1Clues to the evolution of a karst landscape are provided by the profiles and passage levels of solutional caves. Caves and their features tend to be preserved far longer than contemporary surface features, which are more susceptible to weathering and erosion, so cave morphology can be used to help reconstruct the regional geomorphic history. Only those caves that are fed by recharge from the overlying surface are considered here. These so-called epigenic caves are formed by water that acquires its solutional capacity from surface conditions. In particular, its ability to dissolve carbonate rocks is derived from carbon dioxide absorbed from the atmosphere and especially the soil. Organic acids may contribute to the solutional potential, although their role is not so well understood. Fresh infiltrating water is, of course, capable of dissolving evaporite rocks. While passing through limestone or other types of soluble rock, water from diffuse inputs converges and generally emerges at discrete springs located in nearby valleys. The type of flow determines the distribution between vertical zones of karst, as well as the profiles and patterns of cave passages. The vadose zone, above the water table, contains conduits with free-surface streams similar to those on the surface. The phreatic zone, below the water table, contains closed-conduit flow along relatively gentle gradients. In between is the epiphreatic zone, which is flooded during high water and drained during low water, and thus contains both types of flow. As water flows through the initial fissures in a karst massif, it dissolves the surrounding rock and gradually produces well-organized drainage patterns that evolve into cave systems. This is the process of speleogenesis. Such cave systems are controlled by passive parameters (lithologic and tectonic) and by boundary conditions (type of recharge, topographic gradient, base-level position, etc.). Cave patterns depend mainly on geologic structure, type of recharge, and changes in base level. Epigenic caves are those that are fed by meteoric water with solutional aggressiveness derived from near-surface sources, particularly soil CO2. On a world-wide basis, epigenic caves probably account for at least 80-90% of known caves (fig. 1). Not discussed in this chapter are hypogenic caves, which are formed by rising water, or by water that acquires its solutional potential from deep-seated acids, mixing with other waters of contrasting chemistry, or cooling from thermal conditions.
Fig. 1 – Location of caves and karsts areas mentioned in the text.
Fig. 1 – Localisation des cavités et régions karstiques mentionnées dans le texte.
2In traditional groundwater hydrology, the greatest emphasis is on the saturated zone below the water table (or potentiometric surface), because it provides the main source of water to wells. Karst scientists are concerned equally with all aspects of underground flow, because all have a distinct role in the development of caves. The terms ‘unsaturated zone’ and ‘saturated zone’ are rarely used in karst studies, because they can be confused with the state of chemical saturation. Thus the older terms ‘vadose zone’ and ‘phreatic zone’ (above and below the water table, respectively) are preferred. Water in the vadose zone is drawn downward solely by gravity, with no significant build-up of hydrostatic pressure (fig. 2). The downward trend may be interrupted by local ponding where geologic structures cause perched phreatic conditions. Where descending water reaches the water table (the top of the phreatic zone), its flow is controlled by both elevation and pressure head, so that phreatic water follows the hydraulic gradient along the most efficient paths to available outlets in nearby valleys. Caves formed by this water tend to have gentle gradients, typically with local sections that loop down and up in the direction of flow. In karst, the water table is highly irregular both spatially and temporally. All water tables fluctuate up and down with changes in recharge, but in karst the process is more dynamic and has a profound influence on cave patterns. The zone of water-table fluctuation is known as the ‘epiphreatic zone’ in reference to its hydrologic position (Jennings, 1985), or the ‘floodwater zone’ in reference to its karst flow dynamics (Palmer, 1972). Cave passages formed in this zone share many of the characteristics of phreatic passages, with irregular profiles and evidence for high flow velocities (fig. 2).
Fig. 2 – Idealized cross section through a typical multi-stage karst system.
Fig. 2 – Coupe idéale d’un réseau karstique multi-phasé.
Recharge takes place through sinking streams, dolines, and the epikarst to form the tributaries of a branching cave system. Vadose passages (formed above the water table) include shafts and canyons. At the water table, groundwater follows a relatively gentle gradient to springs in nearby valleys. Most phreatic passages are tubular and form at or just below the water table, although many contain vertical loops. During floods (especially in caves fed by rapid runoff), the phreatic passages may be unable to transmit all the incoming water, so complex looping overflow routes form in the epiphreatic zone (zone of water-table fluctuation). Large phreatic passages form when the erosional base level remains at one elevation for a long time, as when erosional benches are formed. As the base level drops and the surface river erodes downward, phreatic passages tend to drain through diversion routes, but the old phreatic passages give evidence of the former base level.
La recharge s’effectue par des pertes, par des dolines et par l’épikarst pour former les affluents d’un système hiérarchisé dendritique. Les conduits vadoses (formés au-dessus de la surface piézométrique) sont des puits et des méandres. Au niveau de la surface piézométrique, l’écoulement s’effectue avec un faible gradient vers les émergences situées dans les vallées proches. La plupart des passages formés en zone noyée ont des sections tubulaires et se localisent à proximité ou juste sous la surface piézométrique, avec localement des boucles plus profondes. En crue (particulièrement dans les cavités alimentées par des écoulements rapides), les conduits noyés ne peuvent transmettre la totalité de l’apport d’eau supplémentaire ; les mises en charge façonnent des boucles complexes de trop-plein dans la zone épinoyée (zone de fluctuation de la surface piézométrique). De vastes conduits se mettent en place lorsque la position du niveau de base est durablement stable, au même titre que les terrasses fluviales. Lors d’un abaissement du niveau de base et d’un creusement de la vallée, les conduits noyés s’assèchent par le biais de soutirages, laissant d’anciens conduits « fossiles » perchés, témoignant de la position d’anciens niveaux de base.
3Most solution caves consist of an array of interconnected passages of various types (Palmer, 1991). Even if a passage is no longer actively forming, its shape and profile make it possible to identify its origin. It is important to distinguish between vadose and phreatic cave development, because the boundary between them is usually a close proxy to the contemporary position of the base level (fig. 2). Cave passages that form in the vadose zone exhibit a continuously downward trend along the steepest available openings in the rock. Shafts are well-like solutional voids that are vertical, or nearly so, where water has descended along a major fracture or an assemblage of intersecting fractures. In some bedrock the transmissive openings consist mainly of partings between beds, and shafts in these strata tend to deepen with time, bed by bed, as the descending water dissolves the floor and exits through a sequence of drains at progressively lower elevations. Canyon passages are typically formed by vadose water that follows the dip of inclined fractures or bedding-plane partings where vertical fractures are not available. As the initial openings grow, their floors are dissolved and eroded downward to form the canyon morphology. Most canyon passages are high, narrow, and sinuous. Many are interrupted along their length by shafts, and so most vadose water descends through rather steep passages that consist of alternating shafts and canyons. The ideal vadose flow path is straight downward along vertical fractures. But where the available vertical openings are too narrow to transmit all incoming water, the excess will overflow along the next-steepest path, which is usually down the dip of a bedding-plane parting or inclined fracture. Water in these inclined routes is able to deepen the passage floors to form canyons. Leakage through fractures in their floors can gradually enlarge alternate routes that grow and capture progressively more water until the original passage is abandoned. At and below the water table, dissolution by groundwater forms tubes (or tubular passages) with lenticular, elliptical, or nearly circular cross-sections. Most tubes originate while entirely water-filled, either continually or periodically. In profile, tubes tend to loop up and down along their length, in response to variations in efficiency of flow paths. Some tubes have a vadose origin where they are floored by resistant beds that inhibit downward enlargement, or where canyons have not had time to be entrenched into the floors. Fissure passages form along prominent vertical or steeply inclined fractures. They are straight and narrow, generally with lenticular cross-sections that are higher than they are wide. As nearby rivers entrench their valleys, the water table gradually drops and phreatic passages may be abandoned by their flow. There are several ways in which the diversion can take place. As the water table drops, the stream may simply cut a canyon in the floor to form a keyhole cross section. Most often when a tube evolves from phreatic to vadose, the vadose water forms an entirely independent lower route that diverges from the original path (fig. 2). At the vadose-phreatic transition (recording the position of the water table), several morphological changes take place in cave passages (Palmer, 1972): (i) the predominant downward trend (canyons and shafts) gives way to tubular and fissure passages; (ii) The overall passage gradient diminishes, with the phreatic section having overall gradients that are rarely more than 1‰ and with vertically looping segments; (iii) The tendency for vadose passages to follow the steepest available paths (e.g., down the dip of partings or fissures) gives way to trends that have no consistent relation to the dip direction; a strike orientation is common. These characteristics are retained by the passages after they are abandoned by their flow. These changes in passage character and profile make it possible to identify former long-term water tables and therefore former stages of rather static base level. Rises in the water table during floods create epiphreatic conditions. Rapid recharge from the surface can be highly aggressive and can form passages that resemble normal phreatic tubes and fissures. This can complicate the identification of former water tables (fig. 2). Distinguishing these from phreatic passages may be difficult, because both passage types tend to have looping profiles and tubular cross sections. However, most epiphreatic passages show evidence for high-velocity of flow, such as small solutional scallops and coarse-grained sediment (Audra, 1994; Häuselmann et al., 2003). In well-bedded strata, a gentle dip can allow vadose passages to be extensive. Many caves never reach the water table and discharge at perched springs in valley walls. Where the dip is steep, vadose passages are short and steep, and phreatic passages tend to contain downward loops that are more frequent and deeper than in rocks with gentler dip.
4Some of the most common cave patterns are shown in fig. 3 (see also Palmer, 2007). Branchwork caves consist of stream passages that converge as tributaries. In general, each major water source, such as a doline or ponor, contributes to a single solution conduit, although more than one input can contribute to a single passage. Stream passages rarely branch in the downstream direction. Closed loops are rare, except where water abandons its original passage for a new one and rejoins an older route farther downstream. Anastomotic caves are composed of curving tubes that intersect in braided patterns that have many closed loops. Nearly all are formed by periodic floodwaters fed by sinking streams or by rapid infiltration through a bare karst surface. Network caves are angular grids of intersecting fracture-controlled fissures. Their patterns are angular and typically rectangular. Many are produced either by uniform seepage through overlying or underlying insoluble rock, or by periodic epiphreatic flooding. Fewer are formed by hypogenic processes, but these include the largest examples. Spongework caves consist of interconnected solution cavities of varied size, which produce a three-dimensional pattern like the pores in a sponge. Most are hypogenic or the product of freshwater-seawater mixing. They form by the growth and coalescing of intergranular pores and other minor openings in rocks that contain no major fractures or partings. Many small caves consist only of single passages. These are simply rudimentary forms of any of the types described above.
Fig. 3 – Common cave patterns in plan view.
Fig. 3 – Principales structures de réseaux karstiques, vues en plan.
See text for explanation.
Voir texte pour les détails.
5A typical cave develops in several stages that grade smoothly from one to the next. At any given time, various passages in the same cave may occupy different stages in this sequence (Palmer, 1991; Ford and Williams, 2007): (i) Initial openings are slowly enlarged by water that is nearly saturated with dissolved bedrock. Its flow is laminar and obeys the laws defined by Hagen-Poiseuille and others. Flow rates are low, and groundwater follows a network of many alternate routes. (ii) As the early routes enlarge, those with the greatest flow become wide enough so that the groundwater retains much of its solutional aggressiveness all the way through to the springs. At about the same time, the flow becomes turbulent, which helps to increase the dissolution rate. This ends the inception phase for these passages. (iii) As the discharge increases through these few favored routes, passages begin to grow rapidly over their entire length. (iv) Cave passages eventually become air-filled, and secondary mineral deposits such as flowstone may start to grow. (v) Growth of a cave passage is slowed or halted in several ways, for example by diversion of water to different routes. (vi) Each passage is eventually destroyed by roof collapse and dissection by surface erosion. Meanwhile, newer ones may form at lower elevations, as long as there is sufficient soluble rock remaining. For a long time the cave remains in a sort of equilibrium between the destruction of old passages and the origin of new ones. Cave origin ceases only when the body of soluble rock is eventually removed entirely by erosion.
6There have been many attempts to construct various models of karst and cave development (Palmer et al., 1999). Conceptual models are based on field observations and the qualitative application of scientific principles. Analytical models rely on quantitative application of the guiding principles, mainly hydraulics and chemical kinetics. Digital models are constructed by finite-difference or finite-element analysis, in which the aquifer is considered to be composed of many small pieces, and the appropriate analytical equations are applied to each by computer software to simulate karst development with time. This topic is included here, because it provides significant information about the times required for cave development, which can be related to rates of surface denudation, and the way in which cave patterns that evolve with time.
7One of the earliest controversies in karst involved the nature of its groundwater flow. A. Grund (1903) viewed it as similar to any other kind of groundwater in porous material, with a discrete water table. F. Katzer (1909) and E.-A. Martel (1921) disagreed, citing evidence from caves where subsurface water in karst follows interconnected conduits, as though in a plumbing system, with no discrete water table. Today, most hydrologists recognize the merits of both models. Soon afterward there were debates as to where cave development took place relative to the water table. W.M. Davis (1930) and J.H. Bretz (1942) proposed that caves form deep beneath the water table, when groundwater flow paths are likely to remain stable for long time periods. A. Swinnerton (1932) contended that caves are more likely to form where groundwater flow is most vigorous, i.e. at and just below the water table (fig. 4). This origin can account for the low-gradient profiles of many phreatic passages. He considered the zone of water-table fluctuation to be most favorable for cave origin. D. Ford (1971), and Ford and R. Ewers (1978) proposed a model based on groundwater flow fields and the spatial density of fissures (i.e., fractures and partings). Fissure frequency is assumed to be low at first, but to increase with time as fissures become wider and more numerous, owing to pressure release during surface erosion and cave enlargement. They envisioned a four-state model with the following sequence (fig. 4): (i) At the lowest fissure frequency only a few phreatic loops develop, and possibly only one. (ii) As fissure frequency increases, multiple loops develop that are shallower than the first, and the water table drops as the permeability increases. (iii) As fissures become still more numerous, both phreatic loops and water-table segments develop. (iv) When fissure frequency is so great that phreatic loops cannot form, cave passages develop almost entirely along the water table. S. Worthington (2004, 2005) questioned the validity of the Ford-Ewers model by noting the development of sub-horizontal caves as much as 100 m below the water table. He also showed statistically that depth of phreatic cave development is proportional to the overall length of flow paths and angle of the stratal dip.
Fig. 4 – The water-table cave hypothesis proposed by A.C. Swinnerton (1932; A). The four-state model of D. Ford and R.O. Ewers (1978; B).
Fig. 4 – Hypothèse des cavités de surface piézométrique selon A.C. Swinnerton (1932 ; A). Le Four-State-Model de D.C. Ford et R.O. Ewers (1978 ; B).
Depending on fissure frequency, various types of caves evolve: low fissure frequency (state 1) produces bathyphreatic caves. With increasing fissure frequency the number of phreatic loops increases (states 2 and 3). High fissure frequency (state 4) results in water-table caves. Extremely low or extremely high fissure frequency does not allow evolution of caves (states 0 and 5, not shown here; after Ford, 1999).
Des types différents de réseaux se développent selon l’intensité de la fracturation : la faible fracturation (stade 1) produit des réseaux noyés profonds. Lorsque la fracturation devient plus dense, la fréquence des boucles noyées s’accroît (stades 2 et 3). Une fracturation dense (stade 4) produit une cavité de surface piézométrique. Des intensités de fracturation extrêmement fortes ou faibles ne permettent pas la formation de cavités (stades 0 et 5, non représentés ici ; d’après Ford, 1999).
8P. Weyl (1958) used strong acids to show that calcite dissolution increases with flow rate. He then used mass-transfer equations to show that caves must form along the paths of highest-velocity groundwater movement. But later, work by L.N. Plummer and T. Wigley (1976) showed that velocity has little effect on dissolution rate except in highly acidic waters. W. White (1977) used the findings of L.N. Plummer and T.M.L. Wigley (1976) to show that as a typical cave passage enlarges, the water emerging from the downstream end becomes more aggressive, gradually at first, and then abruptly. He showed that the sudden rise in aggressiveness takes place at about the same time that the water becomes turbulent. Turbulence allows the water to transport sediment, which is important to the enlargement of surface depressions. He considered these three phenomena to provide a 'kinetic trigger' that must be achieved to allow caves to form. A. Palmer (1991) combined hydraulics with chemical kinetics to explain cave patterns: (i) Early cave enlargement rates depend on the ratio of discharge to flow length (Q/L). Differences in this ratio account for the varied growth rate among the competing flow paths. (ii) Along any path, enlargement rates increase with discharge, but only up to a certain limit. From then on, greater discharge affects the enlargement rate only slightly. (iii) Only a few paths reach cave size, while others stagnate with little further growth. Branchwork caves with relatively few passages are formed. (iv) Maze caves form where Q/L is large along many alternate routes. Epigenic mazes form by recharge through adjacent permeable but insoluble rock (small L and/or uniform Q), or where floodwaters with large discharge are ponded behind constrictions so that water is injected under pressure into all fissures in the adjacent bedrock.
9Using finite-difference modeling, W. Dreybrodt (1990, 1996) and A.N. Palmer (1991) independently determined the breakthrough time (tb) needed for a fissure to reach its maximum growth rate. They both showed that tb decreases with the cube of the initial fissure width, and with roughly the 4/3 power of the hydraulic gradient and the -4/3 power of the flow distance. Thus the most favorable paths for cave development are (by far) the widest initial openings, and (less importantly) the steepest hydraulic gradients and shortest flow distances. Breakthrough time also decreases at higher CO2 partial pressures and lower water temperatures (Palmer, 1991). Breakthrough times on the order of 105 years are typical in epigenic caves. F. Gabrovšek (2000) showed that mixing of waters of varied CO2 content can decrease the breakthrough time, but that large differences in CO2 concentration are necessary. W. Dreybrodt et al. (2005) summarized the previous 15 years of karst modeling. Among their many conclusions are: (i) The main solution conduits tend to concentrate at or near the water table (fig. 5). The presence of relatively wide fractures in the initial system can lead to phreatic loops, but only if they develop prior to those at the water table. (ii) Depth of penetration of solution conduits below the water table is affected very little by aquifer thickness, because the water at depth is mainly saturated with dissolved carbonate minerals.
Fig. 5 – Simulation of vertical profiles of cave development (Dreybrodt et al., 2005).
Fig. 5 – Modélisation de spéléogenèse, vue en coupe (Dreybrodt et al., 2005).
Dissolutional widening is most active in a restricted region below the water table, where maximum flow occurs. Gradually, by headward erosion, the water table becomes almost horizontal, whereas diffuse flow in the phreatic zone simultaneously decreases. In this network, the initial aperture width a0 = 0.009 cm and the dissolved solute entering the system is at 90% saturation. The grey line in upper figure shows a surface recharge of 400 mm/a. There is no flow across the left and bottom boundaries; the river at base level provides a constant-head boundary (right). Flow rate concentrates at the water table; fissure aperture is shown in grey. Successive runs at t = 0, 5, 10, and 20 ka.
L’élargissement par dissolution des fissures se concentre dans des zones restreintes sous la surface piézométrique, où se concentre l’écoulement. Par érosion régressive, la surface piézométrique devient pratiquement horizontale, tandis que l’écoulement en zone noyée profonde diminue de manière concomitante. Dans cette modélisation, l’ouverture de la fissuration est a0 = 0,009 cm, la solution entrant dans le système est à 90 % de saturation. La ligne grise figurée sur le schéma supérieur représente la surface de recharge à 400 mm/a. Il n’y a pas d’écoulement aux limites gauche et inférieure ; à droite, la vallée au niveau de base agit comme limite à charge constante. L’écoulement se concentre progressivement à la surface piézométrique, l’ouverture des fissures est figurée en gris. Représentation des stades successifs de la modélisation à 0, 5, 10, et 20 ka.
10As a surface stream erodes its valley deeper, karst springs that emerge in the valley tend to shift to progressively lower elevations. The cave passages that feed the springs also migrate downward, following the drop in the water table. Therefore, in most caves the highest passages are the oldest and the lowest are more recent. Fluvial base levels change irregularly, often with periods of valley filling, according to changes in sea level, climate, and rates of uplift of the land. Glacial advances and retreats have a large effect, as do adjustments in the pattern of surface rivers. Some cave levels correlate with river terraces (Davies, 1960; Droppa, 1966; White and White, 2001), which suggests cave origin at or just below the water table. But in recent decades, a growing number of researchers have found that the relationship of cave passages to the water table is more complex, as described in the following paragraphs. Rapid valley entrenchment may outpace the cave development, so that many stream passages remain perched above base level. This is most common in mountainous karst. In low-relief karst, it is more common for caves to keep pace with fluvial entrenchment, so that discrete levels, or tiers, are produced (Ford and Williams, 2007; Palmer, 2007). A cave level consists of one or more passages that are confined to a narrow vertical range. This term should apply only to passages that correlate with present or former base levels. Otherwise the term tier or story can be used for the vertical arrangement of major passages, as they do not imply any specific kind of origin. If a stream valley deepens rapidly, cave streams tend to shift to lower routes so frequently that their passages do not remain active long enough to reach a large size. When there is a pause or slowing of entrenchment, the valley bottom broadens into a floodplain, and cave passages at that level have time to acquire large cross sections. These passages constitute a true cave level adjusted to fluvial base level. The level may be represented by a single passage, but the interpretation is greatly supported if several passages in the same cave, or in adjacent caves, all have similar elevations. It is not the average elevation of a passage that is important, but instead the elevation of the present (or former) vadose-phreatic transition. The presence of a vadose passage shows that, while it was forming, the floor of its outlet valley must have been at a lower elevation than the passage. Vadose passages cannot qualify as true cave levels because they form along the descending paths of gravitational water, which are independent of one another. A poorly permeable bed may cause perching of vadose passages at a common elevation, but this is a structural phenomenon that does not relate to the position of base level. A valid interpretation of cave levels, in the geomorphic sense, requires the recognition of vadose-phreatic transition zones (fig. 2). The hydraulic gradient in a water-filled cave passage is so low, except during major floods, that these transition zones are only slightly higher in elevation than the local spring. Although caves enlarge fastest during floods, floodwater dissolution extends over a wide vertical range, with no single dominant elevation, and this process can blur the distinction between cave levels.
11It was stated (Ford and Ewers, 1978) that the depths of phreatic loops diminish in progressively lower passage levels. However, in Mammoth Cave, Kentucky, the deepest known phreatic loop (21 m) is in the lowest and most recent of the major passages (Palmer, 1987). This is the result of thick-bedded, prominently jointed strata at that elevation. Many epiphreatic passages, which form above the low-flow water table under hydraulic pressure, also have irregular profiles with high-amplitude loops. P. Audra (1994) emphasized the influence of the epiphreatic (floodwater) zone for speleogenesis of passages of apparent phreatic origin, and P. Häuselmann et al. (2003) subsequently refined the model, explaining speleogenesis of Bärenschacht (Switzerland) on the base of floodwater fluctuations (see text below). Water chemistry measurements as well as direct observations in caves have shown that floodwaters are much more corrosive and erosive than low waters (Palmer, 2007), and that, for instance, scallop size reveals the floodwater flow velocity (Lauritzen et al., 1983). We thus can ascertain that floodwater effects are very important in speleogenesis and mainly results in looping tubes developing in the epiphreatic zone (see text below). In such caves, high-level passages with large vertical loops are not necessarily the oldest.
12Rises in base level may disrupt the tendency for passages in a cave to become progressively younger with depth. Some possible causes include rising sea level, decreasing stream flow, and glacial depression of the crust. Most of these effects are relatively short-lived, on the order of thousands or tens of thousands of years, but some have endured for millions of years. Even brief episodes can have long-lasting effects. A base-level rise is usually accompanied by sediment filling in valleys to that new level (Audra et al., 2009a). In most cases the sediment is deposited by the river itself (alluviation), although glacial or marine deposits may also be responsible. Cave passages below this level become flooded. They may eventually become sediment-filled, especially if their flow is feeble or is diverted into formerly abandoned passages at a higher level. Paragenesis is the upward dissolution of the ceiling in a water-filled cave passage because of sediment accumulation on the underlying floor (fig. 6; Renault, 1970; Farrant, 2004; Pasini, 2009). The sediment shields the floor from aggressive water, leaving only the upper surfaces of the passage exposed to dissolution. As the ceiling dissolves upward, more sediment accumulates on the floor to maintain the equilibrium among erosion, deposition, and water velocity. Upward migration stops when the tube reaches the water table. Later, as the water table drops, the passage is either abandoned or its sediment is excavated by the cave stream. Paragenesis is often caused by floodwaters, which are highly aggressive, sediment-laden, and abrasive. Because the vertical arrangement of an epigenic cave system has such a close relation to the regional geomorphic history, dating of cave passages can be of great benefit to reconstructing and interpreting that history. The main difficulty is that the caves themselves are voids that cannot be dated directly. However, dating of deposits in the caves can show the minimum ages of cave origin. Chemical deposits (speleothems) are easiest to date, but their ages have little direct bearing on cave origin. Most useful are dates on detrital sediment, because the sediment was presumably carried in by the same water that participated in the last phase of passage enlargement. Paleomagnetic measurements of cave sediments have been successfully used for geomorphic interpretations. However, the method relies on determining the positions of polarity reversals, and continuity of the sediment sequence is required to assure that there are no time gaps (Schmidt, 1982; Sasowsky, 2005; Zupan-Hajna et al., 2010). Measurement of the cosmogenic radionuclides 26Al and 10Be in quartz sediment is most appropriate, because it gives a continuous range of numerical dates, with no gaps; and its useful range extends to about million years, which is sufficient to cover the entire genetic history of most caves (Granger and Fabel, 2005). The following example is of one of the earliest and most complete applications of the method to cave interpretation.
Fig. 6 – Paragenetic ceiling channel (Camelié Aven, France).
Fig. 6 – Chenal de voûte paragénétique (aven du Camelié, Gard).
Photo: J.-Y. Bigot.
Photo : J.-Y. Bigot.
13Mammoth Cave is the longest cave in the world by a wide margin, with at least 600 km of mapped passages (fig. 7). It is located in a low-relief plateau of Mississippian (Lower Carboniferous) carbonates, which are locally capped by low-permeability sandstones and shales. The average stratal dip is about 0.3° toward the northwest. The capped region (Chester Upland) rises to about 250 m above sea level. Stream erosion has breached the caprock, forming irregular flat-topped ridges separated by karst valleys. Most recharge to the cave is through dolines in valley floors, and from the adjacent Pennyroyal Plateau, a vast karst plain to the southeast where the detrital rocks have been completely removed by erosion. The local fluvial base level is the Green River, one of the few permanent surface rivers in the region. It is tributary to the extensive Ohio River, and so the passage layout and sediments of Mammoth Cave hold clues to the erosional history of the entire east-central USA (Miotke and Palmer, 1972; Palmer, 1981). Passages are scattered over a vertical range of about 120 m, and because of frequent diversions, fewer than half still contain active streams. The local rocks are prominently bedded, with few conspicuous faults and joints, and so nearly all passages are guided by the bedding. Typical passages in the cave are vadose canyons that change downstream to tubular phreatic conduits. Dates of passages at various levels have been obtained from 26Al/10Be ratios in detrital sediment (Granger et al., 2001). Upper levels (A and B, above 168 m) are wide canyons and tubes filled to various depths with stream-borne quartz silt, sand, and gravel. Some are completely filled. In others, most of sediment has been removed by stream erosion. The greatest sediment thickness is about 20 m. Passages at these levels have sediment ages of 2.3-3.5 Ma, and of course the passage origins predate the sediment. These passages represent slow fluvial entrenchment during the Pliocene, which was interrupted by periodic aggradation, probably in response to climate changes. The rather flat Pennyroyal Plateau surface formed at this time. At 2.3 Ma ago there was widespread rise in base level to as much as 30 m, which resulted in alluvial fill in valleys and all upper-level cave passages. Some passages were filled partially and others completely. Aggradation with alluvial, colluvial, and lacustrine sediment also took place on the Pennyroyal surface. This event appears to coincide roughly with the first major North American glaciation, although the physical relation, if any, is uncertain. Possible causes for the aggradation may have been climate change or depression of the continental interior by glacial loading. Two major lower levels are present (C and D). They consist mainly of tubes up to 20 m in width. They are smaller than the upper-level passages, partly because groundwater recharge had been fragmented into many sub-basins, and also because pauses in base level were shorter. Overall, these and other low-elevation passages represent fairly rapid Pleistocene entrenchment of the Green River below the Pennyroyal Plateau surface. Prior to this time, karst was probably sparse on the Pennyroyal because of limited entrenchment, but with rapid entrenchment the plateau became a karst plain studded with thousands of dolines. Cave levels at 168 and 152 m consist of tubular passages with little sediment fill. Sediments at these levels date respectively to 1.5 and 1.2 Ma. Passages at other elevations are mainly shafts, narrow canyons, and crawlways, mostly of vadose origin. Some large tubes have formed below the 152 m level (E), but their elevations are not consistent. Each of the major levels in the cave is represented by up to four major passages in different parts of the cave system. Because of the stratal dip, the passages in any given level are located in different beds, so stratigraphic perching cannot account for the levels (Palmer 1987). The sudden abandonment of the large upper-level passages and development of many small passages at lower levels apparently resulted from major changes in the patterns of surface drainage. Until the early Pleistocene, the Ohio River was rather small, not significantly larger than the Green River. Farther east, most of the drainage followed northerly routes. In several stages, early continental glaciers blocked most of the eastern drainage and forced it into the Ohio River. The Ohio became the largest river in the eastern North America and began to entrench its valley rapidly. As a tributary of the Ohio, the Green River did the same. The record of these diversions is preserved in the various levels and sediment dates in Mammoth Cave (Granger et al., 2001).
Fig. 7 – Mammoth Cave, Kentucky, consists of branchwork passages fed by thousands of inputs, mainly dolines and sinking streams (Palmer, 1981).
Fig. 7 – Mammoth Cave, Kentucky, est constituée des conduits dendritiques hiérarchisés, alimentés par des milliers de points de recharge, principalement sous forme de pertes et dolines (Palmer, 1981).
A: Highly generalized profile through the cave showing the four main levels (a – d). Some lower passages (e) are large but do not form consistent levels. Epiphreatic passages are rare because of overflow into closely spaced higher levels. B: Map of major passages, both active and abandoned; the branchwork pattern is obscure. C: Relationship of the cave to the evolution of the Ohio River. When levels a and b were forming, drainage from much of eastern USA went north to the St. Lawrence River (stage 1). Early Pleistocene glaciation diverted the drainage westward to the Mississippi River through the 'Teays River' (stage 2). At that time, the Ohio River extended headward only as far as O on the map. Downcutting of the Ohio and Green River caused level b to be abandoned, and level c formed. Later glacial advances diverted the Teays into the Ohio, making it the largest river in the eastern USA (stage 3). The Ohio and Green cut down further, and level d formed. Continued downcutting allowed lower passages to form (e). A small late Pleistocene base-level rise flooded some passages but formed no major new ones.
A : Coupe très schématisée du réseau montrant les quatre principaux niveaux (a à d). Certains passages inférieurs (e) sont de grande dimension mais ne constituent pas des niveaux s.s. Les conduits épinoyés sont rares car les mises en charge utilisent les conduits immédiatement supérieurs. B : Plan schématique du réseau, incluant les conduits actifs et abandonnés. L’enchevêtrement est tel que la structure dendritique n’est pas visible. C : Relation de la cavité à l’évolution du cours de l’Ohio. Lors du développement des niveaux a et b, l’essentiel de l’est des États-Unis était drainé vers le Saint-Laurent (stade 1). Au Pléistocène inférieur, les calottes glaciaires ont détourné le drainage vers le Mississippi à l’ouest, par l’intermédiaire de la « Teays River » (stade 2). A ce stade, les sources de l’Ohio étaient localisées vers le point « O ». Le creusement de l’Ohio et de la Green River a provoqué l’abandon du niveau b, au profit du niveau c en formation. Les avancées glaciaires ultérieures ont détourné la Teays River vers l’Ohio, qui devient alors le principal organisme fluvial de l’est des États-Unis (stade 3). L’incision consécutive de l’Ohio et de la Green River a formé les niveaux d puis e. Une légère remontée du niveau de base au Pléistocène supérieur a ennoyé les parties aval, sans toutefois créer de nouveau niveau.
14The Gunung Mulu area is located in Sarawak (Borneo, Malaysia), near Brunei. It is one of the major cave areas of the world, with 325 km of cave passages, mostly of large dimensions. Clearwater cave is among the longest caves of the world, with almost 200 km of mapped passages. It displays a stacked vertical sequence of horizontal levels extending up to 500 m above the current resurgence, which is located at a floodplain. Conduit walls are etched by deep notches, in places several kilometers long, which record the base-level stability that allowed the development of extensive water-table caves (fig. 8; Farrant et al., 1995, 2007). These notches record periods of sediment aggradation controlled by the recurrent development of a large alluvial fan issuing from a nearby gorge. Farrant et al. interpret these incision-aggradation cycles as a response to cyclic inland erosion, but since the rivers are about 100 km from the sea and have graded profiles, recurrent abrupt base-level changes due to the glacio-eustatic cycles are probably also involved. U-series and paleomagnetic dating allow calibrating the cave levels. Cave sediments at elevations less than 142 m are of normal magnetic polarity, and several reversals have been recorded above, extending beyond 1 Ma, perhaps up to 2.5 Ma for the highest levels. There is good correlation between the elevation of the notches and the oxygen isotope record (fig. 9). First, it suggests that the long-term rate of uplift had been relatively constant, estimated to be about 0.19 cm/ka. The uplift is interpreted as the result of crustal flexure due to the inland unloading by denudation and correlative offshore sediment accumulation. At least during the last million years, this would have resulted in continuous isostatic uplift (Farrant et al., 1995, 2007). Second, the correlation between notches and the oxygen isotope record also shows that aggradation is climatically controlled. The earliest studies suggest that wetter interglacial periods were responsible for base-level rises and thus aggradation, extensive water table caves and notching, whereas drier glacial stages induced river incision, creation of the lower cave levels, and abandonment of older upper cave levels.
Fig. 8 – A splendid notch in the upstream part of the Clearwater River Passage.
Fig. 8 – Encoche impressionnante dans la partie amont de la rivière souterraine de Clearwater Cave (Sarawak, Malaisie).
Perched above the current river, this notch developed after gravel filling controlled by the aggradation of the outer river. This base-level rise is controlled itself by climatic cycles (photo: J. Wooldridge).
Cette encoche, actuellement perchée, s’est développée suite à un alluvionnement souterrain induit par l’aggradation de la rivière extérieure. Cette remontée du niveau de base est quant à elle contrôlée par les cycles climatiques (photo : J. Wooldridge).
Fig. 9 – Comparison between elevation of cave notches above resurgence level and the benthic δ18O isotope record from ODP site 677 (Shackleton et al., 1990).
Fig. 9 – Comparaison entre l’altitude relative au dessus de l’émergence des niveaux d’encoches et l’enregistrement isotopique δ18O du site ODP 677 (Shackleton et al., 1990).
The chronology is derived from the paleomagnetic timescale and astronomic tuning (Farrant et al., 1995).
La chronologie est proposée sur la base de l’échelle paléomagnétique et des cycles astronomiques (Farrant et al., 1995).
15Vertical cave pattern is mainly controlled by time, by the position of the aquifer (perched vs. dammed), by recharge type (regular vs. irregular), and by base-levels changes (lowering vs. rising; Audra 2001; Palmer and Audra 2004).
16When soluble rocks are first exposed by uplift and removal of any impermeable cover, caves display a juvenile pattern with a steep profile. Because of sparse fracturing, the water table can be steep and located high above fluvial base level (fig. 10A). Initial phreatic paths are later entrenched as the water table drops and the flow becomes vadose. Vadose enlargement, as shafts, meanders, and canyons, often obscures the initial phreatic paths, which are sometimes preserved as ceiling channels. The juvenile pattern often corresponds to the initial phase for most vertical cave passages (shafts). These are common in young, rapidly developing karst (fig. 11) such as: (i) in karst subjected to intense dissolution, after rapid uplift and great rainfall: Nakanai Mountains, Papua New Guinea (fig. 12); (ii) in evaporite rocks where sinkholes deliver water to through-caves directly connected to resurgences: Gébroulaz, France; caves of the Mount Sedom, Israel (Frumkin, 1998); (iii) in carbonate rocks exposed for the first time by glacial erosion, which has removed the impermeable cover (Grand Marchet, France). During cave inception, the mixing of saturated groundwater with aggressive water from points of recharge allows corrosion to take place along a steep water table.
Fig. 10 – Different types of cave profiles, constrained by time, geological structure, or recharge conditions.
Fig. 10 – Différents types de profil des réseaux karstiques, selon les contraintes temporelles, structurales, ou le mode de recharge.
Juvenile pattern: in the first stage of karstification, when high gradient is present, a cave system develops a steep profile corresponding to the path with the least head loss (A). Perched caves: a vadose system develops at the contact of the underlying aquiclude. The mechanical erosion of soft underlying material enlarges galleries. Collapses of the limestone ceiling have partly filled the gallery with boulders (B). An epiphreatic cave: irregular recharge causes backflooding; drains develop throughout the epiphreatic zone, with looping profiles resulting from the influence of structural openings (C, left). A water-table cave: recharge through a poorly permeable cover is diffuse and regular, so the water-table level remains stable with time and the drain develops at the water table (C, right). Influence of base-level change on cave structure (D). Left: cave levels. A base-level drop causes the lowering of the karst drainage. The old drains are abandoned and remain perched. Right: Flooded cave. A base-level rise floods the deep part of the cave system. The main deep passages remain active; the water rises through a phreatic lift (chimney-shaft) and discharges at a vauclusian spring.
Réseau juvénile : au premier stade de la karstification, en présence d’un fort gradient, un réseau incliné se développe selon le cheminement correspondant aux moindres pertes de charge (A). Réseau perché : développement d’un réseau vadose au contact du soubassement imperméable. L’érosion mécanique de ce niveau plus tendre favorise l’élargissement du conduit. L’effondrement du plafond calcaire remplit partiellement le conduit de gros blocs (B). Réseau épinoyé : une recharge irrégulière favorise les mises en charge ; des conduits se développent dans toute la zone épinoyée selon un profil en montagne russe calqué sur les discontinuités structurales (C, à gauche). Réseau de surface piézométrique : la recharge au travers d’une couverture peu perméable assure un écoulement diffus et régulier, si bien que la surface piézométrique ne connait pas de variation verticale significative ; le drain se développe le long de la surface piézométrique (C, à droite). Influence des variations du niveau de base (D). A gauche : réseau étagé. Un abaissement du niveau de base provoque la descente du niveau des écoulements karstiques. Les anciens drains deviennent des étages abandonnés et perchés. A droite : réseau ennoyé. La remontée du niveau de base ennoie la partie profonde du réseau karstique. Les conduits profonds principaux demeurent en activité, l’écoulement remonte par des puits-cheminées et arrive en surface par une source vauclusienne.
Fig. 11 – Examples of juvenile cave patterns, displaying a straight long profile, where a sinkhole feeds through-caves directly connected to the resurgence.
Fig. 11 – Exemples de réseaux juvéniles organisés selon un profil incliné, avec une alimentation par des pertes directement connectées à la résurgence.
Flow is vadose, and no phreatic zone is present. No vertical exaggeration. A: Muruk System (Nakanai Mountains, Papua New Guinea). Intense rainfall (> 10 m/a) generates surface runoff on clay covers, which feeds numerous sinkholes. Huge cave systems develop into recently uplifted soft Miocene limestone. High-discharge rivers flow through large galleries and canyons (fig. 12). B: Gebroulaz Cave (Vanoise, France). Glacial meltwater sinks rapidly into a gypsum body and flows through it in a 350 m-long gallery with a gentle gradient. C: Grand Marchet Sinkhole (Vanoise, France). A small stream sinks into a marble body. The passages develop along metamorphic schistosity and lithologic contacts.
L’écoulement est vadose, il n’y a pas de zone noyée (pas d’exagération de l’échelle verticale). A : Gouffre Muruk (montagnes Nakanai, Papouasie Nouvelle-Guinée). Les précipitations intenses (> 10 m/a) favorisent un écoulement de surface sur les couvertures argileuses, qui alimente de multiples pertes. Des réseaux gigantesques se développent dans les calcaires miocènes tendres récemment soulevés. Les débits gigantesques façonnent de grandes galeries et des canyons souterrains (fig. 12). B : Traversée du Gébroulaz (Vanoise). Les eaux proglaciaires s’enfouissent dans le gypse et le traversent sur une longueur de 350 m, selon un profil modérément incliné. C : Traversée du Grand Marchet (Vanoise). Un petit ruisseau se perd dans les marbres. Les conduits se développent le long des plans de schistosité et des contacts lithologiques.
17Where the aquifer is perched above base level on an underlying aquiclude, there is no significant phreatic cave development (fig. 10B). Shafts and canyons converge to form conduits at the aquiclude top and feed springs along hillslopes. Major springs drain into the heads of pocket valleys (e.g., Diau System, France). Mechanical erosion, aided by detrital sediment, quickly enlarges the main routes by entrenching the underlying aquiclude (especially if it is soft material such as marl). Large galleries develop by collapse of the limestone ceiling and may fill with boulders (fig. 13). Base-level lowering promotes headward retreat of the spring but does not noticeably affect the cave pattern.
Fig. 12 – Muruk System (Nakanai Mountains, Papua New Guinea).
Fig. 12 – Gouffre Muruk (montagnes Nakanai, Papouasie Nouvelle-Guinée).
A large underground river is fed by numerous sinkholes. All along the 10-km-long and 1000-m-deep path from sinkhole to resurgence, the water flows as a torrent in gently inclined passages. Torrential flow dissolves the soft limestone and frequent ceiling collapses produce boulders (photo: J.-P.Sounier).
L’important cours souterrain est alimenté par de nombreuses pertes. Le torrent souterrain, long de 10 km pour une dénivellation de plus de 1 000 m, relie les pertes à la résurgence selon un profil incliné. L’écoulement torrentiel érode les calcaires tendres, produisant des effondrements du plafond attesté par des gros blocs (photo : J.-P. Sounier).
Fig. 13 – The Sarawak Chamber (Gunung Mulu national Park, Malaysia).
Fig. 13 – La salle Sarawak (Parc national de Gunung Mulu, Malaisie).
The allogenic recharge and torrential flow produce strong mechanical erosion. It has progressively removed the center of an anticline composed of shale and sandstones, making the largest chamber in the world (600 m x 400 m x 150 m; Gilli 1993; the white circles show persons for scale.)
La recharge allogène et l’écoulement torrentiel favorisent une puissante érosion mécanique. Le déblaiement du cœur de l’anticlinal, composé de schistes et de grès, a formé la plus grande salle du monde (600 m x 400 m x 150 m ; Gilli 1993 ; les cercles blancs montrent les personnages donnant l’échelle).
Photo: R. Schejbal and E. Gilli.
Photo : R. Schejbal et É. Gilli.
18When the karst aquifer extends below the spring outlet, which is determined by a fluvial or structural base level, the karst is 'dammed' (fig. 10C). In turn, the spring position determines the water-table elevation inside the karst. The main drain becomes established at the water table, at the end of the passage with the lowest head losses. Major passages either follow the water table or contain shallow phreatic loops.
19During high flow, the lowest passages - which outlets are often plugged with surficial deposits - are unable to carry the entire discharge; water table rises, water floods the epiphreatic zone, rises in phreatic lift tubes, and eventually emerges at overflow springs (e.g., Castleguard Cave, Canada; Ford, 1983). During low flow the water follows lesser openings at lower elevations (fig. 10C). In the Siebenhengste Cave System (Switzerland), P. Häuselmann (2002) studied the shifting from gravitational downcutting (forming canyons) to tubular passages formed by closed-conduit flow (fig. 14 and fig. 15). This study allows reconstruction of the vadose-phreatic transition zones corresponding to former water-table positions (Palmer, 2007). He demonstrated that these transition elevations decrease toward the springs and record the top of the epiphreatic zone, i.e. the highest position of the water table. Consequently, this study shows that high-amplitude looping passages form in the epiphreatic zone and are enlarged by aggressive high flows. The vertical amplitude of the epiphreatic loops can exceed 200 m. It indicates the minimum depth of the flooding in the epiphreatic zone. This observation shows that cave levels effectively record the base-level positions, but the cave-level altitudes can be noticeably above that of the corresponding base level. In addition, the amplitude of the loops depends on the vertical amplitude of the epiphreatic zone, and thus on the height and suddenness of flooding during the active period of these passages. Consequently, epiphreatic speleogenesis is due to an irregular discharge, either from storms or from glacial or snow melt, or as the result of concentrated surface runoff into dolines (Audra, 1994).
Fig. 14 – The Hölloch System (Alps, Switzerland) is a maze of active and ancient epiphreatic passages.
Fig. 14 – Le Hölloch (Alpes suisses) est un réseau labyrinthique de conduits abandonnés et épinoyés.
Such tubes flood over more than 200 deep during high water.
Les mises en charges atteignent 200 m de hauteur lors des grandes crues.
Photo: U. Widmer.
Photo : U. Widmer.
Fig. 15 – Cross-section of the southern part of Bärenschacht, with indications of the speleogenetic phases (after Häuselmann et al., 2003; Häuselmann and Granger, 2005).
Fig. 15 – Coupe de la partie sud du Bärenschacht, avec indication des phases de spéléogenèse (d’après Häuselmann et al., 2003 ; Häuselmann et Granger, 2005).
Loops with amplitudes as high as 150 m are visible, and mazes are due to flooding and draining processes. The successive lowering of base level produces several speleogenetic levels. Passages are braided because the amplitude of epiphreatic loops is higher than the altitude difference between each speleogenetic phase. It results in a complex maze connecting active and fossil passages, altogether developing several dozen of kilometers.
Les boucles ont une amplitude atteignant 150 m, les labyrinthes sont dus aux mises en charge et aux vidanges. L’abaissement par étapes du niveau de base est à l’origine des niveaux de conduit. Les niveaux sont entremêlés car l’amplitude des boucles épinoyées est plus grande que l’écart vertical entre chaque étage. Le résultat est un labyrinthe complexe, connectant des conduits actifs et abandonnés, le tout développant plusieurs dizaines de kilomètres.
20When a karst is overlain by a thick semi-permeable cover that acts as a filter, there is little fluctuation in recharge. The regulated seepage induces the regularization of the transfer (fig. 10C). In contrast to bare karst, flooding and the development of an epiphreatic zone are very limited. The main drains concentrate at the water table where the water flow is continuous. Cave systems display low-gradient passages with extensive pools. Similar long profiles are characteristic of through-caves fed by extensive impermeable catchment areas. When such caves reach the mature stage, their passage size is large enough to allow the transfer of all stages of flow, including seasonal peaks. Such through-caves are frequent in monsoonal Southeast Asia (Laos, Vietnam, China, Thailand, Philippines; fig. 16 and fig. 17).
Fig. 16 – Saint-Paul Cave (Palawan Island, Philippines) is a 24 km-long cave system.
Fig. 16 – Saint-Paul Cave (Palawan Island, Philippines) est un réseau de 24 km de développement.
The resurgence opens along the shoreline and the cave extends several kilometers inland. The main drain consists of a large water-table conduit in which the influence of tides is felt as much as 6 km inside the cave (photo: E. Procopio).
La résurgence s’ouvre sur le littoral tandis que le réseau pénètre de plusieurs kilomètres dans les terres. Le collecteur est un conduit de surface piézométrique, où l’influence des marées est perceptible à plus de 6 km de distance (photo : E. Procopio).
Fig. 17 – Cross-section of Saint Paul Cave (Palawan Island, Philippines).
Fig. 17 – Coupe de Saint Paul Cave (Palawan Island, Philippines).
The cave drains a polje at low altitude through a 6 km-long passage along the water table and discharges to the sea (courtesy L. Piccini).
Le réseau draine un poljé situé à une altitude modérée, par un conduit principal long de 6 km développé le long de la surface piézométrique, qui se déverse en mer (d’après L. Piccini).
21In dammed settings the inner organization of karst drainage strongly depends on the base-level position. Any change in base level affects the position of the water table itself and induces a reorganization of the drainage. If fluvial base-level drops stepwise, successively lower phreatic passages develop (fig. 18, fig. 19 and fig. 10C). Pauses in base-level lowering produce cave levels that correlate with river terraces (e.g., at Mammoth Cave, USA). 'Invasion' vadose shafts and canyons extend the vadose zone down to the new water table. Former conduits and springs are abandoned and partly filled with floodwater sediments and secondary minerals. Perched in the vadose zone, old phreatic conduits are cut by new shafts and canyons that feed active conduits. Most of the large cave systems in the world correspond to this type of system with integrated cave levels recording successive base level lowering, respectively: Mammoth Cave, USA; Siebenhengeste, Switzerland; Clearwater Cave, Malaysia; Dent de Crolles, France (fig. 18).
Fig. 18 – The Dent de Crolles system, France, contains 57 km of passages over almost 700 m of depth, below a surface less than 1.5 km2 in area.
Fig. 18 – Le réseau de la Dent de Crolles (Chartreuse), développe 57 km de conduits sur plus de 700 m de dénivellation, sous une surface de moins de 1,5 km2.
The cave consists of vadose shafts and canyons originating from the plateau surface, and which connect to four distinct semi-horizontal levels. The three highest levels are perched fossil galleries within the limestone mass; the lowest one, at the contact with the underlying aquiclude, is active.
Il est constitué de conduits vadoses en puits et méandres provenant de la surface du plateau, qui relient quatre niveaux semi-horizontaux. Les trois plus élevés sont inactifs et perchés dans la masse calcaire, tandis que le niveau inférieur, au contact du soubassement marneux, est actif.
Fig. 19 – The Boundoualou Cave, France.
Fig. 19 – Grotte du Boundoualou (Aveyron).
The upper dry entrance corresponds to an old abandoned level. The lowest one acts as an overflow in high water. The main drain is at the base of the limestone, at the contact of the marly aquiclude shown by the person.
L’entrée supérieure inactive correspond à un niveau ancien abandonné. L’entrée inférieure fonctionne en-trop-plein lors des hautes eaux. Le drain principal est à la base des calcaires, au contact du soubassement marneux où se situe le personnage.
Photo: F. Guichard.
Photo : F. Guichard.
22The depth that phreatic loops can extend below the water table is a matter of debate (e.g., Worthington, 2001). Some water-filled passages have been dived through their springs to depths of several hundred meters, at the limit of present-day techniques, and yet they still continue downward out of sight (Exley, 1994). Most are located along major faults. Lift tubes up to 100 m in relief have been mapped in caves in Sarawak, Malaysia (Farrant et al., 1995). They are now abandoned and resemble vertical shafts. Phreatic loops of comparable depth have been detected in caves of the Sierra de El Abra, Mexico (Fish, 2004). Base-level rise causes flooding of conduits. Some become sediment-filled, but the main flow lines remain active (fig. 10C). New ascending routes, or reactivation of relict conduits, may form phreatic lifts (chimney-shafts) and vauclusian springs (Fontaine de Vaucluse, France). In Mediterranean karsts, the Messinian Salinity Crisis induced first a deepening of the karst systems, then a flooding after the Pliocene transgression, and finally a reorganization of the drains after this base-level rise (fig. 20). This could be a reason why deep phreatic cave systems are so frequent around the Mediterranean basin (fig. 21; Audra et al., 2004). Currently, some cave systems remain flooded and others have been partly or entirely drained after Pleistocene re-entrenchment of the valleys. In partly exhumed canyons, the lower part of the karst has remained flooded since the beginning of the Pliocene, and they discharge as vauclusian springs (Fontaine de Vaucluse type; fig. 22B). In the entirely exhumed canyons, the karst is now drained and the chimney-shafts are fossil (fig. 22C; Audra et al., 2009a). Other causes of base-level rise (eustacy, fluvial aggradation, continental subsidence), less significant in amplitude, have the same effect on per ascensum speleogenesis (fig. 22). The deep karst is flooded and phreatic lifts connect to vauclusian springs. Consequently, there should be a global genetic model for most deep-phreatic systems. Some of them have a hypogenic origin, i.e. a true deep-phreatic origin (Audra et al., 2009b). However, most of them could correspond to a base-level rise inducing the per ascensum speleogenesis, which first flooded the karst and then allowed the development of phreatic lifts, 'chimney-shafts”, and of vauclusian springs.
Fig. 20 – Per ascensum model of speleogenesis during the Messinian-Pliocene cycle (Audra et al., 2009a).
Fig. 20 – La spéléogenèse per ascensum Durant le cycle messino-pliocène (Audra et al., 2009a).
A: During the Messinian, the Mediterranean drying-up caused (1) the entrenchment of canyons and (2) the deepening of karst drainage. B: Pliocene base-level rise occurred in two steps, (3) by marine ingress (dark gray), then (4) by fluvial aggradation (light gray). Deep drainage uses phreatic lifts to emerge as vauclusian springs, recording successive positions of the base level. If the Messinian canyon is located below the current base level, it remains fossil; the karst remains flooded and discharges by a vauclusian spring (Fontaine de Vaucluse type). C: If the Messinian canyon is located above the current base level, the canyon is exhumed and the karst is drained. The current drainage uses (5) the deep Messinian drain; the Pliocene phreatic lifts are abandoned as fossil 'chimney-shafts'.
A : Au Messinien, l’assèchement de la Méditerranée a provoqué (1) l’incision de profonds canyons et (2) l’enfoncement du drainage karstique. B : La remontée du niveau de base au Pliocène s’est produite en deux temps, (3) par l’ingression marine (en gris foncé), puis (4) par comblement alluvial (en gris clair). Les écoulements profonds utilisent des puits-cheminée noyés qui émergent à des sources vauclusiennes dont l’emplacement enregistre les positions altitudinales successives du niveau de base. Si le fond du canyon messinien est en-dessous du niveau de base actuel, il demeure fossilisé par le comblement pliocène ; le karst profond reste ennoyé et drainé par des sources vauclusiennes (type fontaine de Vaucluse). C : Si le fond du canyon messinien est au-dessus du niveau de base actuel, le canyon a pu être exhumé et le karst dénoyé. L’écoulement karstique actuel utilise (5) les anciens drains messiniens profonds ; les puits-cheminées pliocènes sont abandonnés.
Fig. 21 – Deep-phreatic cave systems in Mediterranean France.
Fig. 21 – Les réseaux noyés profonds de la France méditerranéenne.
All cave systems are connected to the Mediterranean or to the Pliocene rias ('flooded valleys'). 1: Messinian incision; 2: Messinian shoreline; 3: Pliocene shoreline; 4: river; 5; Messinian alluvial fan; 6: large karst area; 7: deep phreatic karst system (relative depth reached by scuba diving).
Tous ces réseaux sont en relation plus ou moins directe avec la Méditerranée ou bien avec les anciennes rias pliocènes. 1 : incision messinienne ; 2 : rivage messinien ; 3 : rivage pliocène ; 4 : cours d’eau ; 5 ; cônes alluviaux messiniens ; 6 : régions karstiques principales ; 7 : réseaux noyés profonds, avec indication de la profondeur relative atteinte en plongée.
Fig. 22 – Per ascensum model of speleogenesis, caused by eustatism, glaciation, and tectonics, respectively.
Fig. 22 – La spéléogenèse per ascensum, respectivement produite par l’eustatisme, les glaciations, et la tectonique.
A: Podtraťová jeskyně, Moravian karst, Czech Republic, a 140-m high chimney-shaft, the lowest part of which is flooded below the Beroukna valley (Bruthans and Zeman 2003). It may show a record of the base-level rise of the hydrologic network after pre-Badenian entrenchment. B: The Puits des Bans and the Gillardes Spring (French Alps). The basin fill (glacial, lacustrine, and fluvio-glacial) has blocked the Gillardes Spring. In high water, the Puits des Bans, a 300 m-high chimney-shaft, floods and overflows. C: Lagoa Misteriosa (Brazil), a 200 m deep phreatic shaft, a window in a karst aquifer flooded after continental subsidence of the Pantanal region (survey by G. Menezes).
A : Podtraťová jeskyně (karst de Moravie, République tchèque) est un puits-cheminée de 140 m de hauteur, dont la partie basse est ennoyée sous le niveau de la rivière Beroukna (Bruthans et Zeman 2003). B : Le puits des Bans et la source des Gillardes (Dévoluy). Le comblement morainique et glacio-lacustre du bassin a bloqué la source des Gillardes. En crue, le puits des Bans, un puits-cheminée de 300 m de hauteur, s’ennoie et déverse. C : Lagoa Misteriosa (Brésil), puits noyé de 200 m de profondeur, est un regard sur un aquifère karstique ennoyé par la subsidence continentale de la région du Pantanal (topographie d’après G. Menezes).
23In the currently accepted model of speleogenesis, the Four-State Model of D.C. Ford and R.O. Ewers (1978), looping caves are considered to form in the deep-phreatic (bathyphreatic) zone, as constrained by the opening of fractures. Their origin is reinterpreted here to be generally epiphreatic, in the zone of irregular flow rates. Deep-phreatic caves, in contrast, are generally formed by flooding caused by a rise in base level, or (less commonly) by hypogenic processes. This reinterpretation explains the vertical organization of most epigenic caves (Audra, 2001; Palmer and Audra, 2004; Audra, 2007; Audra et al., 2009a). Base-level change producing stacked interconnected cave levels is responsible for the largest and most complex cave systems in the world (Mammoth Cave: 600 km; Siebenhengste; 200 km; Clearwater Cave, etc.). Base-level rise produces some of the deepest phreatic cave systems (Vaucluse; -308 m), especially around the Mediterranean, which involved the dramatic sea-level rise following the Messinian crisis. Likely future directions for research include both field investigations in order to assess locally the role of constrain parameters and theoretical investigation based on numeric modeling, which could help such assessment at a broader scale.
Many thanks to Dr. Marco Menichetti (university of Urbino), Dr. Jo De Waele (university of Bologna) and the anonymous reviewer for their helpful reviews, which have greatly improved this paper.