Salinity fluctuated widely within ponds and mean salinities varied among ponds (Figure 3; Table 2). All of the 10 ponds sampled monthly over a full year were hypersaline, on average, with half having mean salinities below 100 ppt and half above (Figure 3). Nine of the 17 sampled ponds varied into mesosalinity (20 – 50 ppt ), and one, JOS, became hyposaline () after sustained and heavy rainstorms. Lowest salinities were found in JOS and WB, and variability was lowest in WB.
Highest salinities occurred in BEL and SAL. These ponds contained seawater seeps that continually supplied salts without introducing sufficient seawater to cause flushing. Sodium chloride crystallized over the bottom of SAL during extended dry periods in which pond salinities exceeded 300 ppt, the precipitation point of NaCl. At the same time gypsum crusts formed in BEL, which during dry periods experienced salinities between 175 ppt (the precipitation point for gypsum, CaSO4·2H2O) and 300 ppt (Figure 4). RED was also a highly saline pond, and most of its bottom was covered with a precipitated gypsum pavement. A seep may have existed at RED, but the perimeter of this pond was not fully explored during this study.
During 1995, salinity in all ponds increased between February and August, the driest part of the year, and it declined between August and November, the wettest part of the year (Figure 4). Dilution by rainfall was common to all ponds.
Dilution effects were more precisely illustrated by the hurricane monitoring data (Figure 5 &6). Hurricane Luis (Sept. 4th, 1995), with 7 cm of rain, caused the salinity in BAN to fall 37% in two days (Figure 5). Ten days later, 12 cm of rain fell during Hurricane Marilyn (Sept. 15th, 1995), bringing about a further 23% decline in salinity. Thus over ten days, the salinity of BAN was reduced by 50%, from 120 ppt to 60 ppt. Afterwards, the salinity increased slowly to 87 ppt by December.
Hurricane Georges (September 21st, 1998) affected BAN similarly (Figure 5). It brought 6.5 cm of rain, which diluted the salinity of BAN by 23%, from 83 to 64 ppt. Salinity gradually increased to 71 ppt by October 14th, but another 7.5 cm of rain fell between October 21st and 23rd and caused a further 15% dilution to 60 ppt. Continued rainfall through the following month maintained relatively low salinity in this pond.
Similar effects were seen in 3 other ponds monitored during Hurricane Georges (Figure 6). Salinity in GUA dropped by 27%, from 83 to 61 ppt; salinity in JOS dropped by 65%, from 72 ppt to 25 ppt; and salinity in LON dropped by 67%, from 130 ppt to 43 ppt. After Hurricane Georges, salinity in these ponds gradually increased until the October 21st rains, which diluted GUA by 34%, from 62 to 41 ppt, diluted JOS by 73%, from 45 to 12 ppt, and diluted LON by 21%, from 63 to 50 ppt.
Pond water did not show salinity stratification except for short periods after rainfall. In GUA, 6.5 cm of rain (on 28 and 29 July, 2000) dropped the salinity of the surface water from 75 ppt to 47 ppt. Over the following 36 hours, surface and bottom salinity gradually approached equilibrium at 57 ppt. This pattern was repeated on a smaller scale after a rainfall of 0.18 cm on 19 August, 2000, when the surface water of GUA was diluted to 6 ppt less than bottom water (73 ppt), and equilibrium (71 ppt) was achieved after 9 hours.
Ground water salinity
Salinity in groundwater wells at the vegetation/shore border of FLA averaged 81 ± 4 ppt on 9 August, 1998, while the pond salinity was 89 ppt. No halocline was detected in the groundwater wells. Ten meters behind the shore vegetation where the mangrove fringe began, groundwater salinity (46 ± 0.7 ppt) was closer to that of seawater.
At BLU, groundwater salinity averaged 54 ± 6 ppt on January 23, 1999. Groundwater salinity showed only a small increase to 62 ± 10 over the following 2 months, during which time the pond lost 27 cm in depth and gained 64 ppt in salinity via evaporation (Figure 7).
The rate of long-term evaporation in BAN was 32 m3 H2O day-1ha-1, between 26 March and 26 April, 1995. A corresponding salinity concentration of 125%, from 160 ppt to 200 ppt, was measured in BAN during this drought period.
Short-term evaporation rate in BAN in April, 2000 (Table 3), included a clear day and a cloudy day, was similar (32 to 45 m3 H2O day-1ha-1) to the long-term evaporation rate measured in April, 1995. No salinity change was detected (± 1 ppt minimum detection limit). No evaporation occurred at night.
Short-term evaporation in JOS was very high (150 – 160 m3 H2O day-1ha-1) during daylight hours in April and July (Table 3), corresponding with a 4 ppt increase in salinity (from 124 to 128 ppt) in one afternoon. Evaporation during evening hours was lower (53 m3 H2O day-1ha-1), and no evaporation occurred during the night. By projecting these measurements over a whole day (proportioning daytime, dawn/dusk, and night hours), the whole-day mean evaporation rate was estimated at 64 m3 H2O day-1ha-1.
Inundation and seasonal depth fluctuations
Salt ponds were found to differ in their inundated periods and in their degree of connection with the sea (Table 4). All ponds were shallow, with a ten-pond depth average of 29 ± 20 cm during 1995, though substantial variation existed among ponds (Table 4). Nearly half of the ponds dried completely for between 3 and 9 months annually (temporary ponds), while other ponds never dried (permanent ponds). All ponds received freshwater input via rainfall and runoff, but only eight of 17 ponds received direct seawater input. In the latter group, three types of seawater influence were found. Permanent sea connection, through a narrow channel, occurred in four ponds (BON, FLA, PTP & RED). Two other ponds (BEL & SAL) received seawater input via through-ground seeps that were visible at the shoreline. The remaining two ponds (LON & WB) were periodically connected with the sea during seasonal high tides (June through December, Figure 8).
HAN, LEE and NOR held water only for short periods after rainfall, and they were inundated for a total of 3 to 4 months annually. Bottom sediments were sandy.
BLU, RUN and JOS (Figure 9a) were also greatly influenced by precipitation, but these ponds were continuously inundated for 7 months or more in most years. Bottom sediments in these ponds were predominantly sand, though in RUN and BLU patches of soft mud were also present.
GUA was a temporary (non-permanent) pond until 1990, after which saline water input via an overflow pipe and an effluent pipe from a nearby desalination plant resulted in permanent inundation. Regular input of seawater maintained a relatively stable water level in GUA for most of 1995, though water level dropped 12 cm during the April drought and rose 47 cm after hurricane rains in September (Figure 9b).
WB, with an initial depth of 28 cm in January, was completely dry by April (Figure 9c). The low-volume rains (1 cm) of May 7th and 8th, which inundated JOS (Figure 9a), did not fill WB, presumably because of its smaller watershed area (Table 4). One week after these rains, a high tide of 23 cm above mean low water, which was the first of this magnitude for the year (Figure 8), filled WB to 15 cm (Figure 9c) when seawater breached a low point in the berm separating WB from the sea. Seawater connection through the resulting channel occurred regularly during high tides from June through December, when sea level was higher than earlier in the year (Figure 8). Water level in WB did not fluctuate greatly as long as the connection with the sea was maintained (Figure 9c). The hurricane rains of September had no effect on water level in WB. Direct connection with the sea during the hurricanes presumably allowed rapid equilibration of water level in WB by draining runoff to the sea.
Episodes of seawater inundation, similar to those in WB, were observed over several years (1998 to 2001) in LON, although depth was not monitored in this pond. Three concrete culverts, constructed to allow water to drain under the road that separated LON from the shore on the southern side of the pond, allowed seawater into LON in summer and fall when sea level was highest. The bottom of LON was composed mostly of sand with some patches of organic mud, while that of WB was composed of organic mud.
BAN and SIN were both permanently inundated but had no surface sea connection. The inundated portion of BAN shrunk to approximately 60% of its initial size and lost 22 cm in depth during the driest part of the year (Feb – May, Figure 8d). Hurricanes Luis and Marilyn raised the level of BAN by 26 cm and 13 cm, respectively. Water level fell 9 cm immediately after Hurricane Luis and 13 cm within days of Hurricane Marilyn, presumably via drainage through the sand and coral berm that separated BAN from the sea. The bottoms of both ponds were uniformly covered by organic mud, but unlike SIN, BAN supported benthic microbial communities that formed a firm gelatinous layer over the mud.
BEL and SAL were both permanently inundated and received small but constant seawater input through underground seeps, where water was observed trickling out through the shore sediments and into the ponds. Salinity of water emerging from these seeps was consistently near 40 ppt, which indicated a seawater origin. The berm in the seep area at SAL was composed of coral rubble, while the berm in the seep area at BEL was composed of organic and silty sediments. Depth fluctuations in these ponds were similar to one another and both ponds responded to rainfall and evaporation cycles (Figure 9e &9f). BEL and SAL became shallower and smaller during the drought period in March and April, 1995. The water level at BEL declined below the marked sampling area and thus exposed the depth reference point, hence the zero depth values for April, June, and July in Figure 9e. With increasing sea level between June and August, 1995, (Figure 8) greater seawater input through seeps at BEL and SAL was expected. Contrary to expectation, however, water level in these ponds did not rise in the summer. Rains from the September hurricanes, on the other hand, raised the water level more than 50 cm in both ponds. Bottoms these ponds were composed of thick organic mud deposits.
The Anegada ponds, BON, FLA, PTP and RED, were connected with one another and maintained a direct connection with the sea through a narrow channel near PTP. This channel measured 3.3 m wide and 37 cm deep at its narrowest point. Depth fluctuations in these ponds were generally similar to one another, and the ponds responded to seasonal changes in mean sea level more so than to rainfall and evaporation cycles (Figure 9g–j). Bottom sediments in this group of ponds varied from sand (PTP) to organic mud (BON). Sediments in FLA and RED were covered in most areas by a hard crust composed predominantly of gypsum crystals (CaSO4-2H2O).
Tidal influence on pond water levels
BAN, GUA and JOS showed no evidence of tidal influence on pond water levels (Table 4). HAN, LEE and NOR were presumed to also be non-tidal because they remained dry, except for relatively short periods after rainfall, even during seasonally high tides from May through July.
Water levels in BLU, LON and RUN were influenced by tidal forces (Table 4), but this influence was not always detectable. Tide-induced fluctuations were detected on two out of four measurement periods in BLU, on two out of three measurements in RUN and in three out of four measurements in LON. The largest observed increase in BLU's water level occurred 10 hours after a high sea tide. In contrast, high water in LON and RUN generally occurred within one hour of the high tide. Tidal influence was not detected in LON or WB when, during periods of lower sea level, they lost direct connection with the sea. Tidal water level fluctuations were undoubtedly experienced in LON and WB at times of direct sea connection.
The water level in BEL rose substantially (0.7 cm) in response to a rising tide on 10 July, 2000, the only date on which it was sampled.
A water-level rise of 2.6 cm in SIN was higher than that measured in any other pond, and it occurred in response to a moderate high tide of only 14 cm above m.l.w.
All of the Anegada ponds (BON, FLA, PTP, RED) showed regular water level fluctuations in response to tidal cycles, and the timing these fluctuations closely followed that of tides.