Introduction

Mangroves are recognized for the important ecosystem services they provide, including habitat for many species of birds, fish, and crustaceans (Mumby et al. 2004; Montalto et al. 2006; Nagelkerken et al. 2008), cultural resources (Barbier et al. 2011), and most recently, for the ability to sequester and store carbon (Jardine and Siikamäki 2014). Mangroves also buffer interactions between terrestrial and oceanic environments, protecting land and human development from storm surges (Shepard et al. 2011; Zhang et al. 2012; Kathiresan and Rajendran 2005; Mazda et al. 1997), and by trapping terrestrial pollutants before they reach the nearshore marine environment (Tam and Wong 1999; Clark et al. 1998; Cavalcante et al. 2009; Sthevan et al. 2011; Ismail et al. 2014).

Despite the many valuable ecosystem services provided by mangroves worldwide, their area has declined by 35% since the 1980s (Valiela 2001) and improved mapping at the global scale using remote sensing shows their distribution to be even more limited than other recent estimates suggest (Giri et al. 2011). Urban development, deforestation, shrimp aquaculture, overexploitation for resources, and changes in sea level, temperature, and precipitation are all drivers of mangrove loss (Ellison and Farnsworth 1996; Spalding et al. 1997; Alongi 2002; Gilman et al. 2008). Urbanization is one of the major threats to mangroves and other coastal wetland systems, through direct habitat loss and indirect impacts like changes to hydrologic and sediment regimes that result in introductions of pollutants, including heavy metals, to these coastal systems (Lee et al. 2006, 2014; Fonseca et al. 2013; Zhang et al. 2014). In the U.S. Virgin Islands, years of direct habitat loss and indirect stressors, have resulted in few remaining intact stands of mangroves; this is particularly true on the island of St. Thomas (Platenberg 2006).

On St. Thomas, the largest remaining mangrove system occurs within the St. Thomas East End Reserves (STEER), a marine protected area (Horsley Witten 2013, Fig. 1). However, this system is under considerable stress from land-based sources of pollution, including Bovoni Landfill, boatyards and marinas, sewage effluent from a wastewater treatment plant, and surface runoff from its relatively large inland watershed (11.8 km2, the largest watershed on St. Thomas, Horsley Witten 2013) that is home to over one third of the population of St. Thomas (Horsley Witten 2013). Previous work by Pait et al. (2014) revealed greater heavy metal contamination of marine sediments in parts of STEER closer to the landfill and boatyards (Mangrove Lagoon and northern Benner Bay) for chromium, copper, lead, mercury, nickel and zinc, compared to other areas within STEER’s boundaries. Because of the study’s sampling design, however, the authors could not pinpoint Bovoni Landfill as a definitive source of lagoonal sediment contamination. The authors suggested that both the boatyards and the landfill were likely sources of these heavy metals, but recognized other potential sources including a horse track and ephemeral waterways that could transport upper watershed contaminants into STEER (Pait et al. 2014). In 2006, however, the Environmental Protection Agency (EPA) observed violations of waste management at the unlined landfill, including improper disposal of medical and septic waste, used oil, and lead-acid batteries, as well as the migration of leachate into the adjacent mangrove system (District Court of the Virgin Islands, 2006). Furthermore, in 2012, a representative of The Nature Conservancy in St. Thomas, expressed concern about the mangroves located between the landfill and Mangrove Lagoon due to observed mangrove degradation and apparent loss that had occurred in recent years (Anne-Marie Hoffman, pers. comm.).

Fig. 1
figure 1

An aerial view of Mangrove Lagoon, Bovoni Landfill, and the sample locations within the study area. Arrow in the inset indicates the location of the study area on St. Thomas, USVI

Full size image

Landfills can be significant, well-known sources of heavy metals to the marine environment through surface and groundwater contamination pathways (Jones 2010). Heavy metals like mercury, nickel, copper, chromium, and zinc are problematic because they bioaccumulate within marine organisms, negatively impacting organism function by interfering with metabolic processes (Islam and Tanaka 2004). Coastal habitats, like mangroves, can lessen this effect by intercepting land-derived heavy metals and sequestering them in sediments and vegetation, thereby providing a critical ecosystem service for nearby marine systems (MacFarlane et al. 2007). Knowledge of the hydrological and geomorphological processes within mangrove systems is critical for: (1) understanding their structure (Kjerve et al. 1999; McKee and Faulkner 2000) and ecosystem services (Ewel et al. 1998), (2) identifying the potential spatial and/or temporal variability in the provision of the chemical and sediment transport and buffering capacity (Lee et al. 2014), and (3) developing appropriate conservation and sustainable management practices for these coastal systems (Calderon et al. 2014). Thus, the objectives of this study were three-fold: (1) to document potential heavy metal contamination of sediments and surface and groundwater within the mangrove system abutting Bovoni Landfill, (2) to understand how this mangrove fringe may buffer potential terrestrial impacts to the lagoon through improved documentation of groundwater flow patterns within its boundaries through field observations and modeling, and (3) to explore how the provisioning of this key ecosystem service may change in response to external forcing factors, like wet and dry seasonality, precipitation, and tidal events.

Methods

Study site

St. Thomas is approximately 80 km2 with a population of 51,000 people and receives approximately 2 million visitors a year (Fig. 1; US Census Bureau 2013; U.S.V.I Bureau of Economic Research 2013). Mangrove Lagoon, located in the western-most portion of STEER, supports local tourism businesses, provides habitat for lobsters, conch and sea turtles, and is a nursery for commercially-important fish (STEER 2011). Mangrove Lagoon was declared an Area of Particular Concern (APC) in 1978 by the Territory’s Coastal Zone Management’s Planning Office to preserve it as a significant natural area (DPNR 2003). At that time, the Planning Office found the lagoon to have “exceptional natural value” because of: (1) the anchorage provided to boats during storms, (2) its numerous recreational opportunities, and (3) the protection its mangrove-fringed shores provided against shoreline erosion, flooding, and hurricane waves. Mangrove Lagoon was later incorporated into a larger group of marine reserves and wildlife sanctuaries on the east end of the island, collectively called the St. Thomas East End Reserves (Horsley Witten 2013).

Between Mangrove Lagoon and Bovoni Landfill is the largest remaining intact stand of mangroves on St. Thomas (Horsley Witten 2013, Fig. 1). Our study site is located within this mangrove system. Black mangroves, Avicennia germinans, dominate the inland area closer to the landfill and red mangroves, Rhizophora mangle, dominate the outer fringe that borders Mangrove Lagoon. Over time, the landfill’s historic dumping areas have expanded into the adjacent mangrove system and have decreased the distance between the landfill and Mangrove Lagoon (Horsley Witten 2013). During this study, accessing the mangrove system by land required navigating around large piles of scrap metal into the vegetation bordering the eastern most access road (Fig. 1).

Conditions within the study area vary depending on time of year and the amount of recent rainfall. The Virgin Islands, characterized as a dry tropical climate (Bowden et al. 1970), do not have sharply defined wet and dry seasons (Crossmand and Palada 2003), but on average receive the most rainfall from September through November and the least from January through March (NOAA Average Rainfall, Accessed 11 Apr 2016). The majority of rainfall from May to November results from easterly winds (which may develop into tropical storms and hurricanes), while the majority of rainfall from December to April is the result of cold fronts (Calversbert 1970). Standing water levels in the study area vary depending on location within the mangrove swamp and time of year. Locations closer to the lagoon have standing water year round (or nearly year round), but further inland, the presence of standing water varies. The surface sediment inland is mostly dry or completely dry and cracked during the dry season but contains standing water during the wet season (Fig. 2). The study area is microtidal with a mean range of 0.21 meters (NOAA Tides and Currents 2016).

Fig. 2
figure 2

Comparison between wet and dry seasons at sites 5 (a, c) and 11 (b, d). The top figures (a, b) were taken during January and February while the lower figures (c, d) were taken during November. Photo credits: a, c, and d taken by J Keller, b taken by K Wilson Grimes

Full size image

In this study, ten sites were placed throughout the mangrove system to capture potential hydrologic spatial variation. The ten sites were grouped into three zones based on proximity to Mangrove Lagoon and Bovoni Landfill. The zones serve to highlight the hydrologic and sedimentological differences in different parts of the mangrove swamp. The outer zone encompasses the 5 sites (1, 12, 10, 7, and 6) located close to the mangrove fringe separating the mangrove swamp from the lagoon (Fig. 1). The middle zone contains the two sites (2 and 13) located in the middle of the mangrove swamp. The inner zone is comprised of the three sites (5, 11, and 4) located furthest inland and closest to Bovoni Landfill. Except for the hydraulic head contour maps, all results are presented in terms of these three zones.

Measuring groundwater flow

To determine groundwater flow direction, nineteen groundwater wells (2.54 cm nominal diameter flush threaded polyvinyl chloride (PVC), 30 cm machine slotted screen) were installed across the study area (Fig. 1) by manually driving wells into core holes created with a “Dutch” hand-auger corer. A shallow (1.5 m total length) and a deep (3 m total length) well were placed at each sampling station with two exceptions. At site 7 only a shallow well was installed (due to lack of materials) and at site 5 the deep well had a total length of 2.1 m because it met refusal. Following installation, all wells were developed by surging the water with a sealed plastic pipe and pumping the sediment and water slurry from the well using a hand vacuum pump.

To determine absolute positions and provide reference points for manual surveying, deep wells at four sites (1, 4, 5 and 6) were surveyed using a global positioning system (GPS; Trimble NetR9 GPS receiver and Zephyr II geodetic antenna). GPS data were post-processed using the GPS Precise Point Positioning service run by Natural Resources Canada, Geodetic Survey Division. All remaining wells were manually surveyed to the top of the pipe using a CST/Berger 32x Automatic Level and stadia rod, to calculate the relative elevation differences between wells. Well elevations were verified by comparing optical survey data with GPS data (± 3 cm accuracy). GPS coordinates for all wells were recorded with a hand-held GPS unit (Trimble Navigation Limited).

Hydraulic heads were manually measured using an electrical water meter (Solinst Canada Ltd.) and continuously monitored using data-logging pressure transducers (Solinst Canada Ltd. Leveloggers) to determine vertical and horizontal groundwater flow patterns within the mangrove system. Non-vented pressure transducers provided a nearly-continuous history of hydraulic head over the study period (January 6th to November 11th 2014), while the manually-measured values served as a quality control and were used to reference all data to a common vertical datum. Data loggers recorded water pressure and temperatures every ten minutes. Water pressure data were compensated with atmospheric pressure values recorded with a separate barometric pressure transducer at site 11. The compensated values were offset to align with the manual hydraulic head data.

Hydraulic head contour maps were created to determine seasonal changes in horizontal groundwater flow. The depth to water values in each well were manually measured five times in 2014 (2/13, 3/6, 5/6, 7/1 and 11/11) using an electric water level meter. The depth to water data were subtracted from the surveyed well elevations to obtain hydraulic head data relative to mean sea level. Contoured potentiometric surface maps were created by interpolating hydraulic head data using kriging routines (Childs 2004; Davis et al. 2009; Rabah et al. 2011) available in ArcGIS 10.2 (ESRI 2013). Empirical Bayesian Kriging was used for shallow hydraulic head interpolations, but could not be used for the deep hydraulic head interpolations because of the small sample size (n < 10). In the case of the deep hydraulic head interpolations, simple kriging was used instead.

To test surface water and groundwater for possible contaminates, water samples from both the deep and shallow wells were taken at four sites close to the landfill (2, 4, 5, and 11) and one site close to the lagoon (1) on January 9th, 2014. Surface water samples were also taken at site 5 due to an observed chemical smell during well installation and at a surface ditch containing greenish colored water (Fig. 1). Due to high sediment concentrations, water samples were not filtered in the field, but were filtered in the lab prior to analysis. Samples were collected with no headspace to limit gas exchange, kept in a cooler on ice until refrigerated and processed for pH and specific conductance within 9 h of collection. Samples were shipped to the University of Maine’s Sawyer Environmental Chemistry Laboratory, filtered, and tested for copper (Cu), chromium (Cr), nickel (Ni), lead (Pb), tin (Sn), zinc (Zn), and total dissolved nitrogen (TDN). These six heavy metals are found in sheet metal, wire, chrome plating, appliances, batteries and paint (Agency for Toxic Substances and Disease 2014), all of which are likely to be found in Bovoni Landfill. Available resources limited the scope of the chemical analysis and a small suite of heavy metals were chosen based on their likely presence and their potential to affect organism health. Heavy metal analysis was performed using a Perkin-Elmer Optima 3300XL inductively coupled plasma atomic emission spectrometer and TDN analysis was performed with an ALPKEM flow solution IV autoanalyzer. Specific conductance and/or salinity of surface waters were measured using a portable meter (Hach Company) in the field during well installation. Values were not recorded at sites with very little or no standing water.

To determine vertical fluid velocity in shallow groundwater, we measured the spatial and temporal distribution of temperatures in the subsurface. Groundwater movement affects these distributions, and the use of heat as a tracer to assess groundwater interaction with surface water systems has increased over the past two decades as low cost sensors have become available and data analysis techniques have been refined (Rau et al. 2014). One method uses the decrease in amplitude and lag in the diurnal temperature signal with depth to estimate the vertical groundwater velocity (Stallman 1965). This change in the temperature signal was simulated using a heat transport model and the simulated signal was fit to temperature data by adjusting the vertical groundwater velocity. In this study, high-resolution temperature measurements of groundwater were recorded using data logging temperature sensors (Maxim Integrated Thermochron Ibuttons). Four sensors were installed in a vertical array at depths of 0, 7, 14, and 21 cm. These temperature sensor arrays were placed at sites 1, 2, 5 and 7 and left for one week in February 2014. A one-dimensional finite-difference model, similar to that described by Vandenbohede and Lebbe (2010), was calibrated to temperature data collected at these depths while vertical groundwater velocity and sediment porosity were adjusted in simulations to achieve a best fit.

Sediment properties and stratigraphy

A 3-cm diameter Eijkelkamp “Dutch” hand-auger corer was used to collect 1 to 2 m sediment cores at nine of the ten sites on January 6, 2014. No core was recovered at site 12, due to slippage of the sediment from the core tube. Length of the cores varied depending on sediment composition (soil hardness and consistency). Most cores were about 2 m long, but only 70 cm was collected at site 6, and 150 cm at site 5 because we met refusal.

Cores were split, photographed, stratigraphically described, and sub-sampled at 5 cm intervals for dry bulk density and percent organic matter using loss on ignition techniques (Ball 1964; Craft et al. 1991; Heiri et al. 2001). Sediments were interpreted as mangrove peat (high amounts of peat, roots, branches, and/or fine root hairs) or mud flat/pool (clay-rich sediment largely lacking plant macrofossils) environments. Core sub-samples were dried at 80°C for 8 h until a constant mass was achieved. Dry weight and wet volume were used to calculate dry bulk density, where:

$${\text{Dry bulk density }}\left( {{\text{g}}/{\text{cm}}^{ 3} } \right) = {\text{dry weight }}\left( {\text{g}} \right)/{\text{wet volume }}\left( {{\text{cm}}^{ 3} } \right)$$

Dried samples were ground with mortar and pestle, homogenized over the sample interval, and combusted at 550 °C for 4 h. Percent organic matter by loss on ignition (Ball 1964; Craft et al. 1991) was calculated by mass loss, where:

$${\text{Percent organic matter}} = \frac{{\left( {{\text{dry mass before combustion}}{-}{\text{dry mass after combustion}}} \right)}}{{{\text{dry mass before combustion}} \times 100}}$$

Sediment shear strength is a measurement commonly used in wetlands to understand differences in sediment properties and susceptibility to erosion (Wigand et al. 2014; Turner 2009; Turner et al. 2011). Sediment shear strength was measured in situ at each well site using a 16 × 32 mm Humbolt inspection vane (Elgin, IL) in July 2014. Sediment profiles of shear strength were created by measuring the stress required to force sediment failure (in kilopascals) every 10 cm from 10 to 100 cm depths. Three replications were performed at each site and the mean value was calculated for each depth.

For a more comprehensive understanding of the subsurface structure, stratigraphic interpretations, dry bulk density, water content, organic content, and shear strength profiles for each site were plotted together. Values for dry bulk density, water content, organic content, and shear strength were then categorized as either mangrove peat or mud flat/pool sediments based on our stratigraphic interpretations. Two-sample unequal variance t-tests were performed to compare these values between mangrove peat and mud flat/pool.

To better understand how the subsurface stratigraphy changes throughout the study area, cross-sections of stratigraphic interpretations were created in zones based on proximity to the landfill and Mangrove Lagoon. Three cross-sections (sites 4, 5 and 11; sites 2 and 13; sites 1, 6, 7, and 10) were created to view connectivity across the mangrove swamp while looking for differences among these three zones. A fourth cross-section (sites 5, 13 and 1) was created to more easily examine how the stratigraphy changes from the inner zone toward the lagoon.

Dried sediment samples were analyzed for heavy metal presence using scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS). Other studies have used SEM/EDS to determine the source of sediments in estuaries and examine sediments for heavy metal presence (Zadora and Brożek-Mucha 2003; Rajkumar et al. 2012). In this study, scanning electron microscopy (Tescan Vega II XMU SEM) and energy dispersive spectroscopy (EDAX Genesis system) were used to detect the presence of heavy metal (besides Fe-containing sediments) in selected, dried sediment samples at the University of Maine. At least two samples (one near the top and one near the bottom) from every core were tested for heavy metals (Online Resource 1), while cores from sites 4, 5 and 11 had additional samples tested due to their proximity to the landfill. Elemental analyses were performed on locations within samples based on backscatter analysis; backscatter analysis allows the rapid identification of locations containing elements with high atomic number (Goldstein et al. 1992). Energy dispersive spectroscopy was only performed on sediment particles with high atomic numbers as identified by the backscatter analysis of the SEM.

Results

Groundwater flow and water chemistry

Horizontal groundwater flow directions, determined from manual measurements (Table 1), varied seasonally with shallow and deep groundwater behaving similarly to each other throughout the study period (Fig. 3). In February and March, groundwater flowed into the center of the mangrove system from the direction of Mangrove Lagoon and from the direction of the landfill. In May, flow was from the lagoon toward the landfill. In July, groundwater flowed from the landfill into the mangroves, although the differences among shallow hydraulic heads were smaller and flow was in a more easterly direction compared to the deep hydraulic head (Fig. 3b). In November, flow was from the landfill toward the lagoon in both deep and shallow groundwater.

Table 1 Well locations, elevations, and hydraulic head of each manual measurement
Full size table
Fig. 3
figure 3

Hydraulic head (cm) contours for shallow (a) and deep (b) groundwater with equipotential lines and arrows to indicate direction of flow. Groundwater flowed into the mangroves from both the upland and from the lagoon in February, into the mangroves from the lagoon in May, and into the mangroves from the upland in July and November. The north arrow and scale bar apply for all eight images. Bovoni Landfill is to the southwest of the study area

Full size image

Hydraulic head time-series data revealed differences in groundwater responses between seasons and among zones. At all sites, hydraulic head was higher in the wet season, from May until November, than the dry season. Within seasons, hydraulic head patterns varied among zones. Differences between shallow and deep hydraulic heads were larger in the inner zone (5, 11, and 4) than the outer zone (1, 6, 7, 10, and 12), where these differences were negligible Figs. 4 and 5. At sites 4 and 5, recharge occurred for a large portion of the study period when shallow groundwater was higher than deep groundwater, but at site 11, discharge was more common. In the inner zone, patterns in hydraulic head measured in deep wells lagged behind and exhibited a damped response, compared to hydraulic head measured in shallow wells. Hydraulic head data collected at site 2 and site 13, both located in the middle zone, contained patterns similar to the outer and inner zones, respectively (Online Resource 2).

Fig. 4
figure 4

Hydraulic head time series of both shallow (grey) and deep (black) wells installed in the outer zone, near the lagoon. Hydraulic head plots are stacked based on spatial location of each site (Fig. 1). Continuous pressure transducer values are shown as lines and symbols indicate manually-measured values. The pressure transducers malfunctioned in the deep well at site 6 and from the shallow well at site 12 after July

Full size image
Fig. 5
figure 5

Hydraulic head time series of both shallow (grey) and deep (black) wells installed in the inner zone, closest to the landfill. Hydraulic head plots are stacked based on spatial location of each site (Fig. 1). Continuous pressure transducer values are shown as lines while symbols indicate manually-measured values. When shallow hydraulic head is above deep hydraulic head, recharge is occurring and when the opposite is true, discharge is occurring

Full size image

Comparing hydraulic head time series of sites 1 and 5 to the recorded mean sea level and precipitation revealed differences between the inner and outer zone in hydraulic head responses to external factors (Fig. 6). Shallow and deep hydraulic head at site 1 responded quickly to spring high tides in the beginning of the study period and did not have an obvious response to two large rain events on May 7th and August 22nd. Conversely, site 5 was more responsive to precipitation events than tidal changes (Fig. 6). This site had a quick increase in shallow hydraulic head with a slower increase in deep groundwater during the large rain event in May. However, hydraulic head of neither the shallow nor deep wells at site 5 showed much of an increase after the second large rain event August 22nd.

Fig. 6
figure 6

Time series of hydraulic head (cm) for both shallow and deep groundwater at site 1 (a) and site 5 (b). Lines (grey for shallow, black for deep) represent values recorded with pressure transducers while symbols represent manually measured values. Below these are recorded water levels based on mean sea level from Charlotte Amalie Harbor (c) (NOAA/NOS/Center for Operational Oceanographic Products and Services) and precipitation from the Bovoni weather station (d) (The Weather Channel 2015)

Full size image

Hydraulic head of both shallow and deep groundwater at site 1 (Fig. 7a) responded mainly to sea-level changes during the first large rain event May 7–10 (Fig. 7). In the shallow well, the hydraulic head increased approximately 2 cm on May 7th, but this increase may be from the incoming high tide and not from precipitation. Site 5 (Fig. 7b) did not have the tidal pattern seen at site 1, but had an abrupt 20 cm increase in shallow groundwater and a slow increase of approximately 5 cm in deep groundwater during the rain event. The slower increase in deep hydraulic head suggests the sediments at depth have a low hydraulic conductivity, insulating the deeper well from surface conditions. At site 1, hydraulic head dropped after the large increase on May 7th, while shallow hydraulic head remained high at site 5.

Fig. 7
figure 7

Hydraulic head at site 1(a) and at site 5 (b) with sea-level and precipitation during a rain event May 7th–10th. Grey lines represent shallow hydraulic head and the black lines represent deep hydraulic head. Recorded water levels (c) are based on mean sea level from Charlotte Amalie harbor (NOAA/NOS/Center for Operational Oceanographic Products and Services) and precipitation (d) is from the Bovoni weather station (The Weather Channel 2015)

Full size image

The hydraulic head at these two sites responded differently to the second large rain event August 21–23 (Fig. 8). Shallow and deep groundwater at sites 1 (a) and shallow groundwater at site 5 (b) have similar patterns to the recorded water level, while the deep groundwater at site 5 had no tidal signal. Hydraulic head in all four wells had a slight alteration of their pattern during the peak rain day, August 22nd, but no large response to the precipitation was recorded. At site 1, recharge occurred immediately following the incoming tide, and discharge occurred when the shallow hydraulic head fell faster than the deep hydraulic head. At site 5, hydraulic head did not show an increase in response to rain and recharge occurred throughout the rain event. Deep hydraulic head increased slowly after the rain event, but the recorded mean sea level also appeared to increase during this time.

Fig. 8
figure 8

Hydraulic head at site 1(a) and at site 5 (b) with sea-level and precipitation during a rain event August 21st–23rd. Grey lines represent shallow hydraulic head and the black lines represent deep hydraulic head. Recorded water levels (c) are based on mean sea level from Charlotte Amalie harbor (NOAA/NOS/Center for Operational Oceanographic Products and Services) and precipitation (d) is from the Bovoni weather station (The Weather Channel 2015)

Full size image

Daily fluctuations of hydraulic head occurred from May to November and appear to match the diurnal tidal signal (Figs. 4, 5, 6). In the outer zone, these daily fluctuations in hydraulic head were measured in both shallow and deep wells. In the inner zone, daily fluctuations were observed in shallow, but not deep, hydraulic head measurements. This daily tidal signal only appeared when hydraulic head values were above ~45 cm. Below this threshold, monthly spring high tides affected hydraulic head, but the diurnal signal was not present.

Chromium was found in all water samples, but other heavy metals detected (nickel, tin, and zinc) were only found at site 5 or at the surface ditch (Table 2). The surface ditch sample contained nickel, tin, and the highest concentrations of chromium and TDN. Lead and copper were not found above their detection limits (80 μg/L for Pb and 100 μg/L for Cu) in any sample.

Table 2 Physical and chemical properties of groundwater and surface water samples
Full size table

Small vertical velocities produced by the heat transport model (Table 3), indicate very slow vertical movement of shallow groundwater. Site 5 was the only site to have a positive fluid velocity value, representative of groundwater recharge during the sensor deployment in February. Negative fluid velocities at sites 1, 2 and 7 represent groundwater discharge. Porosity was lowest at site 5, the only inner zone site, although values were similar between all four sites.

Table 3 Fluid velocity, sediment porosity and error (sum of squares) produced by fitting subsurface temperatures to the heat transport model
Full size table

Sediment properties and stratigraphy

Cross–sections of the subsurface stratigraphy from cores revealed clear differences between inner and outer zones. Sediment cores from the outer zone were dominated by mangrove peat (Fig. 9a), while cores from the inner zone had substantial muddy deposits characteristic of mud flat or pool sediments interspersed with sections of peat (Fig. 9b). A cross-section of subsurface stratigraphy of sites perpendicular to the mangrove fringe shows the transition of sediments from the inner to outer zones (Fig. 10), with a decrease in mud flat/pool sediments moving toward Mangrove Lagoon.

Fig. 9
figure 9

Cross-section of core stratigraphy for the inner zone (a) and outer zone (b). Cores are corrected for topographic elevation. Dotted lines in between cores connect interpreted environments (mangrove peat or mud flat/pool

Full size image
Fig. 10
figure 10

Cross-section of core stratigraphy for sites 5, 13, and 7. Cores are corrected for topographic elevation. Dotted lines in between cores connect interpreted environments (mangrove peat or mud flat/pool)

Full size image

The mean values for dry bulk density, percent water content, percent organic content, and sediment shear strength were all significantly different between mangrove peat and mud flat/pool interpreted sediments (Table 4). Sediments interpreted as mud flat/pool had greater dry bulk density values, lower percent water and percent organic contents, and greater sediment shear strengths compared to samples interpreted as mangrove peat. Elemental analyses of dried sediment samples revealed many particles that contained high concentrations of heavy metals at depths between 12 to 112 cm at sites 4 and 5 (Table 5). Elements found include: Bismuth (Bi), Copper (Cu), Cobalt (Co), Lead (Pb), Titanium (Ti), Tin (Sn), and Iron (Fe). The majority of the heavy metals were found between depths of 12 to 22 cm, which contained fine grain sediments classified as mud flat/pool at both sites 4 and 5.

Table 4 Mean values of dry bulk density, percent water content, and percent organic content for both mangrove and mud flat interpreted environments and p-values of two-sample, unequal variance t-tests
Full size table
Table 5 Metallic elements found in sediment samples using a SEM and EDS
Full size table

Discussion and conclusions

Mangrove swamp buffers lagoon from upland landfill

Within the mangrove system, horizontal groundwater flow direction changed seasonally (Fig. 3). From February to the beginning of May (the dry season), groundwater either flowed from the lagoon toward the landfill or flowed from both the landfill and the lagoon toward the center of the mangrove swamp. In July and November (the wet season), groundwater flowed from the landfill into the mangroves and out toward the lagoon (Fig. 3).

Hydraulic head responses to external factors also varied. Sites further inland responded more to precipitation events than tides while tides were more influential for sites located closer to the lagoon. A large rain event in May increased hydraulic head in the inner zone (with little change in the outer zone), but a second large rain event in August did not cause the same increase (Figs. 7, 8). The response of the inner zone to precipitation events may be partly determined by the hydraulic head at the time of the event. During the rain event in August, the hydraulic head of the shallow groundwater at site 5 was approximately 53 cm, compared to 33 cm during the rain event in May. The higher hydraulic head in August may indicate that sea-level rose above a threshold point after May, inundating the swamp, raising hydraulic head, and lessening the impacts of precipitation. This threshold point may be about 45 cm above mean sea level, below which monthly spring high tide signals are observed in hydraulic head, but diurnal signals are not present (Figs. 4, 5, 6). An alternative explanation for the lack of hydraulic response may be that the rain was absorbed by soil that was dry due to high air temperatures and limited rain after the May event.

This threshold in hydraulic head responses to tide suggests the presence of a hydraulic barrier that restricts the interconnection between the mangrove swamp and the ocean from January to May. In the mangrove swamp, the red mangrove fringe (Fig. 1) separating the swamp from the lagoon may trap sediment and limit the hydraulic connection between the lagoon and swamp. Aerial photographs of the study area suggest the presence of a mineral sediment ridge at the interface between Mangrove Lagoon and the mangrove swamp and field investigations from Mangrove Lagoon by kayak confirm its presence. However, in this study, sediments sampled in the outer portion of the mangrove swamp were dominated by peat which is more permeable than inland mineral-rich sediments. Further work investigating the presence and hydrologic impact of the sediment ridge at this location is needed.

Stratigraphic connectivity can be seen throughout the inner zone as all three cores contained mud flat/pool sediments at the surface, followed by a mangrove peat section before transitioning back into a mud flat/pool environment (Fig. 9b). The mud flat/pool interpreted sediments common in the inner zone had significantly higher densities and shear strengths, and lower percent water and organic contents than mangrove peat sediments (Table 4). Significant differences between sediment properties are important as groundwater velocity is affected by the properties of the medium it flows through (Anderson 2007), and sections of denser fine sediments are less permeable than mangrove peat and impede groundwater flow (Harvey and Odum 1990). The denser, finer sediments in the mud flat/pool sections would slow vertical and horizontal groundwater flow, which was observed in the lagged and dampened responses of the deep hydraulic head compared to the shallow hydraulic head in the inner zone (Fig. 5). The alternating mangrove peat and mud flat/pool sediments in the inner zone (Fig. 9b) suggest that the swamp is dynamic and that favorable conditions for mangrove growth have occurred only at certain times in the past, likely due to changes in the water flow regime (sea-level change, alteration of freshwater availability from the upland environment, or storm events) or an increase in sediment supply from the upland.

From our data, we produced a conceptual model (Fig. 11) of the geologic structure of the mangrove swamp depicting idealized groundwater flow patterns during the wet and dry seasons. This model is based on the sediment cores of sites 5, 7, and 13 (Fig. 10) and inferred stratigraphic changes from the inner zone near the landfill toward the outer zone near the lagoon for areas with no data. During the wet season, groundwater recharges into the mangrove swamp and flows from the direction of the landfill toward Mangrove Lagoon (Fig. 11a). During the dry season, groundwater discharges into the swamp (potentially driven by evapotranspiration) with flow from Mangrove Lagoon toward the landfill (Fig. 11b)

Fig. 11
figure 11

Groundwater flow shown in a cross-sectional cartoon view of the mangrove swamp where Bovoni Landfill is to the left and Mangrove Lagoon is to the right. a During the wet season (or after any large rain event) and spring high tides, there is recharge and flow out to Mangrove Lagoon. b During low tides and after long periods of little precipitation (red arrows indicate evapotranspiration), there is discharge and flow into the mangrove swamp from Mangrove Lagoon

Full size image

Mangrove swamp provides key ecosystem service

Spatial trends in the water chemistry indicate chemicals are being transported through surface water and shallow groundwater. Due to the seasonal variation in groundwater flow, chemical constituents from terrestrial sources are likely preferentially transported into the Mangrove Lagoon during the wet season (Fig. 3). However, geochemical trends suggest that the mangrove swamp is acting as a buffer, slowing or preventing heavy metals from entering the lagoon via groundwater or sediments (Tables 3). Groundwater recharge during periods of inundation, coupled with the seasonal shifts in lateral groundwater flow patterns and slow vertical groundwater velocities (Table 4) indicate that any dissolved chemical constituents will be trapped within the mangrove swamp. Our manual measurements indicate that hydraulic heads measured in shallow (1.5 m) sediments are similar to the hydraulic head of the surface water, suggesting that surface waters will have flow patterns similar to those measured in the shallow (1.5 m) sediments. These observations suggest that surface water exchange between Mangrove Lagoon and the mangrove swamp will also be limited and sediments carried by these flows will also be trapped in the mangrove swamp.

Chromium, nickel, and zinc were detected in groundwater samples, with nickel concentrations high enough to potentially impact aquatic organisms. The Criterion Continuous Concentration (CCC) and Criterion Maximum Concentration (CMC) are designated by the Environmental Protection Agency (EPA) as concentrations that can affect an aquatic community through continuous or brief exposure (EPA 2014). Nickel concentrations were above both CCC and CMC values in the surface water sample at site 5, the shallow groundwater sample at site 5, and the surface ditch (Table 6). However, nickel concentrations were not above the detection limit in any other groundwater sample, indicating that the metal is either not traveling widely from its source, or that it is becoming diluted as it is moved through the mangrove system. Zinc was not above its CCC or CMC in any sample. Safe drinking water standards were also surpassed at the surface ditch for chromium and nickel and site 5 for chromium, nickel, and pH (Table 6). The surface water samples represent overland flow that runs into the mangroves. The elevated concentrations of contaminants in these samples could affect not only the mangroves themselves, but also other organisms in the system (e.g., birds that drink from surface water puddles).

Table 6 Criterion Continuous Concentration (CCC), Criterion Maximum Concentration (CMC), and drinking water standards compared to surface ditch samples for Chromium, Nickel, Zinc and pH
Full size table

Any heavy metals that are potentially transported to Mangrove Lagoon would increase the already present risk to local fish and invertebrate communities as heavy metal accumulation has been found to limit fish size, impair DNA in gastropods, and interfere with reproduction in crustaceans (Canli and Atli 2003; Sarkar et al. 2014; Rodriguez et al. 2007). Sediments within Mangrove Lagoon have already been shown to contain high mean concentrations of chromium (26.7 μg/g), copper (48.4 μg/g), lead (18.7 μg/g), mercury (0.061 μg/g), nickel (10.5 μg/g), and zinc (109 μg/g; Pait et al. 2014). Four of these heavy metals were found in either the groundwater or sediments of the mangrove swamp, suggesting that some heavy metals may be transported through the swamp despite our findings, or that they are originating from another source. In this study, all water and sediment samples with high heavy metal concentrations were collected from sites located closest to the landfill and no samples located close to the lagoon (sites 6, 7, 10, 12 and 13) tested above detection limits for heavy metals. The green-colored, leachate-like fluid in the surface ditch, however, had elevated levels of Cr, Ni, and Sn, compared to the mangrove samples that tested positively for these heavy metals, suggesting strongly that the landfill is one source for these contaminants. Additional sampling at higher spatial and temporal resolutions would help determine contamination hotspots within the mangrove swamp and additional migration pathways for these heavy metals.

Continued protection of the mangrove swamp is critical to marine protected area integrity

Continued rapid urbanization of the coastal zone in St. Thomas, U.S. Virgin Islands compromises the capacity of mangrove systems to provide key ecosystem services. Urbanization may have already altered the hydrology of the study area as the red mangroves near the lagoon were found to be inundated nearly year round when mangroves typically have a more restricted inundation by high tides (Lewis 2005). A technical document (Towle 1985) from the 1970s reports similar inundation periods to those found in this study, however there is very limited data about the study area before the development of the landfill. Furthermore, what historic information is available is focused on the Mangrove Lagoon, not the mangrove swamp. Runoff and sedimentation from the landfill seem like likely factors to have altered hydrology in the mangrove swamp, but a lack of data before this study limit inferences.

The mangrove swamp is currently facing impacts from multiple human activities, not just the landfill. For example, fine sediments and heavy metal particles in the inner zone of the swamp are likely due in part to erosion from upland development. Increased sediment input could be one reason for recent mangrove health decline, as excessive sedimentation can result in filling of wetlands (Horner 2000) and alteration of the hydrological regime, causing changes in plant community structure and composition (Lee et al. 2006). Further, changing climate will impact this mangrove swamp, particularly if sea level rises above the hydraulic barrier, increasing the inundation of the system and potentially increasing nutrient discharge from the wetland to the lagoon (Wilson and Morris 2012).

Most pressing, however, is that this mangrove swamp is also facing the threat of direct habitat loss due to upcoming closure plans for Bovoni Landfill. Proposed closure plans (USA v. The Government of the Virgin Islands 2006) will result in cover materials and access roads being placed much closer to, or in, the mangrove swamp. Activities related to the landfill closure should be carefully planned to limit their impact on the mangrove swamp, as this study demonstrates one important ecosystem service that it provides. Removal or burial of part of this system could reduce its ability to act as a buffer, and disturbance of sediments may release trapped heavy metals and allow them to be carried toward the lagoon, a marine protected area containing significant natural and cultural resources (Horsley Witten 2013). Mangrove rehabilitation is an option for this study area, but careful consideration is needed before the start of any effort. Removal of stressors and rehabilitation of mangrove forests have shown to be more successful when started early and with an understanding of the mangrove forest’s characterizations and source of stress (Lewis et al. 2016). Further study of the alterations of the area and decline in mangrove health should be completed before any rehabilitation efforts or further alterations are made to insure the survival and recovery of this important mangrove swamp.