Acta Carsologica 40 (2), 267–273. Summerfield, M.A., 1996. Global Geomorphology: An Introduction to Study of Landforms. Longmann, London. Šušteršič, F., 1984. A simple model of the collapse dolines transformation. Acta Carsologica 12 (1), 107–138. Šušteršič, F., 1994. Classic dolines of classical site. Acta Carsologica 23 (1), 123–154. Šušteršič, F., 2000. Are “collapse dolines” formed only by collapse? Acta Carsologica 29 (2), 213–230. Šušteršič, F., 2002. Collapse dolines and deflector faults as indicators of karst flow corridors. Int. J. Speleol. 31 (1/4), 115–127. Šušteršič, F., 2003. Collapse dolines, deflector faults and collector channels. Speleogenesis and Evolution of Karst Aquifers 1 (3), 1–10. Šušteršič, F., 2006. Relationships between deflector faults, collapse dolines and collector channel formation: some examples from Slovenia. Int. J. Speleol. 35 (1), 1–12. Šušteršič, F., Šušteršič, S., Stepišnik, U., 2003. The Late Quaternary dynamics of Planinska jama, south-central Slovenia. Cave and Karst Science 30 (2), 89–96. Szczygieł, J., Golicz, M., Hercman, H., Lynch, E., 2018. Geological constraints on cave development in the plateau-gorge karst of South China (Wulong, Chongqing). Geomorphology 304, 50–63. Teichert, C., 1947. Contributions to the geology of Houtman's Abrolhos, Western Australia. Proceedings of Linnean Society of New South Wales 71 (3–4), 145–196. Teichert, C., 1950. Late Quaternary changes of sea-level at Rottnest Island, Western Australia. Proc R Soc Vic 59, 63–79. Twidale, C.R., Bourne, J.A., 2000. Dolines of the Pleistocene dune calcarenite terrain of western Eyre Peninsula, South Australia: a reflection of underprinting? Geomorphology 33, 89–105. Vacher, H.L., Mylroie, J.E., 2002. Eogenetic karst from the perspective of an equivalent porous medium. Carbonates Evaporites 17 (1), 182–196. van Beynen, P.E., van Beynen, K.M., 2011. Human disturbance of karst environments. In: van Beynen, P.E. (Ed.), Karst Management. Springer, Dordrecht, Tampa, USA, pp. 379–397. Vincent, J.-B., Bodewits, D., Besse, S., Sierks, H., Barbieri, C., Lamy, P., Rodrigo, R., Koschny, D., Rickman, H., Keller, H.U., Agarwal, J., A'Hearn, M.F., Auger, A.-T., Barucci, M.A., Bertaux, J.-L., Bertini, I., Capanna, C., Cremonese, G., Da Deppo, V., Davidsson, B., Debei, S., De Cecco, M., El-Maarry, M.R., Ferri, F., Fornasier, S., Fulle, M., Gaskell, R., Giacomini, L., Groussin, O., Guilbert-Lepoutre, A., Gutierrez-Marques, P., Gutiérrez, P.J., Güttler, C., Hoekzema, N., Höfner, S., Hviid, S.F., Ip, W.-H., Jorda, L., Knollenberg, J., Kovacs, G., Kramm, R., Kührt, E., Küppers, M., La Forgia, F., Lara, L.M., Lazzarin, M., Lee, V., Leyrat, C., Lin, Z.-Y., Lopez Moreno, J.J., Lowry, S., Magrin, S., Maquet, L., Marchi, S., Marzari, F., Massironi, M., Michalik, H., Moissl, R., Mottola, S., Naletto, G., Oklay, N., Pajola, M., Preusker, F., Scholten, F., Thomas, N., Toth, I., Tubiana, C., 2015. Large heterogeneities in comet 67P as revealed by active pits from sinkhole collapse. Nature 523, 63–66. Waltham, T., 2006. Tiankengs of the world, outside. China Speleogenesis and Evolution of Karst Aquifers 4 (1), 1–12. Waltham, T., Bell, F., Culshaw, M., 2005. Sinkholes and subsidence: karst and cavernous rocks in engineering and construction. Springer-Praxis Books in Geophysical Sciences. Springer, Praxis, Chichester. Webb, J.A., Grimes, K.G., Lewis, I.D., 2010. Volcanogenic origin of cenotes near Mt Gambier, southeastern Australia. Geomorphology 119 (1–2), 23–35. Whitaker, F.F., Smart, P.L., 1997. Groundwater circulation and geochemistry of a karstified bank-marginal fracture system, South Andros Island, Bahamas. J. Hydrol. 197 (1–4), 293–315. White, S., 1994. Speleogenesis in aeolian calcarenite: a case study in western Victoria. Environ. Geol. 23, 248–255. Wilkinson, D., 2015. Eneabba trip 3. The Western Caver 55, 38–39. Xu, J., He, J., Zhang, L., 2017. Collapse prediction of karst sinkhole via distributed Brillouin optical fiber sensor. Measurement 100, 68–71. Xuewen, Z., Waltham, T., 2006. Tiankeng: definition and description. Speleogenesis and Evolution of Karst Aquifers 4 (1), 1–7. Zorn, M., Komac, B., 2015. Natural disasters and social irresponsibility. Geogr. Bull. 87 (2), 75–93.