Upland habitats are converting to salt marsh at unprecedented rates due to sea level rise. While increasing salinity is understood to be the main driver of this conversion of upland to marsh, the factors that influence salinity in the zone of change (i.e., the marsh-forest ecotone) are unclear, especially in the northeastern USA, where marsh migration studies are less common. This study examined spatial and temporal patterns in salinity across the marsh-forest ecotone and potential drivers using a combination of groundwater well measurements and geophysical surveys. Across three study sites, groundwater salinity was influenced by multiple factors, including topography, weather events, and tidal cycles. Electrical resistivity tomography (ERT) showed differing patterns among sites: at the Massachusetts site, there was a sharper transition from low to high resistivity moving from the marsh to the forest. Whereas at the New Jersey and New York sites, there was a low resistivity (high salinity) layer overlaying high resistivity. The presence of this deeper fresh layer contrasts with the ‘salt wedge’ configuration found in open water estuarine systems and suggests that trees may be able to survive saltwater inundation by accessing this deeper reserve of freshwater. Salinity dynamics were occasionally driven by storm surges and hydraulic gradients, but the magnitude and direction of these effects were not consistent across events. Finally, sites with greater soil hydraulic conductivity exhibited lower water tables and enhanced tidal advection. This pattern suggests that well-drained soils characteristic of the Pine Barrens Ecoregion may facilitate efficient drainage but simultaneously heighten susceptibility to saltwater intrusion. These findings underscore the need to consider local hydrologic and soil conditions when predicting the pace of marsh migration and the resilience of coastal forests under rising sea levels.

To examine temporal and spatial variability across the mash-forest ecotone, we used a combination of geophysical surveys (electrical resistivity imaging) and groundwater well salinity meters.

Study Sites

Three locations showing signs of forest retreat were selected within the Atlantic coastal pine barrens ecoregion: Egg Harbor, New Jersey (39.3259°, -74.6502°), Pine Neck, New York (40.8422°, -72.5644°), and Waquoit Bay, Massachusetts (41.5562°, -70.5057°). These sites differ in tide range, slope, and vegetation composition along the marsh-forest transition but share common characteristics of the pine barrens ecoregion, which is known for its sandy, fast-draining soils (Forman 2012).

At Waquoit Bay, the study focused on the Sage Lot Pond marsh, a back-barrier marsh within the Waquoit Bay National Estuarine Research Reserve. Common species along the ecotone included Iva frutescens, Baccharis halimifolia, Toxicodendron radicans, Phragmites australis, Nyssa sylvatica, and Pinus rigida. Pine Neck is located in The Nature Conservancy’s Pine Neck Sanctuary along Shinnecock Bay in East Quogue, NY. The marsh-forest ecotone was dominated by Phragmites australis, mixed with Smilax spp., Iva frutescens, Baccharis halimifolia, and Nyssa sylvatica. The Egg Harbor site, located along the Great Egg Harbor River in the Tuckahoe Wildlife Management Area, was dominated by tall and dense stands of Phragmites australis, with scattered Juniperus virginiana and Nyssa sylvatica.

Topographic slopes across the ecotones were calculated in ArcGIS Pro (Esri Inc., Redlands, CA) by averaging slope values along transects derived from Digital Elevation Models (Molino et al. 2021). Tidal range estimates were obtained using NOAA’s VDatum tool (Tolkova et al. 2023). Waquoit Bay has a tidal range of 0.6 m and an average ecotone slope of 3.45 %. Pine Neck has a 0.8 m tidal range and an average slope of 2.1 %. Egg Harbor has the highest tidal amplitude (1.1 m; Tolkova et al. 2023) and the lowest average slope (1.8 %).

Groundwater Wells

Groundwater wells installed as part of a previous study (Payne et al. in submission) were utilized to measure water table height and salinity. Two wells were used at each site – one in the forest and one in the marsh or Phragmites/scrub shrub zone. Wells were constructed from 1 m sections of 3.8 cm inner diameter polyvinyl chloride (PVC) with a screened interval at 0.5 m – 1.0 m of depth that was covered with pantyhose to prevent infilling of sediment (Jobe and Gedan 2021). Boreholes were dug using a post hole digger or auger and were backfilled with sand after inserting the wells. The top ~15 cm was capped with concrete to prevent gaps acting as preferential flow pathways (Montalto et al., 2006). The well tops extended above the marsh surface by ~10 cm and were open on the top for Payne et al. in submission). To limit artifacts on salinity due to direct precipitation or overtopping, well tops were extended and capped with downward-facing PVC elbows. At each site, the lower well was outfitted with a HOBO U20 water level logger (Onset Corp., Bourne, MA) and a HOBO U24 Conductivity/Salinity logger. The upper well was outfitted with a Meter Hydros 21 Water Level/Conductivity/Salinity meter (Meter Group, Inc. USA, Pullman, WA). Another water level logger was installed in the tidal creek at each site.

Well elevations were measured during leaf-off using a survey-grade Leica GNSS GS14 (Leica Geosystems, Heerbrug, Switzerland). For Pine Neck and Harbor, measurements from 20-40 minute static occupations were post-processed using NOAA’s Online Positioning User Service (OPUS) (https://geodesy.noaa.gov/OPUS/). For Waquoit Bay, elevations were determined from Real-Time Network (RTN) measurements. Reference water elevations were measured periodically throughout the deployment. For the HOBO loggers, data were corrected for barometric pressure using weather station data from the closest airport and converted to NAVD88 using Hoboware Pro (version 3.17.19, Onset Computer Corp., Bourne, MA).

Soil Characteristics

Stratigraphy was assessed by coring the top 1.5 meters of soil, describing the layers and subsampling sediments. At each well site along the east transects, cores were collected using either Russian peat borer for organic sediments (Aquatic Research Instruments, Hope, ID, USA), or an Edelman or Riverside auger (Eijkelkamp, Wilmington, NC, USA) for sandy sediments. Cores were subsampled at varying depths based on the layer thicknesses and amount of sediment available for analyses. For each core, two to five subsamples were analyzed for particle size distribution, three to four for organic content, and one to three for saturated hydraulic conductivity. The goal was to obtain at least four subsamples from the two primary layers: an organic rich surface layer and a deeper outwash sand layer. This contact between organic soils and outwash sands was also assessed for the western transects via probing. 

Soil organic content was measured using Loss on Ignition (LOI), where soils were ignited for four hours at 550°C and the mass loss was used to calculate percent organic matter (Heiri et al., 2001). Particle size distribution was measured in triplicate on 14 – 16 samples per site using a laser granulometer (LS 13-320, Beckman Coulter, Brea, CA, U.S.A.) after pretreatment with concentrated hydrogen peroxide where necessary to remove coarse organic material such as peat and dispersal using a 5 % solution of sodium hexametaphosphate (Gray et al., 2010). Distributions were post-processed with Gradistat.v8 software (Blott & Pye 2001), including bin aggregation to texture classes and statistical description.

Saturated hydraulic conductivity was measured on 10 - 14 samples per site utilizing a benchtop K_sat instrument (Meter instruments, Pullman, WA, USA), which uses the falling head method to measure saturated hydraulic conductivity (K_sat) over a range of 0.01 to 5,000 cm day ^-1 (Trifunovic et al., 2018). The area-normalized flow rate through saturated sediment samples were measured to determine K_sat using the following equation:

 K_sat = (A_burette/A_sample)L*b (Equation 1)

where A_sample and A_burette are the cross-sectional areas of the sample and burette (in sq. cm); L is the sample depth in cm; and b is the coefficient of the fitted exponential function describing the falling head over time (s^-1).

Electrical Resistivity Surveys

To measure subsurface salinity patterns and seasonal variability, we conducted electrical resistivity surveys in the summer and winter of 2023 using a Syscal Pro 96 (Iris Instruments, Orleans, France). Resistivity is the inverse of conductivity and is also commonly used as a proxy for salinity. Resistivity surveys are conducted by transmitting an electrical current through the subsurface and measuring the strength of the returning current using a series of electrodes placed into the ground. For our surveys, electrodes were placed at 1 m intervals along an 80 m transect aligning with the groundwater wells, starting in the marsh and ending in the forest. We used a mixed array to maximize the depth of study in the salty soils. Contact resistance checks were performed prior to the survey start, and electrodes with high values were adjusted or wetted to improve contact with the soil. 

Salinity was measured from groundwater wells and soil samples to validate resistivity measurements. For the summer survey, water samples were collected from 4-5 groundwater wells along the resistivity transect. Wells were pumped out and allowed to refill before the samples were collected and stored in capped bottles. Conductivity and salinity were measured from the samples at the lab using the YSI Pro 30 water quality sonde. For the winter resistivity surveys, salinity was measured from 1.5 m deep soil cores that were collected at each well location using either a Russian peat borer for organic sediments (Aquatic Research Instruments, Hope, ID, USA), or an Edelman or Riverside auger (Eijkelkamp, Wilmington, NC, USA) for sandy sediments. Cores were subsampled at the top, middle, and bottom (exact depths varied based on sediment availability) and salinity was measured on 5:1 (of water to dry soil) soil slurries using a YSI Pro 30 (Xylem Inc., Yellow Springs, Ohio, USA).