Extensive flooded cave systems are developed in a zone 8–12 km inland of the east coast of the Yucatan Peninsula, Quintana Roo, Mexico. In plan, the systems comprise cross-linked anastomosing networks composed of horizontal elliptical tubes (which are actively developing where associated with the present fresh water/saline water mixing zone) and canyon-shaped passages. Both forms are heavily modified by sediment and speleothem infill, and extensive collapse. The pattern of Quintana Roo caves differs both from the mixing chamber form of flank-margin eogenetic caves, and also the dendritic and rectilinear maze patterns of epigenetic continental (telogenetic) caves. Unlike the latter, Quintana Roo caves are formed by coastal zone fresh water/saline water mixing processes. While mixing dissolution is also responsible for development of flank-margin caves, these may be typical of small islands and arid areas with limited coastal discharge, whereas Quintana Roo–type caves are formed when coastal discharge is greater. In the Quintana Roo caves, multiple phases of cave development are associated with glacio-eustatic changes in sea level. Two critical conditions control cave development following lowstands: (1) if the passage remains occupied by the mixing zone and connected to underlying deep cave systems, and (2) for passages above the mixing zone, if active freshwater flow is maintained by tributaries. In the first case, inflow of saline water drives mixing dissolution, enabling removal of the lowstand carbonate fill and continued passage enlargement. In the second, despite limited dissolution in the fresh water, continued removal of uncemented sediments can maintain the cave void. Where neither of these conditions is met, enlargement will cease, and the cave void will become occluded by collapse and sediment infill.

Abstract Dolomitization requires not only favourable thermodynamic and kinetic conditions, but also a fluid-flow mechanism to transport reactants to and products from the site of dolomitization. This paper reviews work that seeks to provide a quantitative framework for conceptual models of dolomitization, using analytical and, particularly, numerical simulation models of fluid flow and rock-water interaction. This approach is starting to yield new insights into the major controls on the rate and pattern of fluid flux, and the resultant dolomitization. Three sets of forces can drive the fluid flow required for dolomitization: elevation (topographic) head of meteoric water and/or seawater; gradients in fluid density due to variation in salinity and/or temperature; and pressure due to sedimentological and/or tectonic compaction. However, in many situations individual flow mechanisms may not operate in isolation. Rather fluid flow will commonly be a product of a number of different drives acting simultaneously. The balance between drives will change over time with variations in relative sea-level, climate, platform geometry and palaeogeography (which collectively comprise the critical boundary conditions). The simplistic prediction of dolomite body geometry from a single driving force may be misleading, as fluid flow will critically depend both on the boundary conditions and the distribution of permeability. Indeed, even for single driving forces, model predictions change significantly as simplistic assumptions are relaxed and these key parameters are specified with increasing realism. The coupled modelling of dolomitization reactions within the flow field is less tractable than that of groundwater circulation because the kinetics of dolomitization are less well understood, particularly at lower temperatures. Dolomitization is likely to occur along a reaction front, where a favourable balance is struck between mass transport and reaction kinetics. For instance, in simulations of geothermal convection dolomitization focuses along the 50–60 °C isotherm. Dolomitization reactions are favoured by higher temperatures in deeper zones, but rates are limited by low flow because of lower permeability. Although flow rates are higher in shallow more permeable carbonates, lower temperatures limit reactions. High flow rates during reflux of platform-top brines give rapid dolomitization. This is associated with porosity occlusion in front of and behind the broad zone of replacement dolomitization driven by anhydrite cementation and overdolomitization, respectively. Lithological heterogeneities strongly affect the pattern of dolomitization, which is highly focused within more permeable beds and those with a higher reactive surface area. While we focus here on dolomitization, models can also provide insights into diagenetic processes such as marine calcite cementation and aragonite, calcite and evaporite dissolution by refluxing brines, and by seawater circulation below the aragonite and calcite compensation depths. However, it is important to be aware of the assumptions and limitations of the numerical model(s) used. Particular attention must be paid to specification of boundary conditions, permeability and reactive surface area. The uncritical application of numerical techniques to particular cases of dolomitization is at best uninformative and at worst misleading. Careful application of these techniques offers great promise for well-constrained field problems, with greater inclusion of natural heterogeneity and time-variant boundary conditions. We also need to model feedbacks between diagenesis and porosity-permeability, and to include platform growth in simulations of slower diagenetic processes.

Abstract In open marine carbonate platform systems there is a potential for dolomitiza- tion by circulation of saline ground waters because of the high magnesium content of sea water and the long residence time of the carbonate sediments in the shallow subsurface. Saline ground water circulation is driven by hydraulic head, which may be generated either by differences in sea-surface elevation caused by tides, winds or ocean currents, or by differences in water density related to salinity and/or temperature. Buoyant circulation occurs where saline ground waters are entrained by the fresh ground water circulation, while the increase in salinity of bank surface waters due to evaporation induces reflux. Finally, thermal contrasts between cold ocean waters and ground waters warmed by the geothermal heat flux gives rise to convective circulation. The scale of any platform, the steepness of its margins, and relative position of sea level are major controls on the type of circulation that may occur. In addition, climate, the distribution of permeability in the carbonate sediments, and geographic position in relation to other land masses and the atmospheric and oceanic circulations are significant. Active circulation of saline ground water occurs beneath North Andros Island on the Great Bahama Bank in the form of major outflows (0.3-3.9xl0 5 m 3 day<sup>-1</sup>) from karstic blue holes along the margin of the Tongue of the Ocean. The elevated salinity of these waters indicates that they derive by reflux from the Great Bahama Bank, and flow eastward under North Andros Island. Ground water temperatures are, however, low, suggesting that cold normal salinity sea waters from depth within the adjacent oceans are also involved in the circulation. It is concluded that this circulation results from the combined effects of differences in temperature, salinity, and elevation potential. Numerical modeling of saline ground water circulation is needed in order to understand the sensitivity of the circulation to the differing controlling parameters, and may enable prediction of the three-dimensional distribution of resultant diagenetic alteration.