Background Information

Historically, the Herring River estuary and flood plain was the largest tidal river and estuary complex on Outer Cape Cod and included about 6 square kilometers (km²; 600 hectares [ha]) of salt marsh, intertidal flats, and open-water habitats (National Park Service and others, 2016). These habitats are critically important for native vegetation and their inherent ecological functions as nurseries for marine animals; nesting and feeding areas for a variety of mammal, bird, amphibian, and fish species; and migratory pathways for anadromous species. Restoration of this resource would enhance its value for wildlife and recreational opportunities, including canoeing, kayaking, fishing, and viewing wildlife in restored wetlands and open-water habitats. Salt marsh restoration, defined as movement towards goals in vegetation, fish, and bird assemblages by restoring previously restricted tidal inputs, has been documented in New England (Warren and others, 2002; Chaffee and others, 2012; Roman and Burdick, 2012).

The town of Wellfleet is preparing to restore the Herring River estuary, which is within the Cape Cod National Seashore and managed by the NPS and has been diked off from the ocean for more than 100 years. The restoration plan consists of replacing the existing dike with a bridge and control gates to restore tidal flushing to large parts of the former salt marsh. Restoration is expected to improve water and soil quality, increase river herring and shellfish habitat, minimize the spread of invasive plant species, and provide improved recreational opportunities, among other benefits (Portnoy, 1999; Portnoy and Allen, 2006; National Park Service and others, 2016). To monitor biogeochemical and hydrologic changes in response to increased tidal exchange, flows would be increased incrementally during an adaptive management phase using adjustable gates at the new control structure on Chequessett Neck Road (Town of Wellfleet, 2019). The project is a partnership between the Cape Cod National Seashore and the Town of Wellfleet.

Diking and subsequent tidal restriction has resulted in loss of substantial areas of (tidal) salt and brackish marsh and increases in wet shrubland, wet and dry deciduous forests, and freshwater marsh in the Herring River Watershed (National Park Service and others, 2016). Habitat degradation in surface waters includes depletion of dissolved oxygen, increases in acidity and aluminum concentrations (that can be toxic to sensitive aquatic biota), and increases in fecal coliform concentrations (Portnoy and Allen, 2006). Diking has profoundly altered the hydrology and chemistry of the marsh upstream of the dike (Soukup and Pornoy, 1986). Portnoy and Giblin (1997a, b) described two distinctly different altered marsh environments after the installation of a dike: one system as diked drained, and a second system as diked and seasonally flooded.

In the diked drained systems, dewatering has resulted in air penetration into previously anoxic sediments that has led to oxidation of pyrite and generation of acidity and release of toxic aluminum concentrations (Portnoy and Giblin, 1997a). Aerobic conditions in the organic sediments facilitated rapid decomposition compared with conditions before installation of a dike. Rapid decomposition has resulted in substantial subsidence of the marsh surface (of 1 meter [m] or more, relative to contemporary sea level) in some places. The marsh surface elevation must increase at a rate greater than or equal to the rate of sea-level rise or risk inundation and loss of vegetation and ecosystem services. The dynamics of marsh elevation change depend on various processes, including sediment transport and deposition on marsh surfaces and the growth of salt marsh vegetation. Diking and drainage has effectively prevented sediment from reaching many areas of the Herring River flood plain, and restoration is expected to increase sediment deposition on marsh surfaces (National Park Service and others, 2016).

Concentrations of nitrogen and phosphorus in the sediments of Herring River have remained high in recent years (National Park Service and others, 2016). Potential sources of nutrient inputs to the watershed include agriculture, fertilized lawns, a golf course, a nearby landfill, septic systems, animal waste, and atmospheric deposition. The inputs have not been quantified. The lack of tidal flushing has allowed nutrients to accumulate in the Herring River watershed. In the diked drained sediments, ammonium (NH₄) has accumulated on sediment ion exchange surfaces, and orthophosphate (PO₄^3-) has accumulated on iron oxides. Column experiments have indicated that restoration of saltwater tidal flushing in these sediments will likely result in mobilization and release of NH₄, PO₄, and iron (Portnoy and Giblin, 1997a). The Wellfleet Harbor aquaculture industry brings in more than 5 million dollars each year. A major nutrient release, even if short-lived, could result in eutrophication leading to hypoxia and adversely affect the industry. This specific concern is one of the major reasons that the restoration plan includes a phased or incremental reintroduction of saltwater to allow for adaptive management (National Park Service and others, 2016).

Description of the Study Area

The drainage area of the Herring River watershed in Wellfleet, Mass., at the dike on Chequessett Neck Road (U.S. Geological Survey [USGS] station 011058798) on Cape Cod, is not precisely defined because the contributing groundwater drainage area is not coincident with the surface-water drainage area (Masterson, 2004) and this basin is groundwater dominated (LeBlanc and others, 1986). The surface-water drainage area is approximately 28.5 km² (fig. 1) based on the USGS StreamStats application (U.S. Geological Survey, 2015) for Massachusetts. The groundwater contributing area for the Herring River has been simulated by use of a three-dimensional numerical model (MODFLOW) to be approximately 19.1 km² (Carlson and others, 2017a, b). In the interpretation of this simulation, the kettle ponds in the northwestern part of the surface water drainage area are not included in the groundwater contributing area.

The Herring River watershed contains a 6-km² (600-ha) estuarine marsh complex that is the largest diked wetland system on Cape Cod. The watershed was diked off from the ocean in 1909, resulting in greatly restricted tidal flushing (Portnoy and Allen, 2006). There are four tributaries (fig. 1) draining subbasins in the Herring River: Bound Brook, Duck Harbor, Mill Creek, and Pole Dike Creek; the four subbasins are briefly described in National Park Service and others (2016). The dike restricts tidal amplitude from about 2.1 m seaward to 0.5 m landward of the structure (Portnoy and Allen, 2006). The dike was intended to control mosquitoes, create additional developable land, and provide for road construction. After the original dike construction, widespread ditching, straightening, and channelizing of the meandering creeks effectively drained most of the remaining salt marshes (National Park Service and others, 2016). Between 1929 and 1933, a golf course was constructed, partly on what had previously been part of the Mill Creek (tributary to the Herring River) flood plain. Additionally, several houses have been constructed in the former flood plain.

The dike structure is an earthen dam about 0.15 kilometer (km) in length that contains a culvert with three approximately 4-square-meter rectangular sluiceways in the center of the dike. Within two of the three sluiceways (center and northwestern sluiceways), hydraulic pressure-controlled flapper gates permit ebb discharge and close following an ebb tide, thereby restricting seawater flow upstream during flood tides (Portnoy and Allen, 2006). The third sluiceway (southeastern sluiceway) is a fixed gate that is partially open. This sluiceway allows ebb discharge and is the only sluiceway that allows seawater to flow upstream during flood tides. With the current [2023] dike configuration, seawater travels upstream about 1,000 m into the estuary and marsh complex (Portnoy and Allen, 2006). The dike structure results in tidal asymmetry such that flood tides average 7.2 hours and ebb tides average 5.2 hours during each complete cycle. The tidal cycle in the ocean is consistent with semidiurnal tides (average periodicity of 12.4 hours) for the Gulf of Maine (Garrett, 1972).

Before diking, the tidal flats and low marsh areas were dominated by Spartina alterniflora (saltmarsh cordgrass), and the high marsh area was dominated by Spartina patens (salt meadow cordgrass). After diking, that vegetation was replaced by brackish and freshwater species, such as Scirpus spp. (sedges), Typha spp., forbs, shrubs, and grasses (Roman, 1987). Diking also resulted in the establishment of invasive plant species in the Herring River watershed (Smith, 2007; National Park Service and others, 2016) and negatively affected the populations of many fish, shellfish, and other invertebrates, birds, mammals, reptiles, and amphibians (National Park Service and others, 2016).

The mean annual temperature observed at the meteorological station in Truro, which is 5.85 km northeast from the Herring River monitoring station, was 12.3 degrees Celsius (°C) from October 2004 through September 2017 (Western Regional Climate Center, 2018). Mean monthly air temperature is usually between 1 and 4 °C from December through February and between 21 and 24 °C from June through August. The mean annual precipitation near Truro was 121 centimeters (cm) for the period January 1983 to December 2017 (National Atmospheric Deposition Program, 2018). Precipitation is fairly evenly distributed among months and generally averages between 80 and 120 millimeters (mm) each month but can be highly variable in any given month (from less than [<] 10 to more than [>] 130 mm). Groundwater recharge on Cape Cod is about 45 percent of total annual precipitation (LeBlanc and others, 1986).

Various characteristics of the basin, as defined by the surface-water drainage area, are described in the USGS StreamStats application (U.S. Geological Survey, 2015). The watershed is entirely underlain by glacially derived coarse-grained stratified drift. Forests cover 71 percent of the watershed surface area, wetlands cover 15 percent, ponds cover 4.6 percent, and developed land (largely the suburban town of Wellfleet) occupies about 13.2 percent. percentages add up to >100 percent because some of the forested land is also classified as wetlands. About 3 percent of the developed land is classified as impervious surface. The mean watershed elevation is 14.4 m, and the total length of all mapped streams (1:24,000-scale) in the watershed is 31.2 km.

In 2021 and 2022, sea water breached the beach and dune system that had kept sea water out of Duck Harbor throughout this and previous study periods, and water quality measurements recorded approximately 1.6 km upstream from the beach detected the presence of this sea water in the Herring River (Tim Smith, NPS, written commun., January 24, 2023). This breaching is associated with rising sea level and is expected to occur more frequently with future increases in sea level during most high monthly spring tides.

Purpose

The purpose of this study was to characterize the natural variability in nutrient and suspended sediment concentrations in flood tide and ebb tide waters at the dike currently [2023] installed on the Herring River at Chequessett Neck Road in Wellfleet in relation to environmental variables before the removal of the dike so that this baseline could be used to assess whether removal results in significant changes. This report includes time series of nutrient concentration data from approximately twice monthly sampling of flood and following ebb tides from June 2020 to December 2021 and nutrient data collected during previous USGS studies. These data are discussed in terms of seasonal variations, differences between flood and ebb tide concentrations, and differences between spring, neap, and midamplitude tides. Ancillary time series for air temperature, precipitation, calculated runoff, solar radiation, sea level, nutrient concentrations in Wellfleet Harbor, and atmospheric deposition of nitrogen are also included. Finally, the results of correlation analyses between nutrient concentrations and the ancillary variables are discussed.

This information is intended for use by resource managers at the Cape Cod National Seashore to facilitate planned adaptive management during phased dam removal and reconstruction. Specifically, the time series data provide a range of observed nutrient, dissolved organic carbon, and suspended sediment concentrations over time and a basis for discussion of variation in constituent concentrations in relation to season and environmental characteristics.

Collection and Processing of Discrete Water Quality Samples

Collection and processing of discrete water quality samples in the current [2023] study followed USGS protocols (U.S. Geological Survey, variously dated). In a previous study, flow-weighted composite samples were collected to obtain samples with nutrient concentrations that were representative of two successive flood or ebb tides (Huntington and others, 2021). Estimation of discharge was not possible in the current [2023] study owing to limitations in project resources; therefore, discrete samples were collected at predetermined times following the beginning of flood and ebb tides. Samples collected at these times were considered to be the most representative of flow-weighted composite concentrations based on analyses comparing discrete samples collected periodically over complete tidal cycles compared with flow-weighted samples where subsamples were collected into a single container over a tidal cycle as described in Huntington and others (2021).

Constituent concentrations varied substantially during tidal cycles and generally followed systematic patterns. Concentrations usually reached a maximum at the end of the ebb tide and the beginning of the flood tide, then declined rapidly to a minimum at the middle of the flood tide, remained near this minimum until early to mid-ebb tide, after which they increased rapidly. Because of this variation in concentration over tidal cycles, discrete grab samples would be representative of the tidal average concentrations if they were collected at times where concentrations were usually similar to the flow-weighted composite samples. Huntington and others (2021, fig. 3) illustrates how nutrient (NH₄, total phosphorus [TP], orthophosphate, and total nitrogen [TN]) concentrations calculated from flow-weighted composite sampling differ from measured individual discrete samples over one sequence of flood followed by ebb tide. The concentrations of orthophosphate on the flood tide were exceptions to this general pattern during the February and May samplings, over which concentrations varied more than during other sequences.

Beginning on February 2, March 22, May 12, and November 4, 2016, samples were collected every 30 minutes for a flood and following ebb tide, and each sample was analyzed for nutrient concentrations. The flow-weighted mean concentrations were calculated from the paired discharge and concentration data. To determine a time following the beginning of the flood or ebb tide to collect a single grab sample that would be most representative of the flow-weighted mean concentration a two-step procedure was used.

In the first step, second order polynomial equations were fit for concentration (x axis) versus time (y axis) for each nutrient for flood and ebb periods. These equations were then solved for the times associated with the calculated flow-weighted mean concentrations. The average times in hours and minutes from the start of each tide from the four dates were calculated for each nutrient. The coefficients of variation for the times associated with the calculated flow-weighted mean concentrations were 12 percent on the flood tide and 16 percent on the ebb tide. The times chosen for sample collection were based on the average times determined for TN and TP. The times based on the polynomial solutions were incremented by 3 minutes for the flood and 6 minutes for the ebb to account for the fact that the first samples were collected after a fixed volume had flowed past the gage (Huntington and others, 2021). The sampling times were determined to be 3 hours and 14 minutes after the start of the flood and 2 hours and 37 minutes after the start of the ebb tide. The beginning of each tide was determined from the time when the surface velocity changed from positive to negative (start of the flood) and negative to positive (start of the ebb). Surface-water velocity was measured by using a contact-free Sommer Messtechnik GmbH RG–30 radar sensor (Sommer Mess Systemtechnik, 2014) on the upstream side (marsh side) of the dike as described in Huntington and others (2021).

In the second step, second order polynomial equations were then fit for time (x axis) versus concentration (y axis) for each nutrient for flood and ebb periods and solved for the concentration associated with the average times calculated above. These concentrations were compared with the calculated flow-weighted mean concentrations to compute the percent error in each case. The errors between the concentrations derived from the polynomials and those derived from the calculated flow-weighted average concentrations generally were between 2 and 10 percent on the ebb tide. The errors were substantially larger on the flood tide ranging from less than or equal to (≤) 20 percent for TN, TP, total dissolved phosphorus (TDP), and orthophosphate to between 50 and 136 percent for NH₄, nitrate (NO₃) plus nitrite (NO₂), and silica (SiO₂). For the ebb tide, the differences between the calculated flow-weighted mean concentrations and the concentrations determined from sampling at 2 hours and 37 minutes after the start of the tide were 7, 14, 5, 1, 4, 34, and 6 percent for TN, TP, TDP, orthophosphate, NH₄, NO₃, and SiO₂, respectively. For the flood tide these differences from sampling at 3 hours and 14 minutes after the start of the tide were 12, 25, 4, 2, 28, 85 and 122 percent for TN, TP, TDP, orthophosphate, NH₄, NO₃, and silicon, respectively.

To quantify differences in concentrations between sampling at fixed times following the beginning of the flood and ebb tides compared with the method of flow-weighted composite sampling in the previous study the data were compared for the same seasonal periods. All constituents sampled between mid-June 2020 to mid-December 2021 were compared with previous data from mid-June 2016 to mid-December 2017 except dissolved organic carbon (DOC) and total dissolved nitrogen (TDN) data from the current [2023] study. Shorter sampling periods were used for the comparison of DOC and TDN because available data records had shorter periods of record.

At the Chequessett Neck Road monitoring station (USGS station 011058798) samples were collected approximately every 2 weeks on flood tides and the subsequent ebb tides from June 2020 through December 2021. Samples were usually collected from spring tides and from neap tides each month. Tides were judged to be spring tides if they were among the five highest amplitude tides and neap tides if they were judged to be among the four lowest amplitude in a given month. In some months, it was not possible to collect samples during spring or neap tides or there were no obvious spring or neap tides, so samples were collected from midamplitude tides.

Sample collections were made using a peristaltic pump through tubing to an orifice located approximately 10 cm above the base of the concrete sluiceway and in the same location as samples were collected as described in Huntington and others (2021). The ISCO pump used in the current [2023] study (Teledyne ISCO, 2015) has been shown to sample sediment concentrations isokinetically (Bogen, 2004). The pump was actuated manually, following U.S. Geological Survey protocol (app. 2; Francesca Wilde, USGS, written commun., 2009), a minimum of 1 liter (L) of water was pumped through the capsule filter to rinse it prior to collection of any samples. After rinsing the raw water exiting the pump was first filtered using 0.45-micrometer (µm) capsule filters for dissolved constituents (NH₄, NO₃ plus NO₂, orthophosphate, TDP, and SiO₂). The filtrate for these constituents were collected in amber opaque 125-milliliter (mL) polyethylene bottles. The raw water exiting the pump was then used to fill amber opaque 125-mL polyethylene bottles for total nitrogen (TN) and total phosphorus (TP) and a sample for filtration using a syringe filter. TDN includes NH₄, NO₃ plus NO₂, and dissolved organic nitrogen. Particulate organic nitrogen (PON) was calculated as TN minus TDN.

A 0.45-μm Supor membrane syringe filter was used to filter samples for DOC and TDN. The syringe filter was used for TDN and DOC because the cartridge filters have been shown to leach organic nitrogen (app. 3; Sorenson and others, 2018; Francesca Wilde, USGS, written commun., 2017). Samples for DOC were collected and stored in 125-mL amber glass bottles. Samples for TDN were collected in amber opaque 125-mL polyethylene bottles. Finally, raw water exiting the pump was then used to fill 3–L bottles for measurement of suspended sediment concentration. All nitrogen, phosphorus, and SiO₂ samples were pumped into amber opaque 125-ml polyethylene bottles and were chilled until analyzed. DOC samples were pumped into 125-mL amber glass bottles. All samples with the exception of the 3–L samples were chilled until analyzed.

Analytical Methods and Quality Assurance

Concentrations of NO₃ plus NO₂, NH₄, TDN, TDP, SiO₂, orthophosphate, TN, and TP were analyzed at the Center for Coastal Studies Laboratory in Provincetown, Mass. DOC was analyzed at the U.S. Forest Service Northern Research Station in Parsons, W.Va. The 3–L samples for suspended sediment concentration were shipped to the USGS Kentucky Sediment Laboratory in Louisville, Ky., for analysis. The analytical methods used for water quality analysis, USGS parameter codes, method references, and minimum detection limits for samples collected at water quality-monitoring stations on the Herring River are summarized in table 1.

Laboratory quality assurance protocols included the analysis of standard reference materials, field replicates, field blanks, and laboratory blanks and reasonable range checks. All water quality and quality assurance data for the current [2023] study are published in the U.S. Geological Survey National Water Information System (NWIS) database (U.S. Geological Survey, 2022). Laboratory analyses of standard reference materials for nutrients and DOC consistently met the data quality objectives established for the current [2023] study. The data quality objectives were relative percent differences (RPDs) of <20 percent. Analysis of field sequential replicate samples shows that median RPD values between the replicates and the corresponding environmental samples for all constituents were usually substantially less than the data quality objectives of ±20 percent.

Field blank results indicated that, in most cases, blank analyte concentrations were below minimum reporting levels or substantially lower than the lowest environmental sample for nutrients, except for concentrations of NH₄. The median value for the concentration of NH₄ field blanks was 0.0033 milligrams per liter (mg/L) as nitrogen, which was above the detection limit but substantially lower than all but two environmental samples that were also below detection limits. Analyses of laboratory-prepared deionized water blanks showed that there was no detectable contamination of samples within the laboratory for any constituent.

Data Collected at the Chequessett Neck Road Monitoring Station in Previous Projects

For analyses in the current [2023] study, nutrient concentration data were aggregated with data from two previous studies. The first study included data collected from November 2015 to September 2017 that was described in Huntington and Spaetzel (2020) and Huntington and others (2021) and are published in NWIS (U.S. Geological Survey, 2022). The second study included data from October 2017 to September 2018. The data from the second study are published in NWIS (U.S. Geological Survey, 2022). Data collection during these previous projects used flow-weighted composite sampling. The data from these previous studies were included in this analysis to increase the period of record to maximize the characterization of the baseline before the removal of the dike.

Measurements to determine if sampling from a single point in one sluiceway was representative of the total volume of water flowing during ebb tides were discussed in Huntington and others (2021). The results of that comparison indicated that, for certain constituents, the automated point-sampling is not fully representative of the water in the cross section and introduces some error. The data from these analyses comparing point sampling to equal-width interval sampling generally met a data quality objective of <20 percent RPD for most analytes (TN, TP, orthophosphate, NO₃ plus NO₂, and SiO₂), but the RPD was >20 percent for other analytes (NH₄ and TDP; Huntington and others, 2021).

Ancillary Environmental Data

Ancillary environmental data were compiled to investigate whether they could explain observed variations in nutrient concentrations. Monthly average surface maximum and minimum air temperature, precipitation, and downwelling surface shortwave radiation were obtained from the gridMET dataset (Abatzoglou, 2013), a dataset of daily high-spatial resolution (about 4-km, 1/24th degree [°]) surface meteorological data covering the contiguous United States. Data from the spatial domain defined by the coordinates that encompassed the Herring River watershed area were retrieved using the area of interest and getGridMet functions in R (Climatology Lab, 2017). The spatial coordinates were x_min = −70.0844, x_max = −69.97111, y_max = 41.97389, y_min = 41.92333, where x refers to longitude and y refers to latitude. Mean temperature was calculated as the simple average of minimum and maximum temperatures. Daily meteorological data were aggregated to monthly data.

Runoff from the Herring River watershed was calculated using the Thornthwaite water balance model that partitions precipitation inputs to evapotranspiration, runoff, or storage by using a monthly accounting procedure described in McCabe and Markstrom (2007). Inputs to the model are the mean monthly surface air temperature (in degrees Celsius), monthly total precipitation (in millimeters), and latitude (in decimal degrees) of the location of interest. The temperature and precipitation inputs were derived from gridMET as described above. The latitude was input as 42°. The model uses other input parameters defined as follows: runoff factor, 50 percent; direct runoff factor, 5 percent; soil moisture storage capacity, 150 mm; rain temperature threshold, 3.3 °C; snow temperature threshold, −10 °C; maximum melt rate, 50 percent. The monthly environmental variables and runoff data are published in Huntington (2023).

Monthly mean sea level data for Woods Hole, Mass. (National Oceanic and Atmospheric Administration [NOAA] station 8447930) were obtained from NOAA (2022a). Mean monthly surface ocean water temperature was obtained from NOAA station 44090 (41.840 N −70.329 W; NOAA, 2022c) supplemented with data from NOAA station 44013 (42.346 N −70.651 W; NOAA, 2022b). Monthly sea level and ocean water temperature data are published in Huntington (2023).

Nutrient concentration data from Wellfleet Harbor station 43 were obtained from the Center for Coastal Studies (2022), which collects samples under the Cape Cod Bay Monitoring Program from the ocean surface once in May and October and twice per month from June through September. Nutrient concentration data obtained from the Center for Coastal Studies are published in Huntington (2023).

Data for atmospheric deposition of nitrogen were obtained from the National Atmospheric Deposition Program station in Barnstable, Mass. (site MA01; National Atmospheric Deposition Program, 2018). This site is 32 km and 218° southwest of the Herring River gage. There are missing data for many months, and the monthly data were only available through December 2020. The NADP uses the following completeness criteria for data acceptance:

Twenty-seven out of 61 months in the record from November 2015 to December 2020 had completeness criteria that were acceptable and were well distributed over the record. Data for atmospheric deposition of nitrogen are published in Huntington (2023).

Correlation Analyses

Kendall’s rank correlation tests were run between nutrient concentrations on the ebb tide and environmental variables including precipitation, surface air temperature, runoff, downwelling shortwave radiation, ocean water temperature, mean sea level, and atmospheric deposition of NO₃ plus NO₂ and NH₄ (Huntington, 2023). Correlation tests were run in Tibco Spotfire software, and the results were summarized based on significance probability levels of <0.01, <0.05 and <0.1; p-values greater than 0.1 were not significant.

Differences in Concentration Based on Sampling Methodologies

The method of sample collection in the current [2023] study involved sampling at a fixed time following the beginning of a single flood or ebb tide and not the flow-weighted composite sampling following two successive flood or ebb tides of the previous studies (Huntington and others, 2021). The analysis showed that it is likely that there would be small differences between methods in measured concentrations on the ebb tides and somewhat larger differences on the flood tides. The larger RPDs are generally associated with low concentrations on the flood tides.

Differences in constituent concentrations between the previous studies and this study could be related to differences in environmental conditions between these periods or differences in sampling methodologies. The following paragraph describes the relationships between selected nutrient concentrations and environmental variables that could explain part of the differences in concentrations between periods. The description of differences in the next paragraph assumes that the differences are explained solely by the different methods.

Average constituent concentrations were higher in the previous study that used flow-weighted composite sampling (Huntington and others, 2021; U.S. Geological Survey, 202243) compared with the current [2023] study that used fixed time sampling for NH₄, TP, silicon, and DOC on the flood and NH₄, TP, TDP, silicon, and DOC on the ebb tides (table 2). Constituent concentrations generally were higher in the current [2023] study using the fixed time sampling method compared with the previous studies using flow-weighted composite sampling for orthophosphate and TDP on the flood tide and NO₃ plus NO₂ on the ebb tide (table 2). For TDN, TN, TDP, and DOC, the RPDs in concentrations were <25 percent between methods on the flood and ebb tides (table 2). For NO₃ plus NO₂ and TP on the flood and for orthophosphate and SiO₂ on the ebb, the RPDs in concentrations were <25 percent between methods. For NO₃ plus NO₂ and TP on the ebb and for orthophosphate and silicon on the flood, the RPDs in concentrations were between 43 and 55 percent between methods. For NH₄, the RPDs in concentrations were 144 percent on the flood and 91 percent on the ebb between methods.

The NO₃ plus NO₂ ebb tide concentrations between the two sampling periods were similar except in September through November (app. 1, fig. 1.1). In September 2021, ebb tide NO₃ plus NO₂ concentrations were more than three times greater compared with the early period (2016–17). Runoff in September 2021 was more than four times greater and precipitation was more than two times greater compared with the 2016–17 period (app. 1, figs. 1.2 and 1.3). The October 2020 and 2021 periods also had higher NO₃ plus NO₂ concentrations and higher runoff. These differences in NO₃ plus NO₂ concentrations from the early to the recent periods could be partially explained by increases in precipitation and runoff. The differences in TN and SiO₂ concentrations between periods (app. 1, figs. 1.4 and 1.5) were also associated with greater precipitation and runoff; however, in these cases, the differences generally were between December 2020 through August 2021 (app. 1, fig. 1.3). The possibility that increases in precipitation and runoff may explain part of the differences in these nutrients between the early and recent periods is consistent with the correlations between these nutrients and precipitation and runoff for the complete dataset (discussed in the “Correlations Between Monthly Average Nutrient Concentrations on the Ebb Tide and Corresponding Monthly Average Environmental Variables” section of this report). Differences between other constituents between periods on the ebb tide do not show consistent relations with the environmental variables studied. There were only very small differences in monthly average air temperature (app. 1, fig. 1.6) and monthly average solar radiation (app. 1, fig. 1.7) between the early and recent periods, so it is unlikely that differences in these variables could explain differences in nutrient concentrations.

Variation in Nutrient and Suspended Sediment Concentrations in the Herring River by Season and Tide Direction

Nutrient and suspended sediment concentration varied by season and tide direction in the Herring River during the period of record. Concentrations of NO₃ plus NO₂, NH₄, TDN, TN, SiO₂, and DOC generally were higher on the ebb tide than on the flood tide. By contrast, concentrations of orthophosphate, TDP, and TP were generally similar between flood and ebb tides.

The concentrations of NO₃ plus NO₂ generally varied between 0.003 and 0.01 mg/L on flood tides and between 0.003 and 0.03 mg/L on ebb tides (fig. 2A). NO₃ plus NO₂ concentrations were similar between flood and following ebb tides in late summer and early fall, but ebb tide concentrations generally increased from late fall, peaking in winter, and decreasing until summer, whereas flood tide concentrations remained low (fig. 2A and B). These patterns resulted in distinct seasonal variation in the difference between flood and ebb tide concentration, although this pattern was less pronounced in the current [2023] study than in the previous studies (Huntington and others, 2021; U.S. Geological Survey, 202243). Variation in NO₃ plus NO₂ concentrations between flood and ebb tides during each season shows that ebb tide concentrations were higher than flood tide concentration in all seasons, but the differences were greatest in spring and winter and least in summer (fig. 3). Variation within a season was greatest in summer and winter for the flood tide and in summer for the ebb tide.

The concentrations of NH₄ generally varied between 0.008 and 0.09 mg/L on flood tides and between 0.02 and 1.0 mg/L on ebb tides (fig. 4A). Seasonal variation in NH₄ concentrations was substantially more variable than for NO₃ plus NO₂ concentrations. In general, NH₄ concentrations were higher on the ebb tide than on the flood tide, but the seasonal pattern of these differences was less consistent than for NO₃ plus NO₂ (fig. 4B). The differences between flood and ebb tides were small but more consistent in the current [2023] study than in the previously published data (Huntington and others, 2021; U.S. Geological Survey, 202243). In contrast to the patterns observed for NO₃ plus NO₂, NH₄ concentrations on the flood tide tended to be higher in summer and fall and lower in spring and winter (fig. 5).

The period of record for TDN was shorter than for TN, NO₃ plus NO₂, or NH₄ because of potential contamination in the earliest record (November 11, 2015, to July 26, 2017). The concentrations of total TDN generally varied between 0.13 and 0.40 mg/L on flood tides and between 0.20 and 0.5 mg/L on ebb tides (fig. 6A). TDN concentrations generally were higher on the ebb tide than the flood tide and higher in late fall and winter than during the spring and summer (fig. 6B).

TDN concentrations were higher on the ebb than the flood tide in all seasons but the difference was smallest in summer (fig. 7). Comparing these TDN data with the data for TN indicates that a substantial fraction of TN is particulate organic nitrogen (PON; >0.45 µm) on the flood tide (33 to 57 percent) and ebb tide (25 to 75 percent). The PON fraction was estimated as the difference between TN and TDN. On the flood tide, the PON fraction is higher at higher TDN concentrations, and on the ebb tide, the PON fraction is higher at lower TDN concentrations.

Concentrations of TN generally varied between 0.25 and 0.5 mg/L on flood tides and between 0.25 and 0.7 mg/L on ebb tides (fig. 8A). TN concentrations generally were similar between flood and ebb tides in summer and fall but were higher on ebb tides than flood tides in winter and spring (figs. 8B and 9). Within-season variation in concentration of TN was greater for ebb tides than flood tides in winter, spring, and fall but smaller in summer. The differences between flood and ebb tides were similar between this and previous studies.

The concentrations of orthophosphate generally varied between 0.007 and 0.09 mg/L on flood and ebb tides (fig. 10A). Orthophosphate concentrations generally increased beginning in late spring or early summer, peaked in mid- to late summer, then decreased to minimum concentrations in winter and early spring. These patterns resulted in distinct seasonal variation with a single peak each year. The differences between flood and ebb tide concentrations generally were small, but ebb tide concentrations generally were greater than flood tide concentrations in fall and winter but were similar in spring and summer (figs. 10B and 11). In the current [2023] study, there was either little or no difference between ebb and flood, or during high summer concentrations, the flood tide concentrations were usually higher (fig. 10A and B).

The concentrations of TDP generally varied between 0.006 and 0.05 mg/L on flood and ebb tides (figs. 12A and 13). The seasonal variation in TDP generally was quite similar to that observed for orthophosphate for flood and ebb tides (figs. 10B and 12B). The differences between flood and ebb tide concentrations generally were small, but ebb tide concentrations generally were greater than flood tide concentrations in fall and spring but similar in summer and winter (fig. 13).

The concentrations of TP generally varied between 0.03 and 0.09 mg/L on flood and ebb tides (figs. 14A and 15). Flood and ebb tide TP concentrations generally increased beginning in late spring, peaked in late summer, then decreased to minimum concentrations in winter (fig. 14A). These patterns resulted in seasonal variation with a single peak each year but with more within-season variability than observed for orthophosphate (fig. 10B). Ebb tide concentrations were slightly higher than flood tide concentrations in spring and fall but were very similar summer and winter, albeit with somewhat larger interquartile ranges (fig. 14B and 15). In the current [2023] study there were either little or no differences between ebb and flood or, during summer when concentrations were highest, the concentrations on the flood tide were usually higher.

The concentrations of SiO₂ generally varied between 0.5 and 1.5 mg/L on flood tides and between 1.5 and 3.0 mg/L on ebb tides (fig. 16A). SiO₂ concentrations were higher on the ebb tide than the flood tide in all seasons by a factor of two or more (figs. 16 and 17). There were no consistent seasonal differences in silica concentrations between flood and ebb tides.

The concentrations of DOC generally varied between 1.8 and 3.5 mg/L on flood tides and between 2.0 and 5.0 mg/L on ebb tides (fig. 18A). DOC concentrations generally were higher on the ebb tide than the flood tide in all seasons (figs. 18B and 19). The largest difference was in the spring, followed by the summer and winter, and the smallest difference was in the fall (fig. 19). DOC concentrations were highest in spring and summer and lowest in winter for flood and ebb tides (figs. 18A and 19), resulting in a seasonal pattern having one peak and one trough each year.

The period of record for suspended sediment concentration was from July 2020 through December 2021 and was insufficient to establish seasonal patterns. Suspended sediment concentrations were low and generally varied between 2 and 9 mg/L during flood and ebb tides (fig. 20A). In most cases, the differences between flood and ebb concentrations were <5 mg/L; however, there were six instances in summer and fall 2021 where ebb concentrations were more than 10 mg/L greater than flood concentrations and two instances in spring and summer 2021 where flood concentrations were more than 10 mg/L greater than ebb concentrations (fig. 20B). In spring, summer, and fall 2021, there were a total of seven ebb and two flood samples with suspended sediment concentrations in the range of 20 to 40 mg/L (fig. 20A). In most of these cases, the ebb samples had substantially higher suspended sediment concentrations than the flood samples (fig. 20B).

Variation in Nutrient Concentration Between Spring, Neap, and Mid Tides

The interquartile ranges and median NO₃ plus NO₂ concentrations were similar for samples collected during spring, neap, and midamplitude tides on flood and ebb tides (fig. 21). The interquartile ranges in NH₄ concentrations were similar, but the median concentrations were higher for samples collected during spring compared with neap tides on flood and ebb tides (fig. 22). The median NH₄ concentrations on the flood tide were 49 percent higher on the spring compared with neap tide, and on the ebb tide, 23 percent higher on the spring compared with neap tide. Samples collected during midamplitude tides were similar to those during spring and neap tides. The interquartile ranges and median TDN and TN concentrations were similar for samples collected during spring, neap, and midamplitude tides (figs. 23 and 24).

The interquartile ranges and median orthophosphate concentrations were similar for samples collected during spring, neap, and midamplitude tides on flood and ebb tides with the exception of higher interquartile range and median concentrations for samples collected on midamplitude ebb tides compared with spring and neap tides (fig. 25). The median orthophosphate concentrations on the ebb tide were 71 and 79 percent higher on the midamplitude compared with spring and neap tides, respectively.

The interquartile ranges and median TDP concentrations were similar for samples collected during spring, neap, and midamplitude tides on flood tides (fig. 26). The interquartile ranges for TDP concentrations were similar for samples collected during spring, neap, and midamplitude tides on ebb tides, but the median concentrations were highest on the midamplitude tides and lowest on the neap tide. On the ebb tide, the TDP concentrations on the midamplitude tide were 92 percent higher than on the neap tide, and the TDP concentrations on the spring tide were 47 percent higher than on the neap tide. On the ebb tide, the TDP concentrations on the spring tide were 31 percent higher than on the neap tide.

The interquartile ranges and median TP concentrations were similar for samples collected during spring, neap, and midamplitude tides on flood and ebb tides with the exception of higher interquartile range and median concentrations for samples collected on midamplitude ebb tides compared with spring and neap tides (fig. 27). The median TP concentrations on the ebb tide were 43 and 45 percent higher on the midamplitude compared with spring and neap tides, respectively.

The interquartile ranges and median SiO₂ concentrations were similar for samples collected during spring, neap, and midamplitude tides on flood and ebb tides with the exception of higher median concentration for samples collected on midamplitude flood tides compared with spring and neap tides (fig. 28). The median SiO₂ concentrations on the flood tide were 43 and 44 percent higher on the midamplitude compared with spring and neap tides, respectively.

The interquartile ranges of DOC concentrations were similar for samples collected during spring, neap, and midamplitude tides on flood and ebb tides (fig. 29). The median DOC concentrations were similar between the spring and midamplitude tides and lower on the neap tide for both flood and ebb tides. The median DOC concentration on the spring tide was 20 percent higher than on the neap tide on the flood tide and 22 percent higher than on the neap tide on the ebb tide.

Nutrient Export

In the previous study of Huntington and others (2021), nutrient export from the Herring River watershed at the dike at Chequessett Neck Road was estimated for each ebb tide where concentration and discharge data were available from November 2015 through September 2017. Nutrient export (fluxes) was estimated based on the flow-weighted average concentrations and the total discharge during ebb tides. Discharge was not measured in the current [2023] study, therefore comparable estimates of nutrient export were not possible. It would be possible to estimate monthly nutrient export (fluxes) during the record where nutrient concentration data are available (November 2015 through September 2018 and June 2020 through December 2021) using a different method. Monthly runoff estimates could be calculated using the Thornthwaite water balance model (McCabe and Markstrom, 2007) with input data accessible from gridMET (Abatzoglou, 2013). These runoff estimates for this period are also available in Huntington (2023). Average monthly nutrient concentrations could be calculated by averaging the measurements during each month using the data available in NWIS (U.S. Geological Survey, 2022). Monthly nutrient export could be estimated by multiplying estimated runoff by concentration.

Correlations Between Monthly Average Nutrient Concentrations on the Ebb Tide and Corresponding Monthly Average Environmental Variables

Monthly average NO₃ plus NO₂, NH₄, TN, and SiO₂ concentrations were positively correlated (p<0.01) with monthly average precipitation (table 3). These positive correlations could be explained by increases in leaching of nutrients from aboveground organic materials and soils by convective mass flow that typically increases with increasing precipitation volume and intensity (Ballard and others, 2019; Winter and others, 2022). Orthophosphate, TDP, TP, and DOC were positively correlated (p<0.01) with surface air temperature. These positive correlations with temperature could be explained by increases in microbial decomposition with increasing temperature (Kirschbaum, 2010) that increase the supply of these nutrients that are available for transport (Freeman and others, 2001; von Lützow and Kögel-Knabner, 2009). NO₃ plus NO₂ (p<0.01) and NH₄ and SiO₂ concentrations (p<0.05) were negatively correlated with surface air temperature. These negative correlations with temperature could be explained by increased uptake by algae in the aquatic environment under higher temperature conditions (Reay and others, 1999; Singh and Singh, 2015).

NO₃ plus NO₂ and SiO₂ (p<0.01) and NH₄ and TN (p<0.05) concentrations were positively correlated with runoff and may be explained by the same processes described above for precipitation. Orthophosphate and TDP were negatively correlated (p<0.01) with runoff. Mechanistic explanations for these negative correlations are not known. Orthophosphate, TDP, TP, and DOC concentrations were positively correlated (p<0.01) with downwelling shortwave radiation. Mechanistic explanations for these positive correlations are not known. NO₃ plus NO₂ (p<0.01) and NH₄ and SiO₂ (p<0.05) concentrations were negatively correlated with downwelling shortwave radiation. These negative correlations with radiation could be explained by increased uptake by algae in the aquatic environment due to higher rates of photosynthesis under conditions of higher solar radiation (Singh and Singh, 2015).

NO₃ plus NO₂ (p<0.01), NH₄ (p<0.1), and SiO₂ (p<0.1) concentrations were negatively correlated with ocean water temperature. Higher ocean water temperatures on the flood tides may be associated with increased algal growth and uptake of NO₃ plus NO₂, NH₄ and SiO₂, resulting in the observed negative correlations (Reay and others, 1999; Singh and Singh, 2015). Orthophosphate, TDP, and DOC (p<0.01) and TP (p<0.05) concentrations were positively correlated with ocean water temperature. Mechanistic explanations for these positive correlations and ocean water temperature are not known.

There was a substantial amount of missing data for atmospheric deposition of NO₃ plus NO₂ plus NH₄; however, using the existing data, there was a weak positive correlation (p<0.1) between atmospheric deposition of NO₃ plus NO₂ plus NH₄ and for NO₃ plus NO₂ concentrations on the ebb tide but no significant correlations for NH₄ or TN. Nutrient concentrations on the ebb tide were not correlated with mean sea level in Woods Hole.

The concentrations of NO₃ plus NO₂, NH₄, TN, TDP, and SiO₂ measured in surface water in Wellfleet Harbor during May through October were not significantly correlated with concentrations of those nutrients on the ebb tide at the Herring River during those months. The concentrations of orthophosphate and TP in surface water in Wellfleet Harbor were weakly positively correlated (p<0.1) with concentrations of those nutrients on the ebb tide at the Herring River during those months.