REPORTS OF THE TIBOR T. POLGAR
FELLOWSHIP PROGRAM, 2019
David J. Yozzo, Sarah H. Fernald and Helena Andreyko
Editors
A Joint Program of The Hudson River Foundation and The New York State Department of Environmental Conservation
March 2021 ii ABSTRACT
Eight studies completed within the Hudson River Estuary under the auspices of the
Tibor T. Polgar Fellowship Program during 2019 have been included in the current volume.
Major objectives of these studies included: (1) investigating the use of gadolinium as a method of wastewater detection, (2) examining changes in groundwater hydrology in
Piermont Marsh during the past twenty years, (3) understanding the developmental histories and sediment characteristics of Hudson River tidal marshes, (4) investigating the presence of the freshwater jellyfish, Craspedacusta sowerbii, in upper New York lake communities, (5) determining the effect of Trapa natans beds on cyanobacterial community composition, (6) analyzing the interaction of Trapa natans beds with pharmaceuticals and pesticides in the
Hudson River, and (7) characterizing and evaluating the temporal and geospatial distribution of microplastics in the Raritan Bay area, and (8) elucidating the role of salinity on fecal bacteria transport in Hudson River sediments.
iii iv TABLE OF CONTENTS
Abstract ...... iii
Preface ...... vii
Fellowship Reports
Use of Gadolinium to Track Sewage Effluent Through the Poughkeepsie, New York Water System Matthew Badia, Neil Fitzgerald, and Christopher Bowser ...... I-1
Is Sea Level Rise Altering Wetland Hydrology in Hudson River Valley Tidal Marshes? Sofi Courtney, Elizabeth Watson, and Franco Montalto ...... II-1
Tidal Marsh Development and Sediment Dynamics at Vanderburgh Cove, Rhinebeck, NY Waverly L. Lau and Brian Yellen ...... III-1
Presence and Trophic Level of Freshwater Jellyfish (Craspedacusta sowerbii), A Cryptic Invader in the Hudson River Basin, NY Jacob Moore and Donald J. Stewart ...... IV-1
The Potential for Harmful Algal Blooms (HABs) in the Hudson River Estuary: Dominant Abiotic Drivers of Cyanobacterial Abundance and Toxicity with the Compounding Influence of the Invasive Water Chestnut Ellie Petraccione, Zion Klos, and Raymond Kepner ...... V-1
Pharmaceutical Transport and Transformation in Trapa natans Beds Cami Plum, Stephen Hamilton, Rebekah Henry, David McCarthy, Emma Rosi, and Michael Grace ...... VI-1
Spectroscopic Characterization and Quantification of Microplastics in the Hudson River Karli Sipps and Georgia Arbuckle-Keil ...... VII-1
Studying the Effect of Salinity and Tide on Fecal Bacteria Transport within Hudson River Estuary Sediments Dong Zhang and Valentina Prigiobbe ...... VIII-1
v vi PREFACE
The Hudson River estuary stretches from its tidal limit at the Federal Dam at Troy,
New York, to its merger with the New York Bight, south of New York City. Within that reach, the estuary displays a broad transition from tidal freshwater to marine conditions that are reflected in its physical composition and the biota its supports. As such, it presents a major opportunity and challenge to researchers to describe the makeup and workings of a complex and dynamic ecosystem. The Tibor T. Polgar Fellowship Program provides funds for students to study selected aspects of the physical, chemical, biological, and public policy realms of the estuary.
The Polgar Fellowship Program was established in 1985 in memory of Dr. Tibor T.
Polgar, former Chairman of the Hudson River Foundation Science Panel. The 2019 program was jointly conducted by the Hudson River Foundation for Science and Environmental
Research and the New York State Department of Environmental Conservation and underwritten by the Hudson River Foundation. The fellowship program provides stipends and research funds for research projects within the Hudson drainage basin and is open to graduate and undergraduate students.
vii Prior to 1988, Polgar studies were conducted only within the four sites that comprise the Hudson River National Estuarine Research Reserve, a part of the National Estuarine
Research Reserve System. The four Hudson River sites, Piermont Marsh, Iona Island, Tivoli
Bays, and Stockport Flats exceed 4,000 acres and include a wide variety of habitats spaced over 100 miles of the Hudson estuary. Since 1988, the Polgar Program has supported research carried out at any location within the Hudson estuary.
The work reported in this volume represents eight research projects conducted by
Polgar Fellows during 2019. These studies meet the goals of the Tibor T. Polgar Fellowship
Program to generate new information on the nature of the Hudson estuary and to train students in estuarine science.
David J. Yozzo Glenford Environmental Science
Sarah H. Fernald New York State Department of Environmental Conservation Hudson River National Estuarine Research Reserve
Helena Andreyko Hudson River Foundation for Science and Environmental Research
viii USE OF GADOLINIUM TO TRACK SEWAGE EFFLUENT THROUGH THE POUGHKEEPSIE, NEW YORK WATER SYSTEM
A Final Report of the Tibor T. Polgar Fellowship Program
Matthew Badia
Marist College Environmental Science and Chemistry Student Researcher [email protected]
School of Science, Department of Chemistry, Biochemistry, and Physics Marist College Poughkeepsie, NY 12601
Project Advisors:
Dr. Neil Fitzgerald School of Science, Department of Chemistry, Biochemistry, and Physics Marist College Poughkeepsie, NY 12601 [email protected]
Christopher Bowser New York State Department of Environmental Conservation Hudson River National Estuary Research Reserve [email protected]
Badia, M., N. Fitzgerald, and C. Bowser. 2021. Use of Gadolinium to Track Sewage Effluent Through the Poughkeepsie, New York Water System. Section I: 1-26 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
I-1 Abstract
There has been an increase in chemical contaminants in waterways around the
world, many of which are released from sewage treatment plants which are incapable of
removing most chemical contaminants and are released into the environment where they
can accumulate. Current methods of wastewater detection include bacterial analysis,
nutrient analysis, chloride analysis, and direct detection; however, each have their own
flaws. An alternative method involves the use of gadolinium, an element used in MRI
procedures. This study investigated the use of gadolinium as a method of wastewater
detection compared to enterococci and chloride analysis. Samples were taken from four
locations located along Hudson River near Poughkeepsie, NY: a sewage treatment plant,
a water quality monitoring station, and the intake and effluent from a water treatment
plant, from June to July 2019. Enterococci were analyzed using the IDEXX Enterolert
system. Chloride content was determined using the Mohr titration method. Gadolinium
was analyzed using Inductively Coupled Plasma Mass Spectrometry. Enterococci
analysis demonstrated high values in wastewater and low levels in the water treatment
plant effluent. The bacteria levels in open water were inconsistent between location and sample day. Chloride concentrations decreased from the sewer treatment plant to the water treatment plant intake, and increased after water treatment, likely due to the decomposition of sodium hypochlorite. Gadolinium analysis showed a content averaging approximately 500 ng/L in the sewage treatment plant effluent, followed by a drastic decrease in open water and water treatment plant intake, averaging 15 and 29 ng/L respectively, and a final decrease during the water treatment process to an average of 5 ng/L. Gadolinium appears to be as effective as chloride analysis for detecting wastewater.
With a single anthropogenic source and low analysis cost, gadolinium has the potential to replace other methods for tracing wastewater in waterways.
I-2 TABLE OF CONTENTS
Abstract ...... I-2
Table of Contents ...... I-3
List of Figures and Tables...... I-4
Introduction ...... I-5
Methods...... I-9
Results ...... I-15
Discussion ...... I-20
Conclusion ...... I-23
Acknowledgements ...... I-24
References ...... I-25
I-3 LIST OF FIGURES AND TABLES
Figure 1 – Map of sample locations ...... I-10
Figure 2 – Enterococci levels throughout the experiment ...... I-16
Figure 3 – Chloride levels for the data obtained by the Mohr method of titration
...... I-17
Figure 4 – Comparison between the chloride levels and gadolinium levels
throughout the experimental period ...... I-18
Figure 5 – Comparison of the three methods of wastewater detection ...... I-20
I-4 INTRODUCTION
The Hudson River Estuary (HRE) extends from the Battery on Manhattan Island,
to the Federal Dam in Troy, New York. The HRE is a vital ecological and economic
resource for the state of New York and the surrounding area. The river itself is 315 miles
in length with its source at Lake Tear of the Clouds in the Adirondack Mountains, before
joining with the Mohawk River at Schenectady, New York. The Hudson River Watershed
covers 13,400 square miles, as far east as Vermont, Massachusetts, and Connecticut, and
as far south as New Jersey (NYSDEC HREP 2009). The HRE is a vital local resource, providing economic and recreational benefits for thousands. The HRE ecosystem is essential for a multitude of species, including the American Eel and the Atlantic
Sturgeon, which is endangered in the Hudson River.
Pharmaceuticals and Personal Care Products (PPCPs) are bio-accumulators, meaning that concentration of the chemical increases in tissues as the chemical moves up the food chain (Stahl 2016). According to the EPA, significant waterways around the country, including the Hudson River, are contaminated with PPCPs, heavy metals, and bacteria (Stahl 2016). A substantial contributor to the contamination of US Waterways is from sewage treatment plants (STPs) (Daughton 2009). PPCPs originate in sewage from the biological waste of people taking pharmaceuticals, or the compound being sent down the drain, either intentionally or unintentionally (Daughton 2009). STPs make efforts to remove contaminants from sewage; however, methods are inadequate for the complete removal of PPCPs and other chemical pollutants (Daughton 2009). Consequently, concentrations of various PPCPs accumulate into the nation’s waterways (Daughton
2009).
I-5 The sewage treatment process is broken down into two steps. In the first step, waste is passed through a screen which removes large objects that would damage equipment. The waste then passes through a grit chamber where sediment and small stones settle to the bottom. The waste then passes into a sedimentation tank where the velocity of the water decreases to allow finer particulate matter to settle. The second step involves removing much of the organic matter. The primary method of this is aeration and activated sludge. This process uses bacteria to breakdown the organic matter into byproducts. The bacteria and microorganisms are reused and the water is passed onto a disinfecting stage. It is common to use sodium hypochlorite for this process (Daughton
2009).
The water treatment process is more sophisticated than the sewage treatment process. Water treatment begins with a process called flocculation. During this step, a coagulant is added to the water which removes sediments and particulate matter. In many instances, the sediment is collected as sludge and is sold to fertilizer companies. The water then is passed through a series of filters containing gravel, sand, and activated carbon. Water then must be disinfected. In the Poughkeepsie Water Treatment Facility, this disinfection is accomplished in three stages. The first stage, water is exposed to ultraviolet light. A solution of sodium hypochlorite is then added to the water for further sterilization. Finally, ozone is passed through the water as a final step in the disinfecting process. Water throughout the process is routinely sampled for basic water chemistry and sent out for chemical analysis to ensure safe drinking water for the community.
Combined Sewage Overflows (CSOs) also contribute to the presence of pharmaceuticals in urbanized waterways (Rechenburg et al. 2006). CSOs occur during
I-6 significant rainfall events when sewer systems become overwhelmed by the volume of water. As a result, the system releases untreated sewage along with the overflow of water into the nearby waterway. During this process, a high concentration of contaminants, including PPCPs and bacteria, is released.
The identification and tracking of these anthropogenic pollutants are imperative in the determination of the potential detrimental effects on human health. Identification of these organic compounds has previously focused on direct quantification using High-
Performance Liquid Chromatography (HPLC), Gas Chromatography-Mass Spectrometry
(GC-MS), and Liquid Chromatography-Mass Spectrometry (LC-MS). All of these methods require complicated extraction procedures and are expensive.
It is possible to use nutrients as a chemical tracer for wastewater. Nitrate and phosphate are commonly used for this purpose (Katz and Griffin 2008). Both nitrate and phosphate arise from the decomposition of organic matter, including human waste. There are several methods used for the detection of nitrate and phosphate ranging from colorimetric methods to electronic probes. These nutrients have several anthropogenic and non-anthropogenic sources. The two most prominent are runoff from agriculture or municipal land using fertilizer, and from runoff from animal waste. Both of these have the potential of producing increased levels of nitrate and phosphate. This becomes an issue in regions such as the Hudson Valley, which have a large agricultural economy.
Bacteria is a commonly used method of wastewater detection. Enterococci are most commonly used (Wéry et al. 2010). These bacteria, depending on the species, form either symbiotic or pathogenic relationships with humans. In either case, these bacteria can be detected in small quantities in sewage treatment plant effluents as the water
I-7 leaving the system is not completely sterile. There are several other sources of enterococci including waterfowl and agricultural animals. As a result, many of the common tests for enterococci detect non-human species.
There are several options for the detection of wastewater using chemical contaminants. Chloride is a popular method for wastewater detection (Wolf et al. 2012).
Chloride has several sources in water. In wastewater, chloride can arise from the human diet in the form of sodium chloride, or from water softeners in the form of sodium or potassium chloride. Chloride can also arise from the use of road salt during winter to melt ice. As the ice melts, runoff brings this salt into waterways. Calcium chloride is commonly used in this purpose. Lastly, chloride can appear in waterways as the result of natural decay of sediments. Heavy metals can also be used as a tracer of wastewater; however, many have non-anthropogenic sources.
Gadolinium (Gd) is a rare earth element whose presence in the waterways has been previously reported as negligible. The primary use of gadolinium is as an Magnetic
Resonance Imaging (MRI) contrast agent. In the body, gadolinium can enhance the quality of images. These compounds are processed by the kidneys with a half-life of 2 hours. Once in the sewer system, gadolinium functions similarly to other PPCPs and are not broken down by the water treatment process (Verplank et al. 2005). By itself, gadolinium is a toxic heavy metal, but gadolinium (III) can also be encased in an organic molecule called a chelate. These chelates are organic molecules that encase the gadolinium ion to protect the human body from the toxicity. This allows for the nearly 1 gram of gadolinium to pass harmlessly thorough the human body. Additionally, the chelate can be modified to a certain portion of the body (Caravan et al. 1999).
I-8 Recent studies have reported the presence of what is referred to as gadolinium anomalies (Verplank et al. 2005; Caravan et al. 1999; Rabiet et al. 2009; Williams et al.
2013). These anomalies are classified as environmental gadolinium levels higher than expected. Anomalies in the expected presence of gadolinium were first published in 1996 and has since been attributed to the contrast used in MRI. A study conducted in Boulder
Creek conducted by Verplank et al. (2005) investigated the chelate itself as opposed to the gadolinium. It was reported that gadolinium chelates are a signature of urban wastewater. Other studies around the world have investigated the impact of municipal water on gadolinium content. These studies have concluded gadolinium anomalies are the result of MRI facilities; however, these studies nearly exclusively focus on surface water, whereas this study focuses on open water, a water treatment plant and a water treatment facility (Verplank et al. 2005; Caravan et al. 1999; Rabiet et al. 2009; Williams et al.
2013).
The goal of this study was to determine the effectiveness of Gd as a tracer for wastewater in the Hudson River. This was accomplished by comparing the results of Gd analysis to the results of chloride and enterococci analysis, two previously accepted methods of wastewater detection.
METHODS
Sampling
Samples were collected from four locations: Arlington Wastewater Treatment
Plant, the Hudson River Environmental Conditions Observing Systems (HRECOS) station located at Marist College, and both the intake and the effluent from the
Poughkeepsie Joint Water Treatment Facility.
I-9
Figure 1: Map of Sample locations showing the Sewage Treatment Plant (STP), the HRECOS station, the intake of the Water Treatment Plant (WTP in), and the Water Treatment Plant effluent (WTP eff).
Site Characterization
The Poughkeepsie Municipal area is serviced by two sewage treatment facilities,
the Arlington Wastewater Treatment Plant and the Poughkeepsie Water Pollution Control
Plant. The Arlington Wastewater Treatment Plant services the City and Town of
Poughkeepsie. The Poughkeepsie Water Pollution Control Plant, located off of Marist
College campus, also serves the Town and City of Poughkeepsie. There are approximately 100,000 residents within the limits of the city and town. The sewer system is designed for each facility to share the sewage from the municipality.
The Arlington Wastewater Treatment Plant serves the approximate 26,000
residents of the Town of Poughkeepsie. While the town does not contain many medical
facilities within the town limits, there are as many as eighteen individual MRI facilities
within the City of Poughkeepsie. The plant averages a discharge rate of 4.6 million
I-10
gallons per day. Samples were taken near the end of the treatment process before water
enters the river.
The drinking water plant is jointly owned by the City and Town of Poughkeepsie
and the Dutchess Water Authority. The facility serves approximately 100,000 people in
the City and Town of Poughkeepsie, the Dutchess County Water Authority, and the
Town of Hyde Park, just to the north. The plant produces nearly 9.5 million gallons per
day (mgd) of treated drinking water with a maximum capacity of around 11 mgd. The
plant is sourced directly from the Hudson River. A pipe submerged 10 meters below the
surface of the river located off Longview Park in Poughkeepsie pumps water from the
river to the facility located slightly uphill.
HRECOS is a network of near real-time environmental monitoring stations. The
network is operated by governmental, academic, and private institutions throughout the
Hudson River Watershed. The stations are located from Battery Park in New York City,
to the Port of Albany, and along the Mohawk River. The stations are equipped with a
variety of sensors that monitor water quality and weather patterns every 15 minutes year-
round. This data is freely available to the general public on the HRECOS website. The
station located at Marist College is jointly operated by the Cary Institute of Ecosystems
Studies, Marist College, and the United States Geological Survey. Unlike most other
stations, the Marist College station samples both surface and from a depth of 10 meters in the hyporheic zone. Samples were taken from the hyporheic zone directly from the Marist pump station.
I-11
Sample Collection
Samples were taken from four locations, the Arlington Wastewater Treatment
Plant, hereby referred to as the sewage treatment plant, the HRECOS station located at
Marist College, referred to as the HRECOS station, and the intake and effluent from the
Poughkeepsie Water Treatment Plant. Samples from the sewage treatment plant were
taken from the final stage of treatment in the open holding pond by means of a dip-stick.
Water samples from the HRECOS station were taken from a holding tub that continuously cycles the sample water. Lastly, the water treatment plant has a sampling lab that samples water directly from various points during the treatment process on tap.
Samples were collected at slack tide, the point at the end of one tide cycle, and the beginning of another where the water does not move in either direction. Five samples were taken at the end of low tide, and five samples were taken at the end of high tide.
Extra samples were taken on the last day of sampling and sent for PPCP analysis.
Samples were collected in plastic bottles cleaned with dilute nitric acid. Sampling
occurred between June and July 2019. The temperature, time of collection, and current
weather conditions were recorded.
Initial Sample Processing
Immediately upon the collection of samples, the pH was measured using a pH
probe. Samples were then taken for enterococci analysis. For heavy metal analysis, 30
mL of each sample was filtered at 0.45 μm using vacuum filtration. These samples were
then acidified to 1% nitric acid (Thermo Fisher Scientific) in acid cleaned storage vessels.
These, and the remaining samples, were refrigerated until analyzed.
I-12
Enterococci Analysis
Enterococci levels were measured using the IDEXX Enterolert system. Enterolert is a US EPA approved method for the detection of enterococci within sewer and water systems. The samples were analyzed immediately after being collected. From each sample, 100 mL was added to the collection vessel along with reagent and shaken. The sample was then added to a Quanti Tray and hermetically sealed. The sample was incubated at 41 degrees ± 0.5 for 24 hours. The samples were viewed under UV, and the fluorescing wells counted. For this method, any blue fluorescence was counted as a positive. The total number of bacteria was then estimated using the most probable number of bacteria. A 95% confidence interval was calculated using the most probable number calculator provided by IDEXX. Enterolert detects a variety of species including
E. faecalis, E. faecium, E. avium, E. gallinarum, E. casseliflavis, and E. durans among other species acknowledged, but not listed by the manufacturer. Enterolert is not capable of detecting individual species of bacteria (IDEXX 2019).
Chloride Analysis
Chloride content in samples was measured by a Mohr Method of Titration. This method involves two steps, a standardization method, then the analysis of each sample.
This is a crude but effective method for the detection of chloride and other halides in water samples by the titration of the sample against silver nitrate with potassium chromate. Solutions of 0.0141 M silver nitrate (Ward’s) and 0.25 M potassium chromate
(Thermo Fisher Scientific) were prepared. For standardization, 0.0200 g sodium chloride
(Sigma Scientific) was dissolved in 100 mL of deionized water. To this solution, 1.0 mL of the potassium chromate was added. This solution was titrated against the solution of
I-13
silver nitrate. This process was repeated until two values were achieved with less than a
1% difference. The endpoint was reached when the color of the solution changed from
bright yellow to a deep orange. From this, the concentration of silver nitrate was
calculated from the average volume of silver nitrate used as follows:
( ) 58.449 = 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑔𝑔 𝑔𝑔 ( ) 3 𝑚𝑚𝑚𝑚𝑚𝑚 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝐴𝐴𝐴𝐴𝐴𝐴𝑂𝑂 3 𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉𝑉 𝐴𝐴𝐴𝐴𝐴𝐴𝑂𝑂 𝐿𝐿
Exactly 100.00 mL of each sample was taken and the pH was adjusted to between
7-10. To this solution, 1.0 mL of the potassium chromate was added, and the solution was
titrated against the now known concentration of silver nitrate. The endpoint is reached
when the solution changes color from yellow to orange. The total concentration of
chloride was then be calculated from the total volume of silver nitrate calculated as follows:
= = 2+ 2+ − 𝑀𝑀𝐴𝐴𝐴𝐴 �𝑉𝑉𝐴𝐴𝐴𝐴𝐴𝐴𝑂𝑂3� 𝑚𝑚𝑚𝑚𝑚𝑚𝐴𝐴𝐴𝐴 𝑚𝑚𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 35.45 = − 𝑔𝑔 − 𝑚𝑚𝑚𝑚𝑚𝑚𝐶𝐶𝐶𝐶 ∗ 𝑔𝑔 𝐶𝐶𝐶𝐶 1000𝑚𝑚=𝑚𝑚𝑚𝑚 − − 𝑔𝑔 𝐶𝐶𝐶𝐶 ∗ 𝑚𝑚𝑚𝑚 𝐶𝐶𝐶𝐶 = 0.1 − 𝑚𝑚𝑚𝑚 𝐶𝐶𝐶𝐶 𝑚𝑚𝑚𝑚 − 𝐶𝐶𝐶𝐶 Gadolinium Analysis 𝐿𝐿 𝐿𝐿
Gadolinium standards were made in concentrations of 1, 2, 10, 100, and 500 ng/L
(ppt) (SPEX CertiPrep). Standards were made in deionized water and acidified to 1%
nitric acid. Gadolinium analysis was conducted using Inductively Coupled Plasma Mass
I-14
Spectrometry (Thermoscientific iCap RQ). The analysis was conducted with Terbium as a reference. Samples were run against a 1% nitric acid blank.
RESULTS
Enterococci
In general, enterococci content varied between sample date and location. Figure 2 shows the enterococci levels with the 95% confidence intervals determined experimentally by IDEXX. Most of the sampling dates demonstrated the highest level of enterococci at the sewage treatment plant, with an average of 4.2 bacteria per 100 mL of sample. The bacteria levels for the HRECOS station and the water treatment plant intake were similar, consistent with predictions given the proximity of the two sampling locations with an average of 1.2 and 4.4, respectively. There were, on occasion, variations to this pattern where the difference between the level of bacteria in the
HRECOS station and the WTP intake was significant, these differences were inconsistent. Consistently over the sampling period, no bacteria were detected in the effluent of the water treatment plant. The highest recorded bacteria level was seen on the final day of sampling at the WTP intake reading at approximately 26 bacteria per 100 mL of sample, far higher than any of the other samples.
I-15
40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 Enterococci (Bacteria/100 mL) 0.0 18-Jun 20-Jun 25-Jun 2-Jul 9-Jul 11-Jul 12-Jul 25-Jul 26-Jul 30-Jul WWTP HRECOS WTP in WTP out
Figure 2: Enterococci levels throughout the experiment. The bacteria level is measured in the number of bacteria per 100 mL seen on the primary vertical axis. The primary horizontal axis shows the date of sample collection. From left to right on a given sample day, the bars can be read as the sewer treatment plant, the HRECOS station, the water treatment plant intake, and the water treatment facility effluent. The error bars represent a 95% confidence interval provided by IDEXX.
Chloride levels followed a more predictable pattern throughout the sampling period shown in Figure 3. The data shows a trend of periodic fluctuations in chloride content; however, the sampling period is too small and too widespread over the summer to conclude a potential pattern. The highest levels of chloride were seen on the second day of sampling. This appeared to correspond to an increase in chloride for all four sampling locations on June 20th. The highest levels of chloride were seen in the sewer treatment plant with an average of 126 mg/L. This is followed by relatively consistent levels of chloride in the HRECOS station and the water treatment plant intake with an average of 26.9 and 26.0 mg/L, respectively. A slight increase in the chloride levels was detected with an average of 39.9 mg/L in the water treatment plant effluent. In general, the data follow the predicted pattern, despite the increase in chloride after the water
I-16 treatment process. It is anticipated these unusual values are due to the addition of sodium hypochlorite during the water treatment process.
160 140 120 100 80 60
Chloride (mg/L) 40 20 0 18-Jun 20-Jun 25-Jun 2-Jul 9-Jul 11-Jul 12-Jul 25-Jul 26-Jul 30-Jul
STP HRECOS WTP in WTP out
Figure 3: Chloride levels for the data obtained by the Mohr method of titration. The primary vertical axis shows the chloride levels measured in mg/L. The primary horizontal axis shows the date of sample collection. From left to right on a given sample day, the bars can be read as the sewer treatment plant, the HRECOS station, the water treatment plant intake, and the water treatment facility effluent.
Figure 4 shows a graph of chloride content measured in mg/L and gadolinium measured in ng/L, or parts per trillion. The purpose of this graph is to compare the effectiveness of the novel method, gadolinium, to the known method of detection, chloride. In general, gadolinium content appears to show a similar pattern as chloride.
Gadolinium content is consistently the highest in the sewer treatment plant effluent. The concentration then decreases at the HRECOS station and the water treatment plant intake.
Unlike chloride, gadolinium levels appear to be more variable between the HRECOS station and the water treatment plant intake. The gadolinium levels then decrease significantly during the water treatment process. The highest level of gadolinium was detected from the sewer treatment plant on July 12th with a concentration of 1150.96
I-17 ng/L. This is much higher than the next highest level which occurred on July 11th at
613.11 ng/L. The maximum values in the HRECOS station was 19.84 ng/L on July 9th and the water treatment plant was 50.95 ng/L on July 9th. The lowest concentration of gadolinium was found in the water treatment plant effluent on July 25 with a concentration of 3.75 ng/L.
Chloride v. Gadolinium 160 1400.00 140 1200.00 120 1000.00 100 800.00 80 600.00 60 40 400.00 Chloride (mg/L) 20 200.00 Gadolinium (ng/L) 0 0.00 18-Jun 20-Jun 25-Jun 2-Jul 9-Jul 11-Jul 12-Jul 25-Jul 26-Jul 30-Jul
STP Cl HRECOS Cl WTP in Cl WTP in Gd WTP out Cl STP Gd HRECOS Gd WTP out Gd
Figure 4: Comparison between the chloride levels and gadolinium levels throughout the experimental period. The primary vertical axis shows the chloride content measured in mg/L. The secondary vertical axis shows the gadolinium content in ng/L. From left to right on a given sample day, the bars can be read as the sewer treatment plant chloride and the sewer treatment plant gadolinium, the HRECOS station chloride and the HRECOS gadolinium, the water treatment plant intake chloride and the water treatment plant gadolinium, and the water treatment facility effluent chloride and water treatment facility effluent gadolinium.
Figure 5 shows a graph of the average gadolinium, chloride, and enterococci content. This graph shows the effectiveness of gadolinium as a tracer compared to the two accepted methods of detection for wastewater. The graph demonstrates changes in gadolinium that correspond to changes in chloride. Gadolinium appears to have a much
I-18 larger dilution factor in open water than chloride. Gadolinium concentrations in the sewer treatment plant average 497.61 ng/L. This value is diluted in open water to an average of
15.19 ng/L at the HRECOS station and 21.39 ng/L at the water treatment plant intake.
This is in comparison to an average of 15.91 ng/L in the HRECOS station and 21.39 ng/L in the water treatment plant intake. This is compared to an average chloride value of
126.23 mg/L in the sewer treatment plant which was only diluted to an average of 26.91 mg/L in the HRECOS station and 26.0 mg/L in the water treatment plant intake. This may simply be due to the naturally higher levels of chloride in the Hudson River.
Additional discrepancies between gadolinium and chloride were seen at the water treatment plant effluent. Gadolinium levels further decrease during the treatment process to an average of 5.9 ng/L; however, chloride levels increased to 39.89 mg/L after the water treatment process. This increase is believed to be due to the water treatment process.
I-19 600 140
120 500
100 400
80 300 mL) 60
Gadolinium (ng/L) 200 40
100 20
0 0 ChlorideEnterococci and (mg/L) (bacteria/100 STP HRECOS WTP in WTP out
Gadolinium Chloride Enterococci
Figure 5: Comparison of the three methods of wastewater detection used in this study. The average of each measure was taken for the individual sampling locations. Gadolinium and chloride content are shown as bars whereas enterococci levels are shown as a line. The error bars represent the average 95% confidence interval for each measurement. The primary vertical axis shows the gadolinium content at each location measured in ng/L (ppt). The secondary vertical axis shows the chloride content measured in mg/L and enterococci levels, measured in bacteria/100 mL. The horizontal axis shows the four sampling locations in order of appearance on the river, the sewer treatment plant, the HRECOS station, the water treatment plant intake, and the water treatment plant effluent.
DISCUSSION
The main issue with the use of enterococci analysis is the broad range of species for which this test targets. There is a total of six species listed on the manufacturer’s website, including E. faecalis, E. faecium, E. durans, E. gallinarum, E casseliflavis, E. avium (IDEXX 2019). E. faecalis and E. faecium are commonly found in the human microbiome (Lebreton et al. 2014). E. durans is frequently seen as a pathogen in foals,
I-20 calves, piglets, and puppies (Orsini and Divers 2014). E. gallinarum and E. casseliflavis
are both suspected to be pathogenic in humans and other mammals (Patel et al. 1993).
Lastly, E. avium is most commonly found in the digestive tract of a variety of species of birds (Monticelli et al. 2018). Many of these species can also be found in symbiosis or as pathogens in a wide variety of species of both mammals and birds. Of all the species, only two are usually found in the human microbiome, while two other species are found in common animals. As a result, it is possible many of the positives represent bacteria that are not a result of human activity. For instance, agriculture is vital to the economy of the Mid-Hudson Valley. Dutchess County is ranked number one in the state for the number of horses, a host of several species of enterococci (Dutchess County Government
2019). Additionally, there is a Canada Geese population of 200,000 within New York
State (NYSDEC 2019), many of which are located around major bodies of water including the Hudson River. The fecal matter of these animals, domestic and wild, may contain various species of enterococci, many of which are detectible by IDEXX
Enterolert. Therefore, it is possible many of the bacteria seen in this study did not originate from humans, but rather animals.
In general, the chloride data follows the expected pattern of the highest values seen in the sewer treatment facility, followed by lower levels located at the HRECOS station and the water treatment plant intake. The higher levels of chloride seen in the water treatment plant effluent were initially confounding. After further research, it was determined the rise in chloride levels were most likely due to the spontaneous decomposition of bleach which is added as a part of the water treatment process. Sodium hypochlorite will spontaneously react with itself to form sodium chloride, oxygen gas,
I-21
sodium chlorate, and small amounts of sodium chlorite. It is predicted this level of
sodium chloride that is formed is enough to increase the total chloride content in the
water treatment plant effluent. While these levels of chloride are elevated, it is important
to note water will not begin to taste salty until a chloride content of approximately 180
mg/L.
The average gadolinium content, when compared to chloride, demonstrates its potential effectiveness as a tracer. Gadolinium levels were found to be the highest in the sewer treatment plant, which was expected as samples from this site should have unaltered levels of gadolinium straight from the source. The large dilution of gadolinium
seen from the sewer treatment plant and the HRECOS station and the water treatment
plant intake. Once leaving the sewer treatment facility, any contaminants will be
homogenized throughout the river due to the sheer volume of the Hudson. It appears that
there is a near 30-fold dilution of gadolinium concentration between the STP and the
open water; however, there is an apparent lower dilution factor for chloride. The
differences in the dilution factors between gadolinium and chloride suggests chloride is
more abundant in the Hudson River than gadolinium. Lastly, in regards to the water
treatment plant, there are discrepancies in the chloride levels. Chloride levels
unexpectedly increased during the water treatment process. The water treatment involves
the addition of sodium hypochlorite, bleach (Church 1994). Bleach can undergo
spontaneous reduction with three pathways proposed in figure 5 (Lister 1956). In each
pathway, sodium chloride is produced which is detectable by the Mohr Method.
I-22 2NaOCl NaCl + NaClO2
NaOCl + NaClO2 NaCl + NaClO3
NaOCl NaCl + 1/2O2
In contrast to the increase in chloride, gadolinium concentrations in the water treatment plant effluent decreased during the treatment process. This can be attributed directly to the water treatment process. During the flocculation step of the water treatment process, a flocking agent is added as noted previously to help facilitate the settling of sediments from the water. The sludge is then collected and can be sold as fertilizer. Chemical analysis from the Poughkeepsie Water Treatment Facility shows higher concentrations of heavy metal such as aluminum, copper, and zinc, higher in sludge than in the effluent. It is expected gadolinium will follow a similar pattern as other heavy metals, settling out of solution with sediment and ending up with the sludge. This may produce issues with the selling of the sludge. High heavy metal concentrations could prove dangerous when used with agricultural crops, or could pollute the environment.
The Poughkeepsie Water Treatment Plant does not have this issue and regularly checks for contaminants and the sludge is regarded as safe.
CONCLUSION
Several organic gadolinium complexes are used as a contrasting agent in MRI procedures. These complexes enhance imaging power of MRI procedures. Gadolinium complexes are extremely stable and are not broken down by the sewer treatment process.
This allows for the potential of gadolinium to be used as a tracer for sewage in areas with high urban density, or with populations that utilize MRI facilities. This study suggests gadolinium is as effective as other methods of detection. Gadolinium levels within the
I-23 study area change in a similar fashion as chloride. Levels of both gadolinium and
chloride are the highest within sewer treatment facilities. These levels decrease
significantly in the open water of the Hudson River. Lastly, gadolinium levels appear to
decrease during the water treatment process. Results from this and other studies suggest
the efficacy of gadolinium as a tracer of municipal wastewater. The data suggests a
positive gadolinium anomaly in the Hudson River. By sampling various points between a
sewage treatment plant and a drinking water path, the gadolinium anomaly was traced to wastewater. Unlike other methods of detection, gadolinium has a single anthropogenic source. Chloride has several sources including water softeners, road salt, and the natural salt front in tidal estuaries such as the Hudson. The only anthropogenic source for gadolinium is sewage treatment plants. Further work needs to be conducted to confirm the effectiveness of gadolinium as a tracer for sewage not only in open water, but through water treatment systems. This study was conducted over a short period of time in a specific circumstance where the primary sewer treatment plant is located relatively close to the primary water treatment facility.
ACKNOWLEDGEMENTS
Special thanks go to the Tibor T. Polgar Fellowship and the Hudson River
Foundation for funding. Dr. Raymond Kepner of Marist College for the use of his laboratory for microbial analysis. Dr Allison Spodek and Karen Wovklich of Vassar
College for the use of their ICP-MS for analysis. Lastly, thanks go to Steve Segna of the
Arlington Sewer Treatment Plant and Dottie DiNobile of the Poughkeepsie Water
Treatment for their assistance in sample acquisition.
I-24
REFERENCES
Caravan, P., J.J. Ellison, T.J. McMurry, and R.B., Lauffer. 1999. Gadolinium (III) chelates as MRI contrast agents: structure, dynamics, and applications. Chemical Reviews 99:2293-2352.
Church, J.A. 1994. Kinetics of the uncatalyzed and copper (II)-catalyzed decomposition of sodium hypochlorite. Industrial & Engineering Chemistry Research 33: 239- 245.
Daughton, C.G. 2009. “Introduction to Pharmaceuticals and Personal Care Products (PPCPS)” United States Department of Environmental Protection (USEPA). https://cfpub.epa.gov/si/si_public_record_report.cfm?Lab=NERL&TIMSType=& count=10000&dirEntryId=142077&searchAll=&showCriteria=2&simpleSearch= 0&startIndex=20001 (accessed December 16, 2019).
Dutchess County Government 2019. Dutchess County Government. Retrieved from https://www.dutchessny.gov/.
IDEXX. 2019. Retrieved from https://www.idexx.com/en/water/water-products- services/enterolert/
Katz, B.G., and D.W. Griffin. 2008. Using chemical and microbiological indicators to track the impacts from the land application of treated municipal wastewater and other sources on groundwater quality in a karstic springs basin. Environmental Geology 55: 801-821.
Lebreton, F., R.J. Willems, and M.S. Gilmore. 2014. Enterococcus diversity, origins in nature, and gut colonization. Enterococci: from commensals to leading causes of drug resistant infection [Internet]. Massachusetts Eye and Ear Infirmary.
Lister, M.W. 1956. Decomposition of sodium hypochlorite: the catalyzed reaction. Canadian Journal of Chemistry 34: 479-488.
Monticelli, J., A. Knezevich, R. Luzzati, and S. Di Bella. 2018. Clinical management of non-faecium non-faecalis vancomycin-resistant enterococci infection. Focus on Enterococcu gallinarum and Enterococcus casseliflavus/flavescens. Journal of Infection and Chemotherapy: 24 237-246.
New York State Department of Environmental Conservation (NYSDEC). 2019. Nuisance Canada Geese. Retrieved from https://www.dec.ny.gov/animals/7003.html.
New York State Department of Environmental Conservation Hudson River Estuary Program (NYSDEC HREP). 2009. Hudson River Watershed Map. Retrieved from https://www.dec.ny.gov/education/63069.html.
I-25
Orsini, J.A., and T.J. Divers. 2014. Equine emergencies: treatment and procedures. Elsevier Saunders, St. Louis, MO.
Patel, R., M.R. Keating, F.R. Cockerill III, and J.M. Steckelberg. 1993. Bacteremia due to Enterococcus avium. Clinical Infectious Diseases 17: 1006-1011.
Rabiet, M., F. Brissaud, J.L. Seidel, S. Pistre, and F. Elbaz-Poulichet. 2009. Positive gadolinium anomalies in wastewater treatment plant effluents and aquatic environment in the Herault watershed (South France). Chemosphere 75: 1057- 1064.
Rechenburg, A., C. Koch, T. Claßen, and T. Kistemann. 2006. Impact of sewage treatment plants and combined sewer overflow basins on the microbiological quality of surface water. Water Science and Technology 54: 95-99.
Stahl, L. 2016. Pilot Study of Pharmaceuticals and Personal Care Products in Fish Tissue. Retrieved September 19, 2016, from https://www.epa.gov/fish-tech/pilot-study- pharmaceuticals-and-personal-care-products-fish-tissue Verplank, P.L., H.E. Taylor, D.K. Nordstrom, and L.B. Barber. 2005. Aqueous stability of gadolinium in surface waters receiving sewage treatment plant effluent, Boulder Creek, Colorado. Environmental Science & Technology. 39: 6923-6929.
Wéry, N., C. Monteil, A.M. Pourcher, and J.J. Godon. 2010. Human-specific fecal bacteria in wastewater treatment plant effluents. Water Research 44: 1873-1883.
Williams, M., A. Kumar, C. Ort, M.G. Lawrence, A. Hambly, S.J. Khan, and R. Kookana. 2013. The use of multiple tracers for tracking wastewater discharges in freshwater systems. Environ. Monit. Assess. 185: 9321-9332.
Wolf, L., C. Zwiener, and M. Zemann. 2012. Tracking artificial sweeteners and pharmaceuticals introduced into urban groundwater by leaking sewer networks. Science of the Total Environment 430: 8-19.
I-26
IS SEA LEVEL RISE ALTERING WETLAND HYDROLOGY IN HUDSON RIVER VALLEY TIDAL MARSHES?
A Final Report of the Tibor T. Polgar Fellowship Program
Sofi Courtney
Polgar Fellow
Biodiversity, Earth & Environmental Sciences Department Drexel University 1900 Benjamin Franklin Pkwy Philadelphia, PA 19103
Project Advisors:
Elizabeth Watson, Ph.D. Department of Biodiversity, Earth & Environmental Sciences Drexel University 1900 Benjamin Franklin Pkwy Philadelphia, PA 19103
Franco Montalto, Ph.D. Department of Civil, Architectural, and Environmental Engineering Drexel University 3141 Chestnut Street Philadelphia, PA 19104
Courtney, S., E.B. Watson, and F. Montalto. 2021. Is Sea Level Rise Altering Wetland Hydrology in Hudson River Valley Tidal Marshes? Section II: 1-36 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
II-1 ABSTRACT
Sea levels are rising globally, and the eastern U.S. is a hotspot for accelerated sea
level rise. Across the mid-Atlantic region, coastal marshes are converting to open water
as marshes lose elevation and tidal channels and ponds expand due to loss of marsh
macrophyte habitat. Marsh groundwater hydrology is a major control on ecological zonation in coastal marshes and is being altered by rising sea levels; thus, it is important
to consider both tidal and groundwater dynamics in order to understand marsh resilience
in the context of climate change. This study examines changes in groundwater hydrology
over 20 years in Piermont Marsh, a mesohaline tidal marsh in the Hudson River Estuary,
New York, by comparing water and marsh surface elevation data collected in 1999 and in
2019. This study found that the frequency of marsh flooding and tidal influence on the marsh water table has increased. Although marsh elevation has gained at a similar rate as sea level rise, the marsh is flooded more frequently because the high tides are increasing more rapidly than sea level, causing an increase in tidal range. This tidal range expansion is also likely deepening marsh creeks which increases tidal gradients to the marsh interior and the influence of tides on the water table. Although the marsh was flooded more frequently, groundwater levels were actually lower relative to the marsh surface in the marsh interior, which might be linked to the expansion of Phragmites australis due to increased evapotranspiration rates. Future research should focus on the impact of P. australis on marsh hydrology, especially in the context of accelerated sea level rise.
II-2 TABLE OF CONTENTS
Abstract ...... III-2
Table of Contents ...... III-3
List of Figures and Tables...... III-4
Introduction ...... III-5
Methods...... III-10
Results ...... III-16
Discussion ...... III-22
Conclusion ...... III-28
Acknowledgements ...... III-29
References ...... III-30
II-3 LIST OF FIGURES AND TABLES
Figure 1 −Expansion of open water in Mid-Atlantic tidal wetlands ...... III-6
Figure 2 −Map of study area ...... III-12
Figure 3 −Average seasonal tidal elevation at the Battery, NY: 1920- 2020 ...... III-15
Figure 4 − Mean sea level at the Battery, NY: 1980-2020 ...... III-17
Figure 5 − Linear trends in tidal elevations at the Battery, NY: 1999 to 2019 .... III-18
Figure 6 − Water table and marsh surface elevations at Piermont Marsh: 1999 . III-19
Figure 7 − Water table and marsh surface elevations at Piermont Marsh: 2019 . III-19
Figure 8 − Maximum and minimum water table elevations at Piermont Marsh . III-20
Figure 9 − Water table elevation relative to marsh surface at Piermont Marsh .. III-21
Figure 10 − Tidal influence on water table fluctuations at Piermont Marsh ...... III-22
Figure 11 –Water table fluctuations at Piermont Marsh: Ponds ...... III-26
Figure 12 –Water table fluctuations at Piermont Marsh: Vegetated...... III-26
Table 1 − Location of stream gauges and groundwater wells ...... III-13
Table 2 − Tidal trends at the Battery, NY ...... III-16
II-4 INTRODUCTION
Sea levels are rising globally (Church and White 2006) and the eastern U.S. is a hotspot for accelerated sea level rise (SLR). Between 1950 and 2009, SLR rates between
Boston and Cape Hatteras were three to four times higher than the global average, a pattern arising not only from geologic factors, but from ocean temperature, density, and circulation patterns (Sallenger et al. 2012). In fact, the rates of SLR acceleration in the eastern U.S. are among the greatest in the world, and in recent years, actual water levels have averaged more than 10 cm greater than predicted heights (Sweet and Zervas 2011).
Across the mid-Atlantic region, coastal marshes are in decline. Analyses conducted for a ‘State of the Estuary’ report indicates that Delaware Bay has lost an acre per day of tidal wetlands between 1996 and 2006 (PDE 2012), and a recent analysis of
Long Island tidal wetlands shows a loss rate exceeding 10% between the early 1970s and early 2000s (Bowman 2015). Similar rates and patterns of losses have also been estimated for Chesapeake Bay and southern New England (Kearney et al. 2002; Smith
2009; Watson et al. 2017), suggesting that these patterns and trends are regionally widespread. These losses of wetland habitat are symptomatic of marsh drowning (e.g.,
Blum and Roberts 2009) and include marsh retreat, edge erosion, marsh island loss, and the development and enlargement of die-back areas found on the marsh platform (Figure
1). While some of these wetland losses are eventually expected to be offset by marsh migration into uplands, intensive development of adjacent areas in the high population density mid-Atlantic region acts as a barrier for upslope transgression of intertidal habitats, exacerbating threats of SLR to coastal marsh survival.
II-5 The widespread loss of marsh habitat loss across the East Coast is being driven by
accelerated SLR. Marsh vegetation losses have been observed to be a function of
elevation. Marshes that sit far below the level of the high tide convert to tidal flats much
faster than higher elevation coastal wetlands (Watson et al. 2014), suggesting increasing inundation as a significant driver of loss. In addition to outright loss (e.g., Hartig et al.
2002), shifts from high to low marsh vegetation have been found to occur where marsh
accumulation falls below rates of SLR (Warren and Niering 1993), illustrating that
current flooding levels are detrimental to the survival of marshes. New evidence from
mesocosm experiments shows that increased inundation reduces marsh macrophyte
productivity (Watson et al. 2017).
Jamaica Bay, New York City Piermont Marsh, Hudson River Barnegat Bay, New Jersey
Figure 1. Expansion of open water in Mid-Atlantic tidal wetlands. Satellite images from the turn of the century and 2016 illustrate an increase in unvegetated extent over time, primarily through the expansion of interior ponds and the widening of tidal channels. Aerial images: Google Earth Pro.
II-6 In addition to increasing flooding in tidal marshes, SLR is changing shallow
groundwater hydrology in coastal systems (Wang et al. 2017; Rotzell and Fletcher 2013).
Tidal groundwater exchange is a vital driver of ecosystem function and ecological
zonation in coastal marshes. In the marsh interior, freshwater shallow aquifers are perched on top of saltwater inputs from ocean systems and are influenced by forcing from tidal creeks that essentially push the freshwater up at high tide (Wilson et al. 2011). The
amplitude of tidally influenced groundwater fluctuations is expressed as tidal efficiency,
which is the fraction of tidal flux in the water table relative to the tidal range in the
oceanic boundary. Due to the restricted length of a tidal period and the low permeability
of marsh sediments, tidal efficiency on the marsh platform decreases with distance from
tidal creeks (Harvey and Odum 1990; Williams et al. 2002; Wilson et al. 2011). Rising
sea levels can have a compounded influence on marsh groundwater elevations and
fluctuations (Bjerklie et al. 2012). This hydrological regime is a major driver of marsh
macrophyte distribution (Morris 1995; Moffett et al. 2012; Xin et al. 2013; Wilson et al.
2015).
As hydrological regime is a driver of marsh macrophyte zonation, changes in marsh hydrology due to SLR can result in vegetation die-off (Smith et al. 2012) which
drives peat collapse and ponding (Delaune et al. 1994). The temporal expansion of marsh pools has also been correlated to hydrological conditions (Schepers et al. 2017; Sandi et al. 2018). The loss of marsh vegetation contributes to marsh instability by reducing sedimentation rates and causing erosion (Coleman and Kirwan 2019). Consequently, macrophyte die-back can result in the formation of increasing areas of open water pools in marsh habitats (Figure 1) which in the context of SLR are unlikely to convert back to
II-7 vegetated land (Mariotti 2016). Highly flooded portions of salt marshes are poor habitats
for Phragmites australis (Chambers et al. 2003) so when native species die back due to increased flooding ponded areas will likely stay bare (Mariotti 2016). While P. australis can be problematic for local plant diversity (Silliman and Bertness 2004) the species may be contributing to greater rates of marsh accretion than Spartina alterniflora and preventing erosion of marsh sediments (Rooth and Stevenson 2000).
Changes in marsh hydrology due to SLR is causing wide scale habitat loss of high priority conservation species such as birds and fish (Bilkovic et al. 2012; Thorne et al.
2015; Hunter et al. 2015). Marsh birds such as the globally threatened saltmarsh sparrow
(Ammospiza caudacuta) depend on high-marsh vegetation for nesting sites. Rising water tables and increased flooding frequency will not only reduce the amount of habitat available to these birds (Hunter et al. 2015), but higher tides will likely result in increased
frequency of nest flooding and egg loss (Bayard and Elphick 2011) and consequent
species decline and extirpation or extinction (Rush et al. 2009; Correll et al. 2016; Field
et al. 2016; Rosencranz et al. 2018). Increased open water area and channel connectivity
in marsh habitats can increase predator access to resident marsh fish (Torio and Chmura
2015) and eventually as marshes convert to open water to the complete loss of salt marsh
pond habitat (Thorne et al. 2015; Crotty, et al. 2017). Consequently, it is important to
understand how groundwater levels in coastal wetlands are being altered by SLR across
multiple scales and ecological groups.
While assessments of wetland vulnerability to SLR have traditionally focused on
wetland elevation relative to the tides and wetland elevation change or predicted changes
using spatial models (Raposa et al. 2016; Watson et al. 2017), this study examines the
II-8 change in the water table and tidal influence over time by leveraging the data from previous research. In 1999, Montalto et al. (2006) mapped the hydrology of a brackish tidal marsh within the Hudson River estuary by assessing a variety of variables including topography, water table elevation, and hydroperiod and found low levels of tidal influence and an increase in water table elevation within the marsh interior. Twenty years later, this study revisited the site and methods used by Montalto et al. (2006) to evaluate the change in marsh hydrology over time due to SLR and its impact on marsh macrophyte distribution and soil stability. In addition to these goals, this study attempts to determine if there are water table characteristics associated with conversion to open water by comparing current areas of marsh die-back with surrounding areas of healthy marsh. The following hypotheses were tested:
1. Hydrology changed at Piermont Marsh from 1999 to 2019.
2. Due to changes in hydrology over time, such as SLR and the expansion of P.
australis populations at Piermont marsh, tidal influence on the water table extends
further into the marsh interior in 2019 than in 1999.
3. The expanding ponds in the center of the marsh have different hydrological
regimes than vegetated areas of the marsh. Specifically, the ponded areas have a
more static water table than vegetated areas, i.e. less tidal influence on the water
table.
II-9 METHODS
Study Site
Piermont Marsh is a tidal marsh located within the Hudson River Estuary in New
York, United States (Figure 2). It is comprised of 417 hectares on the western side of the
Governor Mario M. Cuomo Bridge in Rockland County, about 40 km north of New York
City. The site is comprised of brackish tidal marshland and shallows at the mouths of
Sparkill and Crumkill Creeks and includes intertidal flats and uplands. The marsh experiences diurnal mesohaline tides is bisected by tidal creeks and channels. The marsh interior is dominated by salt marsh ponds in which the spotfin killifish (Fundulus luciae)
was first reported within the Hudson River estuary (Yozzo and Ottman 2003). The marsh
is located at the southernmost edge of the Hudson River National Estuarine Research
Reserve (HRNERR) and is designated as a Significant Coastal Fish and Wildlife Habitat
by the New York State Department of State (NYSDOS 2012) and a Critical
Environmental Area by Piermont Village (NYSDEC 1985). The site’s vegetation is
dominated by Phragmites australis but retains some small patches of native marsh
species in the interior. Local land managers and residents anecdotally report that species
diversity on the marsh has decreased over the last few decades (NYSDEC 2017).
Study Design
Hydrologic measurements were collected over the summer of 2019 on Piermont
Marsh. These data were compared to hydrological measurements made in 1999, which focus on describing the marsh water table from a large tidal channel to the marsh interior
(Montalto et al. 2006). The aim of this study was to compare current tidal flooding and
II-10 groundwater table levels with measures made in 1999, in order to identify how SLR has altered marsh hydrology. In addition, ground water levels were monitored within and adjacent to the expanding open water area in the center of the marsh (Figure 2).
Water Table Measurements
Three tidal gauges were installed within the tidal creek (Figure 2; Table 1). Onset
Hobo 20UL water level loggers were suspended inside perforated pipes and attached to cinderblocks which were placed in the center of the channel. Loggers were programmed to take a water elevation measurement every ten minutes. In order to determine ground water levels throughout the marsh, seven water level loggers were installed along a gradient from the tidal channel to the upland, replicating the measures conducted by
Montalto et al. (2006). The location of Montalto’s wells were established by georeferencing maps from Montalto et al. using Google Earth 7.3.2.5776 and were confirmed visually by Franco Montalto in the field.
Wells were constructed by suspending a pressure transducer within a 7.5 cm diameter perforated PVC pipe lined with screening to prevent sediment from entering the well. A hole was excavated with an augur and the well was placed inside it. The surface of the well was vented to the atmosphere. A concrete collar was installed at the marsh surface around the well in order to prevent the preferential flow of water down the side of the well. Seven wells were installed along the original transect, perpendicular to the creek, and two additional wells were installed along the same transect into the ponded area at the center of the marsh (Figure 2, Table 1).
II-11
Figure 2. Map of study area showing (A) the Hudson River Valley with the location of Piermont Marsh; (B) Vegetation cover at Piermont Marsh which is dominated by the non-native Phragmites australis; (C) the location stream gauges (denoted as circles) as well as groundwater wells along the a transect perpendicular to a tidal channel and in two areas of expanding ponded water at Piermont Marsh. The location of map insets (marked ‘B’ on map A and ‘C’ on map B) is shown.
Wells were installed 5 May 2019, and water levels were downloaded three times between May 2019 and 12 August 2019, using a Hobo Optic USB Base Station. An atmospheric pressure logger was also deployed to correct pressure transducer data for variations in atmospheric pressure. The absolute elevation of the top of each well was measured using RTK-enabled static GPS measurements from Leica GNSS GS14 rover units and static measures using an AX1202 GG base station unit in order to reference water levels to the NAVD88 vertical datum. Reference water levels were measured each time data was collected. The distance from the top of the well to the water surface was
II-12 measured by hand using a meter stick and the time in GMT-4 was recorded. To relate marsh elevation with water elevations, GPS surveys were conducted along the transect using a Leica GNSS GS14 rover unit.
Table 1. Location of stream gauges (SG) and groundwater (GW) wells.
2019 Marsh 1999 Marsh Distance from Type Latitude Longitude Elevation (m, elevation (m, creek (m) NAVD88) NAVD88) 1 SG 0 41.036750° -73.909667° - - 2 SG 0 41.036083° -73.910611° - - 3 SG 0 41.034556° -73.912833° - - 1 GW 1 (levee) 41.036061° -73.910498° 0.72 0.61 2 GW 6 41.036023° -73.910455° 0.72 0.63
3 GW 12 41.035980° -73.910411° 0.73 0.64 4 GW 18 41.035941° -73.910361° 0.75 0.65 5 GW 24 41.035899° -73.910317° 0.77 0.66 6 GW 36 41.035812° -73.910233° 0.72 0.68 7 GW 48 41.035717° -73.910142° 0.80 0.69 8 GW 135 41.035155° -73.909433° - - 9 GW 164 41.034957° -73.909261° - -
Data Analysis
Although rates of SLR at the Battery, NY, are reported at 2.87 mm yr-1, (NOAA
2020), these long-term rates do not capture more recent sea level trends. For this reason,
simple linear trends were calculated using monthly tide station data (from the Battery tide
gauge, NY) to estimate changes that have occurred between 1999 and 2019 in terms of
tidal flooding. Monthly mean sea level, monthly mean high water, and monthly mean low
water were obtained, and trends estimated using linear regression (NOAA 2020). The two
II-13 studies’ position in the 19-year metonic cycle were also examined using monthly water elevation levels (1980-2020) from the NOAA Battery tidal gauge (NOAA 2020).
Pressure transducer data was post-processed using HOBOware Pro (Ver. 3.7.16 ,
Onset Computer Corporation, Bourne, MA) using reference water levels collected in the field and were corrected for atmospheric pressure using the HOBOware barometric compensation assistant. Barometric data was obtained from the Hudson River National
Estuarine Research Reserve, due to inconsistences in data from the barologger (NERRS
2019). Raw water elevation data from 1999 was analyzed in concert with the 2019 data.
Water level data from 1999 were converted from the NVGD29 to NAVD 88 datum using
NOAA VDatum v4.0.1 (NOAA 2019) prior to analysis. The transducer in well seven experienced three brief malfunctions from 30 May to 3 June 2019, which resulted in inaccurate elevation measurements for a total of 19.5 hours. These data were excluded from the analysis. In 1999, Montalto also experienced malfunctions at well seven. These data were corrected by Montalto into smoothed six-hour increments using average water elevation measurements and calculated error and calibrated using regression (Montalto et al. 2006). No other well transducers appeared to have malfunctioned.
Changes in surface flooding were computed using data from channel gauges installed in 1999 and 2019 (5 May – 12 August). Marsh flooding was calculated as the percentage of time that water levels exceeded the average marsh elevation (0.75 m
NAVD in 2019 and 0.65 m NAVD in 1999). The number of tides that flooded the marsh per month and the average maximum flooding depth were identified for 1999 and 2019.
Sub-surface flooding (6 April 1999 – 26 May 1999; 5 May 2019 – 30 June 2019) was calculated as the percentage of time that the groundwater table exceeded thresholds of 5
II-14 and 10-cm below the marsh surface. Although groundwater data were compared over the
same season (spring), which is important due to the strong seasonal variability in sea
level in the Mid-Atlantic (Figure 3), dates could not be completely matched (6 April 1999
– 26 May 1999 vs. 5 May 2019 – 30 June 2019) due to limited data in 1999. Comparing
mean sea level to April and May, to mean sea level during May and June, suggests a minimal difference in sea level over this time period (2 cm).
Figure 3. Average seasonal tidal elevation at the Battery, NY: 1920 to 2020. Monthly tidal variation from at the NOAA Battery tidal gauge (NOAA 2020). These data show the seasonal cycle in tides and illustrate that it is appropriate to directly compare the 1999 and 2019 data despite a one month offset. The seasonal cycle difference between April and May is 2.4 cm and the difference between May and June is 1.6 cm.
Tidal efficiency (TE) is defined as the ratio of the amplitude of groundwater
fluctuations in a coastal aquifer to the amplitude of tidal fluctuations at the ocean boundary. Range of tide and tidal efficiency (TE) were calculated for 1999 and 2019 well
data empirically as:
= 𝑠𝑠𝑤𝑤 𝑇𝑇𝑇𝑇 𝑠𝑠𝑡𝑡
II-15 where is the daily range of water-level fluctuation in a well tapping the groundwater
𝑤𝑤 table and𝑠𝑠 refers to the daily range of tide (Ferris et al. 1962). Because the 1999 channel
𝑡𝑡 gauges were𝑠𝑠 exposed at low tide (Montalto et al. 2006), TE was calculated based on range of tide reported from the NOAA tide gauge at the Battery, NY, (NOAA 2020).
Data analysis and calculations were performed in Excel (Microsoft, version 16.35) and in
R 3.6.1 (R Core Team 2019) using package FSA (Ogle et al. 2020).
RESULTS
Tidal data
Examination of water levels at the Battery shows that measures (in spring 2019 and 1999) occurred during normal and not anomalous tidal periods (Figure 4). Because both studies occurred 20 years apart, and the chief metonic cycle is 18.6 years, both studies occurred at similar points in the cycle. From linear regression, it was estimated that from 1999 to 2019, mean high water at the Battery increased at an average rate of 7.5 mm yr-1, mean sea level increased by 4.5 mm yr-1, and mean low water increased by 2.0 mm yr-1 (Figure 5; Table 2).
Rate of increase Regression Equation Total r2 p (mm yr-1) change Mean monthly high water 7.6 y = 7.56x10-3x – 13.68 0.09 4.38x10-12 15.1 cm Mean monthly sea level 4.5 y = 4.52x10-3x – 8.24 0.08 1.12x10-5 9.0 cm Mean monthly low water 2.0 y = 1.98x10-3x – 3.86 0.02 0.05 4.0 cm Tidal range 5.6 Tidal range = MHW - MLW - - 11.1 cm
Table 2. Tidal trends at the Battery, NY. Rates of mean water elevation change from April 1999 to April 2019, the equations used to calculate rates (x = year; y = water level in meters), and the overall mean change in tidal range. Comparing the channel gauge records with marsh elevations in 1999 and 2019 suggest that tidal flooding has increased. The average marsh elevation along the transect
II-16 in which the wells were emplaced was flooded 3.2% of the time in 1999, and 9.9% of the
time in 2019. In 1999, the marsh flooded on average 12 times per month between May
and August. The average flooding duration was 2.1 ± 0.80 hours (mean ± standard
deviation), and the marsh flooded on average to a depth of 8.7 ± 5.4 cm. In contrast in
2019, the marsh flooded on average 29 times per month between May and August. The
average flooding duration in 2019 was 2.6 ± 1.1 hours, and the marsh flooded on average to a depth of 12 ± 8.8 cm. While marsh flooding was limited to spring tides during both time periods, high tides resulted in marsh flooding about 2.4 times more frequently in
2019 than 1999.
Figure 4. Mean sea level at the Battery: 1980 to 2020. This figure illustrates seasonal changes in water level, upward trends in sea level, and periodic changes in tides due to astronomic and oceanographic forcings (NOAA 2020). A loess smoothing curve (span=0.40) shows multi-decadal sea level fluctuations. Because sampling periods were 20 years apart, they are during approximately the same relative positions within the 18.6 year metonic cycle.
II-17
Sea Level Rise at the Battery: 1999 to 2019
1.75
1.5
1.25
1
0.75
Elevation (m, NAVD88) (m, Elevation 0.5
0.25
0
-0.25 Mar-99 Aug-02 Dec-05 Apr-09 Aug-12 Dec-15 Apr-19
Mean High Water Mean Sea Level Mean Low Water
Figure 5. Linear trends in tidal elevations at the Battery: 1999 to 2019 (NOAA 2020). Mean monthly elevations at the Battery between 1999 and 2019 evaluated with linear regression. Rates of rise in mean high water was 7.6 mm yr-1; the rate of sea level rise was 4.5 mm yr-1, and the rate of mean low water rise was 2.0 mm yr-1 (Table 2).
II-18
Piermont Marsh Water Table Elevation: Spring 1999 1 River
0.5 Levee 6 m 0 12 m 18 m -0.5 24 m
Elevation Elevation (m NAVD88) -1 36 m Marsh Surface -1.5 8-Apr 18-Apr 28-Apr 8-May 18-May Figure 6. Water table and marsh surface elevations at Piermont Marsh: 1999. Water table and marsh surface elevations from April through May 1999. Tidal elevation reference from the NOAA Battery gauge is in black (NOAA 2020), and each Piermont Marsh wells are represented by a gradient from red to purple, reds are closer to the tidal creek and purples are farther from the creek. Marsh surface is shown in green.
Piermont Marsh Water Table Elevation: Spring 2019 River Levee 1 6 m 12 m 0.5 18 m 24 m 0 36 m 48 m -0.5 Pond 1 Elevation Elevation (m, NAVD88)
-1 Pond 2 6-May 16-May 26-May 5-Jun 15-Jun 25-Jun Marsh Surface
Figure 7. Water table and marsh surface elevations at Piermont Marsh: 2019. Water table and marsh surface elevations from May through June 2019. Tidal elevation reference from the NOAA Battery gauge is in black (NOAA 2020), and each Piermont Marsh wells are represented by a gradient from red to purple, reds are closer to the tidal creek and purples are farther from the creek. Marsh surface is shown in green.
II-19 Water table data
Water table elevation, tidal range, and tidal efficiency were all greater in magnitude in
2019 than in 1999 (Figures 8-10). The median maximum water in all wells were at least 10 centimeters higher in elevation in 2019 than in 1999, with the greatest difference at the levee, where water levels were 20 cm greater in 2019 than in 1999. While the median minimum water was higher in all wells in 2019 than in 1999, the difference was of a smaller magnitude than the mean maximum water values. Median minimum water elevation differences ranged from 0.4 cm to 1.3 cm, with the largest magnitudes at the levee and six meters in from the creek. In 2019, median minimum water showed a similar increase in elevation with distance from the creek as in
1999. In both maximum and minimum tides, the interquartile range of the data was generally greater in 2019 than in 1999 (Figure 8).
Figure 8. Maximum and minimum water table elevations at Piermont Marsh. Semidiurnal water table elevations across Piermont Marsh in the spring of 1999 and 2019. 1999 data are in blue and 2019 data are in orange. Dark colors are high water data and light colors are low water data.
II-20 While surface flooding was found to increase consistently over the last 20 years, the
position of water table relative to the marsh varied with channel proximity. In 1999, the mean
water table position was below the marsh surface (e.g. 10-20 cm of depth) adjacent to the tidal
channel, but increased to near the marsh surface (to e.g. 0-2 cm of depth) in the marsh interior. In
2019, the same general trend was found, with lower water tables at the channel edge and
increasing water tables in the marsh interior; however, in comparison with 1999, the percentage
of time that the water table was at the marsh surface in 2019 vs. in the marsh interior decreased
(Figure 9), resulting in decreasing prevalence of root zone flooding.
Figure 9. Water table elevation relative to marsh surface at Piermont Marsh. The percent of daily time in which the water table was within 5 and 10 cm of the marsh surface elevation along the gradient from the channel edge to marsh interior. The values indicate median and inter-quartile range.
The tidal influence on groundwater fluctuations showed a similar trend in both years, that is decreasing with distance from the tidal channel, but the magnitude of tidal influence is higher in all wells in 2019 than in 1999. The greatest difference is at the levee, with a difference of 0.68 and the smallest difference is 48 m from the creek, with a difference of 0.16 (Figure 10).
II-21 Tidal Efficiency Vs. Distance From Tidal Channel
0.16 1999 0.12 2019
0.08 Efficiency 0.04
0.00 0 5 10 15 20 25 30 35 40 45 50 Distance from Creek (m)
Figure 10. Tidal influence on water table fluctuations at Piermont Marsh. A comparison of tidal efficiency in the spring 1999 and 2019. Tidal efficiency is the amplitude of the groundwater fluctuation expressed as a fraction of the tidal range in the Hudson River. 1999 data is in blue and 2019 data is in orange. The wells installed in ponds in the interior of the marsh are experiencing less tidal influence than those wells closer to the tidal creek. While there is a steady decrease of mean tidal efficiency up to 48 meters into the vegetated marsh, the curve seems to have bottomed out in the ponded marsh interior. Pond 1 and pond 2 have almost identical median efficiencies, at 0.03, and the spread of the data at both wells is similar, at 0.04 in the interquartile range. All wells in the vegetated areas had an interquartile range at or more than 0.05.
DISCUSSION
Hydrology has changed in Piermont Marsh over the last 20 years. Tidal range has expanded both in the marsh and at the Battery in the Hudson River, and tidal influence is extending farther into the marsh in 2019 than in 1999. Mean sea level and marsh elevation have risen at similar rates, but the frequency and magnitude of high water flooding the marsh has increased, which is likely driving changes in hydrology and thus the ecological zonation of
II-22 marsh habitats. While marsh flooding has increased over time, the water table was lower relative
to the marsh surface. In addition to SLR, a likely driver of the observed changes in the
groundwater table is the expansion of P. australis into the marsh interior, which could be altering
soil porosity, hydraulic conductivity, and evapotranspiration rates. Finally, the ponded areas in
the center of Piermont Marsh are experiencing less tidal influence than the vegetated areas.
These distinct hydrological regimes may be shifting ecological zonation and consequent peat
collapse and expansion of open water.
Tidal range expansion has primarily been driven by increased elevations of high tides, rather than by changes low tide elevations. In Piermont Marsh, the magnitude of the difference of mean high tides in 1999 to 2019 is greater in all wells than the magnitude of the difference of
mean low tides. For example, 2019 mean high tide at the levee is 19.4 cm higher than the mean
high tide in 1999, but the mean low tide is only 7 cm higher, which is less than the 12 cm
average increase in marsh surface elevation. This trend is apparent along all wells in the transect
and can also be observed at the NOAA Battery gauge in the Hudson River. While sea level at the
Battery rose at a rate of 4.6 mm yr-1 from 1999 to 2019, monthly mean low water has risen by
1.9 mm yr-1 and mean monthly high water has risen by 7.56 mm yr-1, expanding the mean tidal
range by more than 10 cm. Increasing tidal ranges introduce more hydraulic energy in tidal
creeks, which can result in deeper channels with steeper banks (Allen 2000; Williams and Orr
2002; Williams et. al 2002). This creek geometry increases the tidal gradient to the marsh interior, resulting in greater tidal influence on the marsh water table (Allen 2000; Wilson and
Gardner 2006; Wilson et al. 2011). The interactions of increased tidal range, marsh stability, and
resilience to SLR are debated in the scientific community (Osgood 2000; Kirwan and
Guntenspergen 2010; Pickering et al. 2012; Balke et al. 2016; Cahoon et al. 2019).
II-23 Increased tidal range driven by higher high tides are contributing to another major trend observed in this study, which is increased tidal influence on the marsh groundwater table, particularly in the marsh interior. In 1999, tidal influence did not propagate more than 20 meters into the marsh, but in 2019, tidal influence continued far into the marsh interior. As can be observed in Figure 10, in 1999 mean tidal efficiencies, at and past 18 meters, are consistently around 0.015. In 2019 tidal efficiencies are higher at all wells than in 1999, and do not level out until far into the ponded marsh interior. In addition, the pond wells in 2019 experienced tidal efficiencies around 0.025, higher than the interior wells measured in 1999. As a consequence of the change in tidal range and increased tidal efficiency in the marsh interior, the increasing trend of mean high tide with distance that was observed in 1999 is less apparent in 2019. This is because in 1999, only the highest tides were propagating into the marsh interior, which caused the mean to increase due to lack of influence from low tides. In 2019, more tides of all magnitudes are propagating into the marsh interior, causing the mean maximum tide to be fairly consistent across the entire transect; however, the mean minimum tides increase in magnitude with distance from the creek both in 1999 and in 2019, resulting in a perched water table in the marsh interior in both years.
As the frequency and magnitude of water table fluctuations determine the eco- hydrological zonation of marsh macrophyte habitat (Moffett et al. 2012; Xin et al. 2013; Wilson et al. 2015), the observed changes in hydrology are altering plant communities across the marsh.
Another likely factor is the expansion of P. australis into the marsh interior. P. australis could have a compounding impact on marsh macrophyte community distribution, because not only does it crowd out native competitors, but dense P. australis populations also alter groundwater hydrology (Windham and Lathrop 1999; Chambers et al. 2003). While there is not extensive
II-24 research on the impact of P. australis on groundwater hydrology, there is evidence that the P. australis root mat generally increases hydraulic conductivity in marsh sediments (Baird et al.
2004; Saaltink et. al. 2019) and that the presence of extensive stands of P. australis can lower the water table due to increased evapotranspiration (Windham and Lathrop 1999; Windham et al.
2001). Phragmites has been expanding across Piermont Marsh since at least the 1960’s
(Winogrond and Kiviat 1997) and anecdotal evidence and observations of satellite images show that P. australis distribution has greatly expanded at Piermont Marsh over the last 20 years.
The ponded areas in the Piermont Marsh transect are experiencing less tidal influence than the vegetated area. While mean groundwater fluctuations continued to decline along an exponential curve from the levee to 48 meters into the marsh, they appear to have leveled out in the ponded marsh interior. Wilson et al. (2015) identified four primary eco-hydrological zones in
Atlantic salt marshes, characterized by 1) short form Spartina alterniflora and 2) tall form
Spartina alterniflora in the low marsh, 3) Salicornia zone in the high marsh, and 4) a Juncus zone adjacent to the uplands. The hydrological regime observed at Piermont marsh in 2019 resembles the Spartina zones in vegetated areas and the Salicornia zone in the ponds. As defined by Wilson et al. (2015) the Salicornia zone is primarily driven by upward flow of groundwater during neap tides without significant discharge in between high-water events, resulting in hypersaline zones in the marsh interior. This hydrological pattern was observed in the ponded areas at Piermont Marsh (Figure 11) and may be compounded by increased frequency and amplitude of high water and consequent increased tidal influence. Hydrological regime in the intact section of the marsh suggest that the water table is being forced up at high water, but is also draining at low water, allowing for greater flushing in vegetated areas than in the ponded areas (Figure 12).
II-25 1.2 Piermont Marsh Water Table Elevation 2019: Ponded Area
0.8 Creek
0.4 Pond 1
0 Pond 2
-0.4 Marsh Elevation (NAVD 88) (NAVD Elevation Surface -0.8 6-May 21-May 5-Jun 20-Jun 5-Jul 20-Jul 4-Aug Figure 11. Water table fluctuations at Piermont Marsh: Ponds. Water table and marsh surface elevation in the ponded area from May to August 2019. Tidal channel water surface elevations are in grey and wells located inside ponds are in blue. Marsh surface is shown in green. The ponded areas are located far into the marsh interior. Water table fluctuations indicate upward forcing of ground water at high water, but little drainage at l t Creek 1.2 Piermont Marsh Water Table Elevation 2019: Vegetated Area Levee
0.8 6m
12m 0.4 16m
0 24m
-0.4 36m Elevation (m, NAVD 88) NAVD (m, Elevation 48m
-0.8 Marsh 6-May 21-May 5-Jun 20-Jun 5-Jul 20-Jul 4-Aug Surface Figure 12. Water table fluctuations at Piermont Marsh: Vegetated. Water table and marsh surface elevation in the vegetated area from May to August 2019. Tidal channel water surface elevations are in grey, reds data points are closer to the tidal creek and purple are farther from the creek. Marsh surface is shown in green. Water table surface elevation in the vegetated areas indicates both upward forcing from high water and drainage during low water. II-26 Future Research
Tidal range dynamics have changed dramatically from 1999 to 2019 at Piermont
Marsh and at the Battery in the Hudson River. Other studies have found similar trends globally and locally (Pickering et al. 2012; Mawdsley et al. 2014; Balke et al. 2016;
Pickering et al. 2017; Talke et al. 2018) and some tidal datums have been updated to reflect changing tidal dynamics due to accelerated SLR (Bamford 2013; Wang and Myers
2016); however, a comprehensive review of mean high water and mean low water trends in the United States has not been published since 2003 (Flick et al. 2003). A more comprehensive analysis of regional tidal range and SLR should be performed in order to more fully understand changing tidal dynamics in the Mid-Atlantic region and to inform marsh and coastal management into the future.
Further research into the extent and impact of Phragmites australis at Piermont
Marsh is necessary. The authors of this report hope to perform a spatial analysis on P. australis distribution from the mid-1996 to 2015, and to perform porosity tests on sediments from Piermont Marsh in order to better understand how invasion by
Phragmites may alter marsh groundwater levels. Generally, more research is necessary to understand how P. australis alters hydrology in coastal marshes, particularly in regard to hydraulic conductivity and porosity.
Finally, this study is only a comparison of one particular year to another, so while these data do show that hydrology in Piermont Marsh has changed over twenty years, these data do not reveal the rate of change. In addition, extensive statistical analysis of these data is not possible with only two time-steps to compare. Additional years of
II-27 monitoring will clarify the findings of this study and make possible an evaluation of the
statistical merit of these findings.
CONCLUSION
Hydrology changed dramatically at Piermont Marsh from 1999 to 2019. The tidal range in the Hudson River has expanded over the last 20 years primarily due to rapid increase of mean high-water elevations. Larger tidal ranges in tidal creeks are increasing
the tidal gradient into the marsh interior, which results in greater amplitudes of water table fluctuations further from the creek. Tidal influence is now extending further into the marsh interior and while the marsh is flooding more frequently, the water table elevation is lower relative to the marsh surface. Expanding ponded areas in the marsh interior are experiencing tidal forcing during high water but are not draining at low water. Altered hydrology may be impacting the ecological zonation of marsh macrophytes causing vegetation die-back, peat collapse, and the increasing areas of open water across the marsh. SLR, the expansion of tidal range, and increased populations of P. australis are likely important drivers of the changes observed in Piermont Marsh. Further research is recommended on tidal range expansion in the Northeastern United States and on the impact of P. australis populations on marsh hydrology and ecological zonation.
II-28 ACKNOWLEDGEMENTS
The researchers would like to thank the Hudson River Foundation and the Tibor T.
Polgar Committee for providing funding and logistical support for this project. We would also like to thank everyone who helped with this work along the way: The Hudson River
National Estuarine Research Reserve and the New York State Office of Parks, Recreation
and Historic Preservation kindly allowed us access to the study site. Lena Champlain and
Johannes Krause assisted us in the field. Kirk Raper, and the Academy of Natural Science
of Drexel University provided technical assistance and equipment. Iggy, Malakai, and
Nicky were cheerful companions in the field despite the mud and long days. Michael and
Liz Biddle generously allowed us to stay at their house while we performed field work
and Michael Ferrin provided transportation, meals, and unflagging emotional support.
II-29 REFERENCES
Allen, J.R.L. 2000. Morphodynamics of Holocene salt marshes: a review sketch from Atlantic and Southern North Sea coasts of Europe. Quaternary Science Reviews 19: 1155-1231.
Baird, A.J., B.W. Surridge, and R.P. Money. 2004. An assessment of the piezometer method for measuring the hydraulic conductivity of a Cladium mariscus- Phragmites australis root mat in a Norfolk (UK) fen. Hydrological Processes 18: 275-291.
Balke, T., M. Stock, K. Jensen, T.J. Bouma and M. Kleyer. 2016. A global analysis of the seaward salt marsh extent: The importance of tidal range. Water Resources Research 52: 3775-3786.
Bamford, H.A. 2013. Notice of Change to the Nation's Tidal Datums with the Adoption of a Modified Procedure for Computation of Tidal Datums in Area of Anomalous Sea-Level Change. Washington: Federal Information & News Dispatch, LLC.
Bayard, T.S. and C. Elphick. 2011. Planning for sea-level rise: Quantifying patterns of saltmarsh sparrow (Ammoddramus caudacutus) nest flooding under current sea- level conditions. The Auk 128:393-403.
Bilkovic, D.M., M.R. Mitchell, C.H. Hershner, and K.J. Havens. 2012. Transitional wetland faunal community characterization and response to precipitation-driven salinity fluctuations. Wetlands 32: 425-437.
Bjerklie, D.M., J.R. Mullaney, J.R. Stone, B.J. Skinner, and M.A. Ramlow. 2012. Preliminary investigation of the effects of sea level rise on groundwater levels in New Haven Connecticut. Open-file report 2012-1025. Reston, Virginia: US Geological Survey 46p.
Blum, M.D., and H.H. Roberts. 2009. Drowning of the Mississippi Delta due to insufficient sediment supply and global sea-level rise. Nature Geoscience 2: 488- 491.
Bowman, W. 2015. Tidal wetlands trends and conditions assessment, Long Island. Sound Update, (Newsletter of the Long Island Sound Study). Winter 2014-2015:6.
Cahoon, D.R., J.C. Lynch, C.T. Roman, J.P. Schmit, and D.E. Skidds. 2019. Evaluating the relationship among wetland vertical development, elevation capital, sea-level rise, and tidal marsh sustainability. Estuaries and Coasts 42:1-15.
Chambers, R.M., D.T. Osgood, D.J. Bart, and F. Montalto. 2003. Phragmites australis invasion and expansion in tidal wetlands: Interactions among salinity, sulfide, and hydrology. Estuaries 26:398-406.
II-30
Church, J.A. and N.J. White. 2006. A 20th century acceleration in global sea-level rise. Geophysical Research Letters 33: L01603.
Coleman, D.J. and M.L. Kirwan. 2019. The effect of a small vegetation dieback event on salt marsh sediment transport. Earth Surface Processes and Landforms 44: 944- 952.
Correll, M.D., W.A. Wiest, T.P. Hodgman, W.G. Shriver, C.S. Elphick, B.J. McGill, K.M. O’Brien, and B.J. Olsen. 2016. Predictors of specialist avifaunal decline in coastal marshes. Conservation Biology 31: 172-182.
Crotty, S.M., C. Angelini, and M.D. Bertness. 2017. Multiple stressors and the potential for synergistic loss of New England salt marshes. PLoS One 12: e0183058.
Delaune, R.D., J.A. Nyman, and W.H. Patrick, Jr. 1994. Peat collapse, ponding and wetland loss in a rapidly submerging coastal marsh. Journal of Coastal Research 10:1021-1030.
Ferris, J.G., D.B. Knowles, R.H. Brown. and R.W. Stallman. 1962. Theory of aquifer tests: U.S. Geological Survey Water-Supply Paper 1536-E: Elasticity of artesian aquifers, Tidal Fluctuations pp. 85-91. Denver, Colorado: US Geological Survey.
Field, C.R., T.S. Bayard, C. Gjerdrum, J.M. Hill, S. Meiman, and C. Elphick. 2016. High resolution tide projections reveal extinction threshold in response to sea-level rise. Global Change Biology 23:2058-2070.
Flick, R.E., J.F. Murray, and L.C. Ewing. 2003. Trends in United States tidal datum statistics and tidal range. Journal of waterway, Port, Coastal, and Ocean Engineering 129: 155-164.
Hartig, E.K., V. Gornitz, A.S. Kolker, F. Mushacke, and D. Fallen. 2002. Anthropogenic and climate-change impacts on salt marshes of Jamaica Bay, New York City. Wetlands 22: 71-89.
Harvey, J.W. and W.E. Odum. 1990. The influence of tidal marshes on upland groundwater discharge to estuaries. Biogeochemistry 10: 217-236.
Hunter, E.A., N.P. Nibbelink, C.R. Alexamnder, K. Barrett, L.F. Mengak, R.K. Guy, C.T. Moore, and R.J. Cooper. 2015. Coastal vertebrate exposure to predicted habitat changes due to sea level rise. Environmental Management 56: 1528-1537.
Kearney, M.S., A.S. Rogers, J.R. Townshed, E. Rizzo, A. Stutzer, D. Stevenson, and K. Sundberg. 2002. Landsat imagery shows decline of coastal marshes in Chesapeake and Delaware Bays. EOS 83: 173-178.
II-31
Kirwan, M.L. and G. Guntenspergen. 2010. Influence of tidal range on the stability of coastal marshland: Tidal range and marsh stability. Journal of Geophysical Research 115: F02009.
Mariotti, G. 2016. Revisiting salt marsh resilience to sea level rise: Are ponds responsible for permanent land loss? JGR Earth Surface 121: 1391- 1407.
Mawdsley, R.J., I.D. Haigh, and N.C. Wells. 2014. Global changes in mean tidal high water, low water and range. Journal of Coastal Research 70: 343-348.
Moffett, K.B., S.M. Gorelick, R.G. McLaren, and E.A. Sudicky. 2012. Salt marsh ecohydrological zonation due to heterogenous vegetation-groundwater-surface water interactions. Water Resources Research 48: W02516.
Montalto, F.A., T.S. Steenhuis, and J.Y. Parlange. 2006. The hydrology of Piermont Marsh, a reference for tidal marsh restoration in the Hudson River estuary, New York. Journal of Hydrology 316: 108-128.
Morris, J.T. 1995. The mass balance of salt and water in intertidal sediments: Results from North Inlet, South Carolina. Estuaries and Coasts 18: 556-567.
National Estuarine Research Reserve System [NERRS]. 2019. Centralized Data Management Office. https://cdmo.baruch.sc.edu/dges/ accessed 12 November 2019.
National Oceanic and Atmospheric Administration [NOAA]. 2019. Vertical Datum Transformation: Integrating America’s Elevation Data. National Ocean Service. https://vdatum.noaa.gov/welcome.html.
National Oceanic and Atmospheric Administration [NOAA]. 2020. 8518750 The Battery, NY, Tides/Water Levels. Center for Operational Oceanographic Products and Services. https://tidesandcurrents.noaa.gov/waterlevels.html?id=8518750&type=Tide+Data &name=The%20Battery&state=NY
New York State Department of Environmental Conservation [NYSDEC]. 1985. Piermont Pier Critical Environmental Area.
New York State Department of Environmental Conservation [NYSDEC]. 2017. Draft Piermont Marsh Reserve Management Plan. Albany, NY.
New York State Department of State [NYSDOS]. 2012. Coastal Fish and Wildlife Rating Form.
II-32 Ogle, D.H., P. Wheeler, and A. Dinno. 2020. FSA: Fisheries Stock Analysis. R package version 0.8.30, https://github.com/droglenc/FSA
Osgood, D.T. 2000. Subsurface hydrology and nutrient export from barrier island marshes at different tidal ranges. Wetlands Ecology and Management 8:133-146.
Partnership for the Delaware Estuary [PDE]. 2012. Technical report for the Delaware Estuary PDE Report 12-01. 255 pages.
Pickering, M.D., N.C. Wells, K.J. Horsburgh, and J.A.M. Green. 2012. The impact of future sea‐level rise on the European Shelf tides, Continental Shelf Research 35: 1–15.
Pickering, M.D., K.J. Horsburgh, J.R. Blundell, J.J.-M. Hirschi, R.J. Nicholls, M. Verlaan, and N.C. Wells. 2017. The impact of future sea-level rise on the global tides. Continental Shelf Research 142: 50-68.
R Core Team. 2019. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R- project.org/.
Raposa. K.B, K. Wasson, E. Smith, J.A. Crooks, P. Delgado, S.H. Fernald, M.C. Ferner, A. Helms, L.A. Hice, J.W. Mora, B. Puckett, D. Sanger, S. Shull, L. Spurrier, R. Stevens, and S. Lerberg. 2016. Assessing tidal marsh resilience to sea-level rise at broad geographic scales with multi-metric indices. Biological Conservation 204: 263-275.
Rooth, J.E. and J.C. Stevenson. 2000. Sediment deposition pattens in Phragmites australis communities: Implications for coastal area threatened by rising sea- level. Wetlands Ecology and Management 8:173-183.
Rosencranz, J.A., K.M. Thorne, K.J. Buffington, J.Y. Takekawa, R.F. Hechinger, T.E. Stewart, R.F. Ambrose, G.M. MacDonald, M.A. Holmgren, J.A. Crooks, R.T. Patton, and K.D. Lafferty. 2018. Sea-level rise, habitat loss, and potential extirpation of a salt marsh specialist bird in urbanized landscapes. Ecology and Evolution 8: 8115-8125.
Rotzell, K. and C.H. Fletcher. 2013. Assessment of groundwater inundation as a consequence of sea level rise. Nature Climate Change 3: 477-481.
Rush, S.A., E.C. Soehren, M.S. Woodrey, C.L. Graydon, and R.J. Cooper. 2009. Occupancy of select marsh birds within Northern Gulf of Mexico tidal marsh: Current estimates and projected change. Wetlands 29: 798-808.
Sallenger, A.H. Jr., K.S. Doran, and P.A. Howd. 2012. Hotspot of accelerated sea-level rise on the Atlantic coast of North America. Nature Climatic Change 2: 884-888.
II-33
Saaltink, R.M., M. Barciela-Rial, T. van Kessel, S.C. Dekker, H.J. de Boer, C. Chassagne, J. Griffioen, M.J. Wassen, and J.C. Winterwerp. 2019. Drainage of soft cohesive sediment with and without Phragmites australis as an ecological engineer. Hydrology and Earth System Sciences Discussions. https://doi.org/10.5194/hess-2019-194
Sandi, S., J. Rodriguez, N. Saintilan, G. Riccardi, and P. Saco. 2018. Rising tides, rising gates: the complex ecogeomorphic response of coastal wetlands to sea-level rise and human interventions. Advances in Water Resources 114: 135–148.
Schepers, L., M. Kirwan, G. Guntenspergen, and S. Temmerman. 2017. Spatio-temporal development of vegetation die-off in a submerging coastal marsh. Limnology and Oceanography 62: 137-150.
Silliman, B.R. and M.D. Bertness. 2004. Shoreline development drives invasion of Phragmites australis and the loss of plant diversity on New England marshes. Conservation Biology 18: 1424-1434.
Smith, S.M. 2009. Multi-decadal changes in salt marshes of Cape Code, MA: photographic analyses of vegetation loss, species shifts, and geomorphic change. Northeastern Naturalist 16: 183-208.
Smith, S.M., K.C. Medeiros, and M.C. Tyrrell. 2012. Hydrology, herbivory, and the decline of Spartina patens (Aiton) Muhl. in Outer Cape Cod salt marshes. Journal of Coastal Research 28: 602-612.
Sweet, W.V. and C. Zervas. 2011. Cool-season sea level anomalies and storm surges along the U.S. East Coast: climatology and comparison with the 2009/10 El Niño. Monthly Weather Review 139: 2290-2299.
Talke, S.A., A.C. Kemp, and J. Woodruff. 2018. Relative sea level, tides, and extreme water levels in Boston Harbor from 1825 to 2018. JGR Oceans 123: 3895-3914.
Thorne, K.M., K.J.B. Buffington, D.L. Elliot-Fisk, and J.Y. Takekawa. 2015. Tidal marsh susceptibility to sea level rise: Importance of local-scale models. Journal of Fish and Wildlife Management 6: 290-304.
Torio, D.D. and G.L. Chmura. 2015. Impacts of sea level rise on marsh as fish habitat. Estuaries and Coasts 38: 1288-1303.
Wang, J. and E. Myers. 2016. Tidal datum changes induced by morphological changes of North Carolina coastal inlets. Journal of Marine Science and Engineering 4: 79.
II-34 Wang, X.X., R. Li, H.J. Taghadomi, S. Pedram, and X. Zhao. 2017. Effects of sea level rise on hydrology: case study in a typical mid-Atlantic coastal watershed. Journal of Water and Climate Change 8: 730-754.
Warren, R.S. and W.A. Niering. 1993. Vegetation change on a northeast tidal marsh: interaction of sea-level rise and marsh accretion. Ecology 74: 96-103.
Watson, E.B., A. Oczkowski, C. Wigand, A. Hanson, E. Davey, S. Crosby, R. Johnson, and H. Andrews. 2014. Nutrient enrichment and precipitation changes do not enhance resiliency of salt marshes to sea level rise in the Northeastern US. Climatic Change 125: 501–509.
Watson, E.B., C. Wigand, E.W. Davey, H.M. Andrews, J. Bishop J., and K.B. Raposa. 2017. Wetland loss patterns and inundation-productivity relationships prognosticate widespread marsh loss for southern New England. Estuaries and Coasts 40: 662-681.
Williams, P.B. and M.K. Orr. 2002. Physical evolution of restored breached levee salt marshes in the San Francisco Bay Estuary. Restoration Ecology 10: 527-542.
Williams, P.B., M.K. Orr, and N.J. Garrity. 2002. Hydraulic geometry: A geomorphic design tool for tidal marsh channel evolution in wetland restoration projects. Restoration Ecology 10: 577-590.
Wilson. A.M. and L.R. Gardner. 2006. Tidally driven groundwater flow and solute exchange in a marsh: Numerical simulations. Water Resources Research 42: W01405.
Wilson. A.M., W.S. Moore, S.B. Joye, J.L. Anderson, and C.A. Schutte. 2011. Storm- driven groundwater flow in a salt marsh. Water Resources Research 47: W02535.
Wilson. A.M., T. Evans, W. Moore, C.A. Schutte, S.B. Joye, A.H. Hughes, and J.L Anderson. 2015. Groundwater controls ecological zonation of salt marsh macrophytes. Ecology 96: 840-849.
Windham, L. and R.G. Lathrop, Jr. 1999. Effects of Phragmites australis (common reed) invasion on aboveground biomass and soil properties in brackish tidal marsh of the Mullica River, New Jersey. Estuaries 22: 927-935.
Windham, L. M., J.S. Weis, and F. Weis. 2001. Patterns and processes of mercury release from leaves of two dominant salt marsh macrophytes, Phragmites australis and Spartina alterniflora. Estuaries 24: 787-796.
Winogrond, H.G. and E. Kiviat. 1997. Invasion of Phragmites australis in the tidal marshes of the Hudson River. Section VI: 29 pp. In W.C. Neider and J.R.
II-35 Waldman (eds.) Final Report of the Tibor T. Polgar Fellowship Program, 1996, Hudson River Foundation, NY.
Xin, P., J. Kong, L. Li, and D.A. Barry. 2013. Modelling of groundwater-vegetation interactions in a tidal marsh. Advances in Water Resources 57: 52-68.
Yozzo, D.J. and F. Ottman. 2003. New distribution records for the spotfin killifish, Fundulus luciae (Baird), in the lower Hudson River estuary and adjacent waters. Northeastern Naturalist 10: 399-408.
II-36 TIDAL MARSH DEVELOPMENT AND SEDIMENT DYNAMICS AT VANDERBURGH COVE, RHINEBECK NY
A Final Report of the Tibor T. Polgar Fellowship Program
Waverly L. Lau
Polgar Fellow
Department of Geosciences University of Massachusetts Amherst Amherst, MA 01002
Project Advisor:
Brian Yellen Department of Geosciences University of Massachusetts Amherst Amherst, MA 01002
Lau, W.L. and B. Yellen. 2021. Tidal Marsh Development and Sediment Dynamics at Vanderburgh Cove, Rhinebeck, NY. Section III: 1-28 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
III-1
ABSTRACT
Many unknowns still exist regarding the developmental histories and sediment
characteristics of Hudson Marshes. In addition, steep bedrock valley walls along the
Hudson limit locations for tidal marsh development. In order to best direct tidal wetland
creation and restoration work, it is instructive to understand how existing tidal wetlands
have formed and developed along the Hudson. Vanderburgh Cove in Rhinebeck, NY, is
an ideal study site as it contains a rapidly expanding tidal marsh. Vanderburgh Cove has
about 40,000 m2 of emergent marsh, mostly developed at the mouths of two tributaries
that enter the cove, Landsman Kill and Fallsburgh Creek. This study used historical aerial
images, sediment core analysis from x-ray fluoresce data, Loss of Ignition (LOI),
grainsize, and 137Cs age constraints in order to understand the differences in the marsh at
the mouth of Landsman Kill and the mudflat areas of the cove and how the marsh
developed. Aerial photos show rapid marsh expansion between 1978 and 1995. The
sediment cores and sediment traps indicate high accretion rates in the cove and suggest
that there is ample sediment to support future marsh expansion. From this research study,
it is hypothesized that emergent marsh likely establishes when substrate accretes to an elevation of between -0.05 to 0.10 m above sea level (mASL).
III-2
TABLE OF CONTENTS
Abstract ...... III-2
Table of Contents ...... III-3
List of Figures and Tables...... III-4
Introduction ...... III-5
Past Work ...... III-6
Site Description ...... III-7
Methods...... III-10
Field Sampling Methods ...... III-10
Sediment and Sediment Core Analysis ...... III-12
Historical Aerial Imagery and Topographic Analysis ...... III-15
Results ...... III-15
Aerial Imagery ...... III-15
Marsh Cores ...... III-17
Sediment Traps ...... III-18
Mudflat Cores ...... III-19
Discussion ...... III-21
Aerial Imagery ...... III-21
Marsh and Mudflat ...... III-22
Conclusion ...... III-24
Acknowledgements ...... III-25
References ...... III-26
III-3
LIST OF FIGURES AND TABLES
Figure 1 – Vanderburgh Cove watershed map showing study site location in
Hudson River with New York State reference map modified from
Hudson River National Estuarine Research Reserve...... III-8
Figure 2 – Vegetation map of Vanderburgh Cove showing present major
vegetation...... III-9
Figure 3 – Vanderburgh Cove site map ...... III-11
Figure 4 – Historical aerial images of marsh area of Vanderburgh Cove at delta
of Landsman Kill in years 1960, 1978, 1995, and 2016 ...... III-16
Figure 5 – Historical image: marsh comparison 1978 to 2016 ...... III-17
Figure 6 – Marsh core data for VBM1, VBM2, and VBM3...... III-18
Figure 7 – Mudflat core data for VBC4, VBC5, and VBC6...... III-20
Figure 8 – Vanderburgh Cove elevation analysis ...... III-23
Table 1 – Compiled core data for elevation analysis ...... III-21
III-4 INTRODUCTION
Tidal wetlands offer ecosystem services and products which are benefits for the environment (Zedler and Kercher 2005; Schuerch et al. 2018). These ecosystems are beneficial resources to people and as well as to other organisms, providing habitat for various species. Tidal wetlands, such as marshes, can also provide protection along the coastline from natural hazards and are often natural solutions for coastal protection plans
(Shepard et al. 2011); however, rapid sea level rise from anthropogenic climate change threatens tidal marsh sustainability. Many marshes are threatened by limited sediment supply and therefore may not aggrade rapidly enough to keep pace with sea level rise
(Schuerch et al. 2018). In addition, many marshes have been eliminated due to land
reclamation, a process by which shallow marine areas are diked and filled with sediment to create land for development (Squires 1992). With tidal wetlands disappearing, it is
vital that restoration efforts are made to bring back these ecosystems.
There are currently plans in place to restore or create 3,700 acres of tidal marsh in the
Hudson River Estuary (HRECRP 2016; Partners Restoring the Hudson 2018), with 700
acres planned for the tidal Hudson River above Manhattan. Many unknowns still exist
regarding the developmental histories and sediment characteristics of Hudson River
marshes. In addition, steep bedrock valley walls along the Hudson limit locations for tidal
marsh development. Many of the tidal marshes along the Hudson are behind railroad
trestles or dredged material islands. Recent research has shown that many of the marshes
within these artificially sheltered settings developed in response to the human
modification of these shorelines (Yellen et al. 2020; Yellen et al. 2018); however, some
coves have not developed fully into tidal marshes, such as Vanderburgh Cove and Tivoli
III-5
South Bay, while others have developed into tidal marshes, such as Tivoli North Bay. It
is important to understand the differences between sites where marshes have formed and those where marsh has not formed. Understanding this difference would better inform restoration plans and efforts.
Vanderburgh Cove has about 40,000 m2 of emergent marsh, mostly developed at the
mouths of two tributaries that enter the cove, Landsman Kill and Fallsburgh Creek. Most
of the cove remains a shallow intertidal mudflat. Thus, Vanderburg Cove shares some characteristics with Tivoli North Bay, a marsh, as well as Tivoli South Bay, which is
predominantly tidal mudflat. It is hypothesized that the emergent marsh at Vanderburgh
Cove formed due to sheltered cove created by the railroad berm. This study investigates
this site and its deposition rates through time to determine if the construction of the
railroad and initial depths or elevation relative to sea level play a role in marsh
development. In addition, this study examines age of the current marsh in the cove, how
the tributaries and the construction of the railroad trestle have affected deposition rates,
and how that relates to its current state as a mudflat.
Past Work
Research conducted by Dr. Brian Yellen and Dr. Jon Woodruff as part of a NOAA-
funded project, Dams and Sediment in the Hudson (DaSH), examined the effects of dam
removal and sediment release on tidal wetland sustainability in the Hudson River Estuary
(HRNERR 2017). More specifically, the research focused on whether the removal of dams on tributaries to the Hudson could provide needed sediment to marshes. In Tivoli
North Bay, sediment cores were collected and analyzed to understand sediment
III-6 accumulation rates and when the marsh in Tivoli North Bay formed. The preliminary
results of the project were that there were rapid sediment accumulation rates and the
marsh in Tivoli North formed within the last 100-150 years, likely as result of railroad construction. Tivoli South Bay, dominated by mudflat rather than emergent marsh, has slower accumulation rates than Tivoli North, but is still accumulating sediment quickly
(Yellen et al. 2020).
These human-made coves present a unique opportunity to study factors that allow for the creation of emergent tidal marshes. This study on Vanderburgh Cove will also act as an additional study site for the DaSH project and test current hypotheses about Hudson
River marsh formation caused by human infrastructure at a location that shares attributes in common with both of the Tivoli Bays.
Site Description
Vanderburgh Cove is a tidal freshwater area of approximately 0.4 km2. It is located
on the east side of the Hudson River at river kilometer 140, (N41.878, W73.928) (Figure
1). This cove was cut off from the main channel of the Hudson River in 1851 when the
New York Central Railroad was completed (Benoit et al. 1999; Stevens 1926), with two
culverts maintaining tidal connection to the river. The tidal range in this area of the
Hudson River is approximately 1.1m (Ralston and Geyer 2017). There is an emergent
marsh that comprises about 14% of Vanderburgh Cove, in which the predominant
vegetation is Typha angustifolia (narrow-leaf cattail). The remaining area contains 28%
lower intertidal vegetation, which is dominated by Nuphar advena (spatterdock), and the rest of the cove consisting of Trapa natans (water chestnut) (Figure 2).
III-7
Legend Study site ---- Railroad
Figure 1: Vanderburgh Cove watershed map showing study site location in Hudson River with New York State reference map modified from Hudson River National Estuarine Research Reserve. Pink star is Vanderburgh cove. Highlighted in colors from red to blue (Blue represents average elevations at 0 mASL, green is 50 mASL, yellow is 100 mASL, orange is 150 mASL, and red is 200 mASL) and outlined in black is the watershed area with elevations (meters above sea level).
Two tributaries drain directly into Vanderburgh Cove, Landsman Kill and Fallsburgh
Creek. The drainage area of Landsman Kill is 60.8 km2, and the drainage area of
Fallsburg Creek is 10.3 km2, all in an area of moderate to steep slopes (USGS 2016a;
NYSDOS 2012) (Figure 1). Northern hardwood forest covers 71% of the watershed, with
the remainder consisting of agriculture and low-density residential land uses (USGS
2016a). More than 60% of the watersheds for Fallsburg Creek and Landsman Kill has
well or excessively well drained soil; however, the lower portions of the watersheds reach
into glacial lacustrine deposits, which may supply fine-grained material to the marsh
(Soil Survey Staff 2019). Both tributaries have impoundments, which may trap sediment
III-8
and prevent delivery to the cove. This impact is likely small as regional studies have shown that most dams do not trap appreciable amounts of sediment (Ralston et al. 2020).
Figure 2: Vegetation map of Vanderburgh Cove showing present major vegetation. The lower intertidal zone consists mostly of spatterdock (Nuphar advena). Land cover data from Cornell Institute for Resource Information Sciences (IRIS) (2011).
Vanderburgh Cove is an ideal site to evaluate the mechanisms and rates of tidal marsh formation because it combines aspects of the well-studied Tivoli North Bay emergent marsh and Tivoli South Bay mudflat (Benoit et al. 1999; Yellen et al. 2020; Sritrairat et al. 2012). Namely, its combination of tidal flat and emergent marsh makes it a useful case study in how tidal marsh establishes and expands into shallow areas.
III-9 METHODS
3.1 Field Sampling Methods
To understand Vanderburgh Cove and its potential for marsh formation, sediment cores were collected from the existing marsh area and from the open water/mudflat area of the cove during May and June of 2019 (Figure 3). In total, six sediment cores were collected, with three in the emergent marsh at the mouth of Landsman Kill and three from the tidal mudflat. Two successive 1 m drives were collected with a 1 m long, 6.3 cm diameter gouge corer at each site, unless refusal was encountered. Thus, collected cores range from 1.6 m to 2 m in length. Gouge cores provide an uncompacted sample whose length scale can easily be converted to true depths, without the need to correct for compaction during coring. This makes it ideal for evaluating tidal depositional
environments that are sensitive to sea level changes and was adequate for the marsh and
subaqueous mudflat environment in this area.
III-10
Figure 3: Vanderburgh Cove site map. (a)Map of all coring sites: red dots show locations of cores in the mudflat (subaqueous), yellow dots show locations of cores in the marsh, and yellow triangle shows where sediment traps were placed as well as a core location. (b) Zoomed in view of the marsh site with same symbology.
A slide hammer was used at VBC6 core site to drive the gouge corer until refusal.
Some overlap in depth was taken for some core sites (VBM2, VBC6), while in cores such as VBC5 there is a 10 cm gap due to poor recovery. Gouge cores were transferred onto 1 m long, halved PVC pipes.
In addition to sediment cores, three sediment traps were put in place at the edge of the emergent marsh near VBM1 core site (Figure 3). These sediment traps were placed in late May and the sediment that was trapped inside was collected in late June. The trap consisted of a halved 21 L bucket so that the walls of the bucket were approximately 20
III-11
cm in height and 27 cm in diameter. The three traps were placed nearly flush with the
marsh surface with 1-2 cm of bucket wall extending above the marsh surface, but at
slightly different elevations as the marsh surface varies. The sediment was collected by
siphoning out the clear water on top and pouring the sediment-water mixture into 1 L
sample bottles. During this time, a bulk density sample of 13.5 cm by 13.5 cm by 10 cm
was also measured. This was done via a surface box core in which a three-sided metal
box was inserted 20 cm into the marsh. The box was dug out of the marsh then carefully
trimmed with a knife to create a rectangular prism of 13.5 cm width and length, and 10
cm height, representing the sediment at depth of 10 to 20 cm below grade. This was
designed to be able to obtain a fixed volume of sediment including the vegetation and
other organic material and to avoid compaction.
3.2 Sediment and Sediment Core Analysis
Once cores were collected, they were brought back to the University of
Massachusetts, Amherst within the same day. All cores were stored in a 4°C refrigerated storeroom. Gouge cores were split lengthwise with a hack saw blade, and stored in air- tight conditions at 4°C. One half of each of the cores was preserved as an archive, while the other half was analyzed. Once cores were split and opened, they were then visually described to note obvious grain size variation, organic content, and color. This served as initial qualitative analysis for later referral.
All cores were then scanned using an ITRAX x-ray fluorescence (XRF) core scanner to obtain geochemical data in relation to depths. This used a molybdenum tube and was set to run at 30kV and 55 mA with a 10-second exposure and 0.3 mm sample
III-12 resolution (Croudace et al. 2006). The ITRAX scanner images the core, produces x-
radiographs, and measures relative abundances of a suite of elements. Heavy metals such
as lead (Pb) and zinc (Zn) were of particular interest for this study as the onset for these
pollutants can constrain sediment age (Benoit et al. 1999).
Following non-invasive core scanning, 1 cm thickness subsamples were taken
from cores every 10 cm, according to true depths, starting 5 cm from the top, and above
and below visible lithologic transitions. These subsamples were weighed immediately
after sampling and after drying in a 100°C drying oven overnight in previously weighed
crucibles to provide an estimate of sediment porosity, assuming the initial saturation of
sediment pore spaces. Subsamples were burned in a 550°C muffle furnace for four hours and weighed afterwards to calculate loss on ignition, a proxy for percent organics (Dean
1974; Heiri et al. 2001).
Burned samples were gently disaggregated with a mortar and pestle and
subsampled to be examined through a Beckman Coulter LS I3 320 laser diffraction
particle size analyzer with a range of 0.4 µm to 2000 µm. Samples were sonicated for 15 seconds and had a run time of 60 seconds. Grain size distributions are summarized with the median or d50 grain size, which represents the particle diameter that 50% of the
sample by mass is smaller than.
Down-core profiles of radionuclides and heavy metals can be used to place age
constraints on sediment cores. The onset of heavy metals such as lead and zinc can
indicate ages at the beginning of industrial activity, which in this area would have been
railroad construction in 1851 (e.g. Benoit et al. 1999; Kemp et al. 2017). Core VBM1
was sampled for gamma spectroscopy to measure the depth profiles of radionuclides
III-13
137Cs and 210Pb. 137Cs can help determine the ages of modern sediment as the onset of
137Cs correlates to 1954, the year in which atmospheric nuclear testing began, and the
peak of 137Cs correlates to 1963, just prior to the Nuclear Test Ban Treaty (Pennington et al. 1973; Florea et al. 2011; Brandon et al. 2014). 210Pb accumulates in sediments at a
constant rate and thus can be used to understand deposition rates (Crozaz et al. 1964;
Florea et al. 2011). The subsamples of VBM1 were dried in a 100°C drying oven
overnight and later ground with a mortar and pestle. This resulting powder was weighed
and analyzed with a Canberra GL2020R low-energy germanium detector for 137Cs and
210Pb.
The dry bulk density sample collected with the box core was processed using
methods similar to those used for LOI samples. The whole sample was split into two metal trays, weighed to obtain wet mass, and then dried at 100°C to obtain dry mass.
Large vegetative matter was removed and weighed separately, and the remaining sediment was homogenized. A subsample of this remaining mixture was weighed in a crucible. This was later placed in a muffle furnace at 550°C to burn off organics. After four hours in the muffle furnace, the sample was weighed again, with the remaining clastic content used to calculate percent clastics in the modern marsh material at
Vanderburgh Cove.
Sediment trap samples were each poured into metal tins and some water and a spatula was used to rinse out remaining sediment in the bottles. These were then dried at
100°Cfor 48 hours. Once all the water was removed from the samples, they were weighed
and placed into a muffle furnace at 550°C muffle for four hours to remove organics. The
III-14 samples were weighed again in order to obtain mass of clastics delivered to a unit area of marsh during May-June.
3.3 Historical Aerial Imagery and Topographic analysis
Historical aerial images were obtained from the U.S. Geological Survey Earth
Explorer (USGS 2019). Images from April 2001, March 1995, 1978, May 1960, and
1956 were collected and analyzed. All photos were georeferenced to current day satellite imagery using QGIS’s georeferencer and Google’s publicly available satellite base map.
Marsh boundary lines were traced and superimposed onto one another to evaluate changes in the marsh over time. In addition to aerial images, high resolution LiDAR topographic data from 2013 was evaluated, which has a horizontal resolution of 1 m, and a vertical accuracy of 6.5 cm (OCM Partners 2020). These were used in conjunction with the marsh cores to further investigate the depositional environment of the past and elevation thresholds for marsh development.
RESULTS
4.1 Aerial Imagery
Aerial images of the marsh area at the delta of Landsman Kill were taken from various years and compared to one another to understand the progression of the marsh. In
1960, there was no evidence of marsh (Figure 4). In 1978, there was a land mass forming at the mouth of the Landsman Kill with an area of approximately 7,500 m2 (Figure 4). In
1995, there was a larger area of land (approximately 30,000 m2) and in 2016 the edges of
III-15 this land mass may have shifted north a few meters but otherwise remained relatively unchanged (Figure 4).
1960 1978
100m 100m
1995 2016
100m 100m
Figure 4: Historical aerial images of marsh area of Vanderburgh Cove at delta of Landsman Kill in years 1960, 1978, 1995, and 2016. SEE LOCATION IN FIG 3B, INSET BOX. Image Source: Black and white photos (USGS), GoogleEarth (NASA and USGS).
III-16
In Figure 5a, the present-day marsh perimeter is outlined and displayed on the aerial photo from 1978. It is apparent that a large amount of land had accreted in those 38 years. The marsh land at the mouth of Landsman Kill had grown roughly three and a half times in area since 1978.
(a) 1978 (b) 2016
Figure 5: Historical image: marsh comparison 1978 to 2016. Aerial images of marsh area in (a) 1978 and (b) 2016; the red border represents the marsh perimeter in 2016. Image Source: Black and white photo (USGS), GoogleEarth (NASA and USGS).
4.2 Marsh Cores
The depths of sediment cores taken varies at each site corresponding to the refusal depth marking the bottom of the marsh sedimentary unit. VBM1 went to 185 cm in depth, VBM2 to 130 cm, and VBM3 to 150 cm. The marsh cores did not have clear onsets of heavy metals according to the XRF data of Pb, Zn, and K (Figure 6). The onset of 137Cs in VBM1 was at 95 cm and the peak was at 90 cm (Table 1). Considering that
the onset of 137Cs corresponds to the year 1954, and the peak to 1963, the deposition rate of the marsh since 1963 was calculated to be approximately 1.6 cm/year. Immediately above the refusal depths for each core, LOI data show low background levels of organics III-17 of approximately 6%. In VBM1, LOI began to rise upwards in the core at 75 cm, with the highest level in the growing marsh unit within 10 cm of the land surface. VBM2 and
VBM3 show generally higher organic content, with both cores also beginning to increase in LOI at 75 cm, with anomalous peaks at 45 cm in VBM2 and 25 cm in VBM3 (Figure
6). These anomalously high data points correspond to large fragments of vegetation in the cores.
Figure 6: Marsh core data for VBM1, VBM2, and VBM3. For each core, data for Zinc (representing heavy metal pollution), LOI (representing percent organics), and median grainsize are depicted. Depth scales vary for each core.
4.3 Sediment Trap
The three sediment traps provide a snapshot of modern-day deposition rates and
clastic sediment delivery. The total clastic mass deposited in the 572.5 cm2 sediment
traps averaged 97.3 g (SD =1.15g). This clastic mass was converted to sediment depth
using a dry bulk density of 0.714 g/cm3, which corresponded to the observed bulk density
from the box core sample. Using this calculation, the rate of sediment delivery at the edge
III-18 of the marsh and to the vicinity of VBM1 was calculated to be about 0.24 cm/mo, or 2.9 cm/yr, or 1.8 times faster than the deposition rate inferred from the 137Cs peak.
4.4 Mudflat Cores
The sediment cores taken from shallow intertidal mudflat sites were 200 cm at
VBC4 and VBC5. At VBC6, cores were taken until refusal, which was encountered at
164 cm below the substrate due to cohesive clay deposits. At VBC5, probing revealed a
transition to angular sand and gravel at 270 cm depth, grading to sand at 305 cm depth
and refusal at 340 cm in dense, light grey cohesive clay. The onset of heavy metals was
approximately 120 cm in VBC4, 115 cm in VBC5, and 106 cm in VBC6 (Table 2).
Above the metals onset in each core, grain size decreased from sandy silt to silty clay.
Below the metals onset, grain size in each core graded downward from sandy silt to clay
at the base, consistent with field observations of cohesive clay at refusal depth for cores
VBC 4-6. (Table 2). This transition around 110 cm was less clear in LOI data in VBC4,
with an initial drop in organic percentage at that depth and then a gradual rise in organics
towards the surface (Figure 7). From 110 to 175 cm, there was a sandy silt, below which
grey clay was found, which was interpreted as basal material. In VBC5, a similar
decrease in grainsize can be seen at 110 cm (Figure 7). At VBC6, a clear signal of
increasing zinc was observed at 80 cm. Similar to the other mudflat cores, this increase in
heavy metals was accompanied by a decrease in grain size (Figure 7).
III-19
Figure 7: Mudflat core data for VBC4, VBC5, and VBC6. For each core, data for Zinc (representing heavy metal pollution), LOI (representing percent organics), and median grainsize. Depth scales vary for each core.
In trying to understand the conditions of marsh formation, elevation was looked at as a factor for development. Table 1 shows current elevations and depth markers of the core sites along with some age constraints to understand the timeline of Vanderburgh
Cove. These data were compiled to gain insight on past elevations of the marsh and mudflat before the marsh developed. The lower intertidal mudflat within Vanderburgh
Cove ranges in elevation -0.2 to 0.1 mASL.
III-20
Table 1: Compiled core data for elevation analysis. Depths of elevation, time constraints, and marsh origin for each core. Marsh bottom was analyzed through LOI data and core photos, the bottom being where larger/visible organics started. The Land elevation is from 2013 LiDAR data (USGS LiDAR). The elevation threshold is the calculated elevation at which the marsh began to form.
DEPTH BELOW SUBSTRATE Year VBM1 VBM2 VBM3 VBC4 VBC5 VBC6
Cs-137 peak 1963 90 - - - - - Cs-137 onset 1954 95 - - - - - Heavy metal onset 1850 - - - 120 115 80 Depth to refusal - 166 130 - - 340 164 Marsh bottom - -0.63 -0.48 -0.63 - - - Land Elevation (mASL) - 0.65 0.55 0.65 0.05 -0.1 -0.13 Elev. error (m) - 0.2 0.2 0.2 0.065 0.065 0.065 Elev. threshold - 0.02 0.07 0.02 - - -
DISCUSSION
5.1 Aerial Imagery
The historical aerial photos showed a rapid marsh expansion from 1978 to 1995, a
17-year period (Figure 5). From 1995 to today, a 21-year period, the marsh had remained relatively the same, showing the marsh has stabilized at its current size (Figure 4).
Further research is needed to confirm the growth and equilibrium of the marsh, as it is
difficult to accurately determine the state of the marsh, especially in a historical context.
While the historical images were acquired with dates and years, time stamps and tides
levels were unknown. Thus, some images may have been at low tide or at high tide which
could reveal more areas around the marsh or conceal inundated areas of marsh.
III-21 5.2 Marsh and Mudflat
The marsh portion of Vanderburgh Cove is rapidly accreting; however, the
calculated sediment deposition rates must be verified through age constraints of other
cores and more in-depth analysis of VBM1. Through gamma spectroscopy, data of Pb-
210 and Pb-214 for VBM1 was collected and analyzed. This analysis did not show
verifiably coherent trends and thus was not included in the age-constraint analysis.
Additionally, the sediment traps should be kept at the site for longer periods of times and
analyzed further as a one-month time period is not representative of yearly deposition
rates, especially when considering seasonality. The accuracy of this sediment trapping
method needs to be examined further as it is a relatively new method. The discrepancy
between core-derived sediment accumulation rates and the sediment trap results may be
due to preferentially trapping sediment by grain size in the buckets, or due to seasonal
differences in sediment delivery to the marsh; however, the order of magnitude
agreement between the sediment trap-derived and sediment core 137Cs-derived deposition
rates suggests that this marsh continues to trap clastic sediment at rapid rates.
According to the results found in these cores and LiDAR elevation data, the
extant marsh developed when sediment accreted to elevations of approximately -0.05m to
+0.1 mASL (Figure 8). This inference is based on the elevation of the current marsh, and the depth of visible marsh rootlets in cores, noted as the elevation threshold (Table 1). It is suspected that at these elevations, marsh development begins and thus mudflat areas at this elevation will likely become a marsh in the near future. This area of elevation range -
0.05 m and +0.1 mASL also contains a dominant type of vegetation, Nuphar advena or spatterdock (Figure 2). A long-term study could be done to further verify this and the
III-22 relationship of the vegetation, as well as more in-depth analysis of the sediment cores and sediment record of Vanderburgh Cove.
(a) (b)
(c)
Elevation (m) Elevation
Figure 8a,b,c: Vanderburgh Cove elevation analysis (USGS 2016b). (a) Light green highlighted areas are of elevation of -0.05m to +0.1m. (b) Current elevations with color gradient scale from blue to red for elevations -0.1m to +0.8m respectively. The transect for figure 8c is highlighted in a dashed-red line. (c) Elevation profile in red along a transect of all marsh and mudflat cores. Core sites are labeled, and purple rectangles represent cores collected and depict depths collected. Green boxes represent depths in cores at which marsh begins, noted in Table 1 as marsh bottom. Dashed red line represents the hypothesized elevation for marsh onset of 0.1 mASL.
III-23
CONCLUSION
Vanderburgh Cove offers unique insight into marsh development as a tidal freshwater marsh and mudflat. Aerial photos show rapid marsh development after 1978 but the marsh appears to have stabilized since 1995. Sediment cores and sediment traps indicate high accretion rates in the cove, largely through the accumulation of clastic sediments. From this research study, it is hypothesized that initial elevations play a role in determining which areas become a marsh and which do not. Vegetation is also a factor, but further in-depth analysis is needed to understand this relationship. More age constraints on sediment cores would be beneficial in understanding past elevations and a long-term study would be beneficial in understanding the continued growth of this marsh.
It is still unclear why the marsh stopped horizontal growth after 1995. Further research is needed to verify and expand these findings.
III-24 ACKNOWLEDGEMENTS
Thanks to the Hudson River Foundation and the Tibor T. Polgar Fellowship program for funding this project. Thank you to William A. Lee and other donors who supported this research through the William Lee Science Impact Program. The authors would like to thank Kyra Simmons who allowed our field team to enter their property to reach the field site. Special thanks to Julia Casey, Francis Griswold, Mark Butler, Hannah
Baranes and Caroline Ladlow for assistance in the field and in the lab. Thank you to all who have helped make this project happen and my family for supporting me through this.
I would like to give my sincerest gratitude to Dr. Tracie M. Gibson, the Director of the
College of Natural Sciences Office of Student Success and Diversity and Program
Director of the William Lee Science Impact Program among other programs. Dr. Gibson
passed away late October 2019 and was a huge impact in my career and in my life. She
inspired me to continuing pursuing my science career and furthering my education to do
so. She helped me and other students have the opportunity to do research as an
undergraduate and advocated for us every step of the way. This is in dedication to the
loving memory of Dr. Tracie M. Gibson. Thank you.
III-25
REFERENCES
Benoit, G., E.X. Wang, W.C. Nieder, M. Levandowsky and V.T. Breslin. 1999. Sources and history of heavy metal contamination and sediment deposition in Tivoli South Bay, Hudson River, New York. Estuaries 22:167–178.
Brandon, C.M., J.D. Woodruff, J.P. Donnelly, and R.M. Sullivan. 2014. How unique was Hurricane Sandy? Sedimentary reconstructions of extreme flooding from New York Harbor. Scientific Reports 4: 7366.
Cornell Institute for Resource Information Sciences (IRIS). 2011. Hudson River Estuary tidal wetlands 2007. Hudson River National Estuarine Research Reserve (HRNERR) and New York State Department of Environmental Conservation (NYSDEC). Albany, New York
Croudace, I.W., A. Rindby, and R.G. Rothwell. 2006. ITRAX: description and evaluation of a new multi-function X-ray core scanner. Geological Society of London Special Publications 267:51–63.
Crozaz, G.E. Picciotto, and W. De Breuck. 1964. Antarctic snow chronology with 210Pb. Journal of Geophysical Research 69:2597-2604.
Dean, W.E. Jr. 1974. Determination of carbonate and organic matter in calcareous sediments and sedimentary rocks by loss on ignition: Comparison with other methods. Journal of Sedimentary Petrology 44:242–248.
Florea, N., C. Cristache, G. Oaie, and O. Duliu. 2011. Concordant 210Pb and 137Cs ages of black sea anoxic unconsolidated sediments. Geochronometria 38:101-106.
Heiri, O., A.F. Lotter, and G. Lemcke. 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. Journal of Paleolimnology 25:101–110
HRECRP. 2016. Hudson-Raritan Estuary Comprehensive Restoration Plan. U.S. Army Corps of Engineers and the Port Authority of New York and New Jersey.
Hudson River National Estuarine Research Reserve (HRNERR). 2017. Dams and Sediment in the Hudson (DaSH). https://www.hrnerr.org/hrnerr-research/dams- and-sediment-in-the-hudson Accessed August 15, 2019.
Kemp, A.C., B.P. Horton, D. Nikitina, C.H. Vane, M. Potapova, E. Weber-Bruya, S.J. Culver, T. Repkina, and D.F. Hill. 2017. The distribution and utility of sea-level indicators in Eurasian sub-Artic salt marshes (White Sea, Russia). Boreas 46: 562-584.
III-26
New York State Department of State (NYSDOS). 2012. Vanderburgh Cove and Shallows. Department of State, Coastal fish and wildlife rating form. Dutchess, NY.
OCM Partners. 2020. 2013 USGS Lidar: NY post-Sandy, Ulster, Dutchess, Orange Counties from 2010-06-15 to 2010-08-15. NOAA National Centers for Environmental Information.
Partners Restoring the Hudson. 2018. Hudson River Comprehensive Restoration Plan: Recommendations for the New York–New Jersey Harbor and Estuary Program Action Agenda and the New York State Hudson River Estuary Action Agenda. New York, NY. The Nature Conservancy.
Pennington, W., T.G. Tutin, R.S. Cambray, and E.M. Fisher. 1973. Observations on lake sediments using fallout 137Cs as a tracer. Nature 242:324–326
Ralston, D.K., and W.R. Geyer. 2017. Sediment transport time scales and trapping efficiency in a tidal river. Journal of Geophysical Research: Earth Surface 122:2042-2063
Ralston, D., B. Yellen, and J. Woodruff. 2020. Watershed sediment supply and potential impacts of dam removals for an estuary. Preprint.
Schuerch, M., T. Spencer, S. Temmerman, M.L. Kirwan, C. Wolff, D. Lincke, D., C.J. McOwen, M.D. Pickering, R. Reef, A.T. Vafeidis, J. Hinkel, R.J. Nicholls, and S. Brown. 2018. Future response of global coastal wetlands to sea-level rise. Nature 561:231-234
Shepard C.C., C.M. Crain, and M.W. Beck. 2011. The protective role of coastal marshes: a systematic review and meta-analysis. PLoS ONE 6:e27374.
Squires, D.F. 1992 Quantifying anthropogenic shoreline modification of the Hudson River and estuary from European contact to modern time. Coastal Management 20:343-354
Sritrairat, S., D.M. Peteet, T.C. Kenna, R.N. Sambrotto, D.K. Kurdyla, and T.P. Guilderson. 2012. A history of vegetation, sediment and nutrient dynamics at Tivoli North Bay, Hudson Estuary, NY. Estuarine, Coastal and Shelf Science 102:24-35
Soil Survey Staff . 2019. Natural Resources Conservation Service, United States Department of Agriculture. Web Soil Survey. https://websoilsurvey.sc.egov.usda.gov/. Accessed Feb 7, 2019
Stevens, F.W. 1926. The Beginnings of the New York Central Railroad: A History. G. P. Putnam’s Sons, New York.
III-27 U.S. Geological Survey (USGS). 2016a. The StreamStats. Retrieved Feb 7, 2019, from https://streamstats.usgs.gov/ss/
U.S. Geological Survey (USGS). 2016b. USGS NED one meter x59y465 NY Sandy-Ul- Du-Or 2013 IMG 2016. U.S. Geological Survey.
U.S. Geological Survey (USGS). 2019. Earth Explorer. https://earthexplorer.usgs.gov/ (accessed July 2019)
Yellen, B., J.D. Woodruff, D.K. Ralston, C. Ladlow, S. Fernald, and W.L. Lau. 2018. Evaluating the impacts of dam removal on downstream tidal wetlands. Fall Meeting of American Geophysical Union. Washington, D.C.
Yellen, B., J. Woodruff, D. Ralston, C. Ladlow, S. Fernald, and W.L. Lau. 2020. Rapid tidal marsh development in anthropogenic backwaters. Earth Surface Processes and Landforms. In press.
Zedler, J.B. and S. Kercher. 2005. Wetland resources: status, trends, ecosystem services, and restorability. Annual Review of Environmental Resources 30:39-74.
III-28 PRESENCE AND TROPHIC LEVEL OF FRESHWATER JELLYFISH
(CRASPEDACUSTA SOWERBII), A CRYPTIC INVADER IN THE HUDSON
RIVER BASIN, NY
A Final Report of the Tibor T. Polgar Fellowship Program
Jacob Moore
Polgar Fellow
SUNY College of Environmental Science and Forestry Syracuse, NY 13210
Project Advisor: Donald J. Stewart Department of Environmental and Forest Biology SUNY College of Environmental Science and Forestry Syracuse, NY 13210
Moore, J.P., D.J. Stewart. 2021. Presence and Trophic Level of Freshwater Jellyfish (Craspedacusta sowerbii), a Cryptic Invader in the Hudson River Basin, NY. Section IV: 1-25 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation. IV-1
ABSTRACT
The Freshwater Jellyfish (Craspedacusta sowerbii) is an invasive species that is
relatively unstudied and underrepresented in public record due to the sporadic appearance
of its observable medusa life stage. In New York State, Freshwater Jellyfish have been
reported in over 110 bodies of water, which raises concern as prior research suggests
several harmful interactions between C. sowerbii and invaded systems. This study had the
dual purpose of both developing environmental DNA primers for C. sowerbii detection
and investigating the trophic interactions of C. sowerbii with NY lake communities via
stable isotope analysis. Sampling occurred June to September 2019, across ten lakes in
the Hudson River Valley where C. sowerbii had been previously observed. Filtered water
and sediment eDNA samples were collected, plankton tow nets pulled, and Hester-Dendy
settlement plates were deployed to collect stable isotope specimens and test for C.
sowerbii presence. eDNA samples were analyzed using real time qPCR. It was
discovered that qPCR of eDNA in filtered water was the most sensitive detection method, and the least time-consuming method. Due to low catch of C. sowerbii medusae, stable isotope results were mostly inconclusive but may indicate that freshwater medusae do not feed on fish larvae as previously suggested. These findings could be used by scientists or managers who are seeking to track the current distribution of C. sowerbii, or who are interested in better understanding how the appearance of freshwater medusae may impact planktonic communities of interest.
IV-2
TABLE OF CONTENTS
Abstract ...... I-2 Table of Contents ...... I-3 Lists of Figures and Tables ...... I-4 Introduction ...... I-5 Methods...... I-8 Results ...... I-14 Discussion ...... I-21 Acknowledgements ...... I-23 References ...... I-24
IV-3
LIST OF FIGURES AND TABLES
Figure 1 – Field Sites for 2019 sampling season in the Hudson River Watershed I-9
Figure 2 – Hester-Dendy settlement plate design used in Summer 2019 sampling I-10
Figure 3 – 20 µm (bottom) and 750 µm (top) mesh size nets that were towed
horizontally to collect planktonic samples during Summer 2019 ...... I-11
Figure 4 – August water surface temperature measurements across study sites, and
difference between June and August water surface temperature
measurements ...... I-15
Figure 5 – Estimated larval fish densities of 750 µm tow net samples across study
lakes in June 2019 ...... I-16
Figure 6 – Probable Craspedacusta sowerbii polyp specimen from Wolf Lake,
2m depth, as viewed under 250x magnification ...... I-17
Figure 7 – Comparison of effort time required to survey 10 lakes for Craspedacusta
sowerbii among four detection methods ...... I-18
Figure 8 – Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019
plankton tow net samples from Tillson Lake, NY ...... I-19
Figure 9 – Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019
plankton tow net samples from Stillwater Pond, NY ...... I-20
Figure 10 – Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019
plankton tow net samples from Garnet Lake, NY ...... I-20
Figure 11 – Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019
plankton tow net samples from Lake Luzerne, NY ...... I-21
Table 1 – Summary of Summer 2019 Craspedacusta sowerbii detection results I-15 IV-4
INTRODUCTION
Invasive species represent a growing threat to aquatic resources, displacing native
species and altering ecological processes. Invasive impacts on native species radiate out
to people who have social or economic investment in native systems (Pejchar and
Mooney 2009). To combat these negative effects, scientists, managers, and citizens have
invested significant finances and time in invasive species-focused programs; however, invasive species are typically characterized by generalist traits that make them widely successful and difficult to control. The Freshwater Jellyfish (Craspedacusta sowerbii) is
an invasive species originating from the Yangtze River Basin, China, that has spread to
every continent except Antarctica (Dumont 1994). Freshwater Jellyfish are not true
jellyfish but are hydrozoans that exhibit similar life stages, including sessile polyp and
free-swimming medusa stages (DeVries 1992). Freshwater jellyfish have demonstrated
success in both dispersing throughout riverine systems and bypassing geographic barriers
to invade new systems, and they appear to do so without any human intervention.
Most knowledge on the ecology and occurrence of C. sowerbii is derived from
observations of the larger (5-25 mm) free-swimming medusae, rather than the tiny (0.5-2 mm) bottom-dwelling polyp stage (DeVries 1992; Jankowski 2001); however, formation of C. sowerbii medusae in a system is sporadic and unpredictable, and Craspedacusta sowerbii may exist only as polyps for years before they are reported as medusae (Dumont
1994; Fritz et al. 2009). It is commonly hypothesized that temperature is important to the appearance of medusae, although there are conflicting ideas whether rate of temperature increase (DeVries 1992) or simple value of temperature reached (Minchin et al. 2016) is more relevant. In any case, C. sowerbii has likely invaded significantly more freshwater
IV-5 systems than presently recorded and may be influencing those aquatic ecosystems to an unknown extent.
Relatively few studies have been conducted on the ecology of Freshwater
Jellyfish, but feeding experiments and gut-content analyses suggest that the species demonstrates size-selective feeding on zooplankton, and that could shift relative dominance among plankton taxa in a native community (Dodson and Cooper 1983;
Spadinger and Meier 1999; Smith and Alexander 2008). Feeding experiments have also demonstrated that C. sowerbii is capable of killing and eating larval fishes, although it is currently unknown if medusae are significant predators of fish (Dendy 1978; Dodson and
Cooper 1983; Smith and Alexander 2008). Polyp diet has been primarily associated with small crawling invertebrates, but they have been observed feeding on larval fish under experimental conditions (Bushnell and Porter 1967; Dendy 1978). Although their ecological role is currently uncertain, C. sowerbii have the potential to disrupt food webs and harm native aquatic organisms wherever they occur.
In New York State (NYS), there have been reports of Freshwater Jellyfish in over
100 different systems (Peard 2018). Despite the widespread occurrence of C. sowerbii in
NYS, species information is not found on the NYS Department of Environmental
Conservation website (NYSDEC 2019). Presently, C. sowerbii is severely understudied by scientists, and its spread is unchecked by managers; it most likely will continue to spread and impact additional lake ecosystems as increasingly warm summer temperatures open new opportunities for the species.
IV-6
Study Objectives
The primary objective of this study was to test new detection methods for
Freshwater Jellyfish. The hypothesis addressed in this objective is that genetic material sloughed off by C. sowerbii polyps and medusae will be detectable through environmental DNA (eDNA) methods, and eDNA methods will be more sensitive than other detection methods such as plankton net tows and settlement plates. Environmental
DNA techniques detect free-floating genomic materials in water or sediment rather than relying on the capture or observation of physical specimens. As a result, eDNA techniques are generally more sensitive than other detection methods, and rare or cryptic species like C. sowerbii can be detected in bodies of water where they might otherwise go unseen (Rees et al. 2014). Recent research has also shown that cnidarians, such as C. sowerbii, are excellent candidates for eDNA detection (Minamoto et al. 2017). DNA primers suitable for detecting eDNA from C. sowerbii have been developed in collaboration with Dr. Hyatt Green, State University of New York College of
Environmental Science and Forestry (SUNY-ESF) using available sequences and tools from GenBank (see Methods, below). These primers were derived from published sequences on GenBank and assessed using BLAST and OligoAnalyzer software
(GenBank 2019). Species-specific sensitivity tests were conducted, both virtually and in the lab, to reduce the risk of false positives in environmental samples. Successful testing of these primers with environmental samples would provide the NYSDEC and other organizations with the ability to conveniently sample water across the state to detect new invasions of Freshwater Jellyfish.
My second study objective was to use previously unapplied methods to
IV-7
investigate the ecological role of C. sowerbii in invaded systems, particularly in the polyp
stage. The hypothesis addressed in this objective is that stable isotope analysis of δ15N
and δ13C values will indicate that C. sowerbii are feeding on larval fishes as suggested in past studies (Dendy 1978; Dodson and Cooper 1983; Smith and Alexander 2008). Past research has focused on gut content analyses and feeding experiments of medusae, both of which are limited in their ability to represent natural diet over time. Stable isotope analyses of different trophic levels in a system allows assessment of relatively longer- term dietary behavior of animals in their natural habitat (Hamilton et al. 1992).
METHODS
Primer Design
Before testing of C. sowerbii DNA could begin with physical samples, primer sequences had to be derived in silicio from existing sequences available in the GenBank database (GenBank 2019). C. sowerbii DNA sequences were analyzed with the Basic
Local Alignment Search Tool (BLAST) under the discontinuous megablast program. The
C. sowerbii gene sequence selected to develop primers had the accession code
LN901194.1 (mitochondrial genome, partial sequence). Physical characteristics (e.g.,
melting temperature, dimer potential) of candidate primers was further assessed by the
IDT™ OligoAnalyzer tool. Candidate primers were also tested in vitro with DNA
extracted from C. sowerbii tissue and multiple species of Hydra using real-time quantitative polymerase chain reaction (qPCR) analysis to check for undesired amplification. After these tests were concluded, the final primer sequences derived for testing on environmental samples amplify a product length of 146bp and are as follows: IV-8
Forward Primer: 5'- GAA TCA GAA TAG GTG CTG ATA GAG AAT C -3'
Reverse Primer: 5'- CTA ATC ACG GCC TTC CTT CTG G -3'
Field Sampling
Ten lakes within the Hudson River watershed were selected for this study (Figure
1), as 42 sites in the area have already reported sightings of Craspedacusta sowerbii. Five sites were in northern Adirondack areas, and five were located within southern areas of
the Hudson River basin.
Figure 1: Field Sites for 2019 sampling season in the Hudson River Watershed.
Initial sampling began in mid-June 2019 and consisted of assessing vertical
profiles of water quality with a YSI ProDSS multiparameter probe. From late June –
early July 2019, each site was revisited to collect samples and deploy Hester-Dendy IV-9 settlement plate samplers (Figure 2). These 15 cm x 15 cm square settling plates were constructed based on a design tested successfully in the field and may be used as an alternate means to detect C. sowerbii polyps in a lake (T. Peard, personal communication). Each plate sampler consisted of six sanded acrylic plates with PVC spacers on a stainless-steel structure (Figure 2). Three plate samplers were deployed in each lake and georeferenced using a Garmin ™ Striker Plus 4 dual beam transducer
(which also provided surface water temperature).
Figure 2: Hester-Dendy settlement plate design used in Summer 2019 sampling.
During deployment of settlement plates, plankton nets of 20 µm and 750 µm mesh sizes (Figure 3) were towed off the side of a canoe at 1.5 m depth for 5-10 min to collect zooplankton, fish larvae, and any medusae present. Nets were kept at depth with a
1.8 kg cannonball weight. Fish larvae and medusae densities were quantified with the 750
IV-10
µm net using a General Oceanics 2030R mechanical flowmeter mounted in mouth of the net to determine volume sampled (calibrated in water flume at SUNY-ESF, June 11,
2019). Three 10-15 mL samples of surface sediment were also collected at each site with a handmade PVC-Steel gravity corer for eDNA analysis. Sterile technique was followed by soaking sediment sample gear in 10% bleach for at least 15 minutes between sites, and sample blanks were prepared at every site by measuring 15 mL of deionized water with the same gear used to measure sediment samples. Sediment samples were preserved in
3M sodium acetate and 95% ethanol on-site before transport on ice and storage at -80°C.
Figure 3: 20 µm (bottom) and 750 µm (top) mesh size nets that were towed horizontally to collect planktonic samples during Summer 2019.
Late-summer sampling occurred mid-August to mid-September 2019. Additional
750 µm plankton net tows were conducted for 5-10 min to collect large zooplankton, fish
IV-11
larvae, and any medusae present. New vertical profiles of water quality were also
conducted with the YSI probe. Three 2-L whole water samples were collected at each site
and filtered onto 47 mm Whatman glass fiber filters in an effort to detect free-floating eDNA shed by medusae (excepting Glenmere Lake and East Caroga Lake, as they were not accessible at the time). Filter samples were put on ice immediately and stored at -
80°C within 24 hours in Whirl-Pak® bags.
At this time, plate samplers were also retrieved and preserved in plastic bags with
70% ethanol for visual observations under a dissecting scope. If potential C. sowerbii polyps were observed on a plate, they were extracted with tweezers and placed in labeled vials of 95% ethanol at room temperature for later DNA confirmation using the Qiagen
DNeasy™ Blood and Tissue Kit and SYBR qPCR analysis (Qiagen 2019).
eDNA assessment
Sediment eDNA samples were thawed and DNA was extracted using the MP Bio
FastDNA™ SPIN Kit for Soil; eDNA samples on water filters had DNA extracted using the Qiagen DNeasy™ Blood and Tissue Kit. After DNA samples were extracted, total
DNA concentration was measured using an Invitrogen Qubit 4 Fluorometer. Following confirmation of quality DNA, 2 µL of each sample was placed in a well with 23 µL of a mix containing the developed C. sowerbii-specific primer markers on a 96-well qPCR plate. Triplicates were made of every sample on a well. The plate was run with SYBR fluorescence in a QuantStudio 3 machine along with a standard and several sample blanks. Results were exported and analyzed in QuantStudio design and analysis software.
Estimation of Effort Times for Various Detection Methods
IV-12
During activities associated with C. sowerbii detection (plankton net tows,
settlement plates, and eDNA samples), time in the field and the lab was monitored and
recorded a notebook then transcribed onto an Excel spreadsheet. Travel time between
sites was not included to better standardize estimates of effort and is irrelevant as
detection tests were conducted in the same locations. Estimates are for one trained person
operating from a kayak. Effort time estimates (in units of hr elapsed * ten lakes-1) were scaled to ten study lakes to demonstrate the efficiency of eDNA testing at larger sample sizes (due to equipment such as 96-well qPCR plates) in comparison to analysis of settlement plates and plankton net samples that do not have improved efficiency at larger sample sizes.
Stable Isotope Analyses
To conduct stable isotope analyses, C. sowerbii medusae, large zooplankton (from
750 µm tow samples), fish larvae, and small plankton samples (taken from 20 µm net tows and sieved within a 2-500 µm size range) were euthanized with MS-222 (250 mg/ml) and preserved in 95% ethanol on-site. Preserved net tow samples were then brought to the NYSDEC Forest Health Diagnostic Lab in Delmar, NY, for sorting, ID and measurement of specimens. Following sorting, specimens had ethanol rinsed off using deionized water, were dried in a 60°C oven, and pulverized in the laboratory using a mortar and pestle (Feuchtmayr and Grey 2003). After specimen preparation was complete, samples were shipped to the Cornell University Stable Isotope Laboratory
(COIL, Ithaca, NY) for analysis of δ13C and δ15N isotope ratios. δ15N levels were
corrected using atmospheric values as a reference, and δ13C levels were corrected using
the primary reference scale of Vienna Pee Dee Belemnite. Results of these analyses were
IV-13
then plotted on a δ15N versus δ13C scatterplot to compare values at different trophic levels
within a lake. Specimens were only analyzed in lakes where medusae were successfully
collected.
RESULTS Net Tows and Settlement Plates
No specimens of C. sowerbii appeared in plankton net tows until mid-August,
when a few individuals were collected from Tillson Lake, Lake Luzerne, and Garnet
Lake (Table 1). Additional net tows in September yielded additional specimens in all
three lakes previously mentioned, plus two in Stillwater Pond. Estimated medusae
densities ranged from 0.018 (Lake Luzerne) to 0.24 medusae/m3 (Stillwater Pond).
Collected medusae ranged in size from 1.5 - 15 mm. When medusae began to appear in
August, lake surface temperatures were above 22.5 °C (Figure 4). Lakes did not all
appear to consistently increase in temperature from June to August sampling, as East
Caroga Lake and East Stoner Lakes decreased in temperature. Garnet Lake was notable
in having the highest measured surface temperature at 28.15 °C, the greatest number of medusae collected in the nets (11 individuals) and reported bloom conditions in late
September with greatly increased medusa density (Judy Thomson, personal communication).
IV-14
Table 1. Summary of Summer 2019 Craspedacusta sowerbii detection results. + indicates positive detection, - indicates negative detection, N/A indicates site was unavailable on day of sampling.
30 Aug Surface Temp ΔSurface Temp (Aug-June) 25
20
15
10 Surface Surface temperature (C) 5
0 Glenmere White Stillwater Canopus Tillson Wolf Garnet Luzerne E Caroga E Stoner -5 Figure 4: August water surface temperature measurements across study sites, and difference between June and August water surface temperature measurements.
Larval fish density varied between lakes; in White Pond, larvae were collected at a density of 4.47 larvae/m3 whereas no larvae were collected in Wolf Lake (Figure 5).
Larvae were identified as Centrarchidae, including four Micropterus salmoides and 24 unknown Lepomis spp. individuals.
IV-15
Figure 5: Estimated larval fish densities of 750 µm tow net samples across study lakes in June 2019.
During retrieval, several settlement plate samplers were lost, most likely due to
theft, boat propeller damage, or movement due to storm conditions. Visual observation of
settlement plates was a fairly, time-intensive process, and at this time only 12 samplers
have been observed. Of the observed samplers, eight had potential C. sowerbii polyps
(Figure 6), and two were confirmed to have polyps of that species using qPCR analysis
(all triplicates returning amplification above the threshold). From the completed samples,
C. sowerbii polyps were only detected in Wolf Lake and Glenmere Lake using settlement plates; neither of these lakes had medusae present in plankton net tows.
IV-16
Figure 6: Probable Craspedacusta sowerbii polyp specimen identified on settlement plate placed in Wolf Lake, 2 m depth, as viewed under 250x magnification.
Environmental DNA (eDNA) samples
Sediment samples tested for eDNA had only 2 out of 30 samples return with
positive detection (Table 1). These samples were both from Canopus Lake. Filtered water eDNA samples returned positive detection in 11 out of 24 samples. These samples were within East Stoner Lake, Tillson Lake, White Pond, Wolf Lake, Garnet Lake, Canopus
Lake, and Stillwater Pond. Again, several eDNA samples returned with partial detection of C. sowerbii DNA. The reason for these partial detections may be due to the estimated concentration of C. sowerbii DNA, which ranged from 0.0035 to 2.7 DNA copies/µL in sediment extracts and 0.038 to 7.5 DNA copies/µL in filtered water extracts. These relatively low concentrations may have resulted in many samples with amplifications
IV-17
below the threshold value (CT = 0.01).
Estimated Effort Times for Various Detection Methods
Effort times for field collections plus laboratory analyses for possible C. sowerbii
detection was estimated to be highest when using settlement plate samplers at 85.0 hr
elapsed * ten lakes-1, and lowest when using eDNA from filtered water at 38.9 hr* ten
lakes-1, although use of plankton tows was estimated to have 6.83 hr elapsed * ten lakes-1 fewer devoted to field collections (Figure 7). The relatively high effort time needed for use of settlement plates is partially due to the need for two sampling trips: once for deployment of samplers and again for retrieval of samplers.
90.00 Lab Analysis )
1 80.00 - Field Collection 70.00
60.00
50.00
40.00
30.00
20.00
Effort time (hr elapsed * ten lakes ten * elapsed (hr time Effort 10.00
0.00 Plankton Tow (750 µm Settlement Plates eDNA (sediment) eDNA (filtered water) mesh) Figure 7: Comparison of effort time required to survey 10 lakes for Craspedacusta sowerbii among four detection methods: Plankton net tows, settlement plates, environmental DNA (eDNA) bottom sediment samples, eDNA filtered water samples.
Stable Isotope Analyses Results of stable isotope analyses were variable among lakes, with samples from
Tillson Lake having a relatively unique δ13C signature, and elevated small plankton δ15N
levels of 5.26 and 5.56 ppt (Figure 8). Tillson Lake small plankton and separated IV-18 copepods had δ13C 4-5 ppt lower than in medusae. Regarding fish larvae and medusae, in both sites where enough biomass was able to be collected to obtain a value for both, δ13C was within 2 ppt, but in neither case was δ15N higher in medusae than fish larvae (Figure
8; Figure 10). In Garnet Lake, medusae δ15N was within 1 ppt for both larval fish and
Leptodora kindtii, and higher (5.74 ppt) than Bosmina (4.11 ppt) and small plankton
(2.08 – 2.18 ppt) (Figure 10).
Tillson Lake 8.00
7.00
6.00 Small Plankton 5.00 Copepods N N 4.00 Daphnia δ15 3.00 Fish Larvae Medusae 2.00
1.00
0.00 -36.00 -34.00 -32.00 -30.00 -28.00 δ13C Figure 8: Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019 plankton tow net samples from Tillson Lake, NY. Analysis conducted by the Cornell University Stable Isotope Laboratory (COIL), Ithaca, NY.
IV-19
Stillwater Pond 7.00
6.00
5.00 Small Plankton 4.00 N N Daphnia
δ15 3.00 Fish Larvae* Leptodora 2.00 Medusae 1.00
0.00 -32.00 -31.00 -30.00 -29.00 -28.00 -27.00 -26.00 δ13C Figure 9: Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019 plankton tow net samples from Stillwater Pond, NY. Analysis conducted by the Cornell University Stable Isotope Laboratory (COIL), Ithaca, NY. *Fish Larvae sample was below the biomass threshold and may not be accurate.
Garnet Lake 6.00
5.00
4.00 Small Plankton
N N Bosmina 3.00
δ15 Fish Larvae 2.00 Leptodora Medusae 1.00
0.00 -32.00 -31.00 -30.00 -29.00 -28.00 -27.00 -26.00 δ13C Figure 10: Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019 plankton tow net samples from Garnet Lake, NY. Analysis conducted by the Cornell University Stable Isotope Laboratory (COIL), Ithaca, NY.
IV-20
Lake Luzerne
12.00
10.00
8.00 Small Plankton Daphnia 6.00 15N
δ Fish Larvae 4.00 Leptodora
2.00 Medusae*
0.00 -36.00 -34.00 -32.00 -30.00 -28.00 δ13C Figure 11: Scatterplot of δ15N vs. δ13C stable isotope analyses of Summer 2019 plankton tow net samples from Lake Luzerne, NY. Analysis conducted by the Cornell University Stable Isotope Laboratory (COIL), Ithaca, NY. *Medusae sample was below the biomass threshold and may not be accurate.
DISCUSSION
Detection of C. sowerbii proved to be difficult regardless of the method used, even though sampling of lakes was targeted based on location and seasonal appearance of
previously reported medusae blooms. Although previous studies have hypothesized that
elevated temperature is inducive to the appearance of medusae (Minchin et al. 2016)
Overall, eDNA assessment using filtered water was the most sensitive detection
method for C. sowerbii, supporting the previously stated hypothesis. This is likely due to
the season of sampling, as medusae are typically reported to be most abundant in Late
Summer, when water temperatures have been in the mid-20’s °C; however, the low
sensitivity of sediment eDNA samples does not support the hypothesis nor align with
previous eDNA studies that have demonstrated much greater DNA concentrations in
IV-21
sediment samples (Minamoto et al. 2017; Turner et al. 2015). Recommendations for
future use of eDNA assessment of waterbodies for C. sowerbii would suggest focusing on
taking samples during weeks when medusae are likely present, which is with surface
water temperatures having been above 20 °C for several weeks to allow polyps to
produce medusae.
Altogether, these data provide interesting observations on the ecology of C.
sowerbii in NY lakes. For one, δ15N values suggest medusae do not feed at a trophic level
above larval fish, but rather are feeding at a similar planktivorous level along with
Leptodora kindtii (Leite et al. 2002). Results for δ13C also may illustrate that these
medusae were not feeding preferentially on copepods, even though prior feeding experimentation has demonstrated the opposite (Smith and Alexander 2008); however,
the results of this study rely on very few data points due to the sparsity of medusae
samples available, and should not be considered conclusive without additional sampling
and isotope analyses to corroborate these observations.
As professionals and students contemplate options to locate Craspedacusta
sowerbii for research, management, or other purposes, this study may be useful as a guide
to compare effectiveness of various methods of detection. In consideration of the elusive
nature of C. sowerbii, application of the eDNA assay developed and tested in this study
would be a helpful option, as many lakes can be sampled and tested relatively quickly.
Relying solely on eyewitness reports and traditional methods (such as plankton tows) will
likely result in deficient evidence regarding presence of this species and its spread. The
impact that C. sowerbii has on US lakes remains largely unknown, although observations
from isotope results (e.g., the apparent lack of predation on fish larvae) should be useful
to future research on the species.
IV-22
ACKNOWLEDGEMENTS
I would like to offer my appreciation to the Hudson River Foundation for
financial support of this research, and their helpful input during meetings. I would also
like to thank the NYSDEC for assistance in organizing the project; I especially thank
Steven Pearson for support in the field and lab. I also thank the staff and faculty of
Huntington Wildlife Forest for their assistance in researching Wolf Lake. I could not
have completed my project without the help of my family and undergraduate assistants
Emily Froass and Breanna Hummel. I also thank Hyatt Green and Margaret Murphy for their advisement and encouragement of my work. Finally, I thank Donald Stewart for his many hours working with me in field and lab to help overcome a variety of obstacles associated with the project.
IV-23
REFERENCES
Bushnell Jr, J. H., and T.W. Porter. 1967. The occurrence, habitat, and prey of Craspedacusta sowerbyi (particularly polyp stage) in Michigan. Transactions of the American Microscopical Society 86:22-27. Dendy, J.S. 1978. Polyps of Craspedacusta sowerbyi as predators on young striped bass. The Progressive Fish-Culturist 40: 5-6. DeVries, D.R. 1992. The freshwater jellyfish Craspedacusta sowerbyi: a summary of its life history, ecology, and distribution. Journal of Freshwater Ecology 7: 7-16. Dodson, S.I., and S.D. Cooper. 1983. Trophic relationships of the freshwater jellyfish Craspedacusta sowerbyi Lankester 1880. Limnology and Oceanography 28: 345- 351. Dumont, H.J. 1994. The distribution and ecology of the fresh- and brackish-water medusae of the world. pp. 1-12 in Dumont, H. J., J. Green, and H. Masundire, (eds), Studies on the Ecology of Tropical Zooplankton,. Springer, Dordrecht. Feuchtmayr, H., and J. Grey. 2003. Effect of preparation and preservation procedures on carbon and nitrogen stable isotope determinations from zooplankton. Rapid Communications in Mass Spectrometry 17: 2605-2610. Fritz, G.B., M. Pfannkuchen, A. Reuner, A., R.O. Schill, and F. Brümmer. 2009. Craspedacusta sowerbii, Lankester 1880-population dispersal analysis using COI and ITS sequences. Journal of Limnology 68: 46-52. GenBank. 2019. Retrieved from: https://www.ncbi.nlm.nih.gov/genbank/ . Accessed 2019. Hamilton, S. K., W.M. Lewis, and S.J. Sippel. 1992. Energy sources for aquatic animals in the Orinoco River floodplain: evidence from stable isotopes. Oecologia 89: 324-330. Jankowski, T. 2001. The freshwater medusae of the world–a taxonomic and systematic literature study with some remarks on other inland water jellyfish. Hydrobiologia 462: 91-113. Leite, R. G., C.A.R.M. Araújo‐Lima, R.L. Victoria, and L. A. Martinelli. 2002. Stable isotope analysis of energy sources for larvae of eight fish species from the Amazon floodplain. Ecology of Freshwater Fish 11: 56-63. Minamoto, T., M. Fukuda, K.R. Katsuhara, A. Fujiwara, S. Hidaka, S. Yamamoto, K. Takahashi, and R. Masuda. 2017. Environmental DNA reflects spatial and temporal jellyfish distribution. PloS one 12. Minchin, D., J.M. Caffrey, D. Haberlin, D. Germaine, C. Walsh, R. Boelens, and T. K. Doyle. 2016. First observations of the freshwater jellyfish Craspedacusta sowerbii Lankester, 1880 in Ireland coincides with unusually high water temperatures. BioInvasions Records 5: 67-74. IV-24
NYSDEC. 2019. Aquatic Invasive Species in New York State. Retrieved from: https://www.dec.ny.gov/animals/50121.html Peard, T. P. 2018. Freshwater Jellyfish. Retrieved from: http://freshwaterjellyfish.org/ Pejchar, L., and H.A. Mooney. 2009. Invasive species, ecosystem services and human well-being. Trends in Ecology and Evolution 24: 497-504. Qiagen. 2019. DNeasy Blood and Tissue Handbook. Retrieved from: https://www.qiagen.com/mx/resources/resourcedetail?id=6b09dfb8-6319-464d- 996c-79e8c7045a50&lang=en Rees, H. C., B.C. Maddison, D.J. Middleditch, J.R. Patmore, and K. C. Gough. 2014. The detection of aquatic animal species using environmental DNA–a review of eDNA as a survey tool in ecology. Journal of Applied Ecology 51: 1450-1459. Smith, A.S., and J.E. Alexander Jr. 2008. Potential effects of the freshwater jellyfish Craspedacusta sowerbii on zooplankton community abundance. Journal of Plankton Research 30: 1323-1327. Spadinger, R., and G. Maier. 1999. Prey selection and diel feeding of the freshwater jellyfish, Craspedacusta sowerbyi. Freshwater Biology 41: 567-573. Turner, C.R., K.L. Uy, and R. C. Everhart. 2015. Fish environmental DNA is more concentrated in aquatic sediments than surface water. Biological Conservation 183: 93-102.
IV-25
THE POTENTIAL FOR HARMFUL ALGAL BLOOMS (HABs) IN THE HUDSON RIVER ESTUARY: DOMINANT ABIOTIC DRIVERS OF CYANOBACTERIAL ABUNDANCE AND TOXICITY WITH THE COMPOUNDING INFLUENCE OF THE INVASIVE WATER CHESTNUT
A Final Report of the Tibor T. Polgar Fellowship Program
Ellie Petraccione
Polgar Fellow
Department of Environmental Science and Policy Marist College Poughkeepsie, NY 12601
Project Advisors:
Dr. Zion Klos Department of Environmental Science and Policy Marist College Poughkeepsie, NY 12601
Dr. Raymond Kepner Department of Biology Marist College Poughkeepsie, NY 12601
Petraccione, E., Z. Klos and R. Kepner. 2021. The Potential for Harmful Algal Blooms (HABs) in the Hudson River Estuary: Dominant Abiotic Drivers of Cyanobacterial Abundance and Toxicity with the Compounding Influence of the Invasive Water Chestnut. Section V: 1-32 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
V-1 ABSTRACT
Low-current tributary-estuaries and embayments along the margin of the Hudson
River are uniquely at risk for harmful algal blooms of cyanobacteria (cyanoHABs) due to
rising temperatures as a result of climate change. An increased prevalence of cyanoHABs
in near-shore tributaries, and lower-flow estuary extensions of the Hudson River can be
extremely harmful to nearby communities, aquatic organisms and wildlife. High-risk
locations are also susceptible to growth of the invasive species Trapa natans (European
water chestnut). The weekly changes in the background abundance of cyanobacteria in T.
natans blooms and their abiotic drivers were measured at contrasting locations along the
Hudson River. It was hypothesized that observed surface water temperatures and
turbidity would have the strongest control on the background levels of cyanobacteria in
the Hudson River, given the current state of eutrophication. It was also hypothesized that
conditions within the T. natans beds would be more conducive to cyanobacterial growth
than open water. It was concluded that the T. natans beds were significantly warmer and
less turbid than the open water. The nitrogen levels in beds of T. natans fluctuated
throughout the summer. Cyanobacterial community composition within the T. natans
depended on the species tolerance to nitrogen depletion. Possible toxin-producing
Microcystis were dominant in the open water where nitrogen was more available, despite
the less than ideal conditions. Planktothrix, also possibly toxic, was dominant within the
T. natans blooms, but was considered less likely to form harmful blooms in nitrogen
depleted conditions. T. natans beds may be hindering the growth of toxic cyanoHABs.
Mitigation efforts to remove T. natans need to consider the possible role T. natans plays in decreasing the risk of cyanoHABs.
V-2 TABLE OF CONTENTS
Abstract ...... V-2
Table of Contents ...... V-3
Lists of Figures and Tables ...... V-4
Introduction ...... V-6
Methods...... V-12
Site Selection and Description ...... V-12
Data Collection: Sampling and Timing ...... V-14
Measuring Abiotic Factors Along A Transect ...... V-14
Measurement of Nitrate and Orthophosphate Concentrations ...... V-15
Cyanobacterial Counts ...... V-15
Results ...... V-17
Discussion ...... V-23
Time Series and Seasonal Growth of T. natans ...... V-23
Dissolved Oxygen ...... V-24
Surface Temperature, pH, and Turbidity ...... V-24
Cyanobacterial Communities and Toxicity with Influence of T. natans . V-25
Advancing Understanding Through Future Research...... V-27
Implications for Watershed Management ...... V-28
Acknowledgements ...... V-29
References ...... V-30
V-3 LIST OF FIGURES AND TABLES
Figure 1 – Climate change and eutrophication from nutrient-polluted runoff and the effect
on cyanoHABs ...... V-7
Figure 2 – Seasonal variability expressed by maximum and minimum temperatures at
Norrie Point between the months of June and August. Data collected from the
Hudson River Environmental Conditions Observing System (HRECOS) V-10
Figure 3 – Sampling low-current embayments and tributary-estuaries along the Hudson
River ...... V-13
Figure 4 – Description of apparatus ...... V-16
Figure 5 – Cyanobacterial cell counts (cells/mL) and blue-green (BG) Chlorophyll (ppb).
Trendline is represented by the equation y= 0.0048x + 2.1477 with an R2 value of
0.9016...... V-17
Figure 6 – Time-series of abiotic factors at (A) Esopus Creek, (B) Rondout Creek, (C)
Wappingers Creek, (D) Port Ewen and (E) Long Dock Park ...... V-17
Figure 7 – Nitrate concentration (ppm) and T. natans coverage in Wappingers Creek from
July to October ...... V-19
Figure 8 – Correlation matrix of abiotic factors in relation to cyanobacterial
concentrations for all observations (n = 32). Relevant correlation coefficients (r),
P-values, and corresponding plots are displayed ...... V-20
Figure 9 – Correlation matrix of abiotic factors in relation to maximum, minimum, mean
and range of cells/mL at each site (n=5). Relevant correlation coefficients (r), P-
values, and corresponding plots are displayed ...... V-21
V-4 Figure 10 – Difference in abiotic factors between the open water samples and samples
within T. natans bed (n=13). Significant P-values are represented ...... V-22
Figure 11 – Distribution of cyanobacterial species (n = 17 to 36) ...... V-23
Figure 12 – Dominating cyanobacterial species and abiotic drivers within and outside
beds of T. natans ...... V-26
Table 1 – Percent coverage and nitrate concentration at Wappingers Creek ...... V-19
V-5 INTRODUCTION
Cyanobacterial harmful algal blooms (cyanoHABs) have the potential to release
substances that are directly toxic to humans (Salls et al. 2018). One way in which the severity of a cyanoHAB can be estimated is by measuring the amount of toxins called
microcystins, which are released by a variety of bloom-forming freshwater
cyanobacterial genera, including Microcystis, Limnothrix, Phormidium, Nostoc,
Anabaenopsis, Hapalosiphon, and Planktothrix (Cronberg and Annadotter 2006).
Previous studies have shown that exposure to microcystin can promote tumor formation,
interfere with DNA damage repair, and specifically target the liver and the male
reproductive system (Lone et al. 2015; Falconer and Humpage 2001). Microcystins are
hepatotoxins and microcystin-LR is the most potent liver carcinogen yet characterized
(Cronberg and Annadotter 2006). In addition to toxicity, cyanoHAB senescence rapidly
depletes the environment of dissolved oxygen (DO), creating an uninhabitable
environment for fish. This has been shown to cause massive fish die offs and additional
issues in the food web from long-term shading (Rodger et al. 1994; Havens 2008).
The main stem of the Hudson River is well-mixed and constantly moving due to
downstream flow and tidal currents. Outside of the main channel, there are tributary-
estuaries and side embayments that tend to be shallower with greater stagnancy. Shallow
waters off the main stem are less turbulent and are more susceptible in summer to thermal
stratification. The warmer surface layer with higher light levels is ideal cyanoHABs
habit, especially when vegetative cyanobacteria and akinetes already exist. There is
concern that past occurrences of cyanoHABs in extensions of the Hudson River might
lead to increased levels of cyanobacteria in the main stem as well. In tributary-estuaries
V-6 of the Hudson River, incidences of cyanoHABs have increased from one in 2015, to
seven in 2018, according to the New York State Department of Environmental
Conservation (NYSDEC 2018). Sites with a historic record of cyanoHABs tend to have
a higher frequency of blooms across subsequent years (Salls et al. 2018). Climate change is exacerbating the issue, as elevated water temperatures promote cyanobacterial growth and HABs (Fernald et al. 2007). The risk of cyanoHABs may also be compounded by the presence of invasive aquatic species which alter nutrient concentrations, light availability,
water temperature, or other factors in ways that might promote blooms (Figure 1).
Figure 1. Climate change and eutrophication from nutrient-polluted runoff and the effects on cyanoHABs.
The European water chestnut (Trapa natans) is an invasive species with the potential to enhance cyanobacterial growth through mechanisms such as: 1) lowering the total nitrogen to total phosphorus (TN/TP) ratio, 2) lowering light levels, 3) increasing water temperatures, 4) enhancing required trace element availability, or 5) increasing
V-7 either NH3-N, (which will favor non-N-fixing species), or reducing N in general, (which will favor N-fixing species) (Cronberg and Annadotter 2006). An effective way to assess the risk of cyanoHABs in the Hudson River is to study the likelihood of blooms based on background levels of cyanobacteria, the potential abiotic drivers, and the potential
influence of T. natans. T. natans pose a risk to the Hudson River by replacing native
submerged aquatic vegetation (SAV) because the leaves block light and overwhelm the
riverbed with a complex root system (Hummel and Findlay 2006). The floating leaves
photosynthesize, but instead of releasing O2 to the water below like SAV, O2 is released
into the surrounding air (Caraco et al. 2006). The depletion of DO from a lack of photosynthetic activity below the surface can inhibit the growth of diverse ecosystems that are characteristic of Hudson River tributary-estuaries and embayments (Hummel and
Findlay 2006). T. natans are also efficient in denitrification, which can potentially affect cyanobacterial communities (Tall et al. 2018).
The life cycle of cyanobacteria is heavily dependent on the availability of resources (Hense and Backmann 2010). Cyanobacteria can exist at times of resource
limitation in the form of dormant akinetes which can reside in sediments until vegetative
cells are able to flourish under better conditions. In the freshwater systems like the
tributary-estuaries and embayments observed, resource limitation is often due to a lack of
3- - light or low concentrations of orthophosphate (PO4 ) and nitrate (NO3 ) (Hense and
Backmann 2010). Following the deterioration of a bloom, akinetes sink and settle in the
sediment. When this sediment is exposed to solar energy in a nutrient-rich environment,
akinetes mature (Hense and Backmann 2010). This leads to a positive feedback loop
between the pelagic growing stage and benthic resting stage (Hense et al. 2013). A
V-8 greater number of akinetes in the benthic zone potentially leads to a greater number of
vegetative cyanobacteria in the water, yielding more akinetes (Hense 2007). This positive feedback loop compounds the risk of cyanoHABs in subsequent years. The Hudson River is also considered to be eutrophic (Howarth 2011). Cyanobacterial akinetes are reliant on
nutrient availability and light for maturation, but variability in extreme temperatures also
effects maturation (Hense et al. 2013). Research currently shows that the amount of
change in temperature may have an even greater influence on levels of cyanobacteria than the actual temperature (Hense 2007). Seasonal summer temperatures (June-August) were analyzed at a nearby embayment, Norrie Point. Norrie Point is a NYSDEC environmental center and state park located between the five observed sites, roughly 20 miles from both Wappingers Falls and Port Ewen. It was apparent that there has been high variability in seasonal temperatures (Figure 2).
V-9
Figure 2. Seasonal variability expressed by maximum and minimum temperatures at Norrie Point between the months of June and August. Data collected from the Hudson River Environmental Conditions Observing System (HRECOS).
The cyanobacterial threshold of 25 °C is an approximate value based on previous research that found the optimal temperature of cyanobacterial growth is over 25°C in freshwater systems (Paerl 2014). It is apparent that maximum temperatures have consistently risen above the approximate threshold from 2008 to 2016. There is an increased concern for cyanoHABs in embayments, like Long Dock Park, because even a slight increase in temperature has been shown to increase cyanobacterial biomass and dominance (Fernald et al. 2007). This risk is compounded by the eutrophication and seasonal variability of the Hudson River in the form of resuspension events.
Resuspension events are caused by turbidity and mixing which can increase the chance of
V-10 bloom initiation by providing previously settled nutrients to akinetes for maturation
(Davis et al. 2018).
High nutrient inputs, the increased range of temperatures, and warming from climate change are all factors which may increase the likelihood of cyanoHABs in the
Hudson River. There is a need to study the effects of hydrologic controls, nutrient concentrations, and invasive species on background cyanobacterial populations in the
Hudson River. Particularly, focus should be on higher-risk, lower-current areas like tributary-estuaries and near shore embayments, where T. natans often dominate. Given this information, prevention efforts can be explored to protect the Hudson River from future cyanoHABs. While the risk of cyanoHABs and water chestnuts in the Hudson
River Estuary have been separately researched before, there is no information about a potential interaction between T. natans and cyanobacterial community structure. There is also a lack of information about which species occur within Hudson River cyanobacterial communities, their potential toxicity and the presence of aquatic vegetation like T. natans.
It was hypothesized that surface water temperatures and turbidity would have the strongest control on the background levels of cyanobacteria in the Hudson River. It was also hypothesized that conditions within the T. natans beds would be more conducive to cyanobacterial growth than open water. It was also hypothesized that bacterial communities would not differ depending on whether they were within or outside of the T. natans. The objective of this study was to quantify background levels of cyanobacteria, and thus the potential for cyanoHABs, along select tributary-estuaries and other lower- current embayment areas along the edges of the main-stem Hudson River. A secondary
V-11 objective was to relate cyanobacterial abundances to several abiotic factors known to influence them, including surface water temperature, river bed temperature, turbidity, and
N and P concentrations as they vary temporally and spatially. The third objective was to understand the potential role of the invasive water chestnut in influencing cyanobacterial community structure.
METHODS
Site selection and description
Five different study sites along the Hudson River were chosen because they were considered more likely to experience a cyanoHAB due to conditions such as lower turbidity and warmer water temperatures than the high-current main channel of the
Hudson River. Three tributary-estuaries including, Rondout Creek, Esopus Creek and
Wappingers Creek were chosen for observation, as well as two nearshore embayments,
Port Ewen in Kingston, and Long Dock Park in Beacon, which experienced an observed cyanoHAB bloom in August of 2018 (Figure 3).
V-12
Figure 3. Sampling low-current embayments and tributary-estuaries along the Hudson River.
Study sites were chosen to represent a variety of environmental conditions. These conditions include whether the site was immediately downstream of a rural or urban/suburban community and whether it was a larger or smaller area of lower-current water. Rondout Creek has a relatively large estuary area fed by an upstream region that is dominantly rural and agricultural. Alongside this tributary-estuary is Port Ewen, a shallow embayment that also experiences agricultural influence. Long Dock Park is a smaller embayment that is located in an urban/suburban setting. It is the site of a variety of recreational activities like boating and fishing. Esopus Creek is another tributary, which experiences recreational boating and swimming. Wappingers Creek is a relatively large estuary that is located at the mouth of the watershed draining some sections that are heavily suburban, and others that are rural.
V-13 Data collection: sampling and timing
— 3— Sites were reached by canoe to measure hydrologic factors, NO3 N and PO4 P
concentrations, and to obtain samples for background cyanobacterial counts. Sites were
visited every two weeks, as well as at times of extreme temperature or wind events which
have been linked in other regions to facilitating blooms (Davis et al. 2018). A HydroLab
DataSonde® 4a multiprobe was used to measure specific conductance (mS cm-1), salinity
(ppt), pH (std. units), oxygen reduction potential (mV), DO (mg L-1 and % saturation),
turbidity (NTUs), total dissolved solids (g L-1), and surface water temperature (ºC).
Temperature probes were used to measure riverbed temperature in the riverbed littoral zone.
Measuring abiotic factors along a transect
The abiotic factors of three sites were measured along a transect between the T. natans bed and the open water. Collections were done in equidistant increments from the
open water to the middle of the T. natans bed at Rondout Creek, Port Ewen, and
Wappingers Creek. Data was collected via HydroLab DataSonde and the differences
between the bed and the open water were statistically analyzed. Water samples were
collected from the open water, the water within the T. natans beds and PFU samples
within the T. natans beds. A microbial transect was conducted and specific species were
counted with Palmer-Maloney cells and light microscopy.
V-14 Measurement of nitrate and orthophosphate concentrations
- 3- Nitrate (NO3 ) and orthophosphate (PO4 ) analyses were performed in duplicate
for each water sample collected. The samples were collected in 60 mL acid-washed
bottles. The samples were filtered through a 0.45-µm black polycarbonate membrane
— filter. Nitrate-nitrogen (NO3 N) values were determined with the cadmium reduction
method using a Milton Roy Co. Spectronic 1001, read at 500 nm (Eckblad 1978; Pratt
and Kepner 1992). Orthophosphate was determined with the ascorbic acid method using
a Beckman Coulter DU-640 spectrophotometer with reading taken at 700 and 880 nm,
using a 10-cm path-length cell (Eckblad 1978).
Cyanobacterial counts
Cyanobacterial communities were examined in samples collected using artificial substrates, specifically polyurethane foam units (PFUs) suspended in the near-shore area
of the embayment or tributary-estuary. PFUs were attached to a fishing bobber with
string. The bobber floated above the water’s surface, suspending the PFUs within 5 cm of
the surface of the water. The apparatus was anchored by a cinderblock placed on the
riverbed at each sample site (Figure 4).
V-15 Figure 4. Description of apparatus.
PFUs remained in situ for a period of two weeks. Details of the PFU technique are provided elsewhere (Pratt and Kepner 1992). Triplicate PFUs were kept cool following
collection, returned to the lab, and squeezed out into sterile 250 mL plastic containers.
These samples were homogenized and cyanobacteria were quantified using direct counts
on samples placed in a Palmer-Maloney cell. Cyanobacteria from substrates were
tentatively identified to genus based on morphology, using an Olympus BX-51
microscope with differential interference contrast and commonly employed identification keys (Komárek et al. 1998; Komárek and Anagnostidis 2007; Komárek et al. 2013).
Blue-green chlorophyll levels were measured with a FluoroProbe III (FP, bbe
Moldaenke, GmbH) at certain points to ensure the reliability of the PFU counting methodology.
V-16 RESULTS
The methodology of counting via Palmer-Maloney cell from the water sample collected in the PFU was reliable (Figure 5).
Figure 5. Cyanobacterial cell counts (cells/mL) and blue-green (BG) Chlorophyll (ppb). Trendline is represented by the equation y= 0.0048x + 2.1477 with an R2 value of 0.9016.
There was temporal variability across all sites between July and September in percent
- 3-- coverage of T. natans, DO, pH, turbidity, NO3 -N, PO4 P, surface water temperature and
cyanobacterial cells counts (Figure 6A-E).
V-17
Figure 6. Time-series of abiotic factors at (A) Esopus Creek, (B) Rondout Creek, (C) Wappingers Creek, (D) Port Ewen and (E) Long Dock Park
V-18 Besides Esopus Creek, there was an increase in cyanobacterial cells/mL as the
summer progressed into early fall. An inverse relationship between turbidity and cells/mL
was observed across all sites excluding Long Dock Park. A relationship was observed
— between percent coverage of T. natans and NO3 N concentration at Wappingers Creek
(Table 1).
Table 1. Percent coverage and nitrate concentration at Wappingers Creek.
Date of Sampling Coverage (%) Nitrate (ppm) 07/19/19 51 0.1717 8/14/19 75 0.1025 8/27/19 100 0.084 9/12/19 23 0.0928 10/1/19 10 0.1146
- As percent coverage by T. natans increased throughout the summer, NO3 -N decreased, reaching the lowest value of 0.084 when T. natans reached 100% coverage. When T.
- natans began to decay in the fall, NO3 -N began increasing again (Figure 7).
Figure 7. Nitrate concentration (ppm) and T. natans coverage in Wappingers Creek from July to October.
V-19
Figure 8. Correlation matrix of abiotic factors in relation to cyanobacterial concentrations for all observations (n = 32). Relevant correlation coefficients (r), P-values, and corresponding plots are displayed.
A correlation matrix between cyanobacteria concentration and abiotic factors at all sites was developed using R. DO and pH were negatively correlated with cyanobacterial concentrations (Figure 8). There was a statistically significant negative relationship between cyanobacterial densities and both DO and pH (Figure 8). Similarly, a correlation matrix was created with all abiotic factors in relation to the maximum, minimum, mean and range of cyanobacterial densities between all sites (Figure 9).
V-20
Figure 9. Correlation matrix of abiotic factors in relation to maximum, minimum, mean and range of cells/mL at each site (n=5). Relevant correlation coefficients (r), P-values, and corresponding plots are displayed.
T. natans beds had significantly warmer surface water temperatures, lower pH, lower light availability and lower turbidity in contrast to the open water (Figure 10).
These factors were determined to be statistically significant with P-values of 0.0045,
0.0286, 0.0012 and 0.0161, respectively (p < 0.05). The species of cyanobacteria varied
significantly between each of these sites between the open water and within the beds
(Figure 11). At all sites, Planktothrix was dominant within the T. natans beds and
Microcystis was dominant within open water.
V-21
Figure 10. Difference in abiotic factors between the open water samples and samples within T. natans bed (n=13). Significant P-values are represented.
V-22
Microcystis Planktothrix Other
Figure 11. Distribution of cyanobacterial species (n = 17 to 36).
DISCUSSION
Time series and seasonal growth of T. natans
Cyanobacterial cells increased throughout the summer as the T. natans percent coverage increased. T. natans coverage can be explained by the natural life cycle and seasonal growth. As summer continued, percent coverage and cyanobacterial cells
increased due to the higher surface water temperatures and ideal nutrient availability. As
summer turned to fall, percent coverage began to decrease and T. natans dissembled and
died in the cooler weather.
V-23 Dissolved oxygen
Maximum cells were found in areas of higher surface water temperatures, low
turbidity, low DO and low pH. A statistically significant relationship was observed
between maximum cells/mL and these four abiotic factors. Correlation matrices showed
an inverse relationship between DO and maximum cells/mL with a P-value of 0.011 (p <
0.05). This may be explained by the presence of large T. natans beds, which deplete the
environment of DO (Caraco et al. 2006). Due to the floating-leaf structure of T. natans, oxygen created via photosynthesis is vented to the atmosphere as opposed to in the aquatic system (Hummel and Findlay 2006). The cyanobacterial cells continued to increase throughout the summer because of ideal conditions like warmer surface water temperature, rather than as a result of the levels of DO. Cyanobacterial presence may be enhanced by low DO (Krishnamurthy et al. 1989); however, the inverse relationship of
DO and maximum cells/mL is likely due to timing and the activity of T. natans rather than affecting one another.
Surface temperature, pH, and turbidity
Maximum cells/mL increased throughout the summer because surface water
temperature increased in an already eutrophic and lower-flow environment, creating an
ideal habitat for cyanobacteria. This is best described by the positive feedback loop of
surface accumulation, light absorption, and temperature feedback. When the surface
water is warmer, more cyanobacterial cells accumulate on the surface. Increased light
absorption from photosynthetic activity increases the temperature of the water,
accumulating more cyanobacterial cells (Hense et al. 2013). As pH decreased, cells/mL
V-24 increased, but this is likely due to the relationship between pH and temperature (p =
0.01). There was a statistically significant relationship between turbidity and maximum
cells/mL. Observed sites were deemed high risk; slack-water areas of tributary-estuaries
and nearshore embayments where T. natans and cyanobacterial were most likely to
thrive. In these slack water areas of low turbidity, the water forms stratified layers which
allows increased sunlight permeation and warmer surface temperatures. Low turbidity
allows cyanobacteria to accumulate and absorb light with less mixing.
Cyanobacterial communities and toxicity with influence of T. natans
T. natans is an efficient denitrifier, which explains the lower nitrate and inversely mirrored relationship between percent coverage and nitrate concentration throughout the summer. Community composition greatly differed between the T. natans and the open water of the main-stem Hudson River. Microcystis was dominant in the main-stem waters
of the Hudson River, as opposed to the T. natans beds which were dominated by
Planktothrix, rejecting the hypothesis that bacterial communities would stay relatively
the same within or outside of the beds of T. natans. The main-stem of the Hudson River
is not ideal for cyanobacterial growth, as there is higher turbidity and more mixing. The
lack of stratification makes the water cooler and the constant mixing disturbs proper
accumulation; however, plant available nitrogen (PAN) is likely higher in these locations,
because it is not undergoing denitrification in the T. natans. Microcystis cannot fix
atmospheric nitrogen, which may explain the inability to thrive in T. natans. Planktothrix
resided in the ideal habitat for cyanobacterial growth within the slack waters of T. natans
beds with higher surface water temperatures and low turbidity, despite the lack of PAN
(Figure 12).
V-25
Figure 12. Dominating cyanobacterial species and abiotic drivers within and outside beds of T. natans.
Planktothrix has been shown to tolerate nitrogen depletion and low light
intensities (Kurmayer et al. 2016). It has also been shown to monopolize resources and display high resilience, which may explain the dominance in the ideal conditions of the T. natans beds (Kurmayer et al. 2016). While surface water temperature is warmer in the T. natans, light availability (PAR) is lower. Both Planktothrix and Microcystis can produce microcystin (Chaffina et al. 2018). Previous studies have found that elevated nitrate and high light intensities are the main components for significant microcystin production
(Chaffina et al. 2018; Tonk et al. 2005). While both genera are capable of producing toxins, the low availability of PAN in T. natans beds is creating an unideal environment
for Planktothrix to produce Microcystin. The growth of cyanobacteria in these warm, low
V-26 turbidity environments may be high, but it is less likely that Planktothrix will become a
cyanoHAB in the low nitrogen, low light intensity of the T. natans. Environments with
high nitrate can alter cyanobacterial communities, favoring toxic cyanobacteria like
Microcystis which cannot fix atmospheric nitrogen and are intolerant to nitrogen
depletion (Donald et al. 2011). T. natans are more ideal for specific cyanobacterial
growth but prevent significant toxin production by depleting nitrogen. Microcystis resides
in an environment without the influence of T. natans denitrification where nitrogen is
higher, but temperature is lower. Toxin-production is still low in the open water of the
main-stem of the Hudson River because of the high turbidity and low surface water
temperature which is unideal for cyanobacterial accumulation and growth; however, it is
more likely to produce toxins than the Planktothrix in the T. natans because the main-
stem is not nitrogen depleted and not covered with floating leaves that limit light intensity
(Chaffina et al. 2018; Donald et al. 2011).
Advancing understanding through future research
Future research will be necessary to better understand the relationship between
cyanobacterial growth, community composition, and toxicity in relation to T. natans. The effect of nitrate on Planktothrix and Microcystis has been studied, but needs to be better understood in situ with the influence of biotic factors like T. natans in the Hudson River
Estuary (Chaffina et al. 2018; Peck 2020). A study focusing on the quantitative toxicity as
opposed to potential toxicity would aid in a better understanding of the threat
cyanoHABs pose to the Hudson River.
V-27 Implications for watershed management
The concerns around T. natans initially stemmed from aesthetics and recreational
boating, but it was exacerbated by environmental concerns about dissolved oxygen
depletion. In response, a grassroots movement to physically remove the T. natans across
Hudson River tributary-estuaries and embayments gained popularity. It is possible that T.
natans are mitigating toxin formation by Planktothrix and creating a nitrate deprived
environment that is uninhabitable for Microcystis. While T. natans is invasive and may be causing issues for submerged aquatic vegetation, it is also possible that it is reducing the risk for cyanoHABs in lower-flow extensions of the Hudson River.
V-28 ACKNOWLEDGMENTS
A special thank you to the Tibor T. Polgar Fellowship and the Hudson River
Foundation for an unparalleled summer research opportunity. Thank you to Marist
College for the Student Research Grant from Dr. Thomas Wermuth, the Vice President of
Academic Affairs (VPAA) that funded my research materials. Many thanks to my talented advisor, Lucy Holtsnider, for her incredible vision and unwavering positivity.
Lastly, thank you to my student volunteers, and friends, who made this possible:
Samantha Musso, Lexi Kaminski, Carter Schuh, Colleen Bradley and Gabi DeGennaro.
V-29 REFERENCES
Caraco, N., J. Cole, S. Findlay, and C. Wigand. 2006. Vascular plants as engineers of oxygen in aquatic systems. BioScience 56:219-225.
Chaffina, J.D., T.W., Davis, D.J. Smith, M.M. Baer, and G.J. Dick. 2018. Interactions between nitrogen form, loading rate, and light intensity on Microcystis and Planktothrix growth and microcystin production. Harmful Algae. 73: 84-97.
Cronberg, G., and H. Annadotter. 2006. Manual on aquatic cyanobacteria: a photo guide and synopsis of their toxicology. Intergovernmental Oceanographic Commission, Paris.
Davis, L., E. Hofmann, J. Klinck, M.R. Mulholland, and A. Meza. 2018. Understanding environmental controls on Cochlodinium polykrikoides blooms in Lower Chesapeake [Abstract]. AGU Fall Meeting, Washington DC.
Donald, D.B., M.J. Bogard, K. Finlay, and P.R. Leavitt. 2011. Comparative effects of urea, ammonium, and nitrate of phytoplankton abundance, community composition, and toxicity in hypereutrophic freshwaters. Limnology and Oceanography. 56: 2161-2175
Eckblad, J.W. 1978. Laboratory Manual of Aquatic Biology. William C. Brown Co., Dubuque, IA
Falconer, I.R., and A.R. Humpage. 2001. Preliminary evidence for in vivo tumour initiation by oral administration of extracts of the blue‐green alga Cylindrospermopsis raciborskii containing the toxin cylindrospermopsin. Environmental Toxicology 16:192-195.
Fernald, S.H., N.H. Caraco, and J.J. Cole. 2007. Changes in cyanobacterial dominance following the invasion of the zebra mussel Dreissena polymorpha: long-term results from the Hudson River estuary. Estuaries and Coasts 30:163-170.
Havens, K. 2008. Cyanobacteria blooms: effects on aquatic organisms. Advances in Experimental Medicine and Biology 619:733-747.
Hense, I. 2007. Regulative feedback mechanisms in cyanobacteria-driven systems: A model study. Marine Ecology Progress Series. 339:41-47.
Hense, I., and A. Backmann. 2010. The representation of cyanobacteria life cycle processes in aquatic ecosystem models. Ecological Modeling 221:2330-2338.
Hense, I., Meier, H., Sonntag, S. 2013. Projected climate change impact on Baltic Sea cyanobacteria. Climatic Change 119: 391-406.
V-30
Howarth, B. 2011. The Hudson River is the most heavily nutrient-loaded estuary in the world: should we care? Hudson River Foundation. http://www.hudsonriver.org/download/seminars/Howarth_March11.pdf (Accessed January 10, 2020).
Hummel, M and S. Findlay. 2006. Effects of water chestnut (Trapa natans) beds on water chemistry in the tidal freshwater Hudson River. Hydrobiologia 559:169-181.
Komárek, J. 1998. Cyanoprokaryota, Part 1: Chroococcales. Pages 1-548 in Flora of Central Europe [Translated]. Spektrum Akademischer Verlag, Heidelberg.
Komárek, J., and A. Anagnostidis. 2007. Cyanoprokaryota Part 2: Planktothrixles. Pages 1-759 in Freshwater Flora of Central Europe [Translated]. Spektrum Akademischer Verlag, Heidelberg.
Komárek, J. 2013. Cyanoprokaryota: Part 3 Heterocytous genera. Pages 1-1131 in Freshwater Flora of Central Europe [Translated]. Spektrum Akademischer Verlag, Heidelberg.
Krishnamurthy, T., L. Szafraniec, D.F. Hunt, J. Shabanowitz, J.R. Yates, C.R. Hauer, W.W. Carmichael, O. Skulberg, G.A. Codd, and S. Missler. 1989. Structural characterization of toxic cyclic peptides from blue-green algae by tandem mass spectrometry. Proceedings of the National Academy of Sciences 86:770-774.
Kurmayer, R., L. Deng, and E. Entfellner. 2016. Role of toxic and bioactive secondary metabolites in colonization and bloom formation by filamentous cyanobacteria Planktothrix. Harmful Algae. 54: 69-86.
Lone, Y., R. Koiri, and M. Bhide. 2015. An overview of the toxic effect of potential human carcinogen Microcystin-LR on testis. Toxicology Reports 2:289-296.
New York State Department of Environmental Conservation (NYSDEC). 2018. “Harmful Algae Blooms (HABs) Archive Page” NYSDEC. https://www.dec.ny.gov/docs/water_pdf/habsarchive2018.pdf (accessed January 10, 2019).
Peck, D.H. 2020. The role of nitrogen availability on the dominance of the Planktothrix agardhii in Sandusky Bay, Lake Erie. Master’s Thesis. Bowling Green State University, Bowling Green, OH.
Paerl, H.W. 2014. Mitigating harmful cyanobacterial blooms in a human-and climatically-impacted world. Life 4:988-1012.
Pratt, J.R., and R.L. Kepner. 1992. Collecting aufwuchs on artificial substrata. Protocols in Protozoology 9:1-B9.
V-31
Rodger, H., T. Turnbull, C. Edwards, and G.A. Codd. 1994. Cyanobacterial (blue-green algal) bloom associated pathology in brown trout, Salmo trutta L., in Loch Leven, Scotland. Journal of Fish Diseases 17:177-18.
Salls, W., M. Coffer, B. Schaeffer, J. Darling, and E.A. Urquhart. 2018. Assessing the impact of cyanobacterial harmful algal blooms on drinking water intakes across the United States [Abstract]. AGU Fall Meeting, Washington DC.
Tall, L., N. Caraco, and R. Maranger. 2018. Denitrification hot spots: dominant role of invasive macrophyte Trapa natans in removing nitrogen from a tidal river. Ecological Applications 21:3104-3114.
Tonk, L., P.M. Visser, G. Christiansen, E. Dittmann, E.O.F.M. Snelder, C. Wiedner, L.R. Mur and J. Huisman. 2005. The microcystin composition of the cyanobacterium Planktothrix agardhii changes toward a more toxic variant with increasing light intensity. Applied and Environmental Microbiology. 71: 5177-5181.
V-32 PHARMACEUTICAL TRANSPORT AND TRANSFORMATION IN TRAPA NATANS BEDS
A Final Report of the Tibor T. Polgar Fellowship Program
Cami Plum Polgar Fellow School of Chemistry Monash University Clayton, VIC Australia 3800
Project Advisors:
Stephen Hamilton Cary Institute of Ecosystem Studies, Millbrook, New York, 12545
Rebekah Henry Environmental and Public Health Microbiology Laboratory Department of Civil Engineering, Monash University, Australia
David McCarthy Environmental and Public Health Microbiology Laboratory Department of Civil Engineering, Monash University, Australia
Emma Rosi Cary Institute of Ecosystem Studies, Millbrook, New York, 12545
Michael Grace Water Studies Centre, School of Chemistry, Monash University, Australia
Plum, C.I., S.K. Hamilton, R. Henry, D.T. McCarthy, E.J. Rosi and M.R. Grace. 2021. Pharmaceutical Transport and Transformation in Trapa natans Beds. Section VI: 1-52 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
VI-1
ABSTRACT
Pharmaceutical and pesticide pollution is ubiquitous in the Hudson River due to
anthropogenic inputs including agricultural runoff, wastewater treatment and leaky
sewage infrastructure. Understanding the transport and transformation of pharmaceuticals
and pesticides in the river system is fundamental to gaining insight into how these
contaminants affect aquatic biota and, conversely, how the biota may retain or remove
these contaminants. An invasive, floating aquatic plant known as water chestnut or Trapa
(Trapa natans) is abundant along the freshwater reaches. Trapa beds are microbially
active environments that are key contributors to removal of anthropogenic nitrate, a
process essential to maintaining a healthy functioning ecosystem. This project aimed to
test the hypothesis that Trapa beds are hotspots of pharmaceutical accumulation or
degradation as well as other changes in water quality including nitrate removal.
Concentrations of pharmaceuticals and pesticides, as well as the conservative tracer
sucralose, and several traditional water quality metrics, were measured in water flowing
into and out of Trapa beds at two locations. Preliminary results indicate that the beds
alter water quality, most notably by reducing dissolved oxygen concentrations,
particularly at low tide. Rates of respiration and denitrification were quantified in open
water as well as water, biofilms, and the sediment-water interface within the beds.
Respiration and denitrification was measured at elevated levels within the Trapa beds compared to the adjacent water column. These biological processes explain the decreased nitrate and dissolved oxygen concentrations observed, sometimes resulting in anoxia, within the Trapa beds. Concentrations of a number of pharmaceuticals and pesticides were detectable outside and within the beds. The full analysis of how pharmaceuticals and pesticides interact with Trapa beds awaits completion of laboratory measurements of contaminants.
VI-2
TABLE OF CONTENTS
Abstract ...... VI-2
List of Figures and Tables...... VI-4
Introduction ...... VI-6
Methods...... VI-16
Results ...... VI-25
Conclusions and Recommendations ...... VI-42
Acknowledgements ...... VI-44
References ...... VI-45
Appendices ...... VI-49
VI-3
LIST OF FIGURES AND TABLES
Figure 1 – Trapa bed (left) and Trapa plant diagram showing components and
vertical situation within the bed (right) adapted from Caraco et al.,
2006...... VI-7
Figure 2 – Conceptual framework for this study including the four major
hypotheses under investigation ...... VI-11
Figure 3 – Kingston and Norrie Point sampling locations on the Hudson River VI-17
Figure 4 – Map of Kingston Trapa bed (bright green, annual variability)
identifying the three sampling sites along a transect into the bed as
well as the Rondout Creek and Hudson River sampling sites ...... VI-19
Figure 5 – Map of Norrie Point Trapa bed (bright green) identifying the sampling
sites within and exterior to the bed ...... VI-20
Figure 6 – Comparison of % oxygen saturation for the open water of the Hudson
River and within the Trapa bed at Norrie Point from 22 August to
5 September 2019 ...... VI-29
Figure 7 – Temperature for Hudson River and Trapa bed at Norrie Point from
22 August to 5 September 2019 ...... VI-30
Figure 8 – Comparison of denitrification rates over time in the Norrie Point and
Kingston Trapa beds on the incoming tide for samples collected
between 26 June and 3 September 2019 ...... VI-33
Figure 9 – Comparison of respiration rates over time in the Norrie Point and
Kingston Trapa beds only for samples collected on the incoming tide
for samples collected between 26 June and 3 September 2019 ...... VI-34
VI-4
Table 1 – List of pharmaceuticals of interest in this study ...... VI-15
Table 2 – Mean water quality and nutrient variables ± S.D for six sites measured
in this study and five routinely monitored sites for samples collected
between 18 June and 25 September 2019 ...... VI-27
Table 3 – Pharmaceutical and herbicide detection frequency for 23 compounds
measured in site water from three sites near Kingston and two sites in
Norrie Point by Monash University via SPE-triple quad-LCMS-MS . VI-32
VI-5
INTRODUCTION
Hudson River Estuary
The Hudson River Estuary is a singularly important waterbody for New York and the broader region, providing cultural, economic, and environmental values and services
(Levinton and Waldman 2006). The freshwater, tidal part of the estuary extends from
Troy to Newburgh and is 900 m wide and 8.3 m deep, on average. Strong tidal currents
reverse direction every 6 hours and generally keep the water column in the main channel
well mixed both vertically and laterally, and cause the water level to vary on the order of
about a meter.
Despite a long history of industrial and municipal pollution, the Hudson River
today is relatively clean, and serves as a source of drinking water for local populations.
Nevertheless, excessive loading of nutrients from both point- and non-point sources, as
well as the presence of various trace contaminants in the river system including
pharmaceuticals, polychlorinated biphenyls (PCBs) and polyfluoroalkyl substances
(PFAS), continue to be a source of concern. The latter two contaminants are legacies of
past industrial activity whereas pharmaceuticals come from current wastewater inputs,
and their use and inputs to the river may be increasing. A diversity of pharmaceuticals
and personal care products have been found in the Hudson River system (Cantwell et al.
2017; Carpenter and Helbing 2018). The ecological effects of trace contaminants
including pharmaceuticals, as well as their transport and fate in the Hudson River system,
are not well understood.
VI-6
Trapa natans beds contribute to denitrification in the Hudson River
Aquatic vegetation is essential for maintaining water quality and ecosystem health
and can improve water quality through the removal of nutrients and contaminants.
Dominance by invasive plants may not compromise those ecosystem services in spite of
the other negative impacts (Schlaepfer et al. 2011), and can even enhance services like nutrient and contaminant retention if the invasives are more productive than native species (e.g. Martina et al. 2014).
Figure 1: Trapa bed (left) and Trapa plant diagram showing components and vertical situation within the bed (right) adapted from Caraco et al. 2006.
In the Hudson River between Beacon and Troy, many of the shallow backwaters
along the main channel are densely vegetated with the invasive floating macrophyte
Trapa natans, commonly known as water chestnut (Figure 1). These beds are likely to
VI-7
function as biogeochemical hotspots in the river system. In particular, Trapa beds are
recognized to play an integral part of the Hudson River’s ecology due to their
considerable capacity for denitrification, a microbe-mediated process that converts
bioavailable nitrate to inert nitrogen gas (N2). Denitrification in aquatic systems is vital as high nitrogen loads can lead to eutrophication and detrimental algal bloom events which can have implications for public health, economic returns, fishing and aquatic recreation
(Backer 2012). Previous work has estimated that Trapa beds remove ~25% of nitrogen
loads from the Hudson River annually, despite only being active for three months of the
year (Tall et al. 2011). The high capacity for denitrification in Trapa beds compared to
native submersed aquatic vegetation or open water areas is due to Trapa’s floating leaves,
which favor anoxic conditions by limiting light availability and air-water gas exchange
(i.e. reaeration), combined with the tidal system which continuously removes and replaces oxygen- and nutrient-depleted water.
Pharmaceutical pollution
Pharmaceuticals, an emerging class of environmental contaminants, have been detected in freshwater systems worldwide, usually in the ng/L to μg/L range (Hughes et al. 2012; Kolpin et al. 2002; Murdoch 2015; Scott et al. 2014; Watkinson et al. 2009). In
a recent study, 16 pharmaceuticals were detected at up to 72 sites along a 250-km transect
of the Hudson River (Cantwell et al. 2017). The Hudson River catchment contains 114
wastewater treatment plants (WWTPs) that continuously release treated effluent
containing pharmaceuticals (Carpenter and Helbling 2018; NYCDEP 2019). Some
VI-8
pharmaceutical concentrations in close proximity to a WWTP discharge point at the
mouth of Rondout Creek, which drains into the Hudson River at Kingston, have been
detected in the mg/L range (Carpenter and Helbling 2018). Furthermore, many combined
sewer and stormwater overflows discharge directly into the Hudson River and its tributaries, including Rondout Creek. Increasing incidences of extreme rainfalls and stormwater runoff, as well as flooding of WWTPs and other sewage infrastructure due to sea level rise caused by climate change, can result in uncontrollable releases of large
amounts of pharmaceutical-containing untreated effluent (Azevedo de Almeida and
Mostafavi 2016; Flood and Cahoon 2011). This is of concern as pharmaceuticals have the
demonstrated ability to act on non-target organisms, including environmental microbes
(Rosi-Marshall et al. 2013), potentially disrupting fundamental ecosystem services such
as nutrient removal via denitrification.
Environmental interactions and the ultimate fate of pharmaceuticals are poorly
understood. Pharmaceutical pollution includes a broad range of chemical compounds
exhibiting a variety of properties that could cause disruption of important ecological
processes mediated by microbes and algae (Rosi-Marshall et al. 2013; Richmond et al.
2017). For example, some pharmaceuticals remain in the water column while others
accumulate in sediments or biofilms such as those on underwater plant surfaces (Ferrer et
al. 2004; Scott et al. 2014). Accumulation may be problematic due to toxicity exhibited
by increased concentration in microsites of accumulation, and the increased potential for
movement into the food web via particle-feeding aquatic invertebrates and fishes
(Richmond et al. 2018). Furthermore, degradation of some pharmaceuticals yields toxic by-products (Neuwoehner et al. 2010; Santos et al. 2010), and these may have a greater
VI-9
capacity to interfere with microbial communities involved in processes such as
denitrification. Pharmaceutical transport, transformation, and the short- and long-term impacts of pharmaceutical pollution must be investigated in order to understand how these contaminants affect aquatic biota and the ecosystem services they provide, and conversely, how the aquatic biota may retain or remove these contaminants.
While it is thought that Trapa beds are integral to nitrogen removal throughout the Hudson River, their relationship with pharmaceutical pollution is unknown. Tides cyclically force the movement of large volumes of water containing nutrients and pharmaceuticals into and out of the beds. Pharmaceutical retention or removal may be influenced by the chemical properties of the pharmaceutical compounds, and by interactions with the Trapa roots, particulate matter, microbes on sediment surfaces, and biofilms on underwater plant stems and roots. Pharmaceuticals are expected to be present within Trapa beds, but whether they show net decreases on concentrations and how they affect microbial processes within the beds are currently unknown. If pharmaceuticals are removed in Trapa beds, these beds may help mitigate the impacts of pharmaceutical pollution arising from stormwater and combined sewer overflows and the associated public health risks.
There is some evidence that pharmaceuticals at currently measured environmental concentrations can act negatively on important microbial processes. For example, diphenhydramine (an antihistamine), caffeine (a stimulant), cimetidine (a histamine H2 receptor antagonist) and ciprofloxacin (an antibiotic) have each been demonstrated to suppress biofilm respiration (Rosi-Marshall et al. 2013; Robson et al. 2020). Further, the antibiotic amoxicillin has been shown to decrease denitrification rates (Costanzo et al.
VI-10
2005). In the Hudson River Estuary, the effects of pharmaceuticals are of particular interest in relation to denitrification, the microbial process that enhances water quality and mitigates eutrophication, thereby improving the health of the estuary and coastal waters into which it discharges.
Project aims and hypotheses
This study uses a multidisciplinary design to investigate the interactions among pharmaceutical pollution, microbial communities, denitrification rates, and Trapa beds and the implications for water quality in the Hudson River (Figure 2).
Figure 2: Conceptual framework for this study including the four major hypotheses under investigation.
This study will address four main research questions that relate directly to the hypotheses in Figure 2:
1. How does water quality change when main stem river water enters the Trapa
beds?
As noted above, the dense beds of Trapa are predicted to be biogeochemical hotspots compared to the open waters. The plant’s photosynthesis takes place largely
VI-11
above water, and the aquatic metabolism within the beds is expected to be heterotrophic
overall. In other words, microbial and root respiration within the beds will result in net
consumption of dissolved oxygen and promote anaerobic microbial processes including
denitrification. Anaerobic processes increase with lower dissolved oxygen even if the
oxygen in the water column is only modestly depleted because anaerobic microsites in
biofilms and the sediment-water interface become more important with lower oxygen
concentrations.
Spatial and temporal variation is expected within the Trapa bed, with the degree
of changes in water quality dependent on water residence time with the bed and on the
depth of the water column. Therefore, the largest changes are expected at low tide, in the
centers of the beds or in water exiting the centers, and especially when water depths are
shallow (the underwater parts of Trapa expand and contract vertically with changes in
depth).
2. Do Trapa beds remove pharmaceuticals in the Hudson River?
The tidal dynamics of the Hudson River drive the continuous exchange of water
and nutrients between open waters and the Trapa beds. This study compared the
concentrations of pharmaceuticals and their degradation products in water flowing into
and out of Trapa beds over tidal cycles by sampling along transects from open water into the beds at various points in the tidal cycle and across the Trapa growing season (Figure
2). In addition to pharmaceuticals, sucralose (artificial sweetener) was measured as a
conservative tracer to estimate net changes in reactive pharmaceutical concentrations in
the face of uncertain mixing and dilution of waters within the beds. The utility of
sucralose as a conservative tracer of wastewater effluent due to its resistance to
VI-12
degradation has been shown by Soh et al. (2011) and Cantwell et al. (2017), the latter study conducted in the Hudson River.
Decreases in the ratios of a particular pharmaceutical to sucralose between inflows and outflows will reveal the net retention or removal of the pharmaceutical during residence of water within the beds. If degradation products increase in concentration relative to sucralose, their potential effects on the biota will also be considered.
As the Trapa beds are a microbially rich and biologically active hotspot, it is hypothesized that they will remove pharmaceuticals through adsorption and/or degradation. Pharmaceuticals with a lower solubility in water (i.e. greater hydrophobicity) are expected to have stronger particle interactions and partition with organic matter, allowing bioaccumulation in sediments or biofilms or uptake by Trapa roots, resulting in a decreased concentration in the water column. Accordingly, it is expected that lower concentrations of pharmaceuticals, and perhaps higher concentrations of degradation products, will be present in water well within or exiting the bed (i.e., on the outgoing tide). A gradient of pharmaceutical exposure from the main channel into the beds may then alter microbial response and perhaps resistance within the Trapa bed.
Investigating whether Trapa beds remove pharmaceuticals will lay the groundwork for future research to understand how pharmaceuticals affect aquatic biota, and in particular, whether they may affect the rate of microbial denitrification.
This research will provide insights into the role of Trapa beds in pharmaceutical effects and fate in the Hudson River system and will provide the basis for investigating the role of pharmaceuticals influencing microbial community structure (question 3
VI-13
below) and denitrification rates (question 4). This study has been designed and carried
out with sufficient sampling to address the four key hypotheses described in Figure 2;
however, this report will focus on water quality with some brief discussion of
pharmaceuticals. Preliminary rates of denitrification are also presented here. Further
results pertaining to the remaining hypotheses await completion of additional laboratory
analyses. Nevertheless, the remaining questions to be further addressed are briefly
described here:
3. Do pharmaceuticals in Trapa beds influence microbial communities?
Characterization of the microbial community composition is necessary to
determine whether the presence of pharmaceuticals affects diversity and function.
Microbial diversity is essential for an ecosystem’s ability to be resilient in the face of environmental stresses and change.
The Trapa beds within the Hudson River are a hydrodynamically variable environment, with water levels and water quality varying rapidly in space and time over a tidal cycle. For example, a 60% increase in dissolved oxygen concentrations in less than
20 minutes on the turning tide was measured. Accordingly, conditions for microbes was expected to change quickly. Suspended microbial communities from the main channel enter Trapa beds through tidal water movement, but such transient communities would not be adapted to the periodically hypoxic and nutrient-rich conditions in the beds, in contrast to those found in sediment and biofilms within the beds that must be able to survive and function in the highly variable environment. In addition, characterization of the endogenous microbial community will allow further investigation into the effects of pharmaceuticals on individual denitrifying species. In a separate study, this
VI-14 characterization is being achieved by DNA extraction from water, biofilm and plant material for 16S rRNA gene sequencing and whole genome metagenomics.
It is predicted that greater exposure and thus greater microbial resistance to pharmaceuticals will be observed on the outer edges of the Trapa bed compared to the interior. Additionally, a greater diversity of denitrifying microbes is expected in the interior of the Trapa bed where dissolved oxygen, labile carbon, and nitrate concentrations are all subject to greater variability.
4. Do pharmaceuticals in Trapa beds influence the microbe-mediated
biogeochemical process of denitrification?
The dose-response relationship of denitrifying bacteria exposed to commonly detected pharmaceuticals has been investigated. Water incubations were dosed with a cocktail of commonly detected pharmaceuticals at environmentally relevant concentrations (Table 1) in order to measure any effects on denitrification rates.
Additionally, comparison of two sites with different influence of wastewater will allow assessment across an impact gradient of pharmaceutical pollution.
Table 1: List of pharmaceuticals of interest in this study
Drug Class Final concentration ng/L Acetaminophen Pain killer 350 Atenolol Beta blocker 1,000 Caffeine Stimulant 2,500 Cimetidine H2 antagonist 250 Ciprofloxacin HCl Antibiotic 200 Diphenhydramin HCl Antihistamine 1,000 Ibuprofen NaCl Anti-inflammatory 3,000 Metformin Antidiabetic 100,000 Ranitidine HCl H2 antagonist 1,000
VI-15
Respiration was measured simultaneously with denitrification, and microbial communities are being extracted for genomic analysis to explore microbial interactions with pharmaceuticals.
As in the case of microbial community effects, it is predicted that pharmaceuticals will have greater impact on respiration and denitrification on the outer edges of the Trapa bed because of the gradient of pharmaceutical exposure and thus microbial response. It is possible that some denitrifying species are susceptible, but that denitrification is maintained by resistant species. Alternately, competitors that assimilate nitrate may be susceptible, resulting in more nitrate availability for growth of denitrifying species.
METHODS
Experimental Overview
A variety of variables were sampled from each location up to 27 times between 18
June 2019 and 25 September 2019, over various points in the tidal cycle on multiple dates to account for short-term and seasonal variation. Not all variables were measured on every trip. Parameters from the first and final sampling trips, collected on 4 June and 23
October respectively, were not included as no Trapa was present.
VI-16
Site Descriptions
Two areas containing Trapa beds were chosen for investigation in this study, and
are denoted as Kingston and Norrie Point (Figure 3). These sites were selected because they differ in nearby land use and receive different amounts of wastewater effluent, and both have long-term nutrient monitoring data.
Figure 3: Kingston and Norrie Point sampling locations on the Hudson River
VI-17
The Kingston Trapa bed lies south of the Rondout Creek. This bed receives
Kingston WWTP effluent released into the vicinity of the mouth of Rondout Creek.
Three locations along a transect into the center of the Trapa bed were sampled (Figure 4):
• KBA (41°55'10.7"N 73°58'01.9”W) is a site adjacent to the Trapa bed in
which water flows from the Rondout Creek to the Hudson River,
• KBB (41°55'09.6"N 73°58'05.3”W) is located 80 m west of KBA and lies 45
m into the Trapa bed, and
• KBC (41°55'08.5"N 73°58'08.0”W) is the furthest transect point and is
situated 110 m into the Trapa bed and 150 m east of KBA.
Two river monitoring sites with long-term data were also sampled:
• Rondout Creek (RC, 41°55'04.7"N 73°58'55.7"W) in front of Kingston Public
Dock, which is the site of a combined sewer and stormwater outfall. Water
quality data collected by RiverKeeper are available for this location (described
in section 3.2).
• Kingston Hudson River (KHR, 41°56'12.1"N 73°57'38.2”W), which is located
in the main stem of the Hudson River was included in routine sampling. Long
term data for this site are available through the Long Term Research in
Environmental Biology (LTREB) project led by Cary scientists and funded by
the National Science Foundation.
VI-18
Figure 4: Map of Kingston Trapa bed (bright green, annual variability) identifying the three sampling sites along a transect into the bed as well as the Rondout Creek and Hudson River sampling sites. The WWTP effluent is discharged along the north bank of Rondout Creek. Norrie Point was chosen for sampling as it is well characterised from previous studies, easily accessible and routinely monitored by the Hudson River Environmental
Conditions Observing System (HRECOS) station maintained by the Hudson River
National Estuary Research Reserve (HRNERR). A site within the bed (NB,
41°49'54.3"N 73°56'29.9"W) and another in the open water of the Hudson River (NHR,
41°49'54.7"N 73°56'33.6”W) were sampled. NHR is the site of the HRECOS monitoring
station, which is further described in section 3.2 below.
VI-19
Figure 5: Map of Norrie Point Trapa bed (bright green) identifying the sampling sites within and exterior to the bed.
Water quality
Traditional water quality metrics
Temperature and dissolved oxygen (mg/L) were measured in situ using a pre-
calibrated ProODO probe (YSI). Electric conductivity (EC) was measured in situ using
an EC300 probe (YSI).
Depth-integrated samples of the water column were collected using a pump and transported to the laboratory on ice. Fluorescence (Relative Fluorescence Units, RFU) was measured for each sample using a Trilogy fluorometer (Turner designs, model #
VI-20
7200-000), then converted to Nephelometric Turbidity Units (NTU) using a calibration curve constructed from serial dilutions of YSI 6073G Turbidity Standard (126 NTU).
In addition, long-term water quality monitoring data for the main river were obtained from the following sources:
• The New York State Department of Environmental Conservation (NYSDEC)
Hudson River National Estuarine Research Reserve (HRNERR) monitoring
programs including the meteorologic and hydrologic station at Norrie Point
operated as part of the Hudson River Environmental Conditions Observing
System (HRECOS). Long-term data include measurements at 15-minute
intervals of temperature, wind, precipitation, pH, turbidity, chlorophyll
fluorescence, dissolved oxygen etc. Data are available at
https://ny.water.usgs.gov/maps/hrecos/ .
• The Long Term Research in Environmental Biology (LTREB) site at
Kingston, which Cary scientists have monitored bi-weekly since 2000. Long-
term monitoring of water chemistry, nutrient concentrations and
phytoplankton and zooplankton have contributed to assessing the ecological
effects of invasive zebra mussels (Dreissena polymorpha) (Strayer et al. 2006;
Strayer et al. 2008; Fernald et al. 2007; Findlay et al. 2006; Caraco et al. 2000;
Strayer et al. 2014).
• The non-profit environmental organization RiverKeeper makes monthly
observations at 74 sites between the Mohawk River and Gowanus Canal.
RiverKeeper provides counts of Enterococcus, a bacterium that is an EPA-
approved indicator of faecal contamination, as well as measurements of water
VI-21
temperature, turbidity, oxygen etc. This information is useful for beach-goers
or others who interact with the river recreationally and is available at
https://www.riverkeeper.org/.
Nutrient concentrations
Depth-integrated samples of the water column were collected using a pump and
transported to the laboratory on ice. Samples for the analysis of ammonium, filterable
reactive phosphorus, nitrate + nitrite (APN), and non-purgeable organic carbon (NPOC)
were filtered using a 0.2-µm syringe membrane filter (Sartorius). Samples for total
phosphorus and total nitrogen (TP and TN) remained unfiltered. DOC samples were
preserved with 300 µL of 2 M H2SO4 and TP, TN and APN samples were preserved with
500 µL of 2 M H2SO4 and all were stored at room temperature until analysis. NPOC was
analysed on a Total Organic Carbon Analyzer (Shimadzu, model #TOC-VCSH).
Diel variability in dissolved oxygen
To provide an indication of diel changes in dissolved oxygen, which are indicative of the balance between aquatic photosynthesis and respiration, miniDOT oxygen loggers (PME) were deployed from 27 August to 5 September 2019. The miniDOTs were attached to the inside of the roof of a lobster cage filled with bricks and set in the Trapa bed such that the loggers were situated approximately 30 cm above the bottom. Temperature and oxygen measurements were recorded every 5 minutes. These data were compared to the HRECOS oxygen data at the open water off Norrie Point, for the same dates, which were logged every 15 minutes.
VI-22
Tide information was obtained from Tides & Currents Pro (Nobeltec version
3.5.107) using Kingston data based on New York, The Battery (NOAA; 41°55’06N,
073°59’00W, Station ID 1337) and Norrie Point based on New York, The Battery
(NOAA; 41°47’N 073°57’W, Station ID:1335).
Pharmaceuticals
Depth-integrated samples of the water column were collected using a pump and transported to the laboratory on ice in clean 1-L Nalgene bottles. Samples were prefiltered by gravity using 1 µm filter paper, then the pharmaceuticals were pre- concentrated on Oasis HLB 6cc 200mg extraction cartridges (preconditioned with 5 mL of HPLC grade methanol then 5 mL of de-ionised water (Brodin et al. 2013) at a flow rate of 5 mL/min. Cartridges were then refrigerated in the dark to prevent UV degradation. Samples were transported to Monash University, Melbourne, Australia and analysed on a triple quadrupole liquid chromatograph mass spectrometer (SPE-triple quad-LCMS-MS, Agilent) with a 1 ng/L limit of detection. Pharmaceuticals analysed include atenolol, atropine, caffeine, carbamazepine, cetrizine, ciprofloxacin, diclofenac, diphenylhydramine, enrofloxacin, flucanozole, flunixin, fluoxetine, gemfibrozil, imidacloprid, meloxicam, metformin, oxytocin, pimobendan, ranitidine, sertraline, sulfadoxine, sulfamethaxozole, telmisartan, triclosan, trimethoprim and xylazine. In addition, herbicides that were measured include 2,4-dichlorophenoxyacetic acid (a.k.a.
2,4-D), atrazine, DCMU, dicamba, MCPA and triclopyr.
VI-23
Denitrification and respiration
Water incubations to indicate rates
Depth-integrated samples of the water column were collected in a bucket using a pump with its outlet submerged to avoid reaeration. 12.5 mL glass vials with septa
(Exetainers) were filled and capped underwater to exclude air bubbles. Immediately, biological activity in a set of triplicate samples was terminated by adding 0.2 mL ZnCl2
(50% w/v) using a purge needle to eliminate over-pressurization. This time was recorded as T0. All remaining samples were incubated in the dark to preclude photosynthetic activity, at ambient temperature (in a cooler filled with site water). At various time intervals, triplicate control and dosed samples were terminated and stored inverted in the dark until the time of analysis.
This was repeated at each site before and after low tide. Later, slurries of plant biofilm material and sediment were incubated under the same protocol; however, only the results for water collected on the outgoing (ebb) tide, which best reflects conditions within the beds, are discussed in this report.
Membrane Inlet Mass Spectrometry (MIMS) Analysis
Terminated water samples were analyzed for dissolved N2:Ar and O2:Ar ratios at
20°C on a MIMS. Sample temperatures were maintained at 20 °C. Atmospheric pressure
VI-24
was recorded at the time of each sample. A recirculating distilled water bath at 20°C was
used as a standard for air-equilibrated dissolved gas concentrations and measured every
12 samples to correct for any instrumental drift (Reisinger et al. 2016). MIMS readings
were averaged over a 30-second time period for each sample and converted to mg/L. The
“Linest” function in Excel was used to perform a regression of N2 and O2 over time, the
slope of which was taken as the rate (denitrification as µg N2/L-hr produced and
respiration as µg O2/L-hr consumed, respectively).
RESULTS
Water Quality
Average in situ water quality variables (Table 2), except dissolved oxygen, show little variation among the sampling sites at Kingston. Temperature was in a similar range
(24 ± 3 °C) at all Kingston sites, except for those measured by RiverKeeper (KPD-RK
and RPB-RK) which were higher. Conductivity was highest at RC (360 µS/cm) and
decreased from 316 to 296 µS/cm along the transect into the Trapa bed; however, these
differences are not statistically significant. Conductivity was lowest at the K-LTREB site
(254 uS/cm). Turbidity was similar and low (mean ranging from 10 to 19 NTU) at all
Kingston sites.
Dissolved oxygen differed between the open channel and the Trapa beds. On
average, % oxygen saturation was >100% at all open channel sites. The oxygen measured
next to the Trapa bed was similar to that measured in Rondout Creek (97 ± 7% and 98 ±
VI-25
7%, respectively). Oxygen was lower and more variable within the Trapa bed (KBB= 81
± 14% and KBC=92 ± 15%). The minimum oxygen concentration measured in the bed
(47% DO; Appendix A) is considerably lower than the minimum dissolved oxygen measured in Rondout Creek by RiverKeeper or in the main river by LTREB (86% and
95% saturation, respectively).
Temperature and conductivity were similar between sites within the Trapa bed and in the nearby open channel. Turbidity was lower and less variable in the river compared to in the Trapa bed. Temperature was higher in the Norrie Point main channel
(NC-RK 30 ± 2 °C) compared with N-HRECOS and measurements taken in this study
(25 ± 2 °C).
Temperature and turbidity were similar between Kingston and Norrie Point (Table
2). Conductivity was slightly lower at Norrie Point than Kingston, but similar to K-
LTREB. Both Kingston and Norrie Point showed lower % oxygen saturation in the Trapa
bed compared to the nearby open channel; however, within the beds Norrie Point was
found to have a lower average oxygen concentrations (69%) than at Kingston (81% and
92%).
Ammonium, nitrate, phosphate, total nitrogen and total phosphorus concentrations
measured at three sites within the Kingston Trapa bed were lower than concentrations
measured in Rondout Creek. Nitrate concentrations decreased leading further into the
bed, supporting the denitrification capacity of Trapa beds. Despite variation in tidal
stages, results showed less variability than expected. All nutrients had lower
concentrations at the Norrie Point sites compared to Kingston, perhaps reflecting more
local sources in the vicinity of Kingston, including the nearby WWTP.
VI-26
Table 2: Mean water quality and nutrient variables ± S.D. for six sites measured in this study (Kingston - KBA, KBB, KBC & RC; Norrie Point - NHR & NB) and five routinely monitored sites (KLTREB = Kingston LTREB site, KPD-RK= RiverKeeper Kingston Public Dock, PBO-RK=RiverKeeper Kingston Public Beach, N-HRECOS= HRECOS monitor at Norrie Point & NC-RK= RiverKeeper Norrie Mid Channel) for samples collected between 4 June and 25 September 2019 . Summary statistics for each site are presented in Appendix I. Asterisks (*) denoted measurement unit of formazin nephelometric units (FNU).
VI-27
Diel variability in dissolved oxygen
Continuous measurements of dissolved oxygen over nearly two weeks showed concentrations to be lower overall in the Trapa bed (24-hour mean, 44% oxygen saturation) than in the open channel (average 89%), and the 24-hour means are lower than the daytime samples reported above (69% for the Trapa bed and 92% for the open channel, Table 2).
Over the diel cycle, dissolved oxygen was much more variable in the Trapa bed compared to the open channel, ranging from 9% to 89 % saturation within the bed while
the open channel ranged from 76% to 104% saturation. Norrie Point data are shown in
Figure 6.
On four occasions during low tide the miniDOT oxygen logger was exposed to
air, so the minimum % oxygen saturation reported here is conservative and is likely to
have decreased further had it been measurable at the lowest tides. This is supported by
field measurements as indicated in Appendix B where the lowest concentration measured
at site NB was 5% oxygen saturation.
VI-28
120 5
100 4
80 3 % saturation) % 60 2
40 1 (ft) Depth
20 0
Dissolved oxygen ( 0 -1 21-Aug 23-Aug 25-Aug 27-Aug 29-Aug 31-Aug 2-Sep 4-Sep 6-Sep 8-Sep Time
HRECOS NB Tide
Figure 6: Comparison of % oxygen saturation for the open water of the Hudson River and within the Trapa bed at Norrie Point from 22 August to 5 September 2019. NB denotes Norrie Point data collected in this study and HRECOS refers to the Norrie Point monitoring station. Maximum and minimum water depths for each tidal cycle are also shown
Temperatures within the open channel and Trapa bed were similar (Figure 7). The mean water temperature in the bed (25.0 ± 0.9°C) was similar to that in the river (25.4 ±
0.6°C). The mean water temperatures within this two-week period are close to the mean temperature across the entire sampling period (25 ± 2°C); however, higher temperature fluctuations were found within the bed, with a range of 4.7°C compared to the river with a range of 2.4 °C.
VI-29
29 5
28 4
C) 27 ° 3 26 2 25 Depth (ft) Depth 1
Temperature ( 24
23 0
22 -1 21-Aug 23-Aug 25-Aug 27-Aug 29-Aug 31-Aug 2-Sep 4-Sep 6-Sep 8-Sep Time
HRECOS NB Tide
Figure 7: Temperature for Hudson River and Trapa bed at Norrie Point from 22 August to 5 September 2019 where NB denotes data collected in this study and HRECOS refers to long term Norrie Point data. Temperature for HRECOS and NB is on the left y-axis and depth above mean tide is on the right y-axis.
Pharmaceuticals
At the time of this report, 41 of 153 water samples have been analysed for pharmaceutical concentrations, and only preliminary data are available because some additional post-processing is needed to ensure accuracy. Out of 33 compounds analysed,
23 were detected, as outlined in Table 3. Eight compounds (2,4-dichlorophenoxyacetic acid, atenolol, atrazine, caffeine, cetirizine, diphenhydramine, metformin and sulfamethoxazole) were detected in all samples collected from both Kingston and Norrie
Point. For samples collected from Norrie Point, 12 compounds were detected in all samples. Five herbicides (all except atrazine) and one antihistamine (diphenhydramine) detected at Norrie Point had concentrations >10% than those detected at Kingston. 12
VI-30 compounds had concentrations >10% higher at Kingston than at Norrie Point.
Concentrations detected are being correlated with tidal cycles to determine any changes during residence of water in the Trapa bed. The eight compounds analysed but not detected at any site were atropine, diclofenac, enrofloxacin, flunixin, imidacloprid, oxytocin, pimobendan, sertraline, xylazine and meloxicam.
VI-31
Table 3: Pharmaceutical and herbicide detection frequency for 23 compounds measured in site water from three sites near Kingston and two sites in Norrie Point by Monash University via SPE-triple quad-LCMS-MS. Asterisks (*) denote mean concentration >10% higher at Norrie Point. Daggers (†) denote concentrations were >10% higher at Kingston.
VI-32
Denitrification and Respiration
Denitrification and respiration rates in the Trapa bed and open channel at both
Kingston and Norrie Point were measured on three dates throughout the sampling period.
Both sites had very low denitrification (Figure 8) and respiration (Figure 9) rates for the first two samplings, not detectably different from zero, as indicated by error bars crossing over zero on the y-axis.
160 140 120
g/L/hr 100 µ 80 60 40 production production
2 20 N 0 -2014-Jun 24-Jun 4-Jul 14-Jul 24-Jul 3-Aug 13-Aug 23-Aug 2-Sep 12-Sep Time
NB KBB
Figure 8: Comparison of denitrification rates over time in the Norrie Point and Kingston Trapa beds on the incoming tide for samples collected between 26 June and 3 September 2019. Error bars depict uncertainty in the estimate of the slope of the change over time.
VI-33
4000 3500 3000
g/L/hr) 2500 2000 1500 1000 consumption ( µ consumption
2 500 O 0 14-Jun 24-Jun 4-Jul 14-Jul 24-Jul 3-Aug 13-Aug 23-Aug 2-Sep 12-Sep -500 Time
NB KBB
Figure 9: Comparison of respiration rates over time in the Norrie Point and Kingston Trapa beds only for samples collected on the incoming tide for samples collected between 26 June and 3 September 2019. Rates were obtained using the Linest function and error bars depict uncertainty in the estimate of the slope of the change over time.
DISCUSSION
Findings from this study highlight changes in water quality and pharmaceutical
concentrations between the Trapa beds and the open water of the Hudson River. The
documented changes over time scales ranging from diel cycles to the seasonal cycle of
growth and senescence of Trapa. In addition, the study also reveals differences between the Kingston and Norrie Point sites. Differences between the Trapa bed and the open channel are evident in traditional metrics of water quality, particularly dissolved oxygen, but also by pharmaceutical concentrations and rates of respiration and denitrification.
Furthermore, dissolved oxygen and denitrification show evidence for change over the seasonal cycle. This high spatiotemporal variability highlights the need to account for the phenological stage of Trapa as well as tidal cycles, time of day, complex hydrodynamics,
VI-34 and spatial variability to fully comprehend how these beds affect ecosystem processes and are affected by emerging contaminants like pharmaceuticals.
Comparison with data from long term monitoring programs
Water quality data collected in this study are broadly consistent with data from other sampling at Kingston, K-LTREB and KBA, indicating that the open channel next to the Trapa bed is well mixed and that these long-term stations are reasonably representative of water quality and nutrient concentrations in open water adjacent to the
Trapa beds sampled. NHR and HRECOS temperature and oxygen data are in particularly good agreement with measurements in this study. Higher temperatures and dissolved oxygen concentrations in the RiverKeeper data are likely explained by their relatively infrequent sampling, occurring during daylight hours and not accounting for tidal variations.
Variability over a tidal cycle
Continuous measurements with miniDOTs show variability in dissolved oxygen in the Trapa beds over a tidal cycle. As hypothesized, dissolved oxygen in the Trapa bed was always lower than that in the open waters of the river, indicating an excess of respiration over photosynthesis in the beds; however, during the period of 23 August to 8
September 2019, the dissolved oxygen in the bed was highest and most similar to the dissolved oxygen concentrations in the Hudson River during the daytime high tide, when depth of water in the beds exceeded 3.6 feet. High tides decrease density of plant mass,
VI-35 allowing for greater water movement and mixing of water within the Trapa bed, allowing more reaeration and diluting the effects of biological oxygen production or consumption within the bed on oxygen concentrations. Nevertheless, even at the end of the incoming tide, water that has flooded into the bed would have mixed with low-oxygen water that was retained over low tide, resulting in a lower concentration of dissolved oxygen compared to the main channel.
Dissolved oxygen in the Trapa beds also shows diel patterns with higher concentrations in the daytime, reflecting the photosynthetic production of oxygen, but overall the beds showed net oxygen consumption. Although there is the potential for photosynthesis by underwater parts of the Trapa as well as by algae and duckweed within the beds, photosynthesis is likely subject to light limitation in most of the water column.
Root respiration by the Trapa as well as high labile carbon availability from senescing plant parts likely explain the overall excess of respiration over photosynthesis.
Compared to the open water of the river, dissolved oxygen in the Trapa bed was more variable, ranging from 9-89% saturation, compared to the river water’s range of 76-
104%. It is clear that fishes and invertebrates in Trapa beds can experience periodic hypoxic stress given that dissolved oxygen concentrations of < 20-30% saturation are potentially stressful to aquatic biota (Rao et al. 2014). Previous studies show that Trapa beds are a hot spot for biodiversity (Kornijów et al. 2010) supporting diverse invertebrate communities. During this study, a diversity of fauna was observed both in the water column and inhabiting the submerged and emergent portions of the Trapa plants, including fish and invertebrates, showing that despite periodic low dissolved oxygen the
VI-36
Trapa bed is vital for supporting aquatic biota (Kornijów et al. 2010; Findlay et al. 1989;
Yozzo and Odum 1993).
Dissolved oxygen concentrations within the Trapa bed only exceeded 100% saturation for a 90-min period on 22 August (approximately halfway through the incoming tide) and only 1 instance (one data point) on 23 August (approximately halfway through the outgoing tide).
While biogeochemical processes in the Trapa bed may influence the water in the main channel of the Hudson River, it appears that physical processes are effective in re- oxygenating water from the Trapa bed and mixing it with main channel water, which rapidly increases dissolved oxygen concentrations. Thus, hypoxia would be an acute stress that may make it more readily tolerable or avoidable by fauna within the beds.
The overall mean dissolved oxygen concentrations over the two-week miniDOT sampling period (27 August to 5 September 2019) were lower than averages for the entire sampling period (18 June to 25 September 2019). Towards the end of summer, Trapa beds were lower in oxygen, a period corresponding with higher denitrification rates, as expected since anoxic conditions favour denitrification. This may also be a period of enhanced microbial activity as Trapa plants mature and begin to senesce.
Mean water temperatures in the open water of the river over the course of the sampling period were similar at both sites (NHRECOS = 25.4 ± 0.6 °C, NB = 25.0 ± 0.9
°C). Temperature was more variable in the Trapa bed compared to the open water. The open channel of the river is deeper and has a greater thermal mass and is thus not strongly influenced by variability in air temperature. The shallower depth and more protected
VI-37
waters within the bed resulted in greater diel temperature fluctuation compared to the
open waters of the river. Temperature changes are driven more by air temperature
changes.
Although turbidity was similar at all sites, some tidally driven variability was
observed. Suspended particulate matter in the Trapa bed increased on the incoming tide
when particulates that had settled on the benthos and plant material were disturbed by
increased flow. These particulates quickly settled out, but this observation shows the
effectiveness of mixing by tidal water incursions within the bed. On the outgoing tide, a
decrease in water volume resulted in an increased density of plant matter within the water
column, resulting in restricted flow that further trapped suspended material.
Available forms of nitrogen and phosphorus were above critical limiting levels for
aquatic primary production in all sampling sites (Table 2) (Wurtsbaugh at al. 2019;
Guildford and Hecky 2000), consistent with previous studies of Hudson River
environments (e.g., Lampman et al. 1999) and suggesting that light and mixing are more limiting factors for algal and submersed plant growth even in the more stagnant interstitial waters of the Trapa beds. All nutrient concentrations were higher in RC
compared to Trapa bed sites. However, nutrient concentrations were similar within the
three sites in the Kingston bed: KBA, KBB and KBC.
Change over the season
Lower concentrations of dissolved oxygen were recorded at low tide as the season
progressed. The development of more extreme oxygen depletion towards September (e.g.
VI-38
5% saturation measured on 5 September 2019) is expected due to increasing plant growth
as the summer progressed, which limits 95% of light availability to the Trapa bed due to
leaf cover (Tall et al. 2011). In addition, increased growth would encourage greater
microbial activity within the Trapa beds as biofilms develop on plant surfaces and higher
plant densities trap incoming nutrients and particulates, and the plants begin to senesce
soon after seed production.
Lower dissolved oxygen stimulates denitrification, which takes place under
anoxic conditions that would become more prevalent in microsites like biofilms as the water column oxygen concentration decreases. Linear correlation between denitrification and dissolved oxygen has previously been reported (Tall et al. 2011). Only Trapa bed samples collected on the ebb tide are presented in this report. Denitrification and respiration rates were negligible at Norrie Point in 26 June (when there was not a lot of plant biomass or biofilm) but measurable in early August and increased significantly on
22 August.
Site comparison between Kingston and Norrie Point
Dissolved oxygen concentrations and conductivity differed between the sites.
Norrie Point had lower dissolved oxygen saturation (69 ± 43%) compared to Kingston
(81 ± 14%). This may be due to flow patterns and plant density; the bed at Norrie Point is more protected while Kingston has adjacent channels flowing to and from Rondout
Creek. Additionally, the Kingston bed was somewhat sparser even though it was larger.
Differences in water quality between sites may further be influenced by sediment type; the sediments beneath the Norrie Point bed are softer and siltier than those at the
VI-39
Kingston bed. Release of large volumes of sediment gas bubbles (presumably mostly composed of methane) when walking through soft sediment was consistently observed at
Norrie Point, but not at Kingston, supporting the observation of higher rates of microbial
processes at Norrie Point. For example, respiration measured at Norrie Point (3015 ± 570
g O2 consumed/L/hr) was higher than measured at Kingston (873 ± 241 g O2
consumed/L/hr) (Figure 9). Lower dissolved oxygen may be indicative of more plant
activity (as respiration by Trapa roots would contribute to oxygen depletion) and/or
higher microbial activity resulting in increased respiration in the water column.
Conductivity was lower at Norrie Point, which may be indicative of lower wastewater
influence (de Sousa et al. 2014).
Pharmaceutical types and concentrations varied between sites. Finding
pharmaceuticals at concentrations above the detection limit of 1 ng/L is not surprising
given that pharmaceuticals are known to be present throughout the 250 km of the Hudson
River Estuary (Carpenter and Helbling 2018; Cantwell et al. 2017). Additionally, the
WWTP outlet into Rondout Creek is a known pharmaceutical input into the Hudson
River (Carpenter and Helbling 2018).
Pharmaceutical analysis reflects variation in input sources and land-use.
Generally, Kingston had higher concentrations of pharmaceuticals and Norrie Point had
higher concentrations of pesticides. The mean concentration of all herbicides was higher
at Norrie Point than at Kingston while the mean concentration of all pharmaceuticals, except diphenhydramine, was higher at Kingston (when comparing average concentrations at all sites including in and out of the bed; further analysis needed in order to compare sites). This is not surprising given the differing land use in the vicinity. Norrie
VI-40
Point is surrounded by maintained parkland while Kingston is more urbanized and has a
known wastewater and combined stormwater/sewer overflow outfall. Further quality
analysis of the analytical results is required before determining if the mean differences are statistically significant.
The sampling design of this study was such that it incorporated an array of dates
and tidal periods to compare concentrations in the river with those in the Trapa beds and
explore the net effect of the beds on pharmaceutical concentrations over a tidal cycle;
however, at the time of this report only 41 of 153 samples have been analysed. Further
analysis accounting for tidal interactions is required to fully understand how the beds
affect pharmaceutical concentrations. In addition to the pharmaceutical samples described above, Polar Organic Chemical Integrative Samplers (POCIS) were deployed at Norrie
Point for the same period as the miniDOTs. The POCIS data, which are not yet available, will show any change in pharmaceutical concentrations across four sites at Norrie Point over the two week deployment. By showing the net effect of the Trapa bed, this method will account for complex hydrodynamics. Differences in caffeine and carbamazepine across sites show that these relatively conservative compounds may be useful for comparisons within a site but not between sites. Analysis of sucralose, a better conservative tracer of wastewater due to low degradability (Cantwell et al. 2017; Soh et al. 2011; Torres et al. 2011), in the water samples will assist in understanding changes in pharmaceutical concentrations over a tidal cycle. This, together with the pending microbiological assays, will help to answer further questions such as
• Do pharmaceutical concentrations decrease in Trapa bed over tidal cycle?
• Can Trapa be used for pharmaceutical remediation?
VI-41
• How do dissolved oxygen and pharmaceuticals interact to affect microbial
activity and biogeochemical processes in Trapa beds?
Preliminary data show that carbamazepine was stable within the Trapa bed, suggesting that the fractional wastewater influence was similar; however, water samples collected on 18 June show that diphenhydramine (an antihistamine) decreased in concentration during water residence within the Trapa bed at both KBB and KBC sites.
While the mode of action is unknown, it is interesting as diphenhydramine has been shown to negatively affect environmental bacteria by decreasing biofilm respiration
(Rosi-Marshall et al. 2013). The decrease in concentration of diphenhydramine indicates potential pharmaceutical removal by Trapa beds, although this conclusion is preliminary since further post-processing of the pharmaceutical measurements is required.
CONCLUSIONS AND RECOMMENDATIONS
The preliminary results reported here show that Trapa beds function as biogeochemical hotspots in the river system. During residence of river water within the beds, dissolved oxygen tended to be reduced in concentration, sometimes reaching levels that would be stressful to the aquatic biota, particularly at low tides; however this hypoxia is rather ephemeral because the water was soon exchanged as the tide rose.
Water temperatures fluctuated more than in the deeper, open waters. Rates of microbial respiration and denitrification were elevated compared to the adjacent open water.
Nutrient measurements show reductions in nitrate concentrations leading into the bed, which were lower than the adjacent Rondout Creek.
VI-42
Pharmaceuticals and pesticides in water outside and within the Trapa beds were
detected. Some of these compounds appear to have been reduced in concentration during
residence of water in the beds (e.g. diphenhydramine), whereas others seem unaffected
(carbamazepine); however, fewer than half the samples have been analyzed, and those
analyses need more post-processing to ensure accuracy. In the future, measurements of
sucralose, which is perhaps the best conservative tracer of wastewater in this system, will
be analysed.
Preliminary results suggest that concentrations of pesticides were higher at Norrie
Point whereas pharmaceuticals were higher at the site near Kingston. The proximity of
the Kingston site to wastewater treatment plant effluent may explain this difference.
Forthcoming results will provide insight for further investigation of the influence of Trapa on biogeochemical processing and interaction with emerging contaminants; however, it is recommended that modes of interactions, such as plant uptake, adsorption to plant surfaces or breakdown by microbes, is investigated for all pharmaceuticals that are shown to have a net decrease during residence within the Trapa bed. Further, mesocosm experiments may be useful in investigating pharmaceutical dose-response relationships and/or acute vs long-term exposure on microbial community diversity and function.
VI-43
ACKNOWLEDGEMENTS
I appreciate the guidance and mentorship of the Hudson River Foundation throughout this project. Thank you to the Cary Institute for hosting me and providing the invaluable experience of working on the Hudson River. I express sincere gratitude to
David Fischer, Heather Malcolm, Stephanie Robson and Chris Mitchell for encouragement as well as practical and logistical support. I acknowledge and thank my supervisors for their direction and assistance as well as those involved in the
RiverKeeper, LTREB and HRECOS long term monitoring initiatives.
VI-44
REFERENCES
Azevedo De Almeida, B. and A. Mostafavi. 2016. Resilience of infrastructure systems to sea-level rise in coastal areas: Impacts, adaptation measures, and implementation challenges. Sustainability 8: 1115.
Backer, L.C. 2012. Freshwater algal blooms & public health. Lake line/North American Lake Management Society 32: 7.
Brodin, T., J. Fick, M. Jonsson, and J. Klaminder. 2013. Dilute concentrations of a psychiatric drug alter behavior of fish from natural populations. Science 339: 814- 815.
Cantwell, M.G., D.R. Katz, J.C. Sullivan, D. Shapley, J. Lipscomb, J. Epstein, A. R. Juhl, C. Knudson, and G.D. O'mullan. 2017. Spatial patterns of pharmaceuticals and wastewater tracers in the Hudson River Estuary. Water Research 137: 335-343.
Caraco, N.F., J.J. Cole, S.E. Findlay, D.T. Fischer, G.G. Lampman, M.L. Pace, and D.L. Strayer. 2000. Dissolved oxygen declines in the Hudson River associated with the invasion of the zebra mussel (Dreissena polymorpha). Environmental Science & Technology 34: 1204-1210.
Caraco, N.F., J.J. Cole, S.E. Findlay, and C. Wigand. 2006. Vascular plants as engineers of oxygen in aquatic systems. BioScience 56: 219-225.
Carpenter, C. and D.E. Helbling. 2018. Widespread micropollutant monitoring in the Hudson River Estuary reveals spatiotemporal micropollutant clusters and their sources. Environmental Science & Technology 52: 6187-6196.
Costanzo, S.D., J. Murby, and J. Bates. 2005. Ecosystem response to antibiotics entering the aquatic environment. Marine Pollution Bulletin 51: 218-223.
De Sousa, D.N.R., A.A. Mozeto, R.L. Carneiro, and P.S. Fadini. 2014. Electrical conductivity and emerging contaminant as markers of surface freshwater contamination by wastewater. Science of the Total Environment 484: 19-26.
Fernald, S.H., N.F. Caraco, and J.J. Cole. 2007. Changes in cyanobacterial dominance following the invasion of the zebra mussel Dreissena polymorpha: Long-term results from the Hudson River estuary. Estuaries and Coasts 30: 163-170.
Ferrer, I., C.E. Heine, and E.M. Thurman. 2004. Combination of LC/TOF-MS and LC/ion trap MS/MS for the identification of diphenhydramine in sediment samples. Analytical Chemistry 76: 1437-1444.
VI-45
Findlay, S.E., W.C. Nieder, E.A. Blair, and D.T. Fischer. 2006. Multi-scale controls on water quality effects of submerged aquatic vegetation in the tidal freshwater Hudson River. Ecosystems 9: 84-96.
Findlay, S., K. Schoeberl, and B. Wagner. 1989. Abundance, composition, and dynamics of the invertebrate fauna of a tidal freshwater wetland. Journal of the North American Benthological Society 8: 140-148.
Flood, J.F. and L.B. Cahoon. 2011. Risks to coastal wastewater collection systems from sea-level rise and climate change. Journal of Coastal Research 27: 652-660.
Guildford, S.J., and R.E. Hecky. 2000. Total nitrogen, total phosphorus, and nutrient limitation in lakes and oceans: Is there a common relationship? Limnology and Oceanography 45: 1213–1223.
Hughes, S.R., P. Kay, and L.E. Brown. 2012. Global synthesis and critical evaluation of pharmaceutical data sets collected from river systems. Environmental Science & Technology 47: 661-677.
Kolpin, D.W., E.T. Furlong, M.T. Meyer, E.M. Thurman, S.D. Zaugg, L.B. Barber, and H.T. Buxton. 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in US streams, 1999− 2000: A national reconnaissance. Environmental Science & Technology 36: 1202-1211.
Kornijów, R., D.L. Strayer, and N.F. Caraco. 2010. Macroinvertebrate communities of hypoxic habitats created by an invasive plant (Trapa natans) in the freshwater tidal Hudson River. Fundamental and Applied Limnology/Archiv für Hydrobiologie 176: 199-207.
Lampman, G.G., N.F. Caraco, and J.J. Cole. 1999. Spatial and teporal patterns of nutrient concentration and export in the tidal Hudson River. Estuaries 22: 285-296.
Levinton, J.S. and J.R. Waldman. 2006. The Hudson River Estuary, Cambridge University Press. New York.
Martina, J., S. Hamilton, M. Turetsky, and C. Phillippo. 2014. Organic matter stocks increase with degree of invasion in temperate inland wetlands. Plant and Soil 385: 107-123.
Murdoch, K. 2015. Pharmaceutical pollution in the environment: Issues for Australia, New Zealand and Pacific island countries. National Toxic Network 1-36.
Neuwoehner, J., T. Zilberman, K. Fenner, and B.I. Escher. 2010. QSAR-analysis and mixture toxicity as diagnostic tools: Influence of degradation on the toxicity and mode of action of diuron in algae and daphnids. Aquatic Toxicology 97: 58-67.
VI-46
New York City Department of Environmental Protection (NYCDEP) 2019. Waste Water Treatment Plants. https://www1.nyc.gov/html/dep/html/watershed_protection/wastewater_treatment _plants.shtml.
Rao, Y.R., T. Howell, S.B. Watson, and S. Abernethy. 2014. On hypoxia and fish kills along the north shore of Lake Erie. Journal of Great Lakes Research 40: 187-191.
Reisinger, A.J., J.L. Tank, T.J. Hoellein, and R.O. Hall Jr. 2016. Sediment, water column, and open‐channel denitrification in rivers measured using membrane‐inlet mass spectrometry. Journal of Geophysical Research: Biogeosciences 121: 1258-1274.
Richmond, E.K., M.R. Grace, J.J. Kelly, A.J. Reisinger, E.J. Rosi, and D.M. Walters. 2017. Pharmaceuticals and personal care products (PPCPs) are ecological disrupting compounds (EcoDC). Elementa: Science of the Anthropocene 5: 66- 74.
Richmond, E.K., E.J. Rosi, D.M. Walters, J. Fick, S.K. Hamilton, T. Brodin, A. Sundelin, and M.R. Grace. 2018. A diverse suite of pharmaceuticals contaminates stream and riparian food webs. Nature Communications 9: 1-9.
Robson, S.V., E.J. Rosi, E.K. Richmond, and M.R. Grace. 2020. Nominal concentrations of pharmaceuticals alter metabolism, denitrification and diatom communities. Freshwater Science 39:256-267.
Rosi-Marshall, E.J., D.W. Kincaid, H.A. Bechtold, T.V. Royer, M. Rojas, and J.J. Kelly. 2013. Pharmaceuticals suppress algal growth and microbial respiration and alter bacterial communities in stream biofilms. Ecological Applications 23: 583-593.
Santos, L.H., A.N. Araújo, A. Fachini, A. Pena, C. Delerue-Matos, and M. Montenegro. 2010. Ecotoxicological aspects related to the presence of pharmaceuticals in the aquatic environment. Journal of Hazardous Materials 175: 45-95.
Schlaepfer, M.A., D.F. Sax, and J.D. Olden. 2011. The potential conservation value of non‐native species. Conservation Biology 25: 428-437.
Scott, P.D., M. Bartkow, S.J. Blockwell, H.M. Coleman, S.J. Khan, R. Lim, J.A. Mcdonald, H. Nice, D. Nugegoda, and V. Pettigrove. 2014. A national survey of trace organic contaminants in Australian rivers. Journal of Environmental Quality 43: 1702-1712.
Soh, L., K.A. Connors, B.W. Brooks, and J. Zimmerman. 2011. Fate of sucralose through environmental and water treatment processes and impact on plant indicator species. Environmental Science & Technology 45: 1363-1369.
VI-47
Strayer, D.L., J.J. Cole, S.E. Findlay, D.T. Fischer, J.A. Gephart, H.M. Malcom, M.L. Pace, and E.J. Rosi-Marshall. 2014. Decadal-scale change in a large-river ecosystem. BioScience 64: 496-510.
Strayer, D.L., V.T. Eviner, J.M. Jeschke, and M.L. Pace. 2006. Understanding the long- term effects of species invasions. Trends in Ecology & Evolution 21: 645-651.
Strayer, D.L., M.L. Pace, N.F. Caraco, J.J. Cole, and S.E. Findlay. 2008. Hydrology and grazing jointly control a large‐river food web. Ecology 89: 12-18.
Tall, L., N. Caraco, and R. Maranger. 2011. Denitrification hot spots: dominant role of invasive macrophyte Trapa natans in removing nitrogen from a tidal river. Ecological Applications 21: 3104-3114.
Torres, C.I., S. Ramakrishna, C.A. Chiu, K.G. Nelson, P. Westerhoff, and R. Krajmalnik- Brown. 2011. Fate of sucralose during wastewater treatment. Environmental Engineering Science 28: 325-331.
Watkinson, A., E. Murby, D.W. Kolpin, and S. Costanzo. 2009. The occurrence of antibiotics in an urban watershed: from wastewater to drinking water. Science of the Total Environment 407: 2711-2723. Wurtsbaugh, W.A., H.W. Paerl, and W.K. Dodds. 2019. Nutrients, eutrophication and harmful algal blooms along the freshwater to marine continuum. WIREs Water, 6:e1373. Yozzo, D.J. and W.E. Odum. 1993. Fish predation on epiphytic microcrustacea in Tivoli South Bay, a Hudson River tidal freshwater wetland. Hydrobiologia 57: 37-46.
VI-48
APPENDICES
APPENDIX A – Kingston Summary data
KBA Summary n Mean StDev Max Median Min Temperature (˚C) 27 24 2 28 24 20 Conductivity (uS/cm) 18 316 66 491 299 259 Turbidity (NTU) 27 19 17 66 12 6 Dissolved oxygen (mg /L) 27 8.2 0.7 9.3 8.5 6.5 % Oxygen saturation 27 97.8 7.3 108.7 100.9 78.6 NPOC (mg-C/L) 21 4.6 0.7 6.1 4.4 3.6 + NH4 (mg-N/L) 22 0.10 0.07 0.29 0.08 0.02 - NO3 (mg-N/L) 27 0.4 0.1 0.8 0.4 0.1 3– PO4 (mg-P/L) 27 0.03 0.03 0.10 0.02 0.01 TN (mg-N/L) 13 0.8 0.1 1.0 0.8 0.6 TP (mg-P/L) 13 0.08 0.03 0.15 0.07 0.04
KBB Summary n Mean StDev Max Median Min Temperature (˚C) 27 24 2 28 24 21 Conductivity (uS/cm) 18 310 75 485 291 244 Turbidity (NTU) 27 17 25 90 7 1 Dissolved oxygen (mg /L) 27 7 1 9 7 4 % Oxygen saturation 27 81 14 109 85 47 NPOC (mg-C/L) 21 4.5 0.7 6 4 4 + NH4 (mg-N/L) 23 0.07 0.1 0.22 0.06 0.02 - NO3 (mg-N/L) 27 0.3 0.1 0.5 0.3 0.1 3– PO4 (mg-P/L) 27 0.03 0.02 0.09 0.02 0.01 TN (mg-N/L) 13 0.7 0.1 1.0 0.7 0.6 TP (mg-P/L) 13 0.07 0.05 0.15 0.05 0.02
VI-49
KBC Summary n Mean StDev Max Median Min Temperature (˚C) 25 25 2 28 25 21 Conductivity (uS/cm) 16 296 52 427 295 244 Turbidity (NTU) 25 10 10 43 7 2 Dissolved oxygen (mg /L) 25 8 1 9 8 5 % Oxygen saturation 25 92 15 115 92 56 NPOC (mg-C/L) 21 4.3 0.6 6 4 4 + NH4 (mg-N/L) 18 0.06 0.03 0.14 0.05 0.02 - NO3 (mg-N/L) 25 0.3 0.1 0.5 0.3 0.1 3– PO4 (mg-P/L) 25 0.02 0.01 0.05 0.02 0.01 TN (mg-N/L) 13 0.9 0.7 3.3 0.7 0.5 TP (mg-P/L) 13 0.06 0.02 0.09 0.05 0.04
RC Summary n Mean StDev Max Median Min Temperature (˚C) 10 25 2 27 25 22 Conductivity (uS/cm) 10 360 68 492 337 311 Turbidity (NTU) 10 10 4 17 9 6 Dissolved oxygen (mg /L) 10 8.0 0.7 9.3 7.9 6.9 % Oxygen saturation 10 97 7 106 96 86 NPOC (mg-C/L) 7 6 1 9 5 5 + NH4 (mg-N/L) 10 0.19 0.1 0.33 0.19 0.08 - NO3 (mg-N/L) 10 0.5 0.3 1.3 0.4 0.3 3– PO4 (mg-P/L) 10 0.07 0.03 0.12 0.06 0.04 TN (mg-N/L) 6 1.0 0.2 1.2 1.0 0.6 TP (mg-P/L) 6 0.11 0.02 0.15 0.11 0.07
K-LTREB Summary n Mean StDev Max Median Min Temperature (˚C) 8 24 3 29 24 21 Conductivity (uS/cm) 8 258 36 293 269 212 Turbidity (NTU) 8 16 15 50 11 6 Dissolved oxygen (mg /L) 8 8.5 0.6 9.5 8.4 7.6 % Oxygen saturation 8 101 5 108 100 95 NPOC (mg-C/L) 8 3.5 0.2 3.8 3.5 3.2 + NH4 (mg-N/L) 8 0.04 0.03 0.10 0.04 0.01 - NO3 (mg-N/L) 8 0.34 0.07 0.46 0.36 0.25 3– PO4 (mg-P/L) 8 0.02 0.01 0.03 0.02 0.01 TN (mg-N/L) 8 0.8 0.1 0.9 0.8 0.6 TP (mg-P/L) 8 0.05 0.005 0.06 0.05 0.05
VI-50
KPD-RK Summary n Mean StDev Max Median Min Temperature (˚C) 3 31 3 33 31 28 Conductivity (uS/cm) Turbidity (NTU) 3 9 9 19 5 3 Dissolved oxygen (mg /L) % Oxygen saturation 3 101 18 122 96 87 NPOC (mg-C/L)
KPB-RK Summary n Mean StDev Max Median Min Temperature (˚C) 3 30 3 32 32 27 Conductivity (uS/cm) Turbidity (NTU) 3 9 3 13 8 7 Dissolved oxygen (mg /L) % Oxygen saturation 3 100 7 108 101 93 NPOC (mg-C/L)
Norrie Point Summary data
NB Summary n Mean StDev Max Median Min Temperature (˚C) 14 25 2 28 25 21 Conductivity (uS/cm) 9 273 20 311 264 252 Turbidity (NTU) 15 20 20 72 12 0 Dissolved oxygen (mg /L) 13 6 4 12 7 0.5 % Oxygen saturation 13 69 43 143 84 5 NPOC (mg-C/L) 11 4.0 0.2 4.4 3.7 3.7 + NH4 (mg-N/L) 11 0.04 0.02 0.08 0.04 0.02 - NO3 (mg-N/L) 15 0.2 0.1 0.4 0.3 0.1 3– PO4 (mg-P/L) 15 0.02 0.01 0.03 0.02 0.00 TN (mg-N/L) 11 0.6 0.1 0.9 0.6 0.5 TP (mg-P/L) 11 0.07 0.04 0.16 0.05 0.04
VI-51
NHR Summary n Mean StDev Max Median Min Temperature (˚C) 14 25 2 27 25 21 Conductivity (uS/cm) 9 272 20 298 277 247 Turbidity (NTU) 15 14 3 20 14 10 Dissolved oxygen (mg /L) 14 8 1 11 8 7 % Oxygen saturation 14 100 14 127 94 84 NPOC (mg-C/L) 11 4.0 0.3 4.6 3.6 3.6 + NH4 (mg-N/L) 11 0.03 0.02 0.09 0.03 0.02 - NO3 (mg-N/L) 15 0.3 0.1 0.4 0.3 0.2 3– PO4 (mg-P/L) 15 0.02 0.01 0.02 0.02 0.01 TN (mg-N/L) 11 0.64 0.05 0.69 0.65 0.56 TP (mg-P/L) 11 0.05 0.01 0.07 0.05 0.03
N-HRECOS Summary n Mean StDev Max Median Min Temperature (˚C) All data 25 2 29 26 21 Conductivity (uS/cm) recorded 262 38 320 250 170 Turbidity (NTU) every 15 mins 7 4 180 6 0 Dissolved oxygen (mg /L) from 19 Jun 7.6 0.8 11.3 7.5 5.5 % Oxygen saturation to 25 Sep 92 8 132 90 69 NPOC (mg-C/L) 2019
NC-RK Summary n Mean StDev Max Median Min Temperature (˚C) 3 30 2 32 31 28 Conductivity (uS/cm) Turbidity (NTU) 3 7 3 10 7 5 Dissolved oxygen (mg /L) % Oxygen saturation 3 93 7 100 93 85 NPOC (mg-C/L)
VI-52
SPECTROSCOPIC CHARACTERIZATION AND QUANTIFICATION OF MICROPLASTICS IN THE HUDSON RIVER
A Final Report of the Tibor T. Polgar Fellowship Program
Karli Sipps
Polgar Fellow
Department of Chemistry Rutgers University-Camden Camden, NJ 08102
Project Advisor:
Dr. Georgia Arbuckle-Keil Department of Chemistry Rutgers University-Camden Camden, NJ 08102
Sipps, K. and G. Arbuckle-Keil. 2021. Spectroscopic Characterization and Quantification of Microplastics in the Hudson River. Section VII: 1-37 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
VII-1
ABSTRACT
Microplastics (MP), synthetic plastic fragments less than 5 mm in diameter, have become an increasingly large environmental concern for various reasons. MPs have been shown to adsorb contaminants, such as persistent organic pollutants, to their surfaces.
This may allow MPs to act as a medium of transport for pollutants throughout aquatic environments. Additionally, MPs can be ingested by micro and macro organisms and may be toxic. The focus of this project was to characterize MPs using Fourier-Transform
Infrared and Raman spectroscopy and evaluate their temporal and geospatial distribution in the Raritan Bay area.
Surface water samples were obtained from various locations that were chosen to test the hypothesis that MPs accumulate in frontal zones, which may act as a sink for MP and represent areas in which there is an elevated probability of ingestion of MP by microorganisms. In order to evaluate the latter statement, zooplankton samples were also collected and subsequently digested and analyzed for MP content.
Both hypotheses were supported by data collected in this study, in which the percentage of 250-500 µm MPs detected in each surface water sample subset was positively correlated to the measure of how frontal the sampling site was. Additionally,
MPs were detected in zooplankton samples in greater quantities than currently reported in the literature.
VII-2
TABLE OF CONTENTS
Abstract ...... VII-2
Table of Contents ...... VII-3
List of Figures and Tables...... VII-4
Introduction ...... VII-7
Methods...... VII-10
Results ...... VII-18
Discussion ...... VII-32
Acknowledgments...... VII-34
References ...... VII-35
VII-3
LIST OF FIGURES AND TABLES
Figure 1 – River discharge data for 4/11/19 and 4/16/19 sampling events ...... VII-11
Figure 2 – Map of sampling sites for first (top), second (middle) and third (bottom) sampling
events ...... VII-12
Figure 3 – Surface water samples that had been rinsed onto a metal mesh post-oxidation
(bottom). Particles suspected to be MPs set aside for analysis (top) .. VII-15
Figure 4 – Relative amounts of MPs collected per site (4/11/19 sampling event).MPs
classified as “other plastics” include polyamides and epoxy resins .... VII-20
Figure 5 – Relative amounts of MPs collected per site (4/16/19 sampling event). MPs
classified as “other plastics” include epoxy resins ...... VII-21
Figure 6 – Percentage of MPs detected in each sample subset (4/11/19) plotted against the
mean salinity measured at each respective sampling site ...... VII-22
Figure 7 – Percentage of MPs detected in each sample subset (4/11/19) plotted against the
relative standard deviation in salinity measured at each respective sampling
site ...... VII-23
Figure 8 – Average particle counts per sampling site vs. respective salinity values ...... VII-24
Figure 9 – A comparison of total particle counts and percent MPs detected in analyzed
particles for low flow (4/11/19) and high flow (4/16/19) sampling events ...... VII-24
Figure 10 – Raman spectrum of residual exoskeleton (chitin) collected using 532nm
excitation wavelength and 1200 gr/mm grating ...... VII-25
Figure 11 – Zooplankton exoskeleton and a microplastic particle, identified as
polyethylene using Raman spectroscopy. Image captured using 100X
VII-4
objective ...... VII-26
Figure 12 – MPs detected in zooplankton samples. Both images were captured under a
50X objective using a Raman microscope ...... VII-27
Figure 13 – MPs (circled) detected in the chemically digested Acartia tonsa samples. All
images were captured under a 100X objective using a Raman microscope ...... VII-27
Figure 14 – Chemical maps generated from the integration of a characteristic band of
chitin (top left) and a C-H group frequency band (top right) in an FTIR
spectrum of chitin (below). The numerous green/red regions present only in
the right image were analyzed and found to be MPs (PP and PE) and
cellulose ...... VII-29
Figure 15 – Raman spectra of PP (top) and PS (bottom) MPs found in an Acartia tonsa
sample. Both spectra were collected using a 532nm excitation wavelength and
1200 gr/mm grating ...... VII-30
Figure 16 – Distribution of microplastic types in zooplankton samples ...... VII-31
Figure 17 – Distribution of microplastic morphologies in zooplankton samples ..... VII-31
Table 1 – FTIR measurement parameters ...... VII-16
Table 2 – Raman measurement parameters ...... VII-17
Table 3 – Summary of analysis of particles collected during the 4/11/19 sampling event.
Total particles per site are a best estimate rounded to +/- 5 ...... VII-19
Table 4 – Summary of analysis of particles collected during the 4/16/19 sampling event.
The value for total particles for site 1 is a best estimate rounded to +/- 5 ...... VII-20
VII-5
Table 5 – Mean salinity measurements and associated standard deviations for each
sampling site ...... VII-21
Table 6 – A complete summary of MPs detected in zooplankton samples ...... VII-30
VII-6
INTRODUCTION
Plastic pollution in aquatic environments has become an increasingly large concern over recent years. As a whole, the human population produces an average of about 1.5 megatons of plastic waste every year (Boucher and Friot 2017). Although recent work by Ward et al. (2019) suggests that polystyrene may be converted into carbon dioxide through photooxidative pathways, many plastics are thought to not break down completely but, rather, will be subject to degradation processes that will lead to the fragmentation of the macroscopic plastics into microscopic plastic particles (Andrady
2011; Fotopoulou and Karapanagioti 2017; Kubowicz and Booth 2017). These particles are known as microplastics (MPs) and are often defined as being small, plastic fragments that are 5 mm or less in diameter. The tendency for discarded plastic products to ultimately end up in waterways is thought to be one source of MPs in rivers, oceans, and other aquatic systems. MPs have been found in diverse global locations, including the
Arctic (Bergmann et al. 2019). Other concerns include incidental plastic pollution from tires and boats, as well as the introduction of plastic-containing microbeads that were previously present in a number of soaps and facial scrubs, fibers, etc., to waterways through wastewater treatment plants (Duis and Coors 2016).
The contamination of rivers and oceans by MPs may be hazardous to inhabitants of these ecosystems. This also has a detrimental impact on a wide variety of other organisms. Once MPs begin to pollute the environment, they may be consumed by fish, oysters, or plankton, and enter the food chain (Wrighta et al. 2013; Browne et al. 2007;
Claessens et al. 2013). Additionally, MPs may be susceptible to the adsorption of heavy metals (Brennecke et al. 2016) and other pollutants, such as organic contaminants, onto
VII-7
their surface (Rios et al. 2010; Bakir et al. 2014). The MPs then act as a possible
transporter of these pollutants, known as persistent organic pollutants (POPs), which are
thought to cause harm to humans and wildlife. Additives and plasticizers that are used in
plastics have also been known to leach out of the plastic products and have come under
close scrutiny for their possible negative effects on ecosystems (Setala et al. 2014;
Rochman et al. 2013).
In order to gain a better understanding of how MPs enter the water and how,
exactly, they may impact the environment, it is necessary to first identify and quantify the
MPs present in the waterways. This is particularly crucial for high-traffic waterways such as the Hudson River, which is likely to exhibit high levels of MP pollution due to its proximity to urban and high-population areas. While only about 8% of the river basin is urban, MPs have been shown to be pervasive in waterways; therefore, rather than being a localized issue, they can become a widespread problem. Pollution in the Hudson River has widespread and serious implications. Its watershed is home to millions of people and hundreds of various aquatic species, making the environmental impact of the transport of
MPs along the Hudson River of particular importance (New York State Office of the
Attorney General 2015). Additionally, the presence of MPs in the Hudson River may have negative effects on marine life, as this plastic pollution may eventually be dumped into the ocean (Moore et al. 2011; GESAMP 2016).
The purpose of this study was to investigate the geospatial and temporal distribution of MPs in the Raritan Bay area, as well as the effect of hydrodynamic controls on MP concentrations and their potential uptake by zooplankton. It was hypothesized that MPs may accumulate in frontal zones, areas known to accumulate
VII-8 buoyant particles and debris, and that a greater quantity of MPs would likely be observed after high flow conditions (e.g., after heavy rainfall) than after low flow conditions.
Sampling was performed so that the effect of these conditions on MP concentrations could be adequately tested.
The accumulation of MPs in specific areas, such as frontal zones that may act as a
“sink” for MPs, has important ecological implications. Sampling in the confluence of the
Hudson and Raritan Rivers would allow for investigation into such a system. These areas may represent major points of entry of MPs into the food chain. Because of the overlap in size between MPs and the food of many microorganisms, such as zooplankton, it is possible that MPs may be ingested in lieu of/or in addition to food. This has been verified and well-documented in various controlled laboratory studies in which intake was monitored while zooplankton were exposed to both food and environmentally- relevant concentrations of MPs; however, there are far fewer studies that have been conducted on the amounts of MPs found in real world, environmental samples (Botterell et al. 2019).
Therefore, an additional focus of this study was on potential MP uptake by two different species of zooplankton, the common estuarine copepod, Acartia tonsa, and crab larvae, or zoea. These organisms hold a key role in the food web because they are prey for various types of commercially relevant fish. Furthermore, it is critical to gain a better understanding of how MPs may enter the food chain as there is potential for the transfer of MPs to higher trophic levels, possibly posing a risk to terrestrial species and even humans (Karbalaei et al. 2018).
VII-9
It is necessary to determine the chemical composition of the MPs to confirm the
observed particles are indeed plastic. The inadequacy of relying solely on visual
identification when characterizing MPs has been well documented in the literature.
(Ziajahromi et al. 2017; Hidalgo-Ruiz et al. 2012; Renner et al. 2017). Because of their
small size, it is not sufficient to rely on visual identification or optical microscopy alone,
as this can lead to misidentification and overcounting or undercounting of MP particles in
samples. Spectroscopic methods, such as Fourier-Transform Infrared (FTIR) and Raman
spectroscopy, are most commonly used to elucidate chemical information about the
samples and are accepted techniques for analyzing MPs (Renner et al. 2017; Araujo et al.
2018; Kappler et al. 2016). Both methods are non-destructive and require simple post-
processing sample preparation before analysis. For particles smaller than 500 microns, it
is common to use a microscope that is coupled to a spectrometer (Song et al. 2015).
METHODS
Sample Collection
Samples were collected on three different days in order to determine the temporal distribution of MPs in samples. Sampling dates were chosen to capture events of both low (first and second sampling events; July 26, 2018 and April 11, 2019) and high flow
(third sampling event; April 16, 2019). River discharge data corresponding to each April sampling event is depicted in Figure 1.
Sampling sites (Figure 2) were chosen so that a broad range of salinities, from nearly freshwater to ambient ocean water, would be covered. Additionally, as the
VII-10 presence of MPs in frontal zones was of interest in this study, these regions were targeted for some of the sampling sites, particularly for the 4/11/19 sampling trip.
Figure 1: River discharge data for 4/11/19 and 4/16/19 sampling events. Data collected at Bound Brook on the Raritan River.
Surface water samples were collected using plankton nets with mesh sizes of 80 microns and 153 microns. Two replicates were taken at each site, and samples were collected simultaneously by trawling the nets at the surface for 25 minutes at a speed of
10 knots. Surface water samples analyzed in this study were collected during 4/11/19 and 4/16/19 (low and high flow, respectively) sampling dates.
Zooplankton samples were obtained using 0.5 m, 200 µm mesh ring nets fitted with flowmeters at the net openings and filtering cod-ends. Duplicate surface tows were taken at each station. Zooplankton samples were rinsed from the cod-end into glass collection jars and preserved using buffered formaldehyde (10% final concentration).
Samples analyzed in this study were collected during the 7/26/18 sampling event.
VII-11
Figure 2: Map of sampling sites for first (top), second (middle) and third (bottom) sampling events.
VII-12
Surface Water Sample Processing
After collection, surface water samples were prepared for analysis. Samples from each site were rinsed separately with distilled water and sieved sequentially through 500 micron and 250 micron meshes in order to separate the particles by size class. For the purposes of this study, primarily particles between 500 microns and 250 microns in size were analyzed.
Surface water samples were subjected to a wet peroxide oxidation (WPO) in order to digest easily oxidizable organic material present in the samples. This process, which adheres to the protocol (Masura et al. 2015) recommended by the National Oceanic and
Atmospheric Association (NOAA) for preparing environmental samples for analysis of microplastics, involves:
1. Combining samples with a solution prepared using 30% H2O2 solution and
aqueous 0.05 M FeSO4 solution in a 1:1 ratio. The samples were left in the
solution with constant mixing for thirty minutes while the temperature was
maintained at a range between 65o – 70o C.
2. Density separation using a saturated NaCl solution. Buoyant particles were
retained for analysis.
Matrix spikes and field blanks were performed for quality assurance and quality control. 100% recovery of polyethylene beads used to spike the matrix replicate is reported after oxidation. The blanks were analyzed and found to contain some fibers and a few particles that were spectroscopically confirmed to be cellulose. Fibers were excluded from this study when determining MP counts.
VII-13
Zooplankton Sample Processing
Zooplankton samples were sorted to identify and isolate the zooplankton by
species. Dominant individuals were then inspected under a dissecting scope to ensure
there were no external MPs attached to the organisms and then pooled to create samples
of 100, 50, 25 or 10 specimens of a single species, either Acartia tonsa or crab zoea.
An initial round of samples was subjected to an enzymatic digestion according to the protocol developed by Cole et al. (2014) using proteinase-K to destroy biotic material, leaving only MPs for analysis. A second round of samples was subjected to a chemical digestion, according to the procedure developed by Desforges et al. (2015) using concentrated nitric acid (HNO3) for the same purpose. Blanks were also prepared
for each round of samples by treating filters with the reagents used for zooplankton
digestions. In each case, the blanks were confirmed to be free of MPs.
Visual Observations
An initial investigation of the samples was performed using a Zeiss Stemi DV4
stereomicroscope. A total particle count was obtained for each sample. Particles that
appeared likely to be plastic (e.g., were brightly colored, films, or very regularly shaped)
were set aside on glass slides for further analysis. It is well documented that, under
normal laboratory conditions, incidental contamination from microfibers can lead to
overestimations when quantifying microplastics present in samples (Woodall et al. 2015;
Wesch et al. 2017); therefore, fibers were excluded from this study due to the high
possibility of transfer from clothing or lab coats to samples. Glass slides were cleaned
VII-14 with isopropyl alcohol wipes before sample storage to remove any dust or possible contaminants. At all times, samples were handled with tweezers and were kept covered between analyses in order to mitigate contamination as much as possible.
Figure 3: Surface water samples that had been rinsed onto a metal mesh post-oxidation (bottom). Particles suspected to be MPs set aside for analysis (top).
Spectroscopic Analyses
Surface water samples were individually analyzed using Raman and/or FTIR microscopy. The chosen technique varied on a case-by-case basis and was dependent, primarily, upon the physical characteristics of the sample being analyzed. Samples that appeared to be thin or film-like were mostly analyzed using a Bruker LUMOS FTIR microscope (single-point detector) in transmission mode. Conversely, samples that did not appear likely to transmit IR radiation reasonably well but did appear to have reflective properties were analyzed using the FTIR microscope in reflectance mode.
Spectra were collected at various, manually selected points across each particle’s surface in order to confirm uniformity in chemical composition. Measurement parameters used for FTIR analysis are listed in Table 1.
VII-15
Table 1: FTIR measurement parameters.
Number Number Apodization Mode Substra Backgroun Sample Resolution Function te d Scans Scans CaF2 Blackman- Transmission window 64 64 4 cm-1 Harris 3 Term
MirrIR Blackman- Reflection slideb 64 64 4 cm-1 Harris 3 Term
b Glass microscope slide coated with two thin layers of silver (Kevley Technologies)
In many cases, neither of these options allowed for the collection of reasonable
spectra. These samples would then be analyzed using a Horiba XploRA Raman
microscope. A significant drawback of this technique was interference from fluorescence
that can be caused by trace organics that are present in environmental samples, as well as pigments and additives added to commercial plastics. Additionally, the measurement parameters set for each scan are very sample-specific, which may necessitate multiple
scans to optimize the parameters, leading to lengthy analysis times. Sets of typical
measurement parameters are listed in Table 2. Efforts were made to collect both FTIR
and Raman spectra for each sample. These two techniques are complementary to one
another and, when used together, can allow for a more accurate identification of unknown
particles than when a single technique is used alone.
After digestion, zooplankton samples were analyzed using both a Raman
microscope and a Bruker Hyperion 3000 FTIR microscope equipped with a liquid
nitrogen cooled 64X64 element focal plane array (FPA) detector. Measurement
VII-16
Table 2: Raman measurement parameters.
Sample Spectral Excitation Slit, Confocal Microscope Acquisition Time, Description Grating Wavelength Hole Size Objective Accumulation
532nm 100um, 100um Thin fragments, 1200 gr/mm 638nm 10X 20 seconds, 20 films 532nm 100um, 300um Thick fragments, 1200 gr/mm 638nm 50X 20 seconds, 20 pellets 50um, 100-300um 10X Colored/fluorescent 600 gr/mm 785nm 50X 20 seconds, 30
parameters used on the Raman microscope closely resembled those outlined in Table 2, but the smaller size of the MP in the zooplankton samples required the collection of spectra through a 50X or 100X microscope objective. The use of an FTIR microscope with FPA detector aided in detecting MP that were not readily visually identified, as broad regions of the sample filter could be scanned as opposed to a single, selected point.
All zooplankton samples were run on 25 mm (diameter) anodized aluminum
(Anodisc, Whatman) filters through which the samples were rinsed during digestion.
This limited FTIR analysis as these filters were suitable for analysis in transmission mode only and were only IR transmissive to approximately 1200 cm-1; however, collection in this region (3600 – 1200 cm-1) was still sufficient to ascertain broad MP identifications as the entire group frequency region was covered.
FTIR spectra were cross-referenced against a spectral library using the library search function in Bruker’s OPUS software (Version 7.2) in an attempt to produce a match and successfully identify the polymer. BioRad’s KnowItAll→ software was also used, primarily in matching Raman spectra. When a suitable match could not be
VII-17 produced, band assignments were made by correlating specific functional groups to their relevant characteristic group frequencies. Samples were then broadly classified based on the functional groups and linkages present for each polymer.
Data analysis
The normality of the log-transformed particle number data was confirmed with a
Shapiro-Wilk test, but the data did not have homogeneous variance (confirmed via a
Levene's test); therefore, a one-way ANOVA procedure was applied to test for significant
differences in the total particle number by sampling site. Correlation testing was
performed using a linear regression.
RESULTS
Surface Water Samples
Total particle counts (e.g., total number of individual solids remaining after
sample oxidation) yielded a broad range of particle quantities collected among the
sampling sites. These total counts are approximations of actual values. Due to the
microscopic size and large number of particles present in each sample dish, it is
extremely difficult to accurately determine these values with no uncertainty. The
variability in morphologies and physical characteristics of the particles that caused their
microscope images to lie in different focal planes prevented the use of imaging software
that could be used for counting, as a single representative image of the entire sample dish
could not be rendered. Significant differences in total particle number were not observed
VII-18
between sampling sites (p=0.43) due to the variation between replicate samples; however,
the highest number of particles were observed at sites 2, 3 and 4.
For samples that had very high particle counts, only a subset of particles from
each replicate was analyzed using FTIR and Raman spectroscopy due to time constraints.
The percentage of particles analyzed that were spectroscopically confirmed to be MPs
was calculated for each sampling site (Tables 3 and 4). For the 4/11/19 sampling event,
sampling sites 3, 4, and 5 were found to have very similar percentages of MPs, higher
than those of sampling sites 1 and 2. For the 4/16/19 sampling event, sampling site 2 was
found to have a higher percentage of MPs than sampling site 1.
Table 3: Summary of analysis of particles collected during the 4/11/19 sampling event. Total particles per site are a best estimate rounded to +/- 5.
Sampling Site/ Total Total Percent Location Particles Analyzed MP 1 Ambient Ocean 115 50 60% Waters (Net 1 = 62; Net 2 = 54)
2 305 Hudson Plume (Net 1 = 271; Net 2 = 35) 75 68%
3 285 Bay (Net 1 = 234; Net 2 = 52) 60 75%
4 485 Hudson-Raritan Front (Net 1 = 421; Net 2 = 65) 70 74%
5 55 Mouth of the Raritan (Net 1 = 34; Net 2 = 21) 55 76%
VII-19
Table 4: Summary of analysis of particles collected during the 4/16/19 sampling event. The value for total particles for site 1 is a best estimate rounded to +/- 5. Sampling Site Total Particles Percent Particles Analyzed MP 1 Hudson-Raritan Front 155 60 63%
2 Mouth of the Raritan 62 62 74%
MPs across all sampling sites were predominantly polyethylene (PE) and polypropylene (PP) for both sampling events. PE comprised over 45% of the MPs found in 4 out of 5 sampling sites for the first sampling event (Figure 4), while PP was more dominant for the second sampling event (Figure 5). Other plastics were also found, although in much smaller quantities, including polystyrene (PS), polyesters, polyamides, and epoxy resins. A number of non-plastics were also characterized, including inorganic minerals, such as SiO2 (sand) and CaCO3, and biopolymers, such as cellulose.
Figure 4: Relative amounts of MPs collected per site (4/11/19 sampling). MPs classified as “other plastics” include polyamides and epoxy resins.
VII-20
Figure 5: Relative amounts of MPs collected per site (4/16/19 sampling event). MPs classified as “other plastics” include epoxy resins.
In order to test the hypothesis of MP accumulation in frontal regions, these MP percentages from the first sampling event were compared to hydrodynamic data (Table 5) that were collected at each sampling site. The percentage of MPs was found to be negatively correlated to salinity (Figure 6).
Table 5: Mean salinity measurements and associated standard deviations for each sampling site. Mean Salinity Relative Standard Sampling Site Site Description (PSU) Deviation 1 Ambient Ocean 29.9 0.21 Waters 2 Hudson Plume 25.5 0.32
3 Bay 23.1 0.97
4 Hudson-Raritan Front 20.6 0.58
5 Mouth of the Raritan 16.32 1.1
VII-21
Figure 6: Percentage of MPs detected in each sample subset (4/11/19) plotted against the mean salinity measured at each respective sampling site.
The calculated relative standard deviation in salinity measurements taken at the time of sampling was used as an indication of how frontal each sampling site was, where a higher standard deviation indicated a more frontal site. On this basis, it was observed that sampling site 5 was most frontal, followed by sites 3 and 4. Sites 1 and 2 were the least frontal. The percentage of MPs exhibited a direct relationship with the relative standard deviation (Figure 7).
VII-22
Figure 7: Percentage of MPs detected in each sample subset (4/11/19) plotted against the relative standard deviation in salinity measured at each respective sampling site.
The numbers of particles collected at each site were also examined. Because of the high variability in particle counts between replicates at a single site, the average number of particles collected was used for comparison. As shown in Figure 8, these values did not vary linearly with salinity, as would be expected. This may indicate that this is more complex than simple mixing of end members (Chant, personal correspondence).
Comparisons were also made between sampling events in terms of total particles collected and percentage of MPs per site in order to investigate the temporal distribution of MPs. No trend could be observed in total particle counts, but a greater percentage of
MPs was observed for the 4/11/19 (low flow) sampling event (Figure 9).
VII-23
Figure 8: Average particle counts per sampling site vs. respective salinity values.
Figure 9: A comparison of total particle counts and percent MPs detected among analyzed particles for low flow (4/11/19) and high flow (4/16/19) sampling events.
VII-24
Zooplankton Samples
The use of both Raman and FTIR microscopy allowed for the detection and identification of MPs in nearly all samples. It was found that a chemical digestion using
HNO3 allowed for greater ease of subsequent analysis than an enzymatic digestion using proteinase-K. Samples that were treated with proteinase-K contained large amounts of residual biotic material, believed to be fragments of exoskeleton that were not removed during the digestion process. These fragments, which, as seen in Figure 10 below, were spectroscopically confirmed to be chitin, complicated the analyses as they were difficult to visually discern from potential MPs. Additionally, in some cases, the exoskeleton appeared to overlap the MPs, shown in Figure 11. This covering of the MPs presents the risk of underestimating the number of MPs in the samples.
Figure 10: Raman spectrum of residual exoskeleton (chitin) collected using 532nm excitation wavelength and 1200 gr/mm grating.
VII-25
Figure 11: Zooplankton exoskeleton and a microplastic particle, identified as polyethylene using Raman spectroscopy. Image captured using 100X objective.
The use of an FTIR microscope with a focal plane array (FPA) detector alleviated
this difficulty as the detector allowed for the automated collection of spectra across areas
up to a 2000 micron X 1000 micron grid in this case. The generation of a chemical map
allowed for easy differentiation between exoskeleton and potential MPs; however, due to
the nature of the sample substrate (material on which the sample was placed), this
technique was limited to running samples in transmission mode, and the quality of the
spectra was dependent upon the fact that particles in a given area were in the same focal
plane and the samples’ ability to transmit IR radiation reasonably well to the detector.
In contrast to the enzymatically digested samples, those that were digested using
HNO3 exhibited very little to no residual exoskeleton. As such, it was considerably more
straightforward to visually identify some of the MPs in the samples. Examples are given
in Figure 12. In all cases, visual sorting was used solely to determine which particles
should be examined further and was followed by Raman and/or FTIR spectroscopic
analyses in order to characterize the chemical compositions of the particles.
VII-26
Figure 12: MPs detected in zooplankton samples. Both images were captured under a 50X objective using a Raman microscope.
Many of the MPs in the samples were not readily identified. This was due,
primarily, to the colors, sizes and morphologies of the MPs. A significant portion of the
MPs that were detected were less than 100 microns in size and were colorless beads or
films, shown in Figure 13, that contrasted very little with the sample substrate.
Figure 13: MPs (circled) detected in the chemically digested Acartia tonsa samples. All images were captured under a 100X objective using a Raman microscope.
Use of the FTIR microscope (FPA detector) was particularly effective in differentiating between exoskeleton (chitin) and potential MP, as well as detecting MPs
VII-27
that were not easily visualized on the sample filter. Once data had been collected, a
chemical map could be generated by integrating key bands in the spectra. A chemical
map uses color coding to display relative intensities of the integrated band for all spectra
collected across the grid. Because an overwhelming majority of organic polymers,
including biopolymers such as chitin, contain C-H bonds, the corresponding region of the
FTIR spectrum (approximately 3000-2800 cm-1) was integrated first, thereby identifying
any organics. Next, a band characteristic of polyamides (approximately 1660-1600 cm-1),
the functional group to which chitin belongs, was integrated. This band would not be
expected to be present in common polymers, with the exception of nylons, so a side-by- side comparison (Figure 14) of the two maps generated from the integrations made it possible to identify particles that were organic but not likely to be chitin. The spectra of these particles were then individually analyzed. Most were found to be MP or cellulose.
Spectra of particles believed to be chitin were also examined to eliminate the possibility that they could be nylons or some other polyamide.
VII-28
Amide C=O stretch Aliphatic C-H stretching bands
Figure 14: Chemical maps generated from the integration of a characteristic band of chitin (top left) and a C-H group frequency band (top right) in an FTIR spectrum of chitin (below). The numerous green/red regions present only in the right image were analyzed and found to be MPs (PP and PE) and cellulose.
The most abundant MP across 75% of the samples was determined to be polyethylene (PE). The Raman spectra of PE and PS MPs found in an Acartia tonsa sample is given in Figure 15. Complete findings are presented in Table 6 and Figure 16.
The dominant MP morphology encountered was fragments (Figure 17).
VII-29
Figure 15: Raman spectra of PP (top) and PS (bottom) MPs found in an Acartia tonsa sample. Both spectra were collected using a 532nm excitation wavelength and 1200 gr/mm grating.
Table 6: A complete summary of MPs detected in zooplankton samples.
Number Digestion Total MPs Types of MPs Species Zooplankton in Method Detected Detected Sample Crab zoea 25 Enzymatic 2 PP Crab zoea 50 Enzymatic 1 PP Crab zoea 100 Enzymatic 1 PE Acartia tonsa 50 Enzymatic 1 PE Acartia tonsa 100 Enzymatic 3 PP, PE Acartia tonsa 100 Chemical 7 PP, PS, PE, polyamide Acartia tonsa 100 Chemical 6 PE, PS Acartia tonsa 100 Chemical 7 PE, PP, polyester Crab zoea 100 Chemical 3 PE, polyester
VII-30
Figure 16: Distribution of MP polymer types detected in zooplankton samples.
Figure 17: Distribution of MP morphologies detected in zooplankton samples.
VII-31
DISCUSSION
The findings of the surface water portion of this study support the hypothesis that
250-500µm MPs are present in higher quantities in frontal zones compared to less frontal
sites. The results from the first sampling event show that a greater percentage of MPs was present in freshwater than more saline water. The most frequently observed MPs, polyethylene and polypropylene, are two of the most industrially relevant thermoplastics used in a variety of applications from packaging to plastic bags and soda bottles. These observations may be important pieces of information that can be used to guide the development of mitigation strategies designed to reduce plastic pollution in waterways.
Identification of the types of MP observed (spectral searching) was complicated, in some cases, by what appeared to be surface oxidation on the plastics. Even though both spectral libraries that were used contain oxidized polyethylene and polypropylene in their databases, these matches were often not among the top hits. Excessively noisy or complicated spectra also fared poorly when sent through a spectral library search, and these spectra often required manual interpretation to identify the MP.
The results from the zooplankton portion of this study support the hypothesis that
MPs may be ingested by zooplankton in the environment. These results are significant; a similar study by Desforges et al. (2015) reports having found 24 MPs among 906 digested zooplankton, while, in this case, an average of 6 MPs were found in 100 digested zooplankton, among the chemically digested samples. Additionally, this is potentially the first study of this kind conducted on copepods in the Hudson River.
Similar to the surface water samples (250-500µm), the dominant plastic observed in the zooplankton samples was polyethylene, followed by polypropylene, despite the difference in size. Observation of MP in the zooplankton extracts is also consistent with
VII-32
the hypothesis that zooplankton may be ingesting MPs present in the water that may be
confused with food because of the overlap in size ranges. MPs detected in the
zooplankton samples ranged in size from 200 microns to 25 microns, with a large
majority falling under 50 microns. The 200 micron MP found in one of the samples is
likely too large to have been ingested by the zooplankton and may have been overlooked
during the initial inspection of the zooplankton for any external plastics.
These findings will guide future zooplankton analyses. Samples collected in April
2019 will be analyzed for MP content and compared to results from the surface water samples for those respective dates that were analyzed as part of this study. This future work will have the ultimate goal of assessing not only elevated levels of MPs in frontal zones, but also increased ingestion of MPs by zooplankton in frontal zones.
VII-33
ACKNOWLEDGEMENTS
A special thank you to the Hudson River Foundation and Tibor T. Polgar Fellowship committee for the opportunity to complete this project, as well as Dr. Georgia Arbuckle-
Keil for guidance and supervision for this research. Funding to support the sampling and sample preparation was provided by a grant from NJ Sea Grant to Nicole Fahrenfeld
(Dpt. of Civil and Environmental Engineering), Robert Chant, and Grace Saba (Dpt. of
Marine and Coastal Sciences). This was a collaborative project that drew from the expertise of Dr. Chant, Dr. Saba and Dr. Fahrenfeld, who are experts in estuarine physics, coastal marine organismal ecology and fate and transport of MPs, respectively. Thanks to Kendi Bailey and Shreya Patil (with support from the Rutgers Douglass Project
Summer STEM) for extraction and oxidation of the MP from the water column and
Kasey Walsh for collection, identification, and extraction of MP from the zooplankton.
VII-34
REFERENCES
Andrady, A. 2011. Microplastics in the marine environment. Marine Pollution Bulletin 62:1596-1605.
Araujo, C., M. Nolasco, A. Ribeiro, and P. Ribeiro-Claro. 2018. Identification of microplastics using Raman spectroscopy: Latest developments and future prospects. Water Research 142:426-440.
Bakir, A., S.J. Rowland, and R.C. Thompson. 2014. Enhanced desorption of persistent organic pollutants from microplastics under simulated physiological conditions. Environmental Pollution 185:16–23.
Bergmann, M., S. Mutzel, S. Primpke, M. Tekman, J. Trachsel, and G. Gerdts. 2019. White and wonderful? Microplastics prevail from the Alps to the Arctic. Science Advances 5:1-10.
Boucher, J., and D. Friot. 2017. Primary microplastics in the oceans: A global evaluation of sources. Gland, Switzerland: IUCN
Botterell, Z, N. Beaumont, T. Dorrington, M. Steinke, R. Thompson, and P. Lindeque. 2019. Bioavailability and effects of microplastics on marine zooplankton: A review. Environmental Pollution 245: 98-110.
Brennecke, D., B. Duarte, F. Paiva, I. Cacador, and J. Canning-Clode. 2016. Microplastics as vectors for heavy metal contamination from the marine environment. Estuarine, Coastal and Shelf Science 178:189-195.
Browne, M., T. Galloway and R. Thompson. 2007. Microplastic- an emerging contaminant of potential concern? Integrated Environmental Assessment and Management. 3:559-566.
Claessens, M., L. Van Cauwenberghe, M.B. Vandegehuchte, and C.R. Janssen. 2013. New techniques for the detection of microplastics in sediments and field collected organisms. Marine Pollution Bulletin 70:227–33.
Cole, M., H. Webb, P.K. Lindequn, E.S. Fileman, C. Halsband, and T.S. Galloway. 2014. Isolation of microplastics in biota-rich seawater samples and marine organisms. Scientific Reports 4:4528.
Desforges, J.P., M. Galbraith, and P.S. Ross. 2015. Ingestion of microplastics by zooplankton in the northeast Pacific Ocean. Archives of Environmental Contamination and Toxicology 3:320-330.
VII-35
Duis, K. and A. Coors. 2016. Microplastics in the aquatic and terrestrial environment: sources (with a specific focus on personal care products), fate and effects. Environmental Sciences Europe. 28:2.
Fotopoulou, K., and H. Karapanagioti. 2017. Degradation of various plastics in the environment, The Handbook of Environmental Chemistry, Hazardous Chemicals Associated with Plastics in the Marine Environment, Vol. 78, Springer Cham.
The Joint Group of Experts on Scientific Aspects of Marine Protection (GESAMP). 2016. Sources, fate and effects of microplastics in the marine environment: part two of a global assessment. In: Kershaw P.J., and C.M. Rochman (eds) IMO/FAO/UNESCO IOC/UNIDO/WMO/IAEA/UN/UNEP/UNDP Joint Group of Experts on the Scientific Aspects of Marine Environmental Protection Rep. Stud. GESAMP No. 93, p 220.
Hidalgo-Ruiz, V, L. Gutow, R. Thompson. and M. Thiel. 2012. Microplastics in the marine environment: A review of the methods used for identification and quantification. Environmental Science & Technology. 46:3060–3075.
Kappler, A, D. Fischer, S. Oberbeckmann, G. Schernewski, M. Labrenz, K. Jochen- Eichhorn, and B. Voit. 2016. Analysis of environmental microplastics by vibrational microspectroscopy: FTIR, Raman or both. Analytical and Bioanalytical Chemistry. 29:8377-8391.
Karbalaei, S., P. Hanachi, T.R. Walker, and M. Cole. 2018. Occurrence, sources, human health impacts and mitigation of microplastics pollution. Environmental Science and Pollution Research 25:36046-36063.
Kubowicz, S., and A.M. Booth. 2017. Biodegradability of plastics: challenges and misconceptions, Environmental Science & Technology. 51:12058-12060.
Masura, J., J. Baker, G. Foster, and C. Arthur. 2015. Laboratory Methods for the Analysis of Microplastics in the Marine Environment: Recommendations for Quantifying Synthetic Particles in Waters and Sediments. NOAA Technical Memorandum NOS-OR&R-48.
Moore, C, G. Lattin, and A. Zellers. 2011. Quantity and type of plastic debris flowing from two urban rivers to coastal waters and beaches of Southern California. Journal of Integrated Coastal Zone Management. 11:65-73.
New York State Office of the Attorney General.2015. Discharging microbeads to our waters: an examination of wastewater treatment plants in New York: https://ag.ny.gov/pdfs/2015_Microbeads_Report_FINAL.pdf
VII-36
Renner, G, T. Schmidt, and J. Schram. 2017. Characterization and quantification of microplastics by infrared spectroscopy. Comprehensive Analytical Chemistry 75:67-118.
Rios, L, P. Jones, C. Moore, and U. Narayan. 2010. Quantification of persistent organic pollutants adsorbed on plastic debris from the Northern Pacific Gyre’s “eastern garbage patch.” Journal of Environmental Monitoring 12:2226-2236.
Rochman, C, E. Hoh, T. Kurobe, and S. Teh. 2013. Ingested plastic transfers hazardous chemicals to fish and induces hepatic stress. Scientific Reports. 3:3263-3270.
Setala, O, V. Fleming-Lehtinen, and M. Lehtiniemi. 2014. Ingestion and transfer of microplastics in the planktonic food web. Environmental Pollution. 185:77-83.
Song, Y.K., S.H. Hong, M. Jang, G.M. Han, M. Rani, J. Lee, and W.J. Shim. 2015. A comparison of microscopic and spectroscopic identification methods for analysis of microplastics in environmental samples. Marine Pollution Bulletin 93:202-209.
Ward, C, C. Armstrong, A. Walsh, J. Jackson, and C. Reddy. 2019. Sunlight converts polystyrene to carbon dioxide and dissolved organic carbon. Environmental Science and Technology Letters 6:669-674.
Wesch, C, A.M. Elert, M. Worner, U. Braun, R. Klein, and M. Paulus. 2017. Assuring quality in microplastic monitoring: about the value of clean-air devices as essentials for verified data. Scientific Reports 7:5424.
Woodall, L.C., C. Gwinnett, M. Packer, R.C. Thompson, L.F. Robinson, and G.L. Paterson. 2015. Using a forensic science approach to minimize environmental contamination and to identify microfibres in marine sediments. Marine Pollution Bulletin 95:40-46.
Wrighta, S.L., R.C. Thompson, and T.S. Galloway. 2013. The physical impacts of microplastics on marine organisms: a review. Environmental Pollution 178:483– 492.
Ziajahromi, S, P.A. Neale, L. Rintoul, and F.D. Leusch. 2017. Wastewater treatment plants as a pathway for microplastics: development of a new approach to sample wastewater-based microplastics. Water Research 112:93-99.
VII-37
STUDYING THE EFFECT OF SALINITY AND TIDE ON FECAL BACTERIA TRANSPORT WITHIN HUDSON RIVER ESTUARY SEDIMENTS
A Final Report of the Tibor T. Polgar Fellowship Program
Dong Zhang
Polgar Fellow
Department of Civil, Environmental and Ocean Engineering Stevens Institute of Technology Hoboken, NJ 07030
Project Advisor:
Valentina Prigiobbe Department of Civil, Environmental and Ocean Engineering Stevens Institute of Technology Hoboken, NJ 07030
Zhang, D. and Prigiobbe, V. 2021. Studying the effect of salinity and tide on the fecal bacteria transport within Hudson River Estuary sediments. Section VIII: 1-26 pp. In D.J. Yozzo, S.H. Fernald, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2019. Hudson River Foundation.
VIII-1 ABSTRACT
Water contamination by fecal bacteria is a serious issue in coastal urban areas.
Fecal bacteria can be discharged accidentally into surface water bodies through run-off or combined sewer overflows. Once they are in the water, they can be transported downstream with the flow as well as settle towards the bottom and potentially migrate within the sediments of the river bed. The fate and the transport of bacteria in porous media is mainly governed by attachment and detachment processes at the solid-liquid interface. The salinity/ionic strength is one of the most important factors as it controls the electrostatic interactions between the microorganisms and the porous medium surface and, therefore, the speed at which the microorganisms migrate. Salinity can be significant in estuarine river water, and it varies with daily tidal cycle as well as on the seasonal time scale due to precipitation and runoff. Here, a study is presented that combines field measurements with laboratory tests to elucidate the role of salinity on fecal bacteria transport in sediments of the Hudson River Estuary. A one-dimensional (1D) transport model for fecal bacteria (Escherichia coli) was developed and used to describe the laboratory experiments. Column-flood experiments of E. coli transport through sand were carried out to calibrate the model and test the effect of salinity change. In the experiments, the concentration front of bacteria show the formation of a spike as the salinity of the water drops. Such a transport behavior can be well described by the developed mathematical model. Detachment of bacteria due to a reduction in salinity could be important in river sediments during the tidal cycle and be the reason behind an increase in bacterial concentration observed in river water.
VIII-2
TABLE OF CONTENTS
Abstract ...... VIII-2
Table of Contents ...... VIII-3
Lists of Figures and Tables ...... VIII-4
Introduction ...... VIII-5
Methods...... VIII-8
Modeling ...... VIII-8
Laboratory Experiments ...... VIII-9
Field Sampling ...... VIII-12
Results ...... VIII-13
Discussion ...... VIII-20
Acknowledgments...... VIII-23
References ...... VIII-24
VIII-3
LIST OF FIGURES AND TABLES
Figure 1 – Schematic of the transport mechanisms affecting the migration of
fecal bacteria in sediments ...... VIII-7
Figure 2 – FIB concentrations measurements in the Hudson Estuary area ...... VIII-8
Figure 3 – Scheme of the set-up used for the tests in this work ...... VIII-11
Figure 4 – Map of study area (left) showing the location of sampling point in
Hoboken and photo (right) of sampling location ...... VIII-12
Figure 5 – Field sediments ...... VIII-13
Figure 6 – Particle size distribution of sediments sample and Ottawa sand ...... VIII-13
Figure 7 – E. coli concentrations, salinity, and river level during the sampling
Period ...... VIII-15
Figure 8 – Measured (dots) and simulated (black line) breakthrough concentrations
of E. coli for different salinity conditions ...... VIII-17
Figure 9 – Parameter values as a function of salinity estimated by fitting the
experimental data of experiments 1 through 5 ...... VIII-19
Figure 10 – Breakthrough curves measured during experiments ...... VIII-20
Table 1 – Characterization of river water samples taken along the Hudson River
Estuary at Maxwell Cove, Hoboken, New Jersey ...... VIII-14
Table 2 – Initial and injected conditions applied during the column-flood tests . VIII-16
Table 3 – Fitted model parameters obtained using the transport model ...... VIII-18
VIII-4
INTRODUCTION
Water contamination by fecal bacteria is a serious issue in coastal urban areas. The presence of bacteria in beach sand and coastal water is a risk to human health (Heaney et al. 2009; Halliday and Gast 2010; Hassard et al. 2016). Recently, some studies have shown high fecal bacteria concentration in the surface waters of urban estuaries. An investigation by Riverkeeper (2018) has reported that along the lower Hudson River, some tributaries, which flow through highly populated areas, had high concentrations of fecal indicator bacteria. In some cases, their concentrations exceeded the U.S.
Environmental Protection Agency (EPA) quality standards.
The understanding of the transport of fecal bacteria within estuarine sediments is critical for the prediction of their migration and fate. Fecal bacteria can be accidentally discharged into rivers or streams through combined sewer overflows or runoff and contaminate surface water (Clary et al. 2014). An investigation in Chicago by Drury et al.
(2013) and a study by Lu et al. (2014) have shown an increase in fecal bacterial diversity within estuarine sediments downstream of a wastewater treatment plant in comparison to the population in sediments upstream of the plant. Yamahara et al. (2007) show that once the fecal bacteria are discharged into surface water, they can settle onto the sediments and be trapped within the hyporheic zone. As the tide rises in river estuaries, the hydraulic gradient between the river level and the water table of the shallow aquifer increases leading to the infiltration of the bacteria into the sediments (Mulligan and Charette 2009).
Upon infiltration into the hyporheic zone, various mechanisms can affect bacteria migration, namely, attachment/detachment, straining, and decay (Brunke 1999; Grant et al. 2011).
VIII-5
Salinity can be significant in estuarine river water, and it increases with rising
tide. For example, measurements along the Hudson River Estuary nearby Battery Park
(NY) change between 10 and 22 ppt. At higher salinity levels, attachment of bacteria onto
the porous medium surface is favored. Several studies (Schijven and Hassanizadeh 2000;
Becker et al. 2004; Torkzaban et al. 2008; Bradford and Torkzaban 2015; Sasidharan et
al. 2017) have been conducted to investigate the effects of water composition on bacterial
attachment/detachment. They concluded that salinity/ionic strength is one of the most
important factors controlling microorganism migration. Salinity affects the electrostatic
interactions between the microorganisms and the porous medium surface (Bradford et al.
2012). Salinity can compress the double layer of the surface, and the van der Walls forces
of attraction become more important (Cao et al. 2010). Microorganisms have strong
attachment under high ionic strength (Wang et al. 2013). The value of pH is also critical in microorganism transport as the pH can change the porous medium surface charge favoring electrostatic attraction between the microorganisms and the porous medium surface (Kim et al. 2008).
Water level and composition in river estuaries are very sensitive to the tide.
During rising tides the salinity of the water increases, favoring attachment of microorganisms onto the sediments, whereas on falling tides the salinity decreases and
VIII-6
detachment is favored (Figure 1). Although several studies have been performed to
investigate the effect of salinity/ionic strength on the transport of microorganisms
through porous media, some of them focus only at low salinity conditions (i.e., between 1
and 4 ppt). Moreover, most of the works (e.g., Gargiulo et al. 2008) neglect the presence
of soluble organic matter in porous medium, which can be significant within sediments.
In addition, only few studies (e.g., Gao et al. 2015) quantitatively analyze the effect of the
tide on the distribution of fecal bacteria concentration in urban estuaries.
Figure 1: Schematic of the transport mechanisms affecting the migration of fecal bacteria in sediments. (a) high tide, (b) low tide.
Figure 2 shows the concentrations of Fecal Indicator Bacteria (FIB) as a function
of rainfall four days prior the measurements for locations at the estuary of the Hudson
River, namely, two in Hoboken (NJ) and two in New York City (NY) (Riverkeeper
2018). Parts (b) through (e) show that in dry-weather conditions the FIB concentrations exceed the water quality standard regulated by the U.S. EPA, as indicated by the red dashed line. The grey area in the diagrams represents dry-weather condition. The hypothesis was tested that dry-weather FIB events are due to bacteria resuspension upon detachment from river sediment surface. To verify this hypothesis, 1) the effect of
VIII-7 salinity on the transport of fecal bacteria was studied through sediments by performing column-flood tests in the laboratory that then we described with a transport model and 2) surface water samples were collected from the Hudson River Estuary nearby Hoboken to measure E. coli concentration and salinity during dry weather conditions.
Figure 2: FIB concentrations measurements in the Hudson Estuary area.
METHODS
Modeling
Following a previous study (Cao et al. 2010), a 1D model for microorganism transport through porous media combining the mass conservation equations of microorganisms (e.g., bacteria or viruses) and salinity was developed (Zhang et al. 2019),
VIII-8
∂c ρ ∂SI ∂SII ∂c ∂2c + b + + v -D = 0 (1) ∂t n ∂t ∂t ∂x ∂x2 � � ∂c ∂c ∂2c s + v s - D s = 0 (2) ∂t ∂x ∂x2
where t > 0 is the time, s; 0 < x < L is the longitudinal coordinate, m, with L the length
of the domain; c is the bacteria concentration, cfu/mL; cs is the salinity given in part per
3 thousand, ppt; n is the porosity, (-); ρb is the bulk density of the porous medium, g/cm ; v
is the interstitial flow velocity, m/s; SI and SII are the concentrations of microorganisms at
the two different types of sites on the porous medium surface, cfu/g, generally associated
to reversible and irreversible attachment mechanisms, respectively,
ρ ∂SI ρ b = kI c - b kI SI (3) n ∂t att m n dⅇt
II II ρb ∂S II S = katt 1- c (4) n ∂t Smax � � I II I where k att and k att represent the attachment coefficients, 1/min; k det is the detachment
coefficient, 1/min; and Smax is the maximum concentration attached on the porous
I medium surface, cfu/g. The mathematical relationships between model parameters (k att,
I II k det, k att and Smax) and salinity (cs) were derived on the basis of lab experiments carried
out at constant salinity. The model was discretized with a finite difference scheme and
solved iteratively in MATLAB (MathWorks 2018).
Laboratory Experiments
Column-flood experiments were conducted to investigate the transport of fecal
bacteria through sediments. In these tests, both Ottawa sand and sediments collected from
the field (Maxwell Cove Beach) were used. Ottawa quartz sand (U.S. Silica) was selected
for the tests to understand the effect of mineral coating present in the sand of the natural
VIII-9
sediments. Sodium chloride was used as a conservative tracer in the column experiments
to estimate the longitudinal hydrodynamic dispersion. Labware and columns (Omnifit,
U.S.) used in these tests were pre-sterilized by autoclaving. Escherichia coli (ATCC
13706) was used to run these tests. The bacterial suspensions were created in brine of
varying concentrations, namely, 1, 2.5, 5, 7.5, and 10 ppt. The brine was prepared using
analytical grade NaCl and Milli-Q water (DI water) around pH = 5.8. The injected
concentration of E. coli was around 4×107 cfu/mL. Ottawa sand was dry-packed into a
glass column (a length of 12 cm with an inner diameter of 1 cm), then saturated with
water to measure gravimetrically the porosity, which varied between 0.35 and 0.4. The
columns were wet packed using clean sand while vibrating the column to liberate
entrapped air. After packing, the column was flushed with 120 pore volumes (pore
volume injected, PV = Q·t/Vv, with Q the flow rate and Vv the pore volume) of DI water
prior to running the test. PV is a conventional unit to report breakthrough curves measured during column-flood test. PV allows to understand how much faster solutes or particles move with respect to the flow velocity. If a concentration starts to rise around 1
PV, it indicates that the speed of the species is equal to that of water. Breakthrough before 1 PV indicates the presence of short-cuts through walls or fractures. Breakthrough after 1 PV indicates the presence of processes, such as attachment or adsorption, which retard the transport process through the porous medium.
The bacterial suspension was injected into the sand-packed column by using a peristaltic pump. The effluent was characterized online with a UV-Visible spectrophotometer (Cary-60, Agilent, U.S.) using a 1 cm flow-through cell (at 600 nm).
Online measurements were performed in order to identify sharp features of the bacteria
VIII-10
concentration fronts. Prior to each test, the UV-Vis spectrometer was calibrated by
correlating the wavelength of 600 nm with the offline counting of bacterial cfu. The
effluent was also collected by an automatic fraction collector at regular time intervals to
confirm UV-Vis measurements. The salinity at the outlet was measured by an online
conductivity sensor (Endress+Hauser, U.S.). Temperature and pH were also monitored
during the experiments. The measured conductivity was converted to salinity through a calibration curve. The scheme of the laboratory set-up is shown in Figure 3.
Figure 3: Scheme of the set-up used for the tests in this work.
The column-flood tests were run under two different conditions, namely, constant salinity and variable salinity. The tests performed under constant salinity condition were
I I II designed with the purpose to estimate the model parameters (i.e., k att, k det, k att and Smax)
by minimizing the difference between the model predictions and the data. At the
beginning of each test, a background brine was injected into the column at a flow rate of
0.5 mL/min for approximately 2.75 hours (corresponding to 25 PVs). Then, a solution
VIII-11 containing E. coli with the same salinity was injected into the column for about 20 hours
(200 PVs). After this time, the initial solution was injected again for 5.5 hours (50 PVs).
In order to explore the effect of salinity change on the fecal bacteria transport in Ottawa sand and sediments, a low-salinity brine of 1 ppt was injected through the column initially, then a suspension of E. coli with a salinity of 10 ppt was injected for 20 hours
(200 PVs). Finally, the condition was reversed, and the initial brine provided again for around 10 hours (1100 PVs).
Field Sampling
Water samples were taken along the Hudson River Estuary near Maxwell Place in the City of Hoboken (Figure 4). The samples were collected 10-feet from shore and approximately one foot below the water surface twice a day, namely, during high and low tide. Samples were analyzed for pH, salinity, and E. coli. The sampling and characterization were carried out according to surface water sampling requirements established by U.S. EPA (2002) and EPA standard Method 1604, respectively.
Figure 4: Map of study area (left) showing the location of sampling point in Hoboken and photo (right) of Maxwell Cove beach.
VIII-12
Sediment samples were also taken at the same location. The particle size distribution of the sand was determined with sieving analysis (ASTM method D6913). Upon characterization, the sediments were used to run column-flood tests under variable salinity conditions.
Figure 5: Field sediments
RESULTS
Sediment Characterization
An image of the sediments upon drying is provided in Figure 5. It shows the typical color of natural sand with some mineral oxide coating that provide a browner color. Figure 6 depicts the particle size distribution (PSD) of the sediments and Ottawa
sand. The two PSD curves are very 100 Ottawa sand similar. They have a coefficient of Sediments 80 uniformity (Cu) equal to 1.66 and a mean
60 particle size D50 = 0.35 mm. According
to the size distribution, the sediments was 40 characterized as fine sand in the Unified Percent of passing (%) 20 Soil Classification System (USCS).
0 1 0 -1 -2 10 10 10 10 Characterization of River Water Particle size (mm) Samples Figure 6: Particle size distribution of the sediments and Ottawa sand used in The data related to the analyses of this work. the water samples are listed in Table 1
VIII-13
and shown in Figure 7 together with the values of the tidal elevation. However, for a
more detailed description of bacteria concentration trend with the river level either a
longer monitoring time or a more frequent sampling time is required and this is a goal for
future work.
Table 1: Characterization of river water samples taken along the Hudson River Estuary at Maxwell Cove, Hoboken, New Jersey.
Sampling Tide E. coli Salinity Date pH Time (High/Low) (cfu/100mL) (ppt)
7/17/2019 10:25 AM High 41 7.3 18 03:23 PM Low 53 7.56 13 7/19/2019 12:00 PM High 72 7.47 17 04:53 PM Low 112 7.6 11 7/25/2019 09:12 AM Low 17 7.56 12 03:00 PM High 36 7.71 14 7/26/2019 10:32 AM Low 23 7.89 10 04:07 PM High 22 8.06 11 7/30/2019 08:41 AM High 4 7.71 18 01:40 PM Low 10 7.58 19
VIII-14
Figure 7: E. coli concentrations, salinity, and river level during the sampling period. Column-flood Tests
The conditions applied during the experiments are listed in Table 2. Here, salinity, temperature, and pH before and after the injection of the bacteria are reported.
Experiments 1 through 6 were conducted under constant salinity conditions, whereas experiments 7 through 9 were run stabilizing initially the column with a low salinity brine and then injecting a high salinity brine with a suspension of bacteria until steady- state. Once the column reached steady-state (i.e., the concentration of the bateria and the salinity in the outflow reached a constant value), the initial low salinity brine was injected again. The maximum salinity value was selected within the range of the river salinity
(i.e., lower than 20 ppt). The column used for these experiments was filled with Ottawa sand and the sand collected at the Maxwell Cove beach, respectively.
VIII-15
Table 2: Initial and injected conditions applied during the column-flood tests.
Initial Composition Injected Composition Exp. Soil Salinity T Salinity T E. coli pH pH No. (ppt) (ºC) (ppt) (ºC) (cfu/mL) 1 Ottawa 1 23.2 5.57 1 22.6 5.43 4.4×107 2 Ottawa 2.5 27.2 5.72 2.5 26.3 5.47 3.72×107 3 Ottawa 5 22.3 5.72 5 22.7 5.81 4.14×107 4 Ottawa 7.5 22.6 5.65 7.5 22.9 5.87 4.37×107 5 Ottawa 10 22.9 5.67 10 24.8 5.91 4.81×107 6 Ottawa 15 22.5 5.77 15 23.0 5.87 4.57×107 7 Ottawa 1 22.3 5.74 10 23.0 6.10 3.82×107 8 Sediments 1 22.1 5.61 10 22.5 5.97 4.24×107 9 Ottawa 1 21.9 5.57 15 22.0 5.92 4.16×107
Measured and simulated E. coli breakthrough curves for experiments 1 through 6 are shown in Figure 8. Here, the normalized effluent concentration of E. coli (c/cinjected) is plotted against the number of PVs (pore volumes). In all cases, the bacteria break through the column at 1 PV (0.11 hours).
VIII-16
Figure 8: Measured (dots) and simulated (black line) breakthrough concentrations of E. coli for different salinity conditions. The blue lines represent the salinity of effluent. Experiments (a) 1 , (b) 2 , (c) 3 , (d) 4, (e) 5, and (f) 6.
VIII-17
To calibrate the transport model, the measurements were optimized by the model
I II I and the attachment (k att and k att) and the detachment (k det) coefficients as well as the
maximum concentration attached on the Ottawa sand (Smax) were estimated. The
optimization of the measurements consisted in minimizing the sum of the squared difference between the data and the model simulations. The dispersion coefficient (D) required in these simulations was estimated by solving equation (2) for a tracer test using
NaCl. The estimated values of the model parameters are listed in Table 3.
Table 3: Fitted model parameters obtained using the transport model. The * parameter S max is equal to Smax/Nbacteria, where Nbacteria is the number of bacteria in a unit volume of injection (cfu/mL).
I I II * Salinity k att k det k att S max (ppt) (min-1) (min-1) (min-1) (g-1) 1 0.42 0.86 0.023 0.97 2.5 0.41 0.23 0.022 1.09 5 0.31 0.34 0.042 2.84 7.5 0.2 0.11 0.078 4.38 10 0.26 0.13 0.081 5.10 15 0.22 0.07 0.088 7.20
The values of the parameters are shown in Figure 9 together with the
mathematical relationships derived to describe their trend as a function of salinity. The
I I II values of k att and k det decrease with salinity; contrarily, the values of k att and Smax
I increase with salinity. Moreover, the value of k att is one order of magnitude larger than
II that of k att suggesting that irreversible attachment mechanism described by eq. (4) is the dominant attachment mechanism.
VIII-18
Figure 9: Parameter values as a function of salinity estimated by fitting the I I experimental data of experiments 1 through 5. (a) k att (b) k det, (c) II k att, and (d) Smax. The dots are the estimated values and the lines are the interpolations.
The calibrated model was used to predict experiments 7 through 9 run under
variable salinity and the results are shown in Figure 10 together with the measurements.
A spike in bacteria concentration is predicted by the model upon decrease of salinity. The
measured bacteria concentration fronts show a small spike when salinity drops from 10 to
1 ppt in both experiments 7 and 8 and a much larger one in experiment 9 where salinity
changes from 15 to 1 ppt. This is because the attached bacteria concentrations (SI and SII
in equations 1 through 4) increase with salinity becoming important around 15 ppt in this work
VIII-19
Figure 10: Breakthrough curves measured during experiments (a) 7 (b) 8 and (c) 9.
DISCUSSION
Characterization of River Water Samples
Results of the water sample analyses indicate that the salinity is between 10 and
20 ppt and the concentrations of E. coli per 100 mL water in 50 percent of samples was high (i.e., larger than 30 counts/100mL), according the EPA standard Method 1604, but the maximum concentration never exceeds the limit of 126 cells/100 mL for recreational waters. The salinity and the bacteria concentration values in the river water samples were used as a guide for the conditions to be applied in the laboratory during the transport test.
VIII-20
Column-flood Tests
In the breakthrough curves shown in Figure 8, it is possible to see that the bacteria
break through the column at 1 PV (0.11 hours) indicating that the bacteria travel at the
average flow velocity. The concentration then stabilizes around a constant value equal to
the injected concentration (i.e, c/cinjected equal to 1). When the salinity is larger than 5 ppt, the concentration front is smoother. This is because of the attachment process that is more important at this salinity retards the transport of the bacteria through the column.
Such as behaviour was already observed in previous works where the transport of E. coli and Enterococcus faecalis through sand was investigated (Chen and Walker, 2012).
Figure 10 reports the measurements and the model simulations for experiments where the salinity of the two injected solution was different, i.e., changed between 1 and
10 ppt in experiments 7 and 8 and between 1 and 15 ppt in experiment 9. The model agrees very well with the data and predicts the formation of the concentration peak when salinity is decreased (around 180 PVs). The measurements in exp. 7 and 8 as well as the good agreement of the model with the data suggest that the transport behavior of bacteria
through Ottawa sand and natural sediments is similar, and therefore the mineral coating
and other impurities on the natural sediments have a minor effect. When salinity drops
from 15 to 1 ppt, a more significant spike arises indicating that attachment of bacteria on
the sand was more significant at this salinity condition. The effect of salinity on E. coli
transport was investigated experimentally by other authors (Kim et al. 2009; Kim and
Walker 2009; Bai el al. 2016) and the formation of a concentration spike due to salinity drop was observed by Foppen et al. (2007) for a change of salinity from 0.27 to 0 ppt.
However, in that study the authors do not describe the experimental data with a model
VIII-21
because the model developed in that work does not account for a relationship between the
model parameters and salinity which is critical to predict such a behavior. In that study,
the authors consider the parameters constant even under variable salt concentration. The
model proposed in this project instead provides a tool that allows to predict the effect of
transient changes of salinity due to a tide on bacteria transport through sediments because
of the derived relationships between the model parameters.
Conclusions
A coupled model of bacteria and salinity transport through porous media was
developed, calibrated, and used to predict column-flood experiments under variable
salinity conditions. The model and the measurements agree and both show the formation
of a spike of bacteria concentration when sand stabilized with a brine of salinity greater
than 10 ppt containing bacteria and then flooded by low salinity water. This may suggest
a negative feedback of tidal fluctuations in urban estuaries where fecal contamination
may be present. For example, during high tide the salinity of the water increases, favoring attachment of microorganisms onto the sediments, whereas during low tide the salinity decreases and detachment is enhanced. Future research will focus on extending this work to account for the ratio between the attachment/detachment reaction kinetics and the tidal cycle on bacteria accumulation and resuspension.
VIII-22
ACKNOWLEDGMENTS
I would like to acknowledge the Tibor T. Polgar Fellowship Program of the
Hudson River Foundation for the support of this study, the Department of Civil,
Environmental and Ocean Engineering at Stevens Institute of Technology for providing the laboratory for the measurements, the American Chemical Society Petroleum Research
Fund (ACS-PRF) grant PRF# 57739-DNI9 for partial support of this project. Thanks to
Dr. Valentina Prigiobbe for her guidance and support in the project.
VIII-23
REFERENCES Bai, H., N. Cochet, A. Pauss and E. Lamy. 2016 Bacteria cell properties and grain size impact on bacteria transport and deposition in porous media. Colloid and Surface B: Biointerfaces 139: 148-155. Becker, M.W., S.A. Collins, D.W. Metge, R.W. Harvey, and A.M. Shapiro. 2004. Effect of cell physicochemical characteristics and motility on bacterial transport in groundwater. Journal of Contaminant Hydrology 69: 195-213. Bradford, S.A., and S. Torkzaban. 2015. Determining parameters and mechanisms of colloid retention and release in porous media. Langmuir 31: 12096-12105. Bradford, S.A., S. Torkzaban, H. Kim, and J. Simunek. 2012. Modeling colloid and microorganism transport and release with transients in solution ionic strength. Water Resources Research 48: 1-10. Brunke, M. 1999. Colmation and depth filtration within streambeds: retention of particles in hyporheic interstices. International Review of Hydrobiology 84: 99-117. Chen, G. and S.L. Walker. 2012. Fecal indicator bacteria transport and deposition in saturated and unsaturated porous media. Environmental Science and Technology 46: 8782-8790. Cao, H., F.T.C. Tsai, and K.A. Rusch. 2010. Salinity and soluble organic matter on virus sorption in sand and soil columns. Groundwater 48: 42-52. Clary, J., R. Pitt, and B. Steets, eds. 2014. Pathogens in Urban Stormwater Systems. Final Report to the American Society of Civil Engineers. Denver, Colorado. Drury, B., E. Rosi-Marshall, and J.J. Kelly. 2013. Wastewater treatment effluent reduces the abundance and diversity of benthic bacterial communities in urban and suburban rivers. Applied Environmental Microbiology 79: 1897-1905. Foppen, J.W., M. van Herwerden, and J. Schijven. 2007. Transport of Escherichia coli in saturated porous media: Dual mode deposition and intra-population heterogeneity. Water Research 41: 1743-1753. Gao, G., R.A. Falconer, and B. Lin, B. 2015. Modelling the fate and transport of faecal bacteria in estuarine and coastal waters. Marine Pollution Bulletin 100: 162-168. Gargiulo, G., S.A. Bradford, J. Simunek, P. Ustohal, H. Vereecken, and E. Klumpp. 2008. Bacteria transport and deposition under unsaturated flow conditions: The role of water content and bacteria surface hydrophobicity. Vadose Zone Journal 7: 406-419. Grant, S.B., R.M. Litton‐Mueller, and J.H. Ahn. 2011. Measuring and modeling the flux of fecal bacteria across the sediment‐water interface in a turbulent stream. Water Resources Research 47: 1-13.
VIII-24
Halliday, E., and R.J. Gast. 2010. Bacteria in beach sands: an emerging challenge in protecting coastal water quality and bather health. Environmental Science and Technology 45: 370-379. Hassard, F., C.L. Gwyther, K. Farkas, A. Andrews, V. Jones, B. Cox, H. Brett, D.L. Jones, J.E. McDonald,. and S.K. Malham. 2016. Abundance and distribution of enteric bacteria and viruses in coastal and estuarine sediments—a review. Frontiers in Microbiology 7: 1692. Heaney, C.D., E. Sams, S. Wing, S. Marshall, K. Brenner, A.P. Dufour, and T.J. Wade. 2009. Contact with beach sand among beachgoers and risk of illness. American Journal of Epidemiology 170: 164-172. Kim, S. B., S.J. Park, C.G. Lee, N.C. Choi, and D.J. Kim. 2008. Bacteria transport through goethite-coated sand: Effects of solution pH and coated sand content. Colloids and Surfaces B: Biointerfaces 63: 236-242. Kim, N.H. and S. Walker. 2009. Escherichia coli transport in porous media: Influence of cell strain, solutions chemistry, and temperature. Colloid and Surface B: Biointerfaces 71: 160-167. Kim, N.H., S.A. Bradford, and S.L. Walker. 2009. Escherichia coli 0157:H7 transport in saturated porous media: role of chemistry and surface macromolecules. Environmental Science and Technology 43: 4340-4347. Lu, H., K. Chandran, and D. Stensel. 2014. Microbial ecology of denitrification in biological wastewater treatment. Water Research 64: 237-254. MathWorks. 2018. MATLAB Version R2018b. Natick, Massachusetts: The MathWorks Inc. Mulligan, A.E., and M.A. Charette. 2009. Elements of Physical Oceanography: A Derivative of the Encyclopedia of Ocean Sciences. Elsevier Science, Amsterdam. Riverkeeper. 2018. “Hudson River Estuary Data,” Riverkeeper: https://www.riverkeeper.org/water-quality/hudson-river/ (Accessed July 10, 2018). Sasidharan, S., S. Torkzaban,, S.A. Bradford, P.G. Cook, and V.V. Gupta. 2017. Temperature dependency of virus and nanoparticle transport and retention in saturated porous media. Journal of Contaminant Hydrology 196: 10-20. Schijven, J.F., and S.M. Hassanizadeh. 2000. Removal of viruses by soil passage: Overview of modeling, processes, and parameters. Critical reviews in Environmental Science and Technology 30: 49-127. Torkzaban, S., S.S. Tazehkand, S.L. Walker, and S.A. Bradford. 2008. Transport and fate of bacteria in porous media: Coupled effects of chemical conditions and pore space geometry. Water Resources Research 44: 1-12.
VIII-25
United States Environmental Protection Agency (US EPA). 2002. Method 1604: Total Coliforms and Escherichia coli in Water by Membrane Filtration Using a Simultaneous Detection Technique (MI Medium) Wang, Y., S.A. Bradford, and J. Šimůnek. 2013. Transport and fate of microorganisms in soils with preferential flow under different solution chemistry conditions. Water Resources Research 49: 2424-2436. Yamahara, K.M., B.A. Layton, A.E. Santoro, and A.B. Boehm. 2007. Beach sands along the California coast are diffuse sources of fecal bacteria to coastal waters. Environmental Science and Technology 41: 4515-4521. Zhang, D., M. Zabarankin, and V. Prigiobbe. 2019. Modeling salinity-dependent transport of viruses in porous media. Advances in Water Resources 127: 252-263.
VIII-26
2019 Polgar Fellows
From left to right: Jacob Moore, Dong Zhang, Sofi Courtney, Matthew Badia, Ellie Petraccione, Waverly L. Lau, Cami Plum, Karli Sipps
Special thanks to Sam Gordon for assistance in formatting manuscripts.