How underwater caves form: karst, dissolution and drowned passages

Every underwater cave a diver visits is a record of water moving through rock over very long periods, and of sea level rising and falling around it. Knowing how these caves form helps explain why a Mexican cenote, a Florida spring and a Czech abyss look and behave so differently, and why the same passage can hold fresh water, salt water, sediment and old stalagmites all at once.

Karst and dissolution

Most diveable caves form in karst: landscapes built on soluble rock, mainly limestone and dolomite, where water dissolves the rock and drains underground instead of flowing in surface rivers ^[5]. Rain picks up carbon dioxide from the air and soil and becomes a weak acid. In Quintana Roo, the Quintana Roo Speleological Survey (QRSS) describes how rainfall, made slightly acidic by atmospheric carbon dioxide and by tannic acids from the soil, percolates quickly through very porous rock to a shallow water table and then flows towards the Caribbean along bedding planes and fractures, enlarging them into conduits ^[3]. Florida's springs drain the same way: the Floridan aquifer is a thick sequence of Tertiary carbonate rocks, and the US Geological Survey (USGS) notes that karst aquifers like it are highly productive but also extremely vulnerable to contamination ^[5].

Dissolution does not stop once a passage exists. Where two waters of different chemistry mix, the mixture can dissolve more limestone than either water alone. This mixing-zone dissolution is central to the coastal caves of the Yucatán, as described below.

Young rock and old rock: eogenetic and telogenetic karst

Geologists H. L. Vacher and John Mylroie drew a widely used distinction in 2002 ^[1]. Eogenetic karst forms in young limestone that is still close to where it was deposited and has never been deeply buried. Its porosity is a double system: open channels and preferred pathways threading through a sponge of tiny spaces between grains. Telogenetic karst forms in old limestone that was buried, compacted and later uplifted; its porosity is mostly conduits running along networks of fractures ^[1].

The difference matters underwater. In eogenetic rock, caves tend to form at boundaries: the contact between limestone and impermeable rock below it, the water table, and the coastal boundary between fresh and salt water, where so-called flank-margin caves form as mixing chambers ^[1]. The USGS describes south Florida's Biscayne aquifer as eogenetic karst with three kinds of porosity (matrix, touching vugs and conduits) ^[5]. Telogenetic caves instead belong to integrated conduit networks guided by fractures ^[1], which is closer to the setting of deep continental shafts such as the Hranice Abyss.

The Yucatán: caves at the halocline

The flooded systems of eastern Quintana Roo, such as Sac Actun and Ox Bel Ha, form a distinctive type. A 2006 study by Peter Smart, Patricia Beddows, Jim Coke and colleagues described extensive networks lying 8 to 12 km inland of the coast, made of cross-linked horizontal elliptical tubes and canyon-shaped passages, heavily modified by sediment, speleothems and collapse ^[2]. They argued that these caves differ from both flank-margin caves and continental (telogenetic) mazes: they are dug by coastal mixing of fresh and saline groundwater where coastal discharge is large, and their passages are actively enlarging where they coincide with today's mixing zone ^[2].

A halocline is the sharp boundary between lighter fresh water floating above denser salt water. A 2011 review of the Yucatán aquifer noted that cave depth tends to follow the depth of the halocline, so caves near the coast lie at roughly 10 to 15 m and deepen inland ^[4]. That review, drawing on the 2006 study, cited maximum recorded depths of up to 72 m ^[4]; the QRSS deep-cave list now records more than 119 m in Sac Actun, a reminder that published figures change as exploration continues ^[15].

Sea-level change and drowned caves

Many of today's flooded caves were dry for long stretches. During the last glacial period sea level stood far lower: the QRSS summarises published work showing a rise from about 125 m below present beginning about 18,000 years ago ^[3], and a study of a Bahamian blue hole notes that sea level during the last glacial interval was as much as about 130 m lower ^[9]. With the sea down, the water table fell, cave ceilings lost the support of water and some collapsed, opening the windows now called cenotes ^[3]. Inland blue holes in The Bahamas are likewise voids in carbonate rock formed by dissolution, undercutting and collapse that extend below sea level ^[9].

While passages were air-filled, dripping water built stalagmites, stalactites and flowstone. These speleothems cannot grow underwater, so a drowned stalagmite is evidence that its passage was once dry. At Hoyo Negro in Sac Actun, calcite growths formed on bones until rising water covered the chamber floor; uranium-thorium dating showed these growths stopped forming below 42 m below sea level at about 9,600 years ago ^[6]. Sediment cores from the connected cenotes indicate the floor of Hoyo Negro was flooded by at least 9,850 calibrated years before present and that the pit could no longer be entered by people or animals from about 8,100 years ago ^[7]. A 2021 dataset dated 14 speleothems from submerged caves around Tulum, at elevations between 0 and 28.7 m below sea level, with 48 uranium-thorium ages used to reconstruct sea level between about 9,000 and 421,000 years ago ^[8].

The cave also records sea level in its shape. The Smart study links multiple phases of passage growth to glacial sea-level changes, and explains that after a low stand a passage keeps enlarging only if it stays in the mixing zone or keeps an active freshwater flow; otherwise it fills with collapse and sediment ^[2].

Florida: springs in Tertiary limestone

North Florida's spring caves are cut into Tertiary limestones of the Floridan aquifer ^[5]. The park history for Peacock Springs explains its cave system as limestone formed from crushed shells, sand and coral laid down when Florida lay under the sea in the Eocene, later dissolved by acidic rain ^[14]. At Wakulla Springs, the Wakulla 2 project's geology summary places the cave passages in the Oligocene Suwannee Limestone, notes a harder dolomitic layer at about 65 m that may have slowed further downcutting, and reports that the 1998-1999 3D map showed much of the cave floor sitting at about 90 m depth ^[10].

Deep shafts: Hranice and Pozzo del Merro

Some of the deepest flooded caves are near-vertical shafts rather than horizontal tunnels. The Hranice Abyss in the Czech Republic had a reached depth of 473.5 m (including its dry upper section) when a 2020 study used gravity, resistivity, magnetotelluric and seismic surveys to argue that it may extend to about 1 km and that it formed from the top down (epigenic) rather than by rising acidic water (hypogenic) ^[11]. Other geologists challenged that interpretation in 2021, arguing that later faulting invalidates the 1 km estimate ^[12]. The disagreement is unresolved in the sources, and the bottom has not been reached.

Italy's Pozzo del Merro, north-east of Rome, is a flooded collapse sinkhole: dissolution created a large underground void whose roof later fell in, leaving a funnel about 150 m across at the surface ^[13]. In 2002 a remotely operated vehicle reached 392 m without clearly identifying the bottom ^[13].

Why this matters to divers

Geology explains what divers see: silt and clay banks left by old sediment fills, breakdown rooms where ceilings collapsed, haloclines that blur vision as waters mix, and decorated passages that grew when the cave was dry. It also explains fragility. Formations that took thousands of years to grow cannot regrow underwater, and the porosity that makes karst such a rich aquifer is the same property that lets pollution travel quickly ^[5]. The articles on groundwater, archaeology and threats in this section build on these basics.