BVI salt ponds were all shallow and mostly hypersaline, and their waters were generally well mixed. They exhibited large variations in salinity, both temporally and spatially, among ponds.
Rainfall controlled temporal salinity fluctuations (Figures 4, 5, 6), and it accounted for most, if not all, freshwater inputs to ponds. Groundwater appeared to be an unimportant source of freshwater input, as evidenced by the constant salinity of groundwater near ponds despite changing pond salinities (Figure 7). This finding agreed with observations from St. John, USVI . We observed a decline in groundwater salinity with distance from the pond edge, a pattern typical of the interfaces between salt flats and mangroves and presumably caused by the greater influence of fresh water with distance from an evaporative basin .
Evaporation was an important force in all ponds, influencing both salinity and depth. Through-ground drainage of pond water to the sea did not normally occur, except when a pond was excessively full, as was observed in BAN after two consecutive hurricanes. Long-term and short-term methods of measuring evaporation yielded similar results, indicating that the mean evaporation rates in ponds during dry weather was between 32 and 64 m3H2O/day-ha, all (or nearly all) of which occurs during daylight hours. The maximum change in pond depth during dry weather was -0.5 cm/day, a value similar to the rate of evaporation (-0.6 cm/day) reported for a hypersaline lagoon in the Red Sea .
Pond water levels declined during dry periods, exposing large areas of pond sediments, while rainfall caused ponds to fill and expand. This effect was less pronounced in ponds with direct sea connections (Figure 9g–j). The buffering effects of seawater connection did not extend to salinity, as even the permanently sea-connected Anegada ponds became highly saline (>100 ppt) during dry periods (Figure 4g–j). This indicates that seawater entering the channel to the Anegada ponds did not flush concentrated waters out of the ponds. The phenomenon of hypersalinity despite communication with the sea has also been described in a Brazilian lagoon  and in a Saudi Arabian lagoon .
Salt ponds exhibited a range of hydrological differences. This variability allowed us to identify several groups of ponds, using inundation period and surface (visible) connection with the sea as primary descriptors. Below, the hydrological characteristics and characteristic salinity ranges are described for each of these groups.
1. Permanent ponds with direct sea connection (BON, FLA, PTP, RED):
These ponds, all situated on the flat coral island of Anegada, formed the largest wetland system in the BVI. They were inter-connected in addition to being connected with the sea through a narrow channel. Water levels in ponds with permanent direct sea connection were controlled by sea level fluctuations and by rainfall and evaporation. Salinity was variable but generally high (mean annual salinities ranged from 93 to 184 ppt; Figure 3). Rainfall and evaporation appeared to be far more important in forcing salinity fluctuations than was seawater flushing.
2. Permanent ponds with seawater seeps (BEL, SAL):
A constant seep of seawater trickled into these ponds through the shore sediments. Water levels were determined primarily by rainfall and evaporation, though small daily fluctuations in water level corresponded with tidal cycles. Ponds with seawater seeps experienced some of the highest salinities of all ponds (mean annual salinities ranged from 165 to 269 ppt; Figure 3). Gypsum and sodium chloride were deposited by precipitation in these ponds during the dry season.
3. Temporary ponds with periodic direct sea connection (LON, WB):
Ponds with periodic direct sea connection were filled by sea overwash during seasonally high tides between May and December (Figure 8). During these periods of sea connection, water level was controlled solely by tidal fluctuations. When tides were low, sea communication was broken, and then water levels were controlled by rainfall and evaporation cycles. Salinity fluctuations in these ponds were similar in magnitude to those in other temporary ponds. However, seawater flushing during the dry season reduced pond salinities, causing the salinity fluctuations in these ponds to differ in timing from those in other temporary ponds (Figure 4).
4. Permanent ponds with no surface sea connection (BAN, SIN):
Two ponds with rather different hydrologies are included here. Water level in BAN was influenced by rainfall and evaporation but not by tidal fluctuations. SIN, in contrast, exhibited large tidal fluctuations (Table 4). Unlike BAN, SIN maintained a relatively constant area of inundation throughout the dry season. Salinity in BAN was generally high and variable (135 ± 48 ppt; Figure 3), while that of SIN was consistently near seawater (35 – 51ppt; Table 2). In all of these respects, BAN was more similar to other ponds than SIN was.
SIN's unique hydrology, characterized by high-amplitude tidal fluctuations, permanent inundation and near-seawater salinity, suggests that this pond maintained underground connection with the sea, probably through its coral berm. Such hydrology is similar to that of the anchialine ponds of Bermuda, which maintain sea connection through underground caves [21,26].
Consequently, this category probably represents two distinct types of permanent ponds: those with underground sea connection (e.g. SIN) and those with no sea connection (e.g. BAN).
5. Temporary ponds without surface sea connection (BLU, HAN, JOS, LEE, NOR and RUN):
Inundation period in this group of ponds ranged from three to eight months per year. Inundation was controlled by rainfall and evaporation. Seasonally high tides did not force water into these ponds. Recorded salinities ranged from hyposaline (for brief periods after hurricane rains) to three times the seawater salinity (Table 2). In half of these ponds (BLU, JOS, RUN), tidal influence on water level, once inundated, was detectable, suggesting that their bottoms were at or below sea level. The other half of these ponds (HAN, LEE and NOR) were the smallest and shallowest ponds studied, and they had the shortest inundation periods (three to five months).
These hydrologically distinct groups of ponds reflect a trend from high connectivity with the sea to complete isolation from the sea. This trend is also apparent when the geomorphologies of bays and lagoons are considered. Completely open bays, or bights, lie at one extreme, followed by bays that are partially restricted by shallow coral reefs. Where mangroves colonize the exposed tops of these coral reefs, sediment accretion by mangrove roots leads to the construction of berms, which restrict wave energy within lagoons. Salt ponds lie at the other extreme, where a berm completely (in most cases) separates the inland water body from the sea.
The existing hydrologic variability observed among BVI salt ponds suggests a pattern of geologic evolution from shallow coastal marine waters to salt ponds and eventually to dry land. BVI salt ponds thus provide present-day examples of natural wetland transformations that have previously been described by Dix et al.  in their geologic analysis of marine sediments from a Bahamian salt pond. A possible route through the observed stages of salt ponds is diagrammed in Figure 10. A pond with direct sea connection, depending on its geomorphic and sediment characteristics, might evolve into any one of four different types of ponds: 1) a permanent pond with underground sea connection, 2) a permanent pond with a seawater seep, 3) a permanent pond with no sea connection, or 4) a temporary pond with periodic direct sea connection. Of these, a permanent pond with a seawater seep might transform into one with no sea connection as the seep becomes blocked by sediments; a permanent pond with no sea connection may, in turn, become a temporary pond as it collects sediments until its bottom lies above sea level; and a temporary pond with periodic direct sea connection may become one with no sea connection as the berm through which sea connection is established is stabilized by mangrove roots and accreted sediments. The fate of a permanent pond with underground sea connection to become a temporary pond with no sea connection is unclear, as this type of pond was unique in our data set. Regardless, senescing ponds should all reach the stage of a temporary pond with no sea connection, and as these ponds fill with erosional sediments and as salts are washed out by rainfall, dry forest trees will encroach along the borders into the former pond.